701: I Improved My Town

23 Jul 2026 · 2 h 11 min · 45 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

A member special plus a long “tech life” update. The host describes a real outage in a small data center (transcription servers down) and the hardware/power fix needed. The episode also covers Sonos/Unifi networking changes (disabling SonosNet), Apple icon “squircle jail” debate, and several news/reader segments (PlayStation disc strategy, rising CD sales, Apple Music price increases).

Guests (and backgrounds)

No formal guest list. The episode is primarily hosts discussing among themselves. Named participants include John (tier list app/vibe coding), Marco (Ubiquiti/Sonosnet discussion context), Casey (emoji naming attempts), and Louis Mantia (author of the “squircle jail” blog post). Louis is referenced as a blog author, not as an in-studio guest.

Key claims

  1. Data center rack outage was caused by a failed automatic transfer switch (ATS) / power switching unit; replacing it is required.
  2. SonosNet can now be disabled in Sonos apps; this should reduce network issues on modern routers and on Ubiquiti setups, without affecting surround latency links.
  3. “Squircle jail” improves bad icons and removes odd dock sizing, but forces best icons to look worse and punishes existing apps.

Notable examples

  • The host’s rack had power off despite other racks working; ATS showed blinking lights and was likely board-failed.
  • Sonos surround question: wired soundbar + wireless surrounds/sub still works; disabling SonosNet doesn’t break the low-latency surround link.
  • Squircle jail example: games that previously filled the full icon area get shrunk into the squircle, and tiny icon elements can be clipped/punished.

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Data Center Troubles

0:05 to 0:50

The host shares an incident about offline transcription servers and power issues at a data center.

“I'm just going to always say New Jersey every time you say that until you stop.”

Diagnosing the Power Failure

0:50 to 4:00

The host describes the diagnosis process and issues with power supply and automatic transfer switch.

“They have compliance and security requirements and everything.”

Seeking Recommendations

4:00 to 5:20

The host expresses dissatisfaction with a power supply and seeks recommendations for replacements.

“It's six months old, so I'm going to order something else.”

Emoji Tier List Special

5:20 to 7:30

The hosts discuss their recent member special about ranking emojis and the process involved.

“John, can you explain to us how we handled this, how you assembled this for us, please?”

Refinements in the Tier List App

7:30 to 9:30

Discussion on improvements made to the emoji tier list app for better usability and features.

“Oh, and I will say that, you know, again, like all the tier list things that I updated the tier list app for this one.”

Trimming Video with New App

9:30 to 12:30

The host shares experiences about using a newly created app for lossless video trimming.

“but I trim off the beginning part where we're talking to each other in the end part after we're done.”

Improving Sonos Experience

12:30 to 14:01

The host discusses improvements made to their Sonos setup and issues experienced with Ubiquity networking.

“And it all worked pretty much flawlessly.”

SonosNet and Modern Networking

14:01 to 18:08

Learn about SonosNet's impact on network performance and recent updates.

“And I tried that, and it seemed to help a smidge, but it didn't really help that much.”

The Case for and Against Squircle Jail

18:09 to 22:00

Explore the arguments for and against Apple's squircle icon design rule.

Implications of Apple's Design Choices

22:01 to 28:00

Discuss the implications of Apple's design decisions on developers and users.

“And this was this is the reason they went with this behavior.”
Show all 45 chapters

Apple's Icon Design Changes

28:00 to 33:00

Discussing the recent changes in Apple's icon design and their implications for developers and users.

“Yeah, and they were so mean about it too.”

Apple's Icon Design Changes

33:01 to 34:43

Discussing the recent changes in Apple's icon design and their implications for developers and users.

“doesn't work now i need to workshop that one we are sponsored this week by lisa let me tell you a true story.”

Sony's Digital-Only Future and CD Resurgence

35:03 to 42:00

Exploring the implications of Sony's decision to stop selling PlayStation discs and the surprising rise in CD sales.

“Okay, so last week on Overtime, we discussed how Sony is no longer shipping PlayStation discs or game discs after 2028.”

Apple Service Price Increases

42:00 to 43:50

Discussion on Apple One service price changes and inflation concerns.

“Family goes up$2 to$27.95 and Premiere goes up$2 to$39.95.”

FileMaker Usage and Alternatives

43:50 to 46:40

Exploration of why businesses continue using FileMaker and potential alternatives.

“just things that have RAM and storage that are increasing in price.”

Transitioning from FileMaker

46:40 to 48:40

A case study of a user transitioning from FileMaker to Claude Code.

Community Feedback on FileMaker

48:40 to 52:10

Review of listener feedback on FileMaker and community interactions.

“definitely piqued my interest, and I presume Marco and John's as well.”

Apple Silicon Extreme Chips

55:12 to 56:00

Discussion on the development of Apple's extreme chips and their potential.

“We have, speaking of Gurman, we have a Gurman sandwich.”

Discussion on Apple's Extreme Chips

56:00 to 58:50

Exploration of the development status of Apple's m1, m2, and m3 extreme chips.

“I think it was an obvious kind of like suffix with the suffixes that we knew to call, you know, they got the Pro, the Max, the Ultra.”

The Fate of the Mac Pro

58:50 to 1:01:20

Insights into the development and cancellation of various Mac Pro models.

EU Regulation Changes and Impact on Apple

1:01:20 to 1:04:40

Discussion on the EU's changes to battery removal regulations affecting Apple products.

“And the M-Drylter Mac Studio was already pretty weird because it was a generation older chip introduced.”

Challenges of Replaceable Batteries in Devices

1:04:40 to 1:10:05

An examination of the complexities and implications of making batteries replaceable in consumer electronics.

The Complexities of Replaceable Batteries

1:10:05 to 1:12:08

Discussion on the challenges and implications of making batteries user-replaceable in devices.

“So it really does start to become pretty cumbersome in a product that's very small, like a watch and like an earbud.”

Regulations and the Monkey's Paw

1:12:08 to 1:15:49

Exploring the challenges of regulations and their unintended consequences through the metaphor of the monkey's paw.

“And there might be really good reasons that you might not be considering or you might not know about.”

Rumors of an OLED iMac

1:15:49 to 1:19:46

Insights into the recent rumors surrounding the development of an OLED iMac and its implications.

3D Printing Applications in Everyday Life

1:19:46 to 1:21:33

Discussion on the use and benefits of 3D printing for practical solutions and fun projects.

“But what a weird place for OLED to appear.”

3D Printing Solutions for Everyday Problems

1:24:00 to 1:26:30

Learn how creative 3D printing can solve practical issues in daily life.

Custom Innovations for the Beach Community

1:26:30 to 1:28:40

Discover how custom solutions can enhance community living and personal convenience.

“And it has like a special storage box to store the square ice cubes in the freezer.”

Exploring 3D Printer Options

1:28:40 to 1:31:10

Get insights into the best 3D printers for home use and their features.

“If you're willing to be a little bit flexible on the design details, try me.”

Transitioning from Tinkercad to Onshape

1:31:10 to 1:33:55

Understand the evolution of 3D modeling software from beginner to advanced.

Transitioning from Tinkercad to Onshape

1:33:58 to 1:34:26

Understand the evolution of 3D modeling software from beginner to advanced.

“One of the many perks of membership is ATP Overtime, our weekly bonus topic.”

Analyzing AirPlay Bugs in Overcast

1:38:00 to 1:39:58

Learn about the challenges of debugging AirPlay issues in Overcast's audio stream.

“Okay, anyway, before the Overcast rewrite, which I guess is now launched about two years ago, maybe, what I had written for the AirPlay 2 system basically worked.”

The Excitement of Reproducible Bugs

1:39:58 to 1:42:05

Discover the thrill programmers feel when they identify a reproducible bug.

“Audit the AirPlay code, try to figure out why, and fix it.”

Investigating Audio Engine Issues

1:42:05 to 1:44:54

Follow the journey of diagnosing audio playback issues and testing solutions.

Speed Control and Playback Problems

1:44:54 to 1:47:37

Understand how speed control affects audio playback and related glitches.

“So the way Overcast has done the audio engine so far, I write the whole pipeline that deals with smart speed, voice boost, and the sample decoding from the file and everything.”

Developing a Custom WSOLA Algorithm

1:47:37 to 1:52:00

Learn about implementing a custom WSOLA algorithm for improved audio playback.

“so i'm like how how can you work around this so fable churns away and churns away and it's like okay the problem is that apple's airplay system library when you use the playback speed on this file.”

Implementing Advanced Audio Features

1:52:00 to 1:54:25

Learn about integrating smart speed and voice boost into Overcast for Apple Watch.

“I had to modify a couple of things about the UI because it made some crappy UI to toggle on and off in the watch screen.”

The Role of AI in Bug Fixing

1:54:25 to 1:55:48

Discover how AI assists in identifying and fixing bugs in software development.

“But its ability to find and fix bugs, it's shocking.”

Overcoming Development Challenges

1:55:48 to 1:57:46

Hear about the struggles and breakthroughs in maintaining Overcast amidst persistent bugs.

AI's Impact on Software Quality

1:57:46 to 2:01:29

Explore the implications of AI tools in improving software quality and productivity.

Encouraging Programmers to Utilize AI

2:01:29 to 2:03:31

Get inspired to use AI tools to tackle programming bugs and improve software.

“And so to see in one year a huge list of performance improvements, bug fixes, like little nice to haves being finally added, that's because they're getting assistance.”

Personal Experiences with AI Bug Fixing

2:03:31 to 2:06:00

Hear personal anecdotes about the effectiveness of AI in fixing bugs in applications.

“I forget exactly what the situation was, but I know that I am now paying for Fable in a way that I wasn't just a few days prior.”

Exploring Bug Fixes with AI Models

2:06:00 to 2:09:55

Learn about the challenges and insights in using AI for debugging code.

“to go through and understand top to bottom, inside and out, what's being done to what I still consider to be my code.”

AI's Limitations: Not Converging on Perfection

2:09:55 to 2:10:19

Discover the limitations of AI in achieving error-free code.

“But right now, in my experience with the top end models, they do not converge on perfection, just like humans don't converge on perfection.”

Cultural References and Humor

2:10:19 to 2:10:53

Engage with the hosts' discussions about pop culture and humor in tech.

“And what's also a shame is that we don't use curse words on the show or in our titles most of the time.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:00Marco Arment:Coincidentally, I was going on the ferry back to the mainland, John, yesterday. I'm just going to always say New Jersey every time you say that until you stop. New Jersey? So as I'm on the boat, I get an alert that all of my transcription servers in the data center are offline. Oh, that's fun. So all the Mac minis. And I'm like, huh. Now, fortunately, they aren't literally all in that one data center. I have a few that are at CyberLink, which is MacMiniVault, I believe, was their name for that. But CyberLink is like the company behind it. So I have a few at CyberLink and I have a few in my house.

0:35Marco Arment:And because transcripts are prioritized based on the show's popularity and I have something like 12-ish total Macs that are not in the data center, I can go without some of them for a while. And you might not notice because the popular shows are still getting transcribed. and so anyway so so i'm on the boat and i and i'm you know i get all this alerts i'm like huh okay so i i do some quick diagnosis of like you know can i can i remote desktop to any of them can i access any of them no um i open up the uh unify uh ubiquity portal thing of like their site manager like and it says the entire unify console is offline hmm that's not great so i'm like well looks like i'm stopping at the data center on the way home so i show up and you know i go into the cage and everything is off like the power is off all of the power lights are off to everything including the console and everything yeah hold on let me stop right there to get into

1:33John Siracusa:the cage you have to pass through humans right so when you say oh no definitely not i have to pass

1:38Marco Arment:through with like a key card a data center does not typically have a ton of humans walking around um at least not you know it's like there's like one guy at a security desk at the beginning who you know you like wave to and that's it like but then okay yeah but between me between the entry door uh to the building and the room that contains the servers there's like four or five key card doors i see and you need the key card and the pin for most of them so like it's it's it's a good amount of security because you know like data center like who rents servers in a small data center like this. It's like banks and government offices.

2:15Marco Arment:They have compliance and security requirements and everything. So there's a lot of security and logging and everything.

2:22John Siracusa:But what I'm driving at is when you say that the cage was off, you're not talking about the entirety of the data center. You're talking about your particular rack in the data center? Yes.

2:33Marco Arment:Everyone else was fine. So my rack is just everything's off. And I'm like, that's not great. so you know i i opened the door and even you know the door also had like there's like a big like swinging like cage metal door that also has the keypad scanner on it front and back so you know nobody else who has access to the data center room with servers in it can access my servers unless they have my key card you know like so it's and i can't access everyone else's which again for security that's that's a good thing so anyway um so i figure out that um remember i had mentioned when i was describing how i had to learn all this stuff and all you know all the stuff i didn't know i mentioned a couple things one thing i mentioned is that data centers all use 220 volts which i didn't know so everything has to be like dual voltage which most stuff is which is fine um and also that it's fed from two different banks of power yeah and you have to have a transfer switch anything that does not have two power supplies um you have to have an automatic transfer switch that that like intelligently basically like keeps everything plugged into one source and if that source goes away it instantly switches to the other source kind of like a ups with no battery just switching between two different power sources um well my ats was dead and so the switch the power switch thing is just i go and it's like blinking some weird lights and it's totally dead like well that's the problem okay so long story short i have to replace that now you know i had gotten some like cyber power thing and it was like it was about a thousand dollars which i thought would be decent i'm kind of thinking maybe it's not

4:06Casey Liss:decent now well did you try and plugging it and plugging it back in from both power sources of

4:11Marco Arment:course that didn't fix it well it's like blinking these weird you know it's basically blinking it's blinking all of its lights at once and what all the documentation and ai stuff i could find said it was basically like yeah some part of the board probably failed great so anyway not super happy with cyber power and that i'm wondering if anybody out there has any specific recommendations for what I can replace it with.

4:30Casey Liss:It's not still under warranty? It probably is.

4:33Marco Arment:It's six months old, so I'm going to order something else. Right now I'm borrowing one from the data center until I get something else because also all these things have lead times to order. I'm curious. It seems like this is actually not that common of a need because there aren't that many options out there. I think most people who are racking servers like this generally use servers with dual power supplies so they can just plug directly into both power sources and they're fine. I'm a little curious if anybody has any recommendations for you know i need to plug in you know eight c13 plugs i think they are the the ic little you know the three-prong thingies i think i think they're c13s um anyway i need to plug in eight of those to something and they take power from two c20s so uh we'll see if anybody has anything specific let me know otherwise i'm going to order some trip light thing.

5:21All right.

5:22John Siracusa:We have a new member special. We did ATP tier list emoji. John, can you explain to us how we handled this, how you assembled this for us, please?

5:34Casey Liss:Yeah. So obviously we didn't do all emoji because there's just too many. We did a whole bunch though. And all three of us just basically submitted lists of emoji that we wanted to put into tier lists. And that's how we came up with the list. There was a little bit of overlap, but actually not that much. and this was also another opportunity for me to try out my tier list app last time a lot of people requested a feature where you could see the tile that we are tiering at a larger size and so that feature was integrated and it was very useful for the emoji although like i said on the the episode i could only get emoji at 160 by 160 pixels i don't maybe that's as big as they are in mac os but i don't know anyway well there was they were scaled up past 160 by 160 so you could see them better in the video and yes like all our tier list videos there is in fact our tier list specials there is in fact a video on youtube but a secret url that you should not share with non-members but if you are a member in the show notes at the very top will be a link to the video so you can watch us put emoji into tier list and as usual um the episode always all these member special episodes but especially the tier list ended up leading us into weird directions i feel like this one had like a lesson at the end that I didn't anticipate going in.

6:47Casey Liss:So maybe you're not interested in what we think about different emoji, but I think we did cover lots of aspects of what makes a good and bad emoji. And at the end, there's a little lesson of all this. Let's say something that I learned from the special, and maybe you'll learn it too. And I have gotten a lot of feedback from people who say they listened to the episode too and were also surprised that they had different ideas about what emojis were supposed to mean and be used for than we did.

7:12John Siracusa:Fair enough. Marco, if you are not a member and wanted to become a member to hear this absolute incredible podcast episode and watch the absolute cinema that is John's YouTube video, how do you go about doing that?

7:25Marco Arment:ATP.FM slash join and just give us eight bucks a month. That's it. It's really simple.

7:30Casey Liss:Oh, and I will say that, you know, again, like all the tier list things that I updated the tier list app for this one. During this episode, I thought this is great. I've named all of the little, you know i've named the emojis with the apple name that they give them so we can read them on the show they just appear in little tooltips that have like a modifier key that i can hold down and mouse over them and then it was just the tooltips for being so annoying about appearing and we had to fight with that the whole episode well you'll be glad to know as soon as the episode ended i fixed my tier list app much more reliable and easier to read tooltips basically the problem was i was using the uh the browser native tooltips if you just do like title equal attribute on a thing but just fine that works it's like native you get it for free but browsers tend to have a delay before they show that like if you mouse over something with the title equals whatever attribute it will show the tooltip eventually and that eventually thing is so flaky and weird and i guess i was just not being patient enough or whatever but anyway i replaced that entirely with a better system so next time zoomed in tile so you can see them and also tooltips that work and then the next episode that we do is here list i'll find out the next feature that my app needs that it doesn't have.

8:36John Siracusa:Right? Now, remind me, I'm genuinely asking, I'm not trying to snark, that was almost entirely vibe-coded? Is that right? 100%.

8:42Casey Liss:Haven't touched any of the code. It was just entirely vibe-coded. It was one of my first vibe-coded things.

8:47Marco Arment:Gotcha. And you will not hear anything about those tooltips in the audio version because of the magic of editing.

8:51Casey Liss:Yeah.

8:52Marco Arment:It's in the bootleg, though.

8:54Casey Liss:Oh, you will hear Casey try to come up with the name of all of the emoji and then me jump in with the actual name once I get the tooltips to actually work.

9:02John Siracusa:That's exactly how it went. It really is. All right. And then with regard to the member special is the perfect segue. Uh, you apparently made a new vibe coded app or no. Yeah.

9:12Casey Liss:I talked about this on an earlier episode that I, I've had this, you know, I not thought I do any video editing, but when we do these member specials, we do the tier list where we have a video of it. And again, it's just a video of the tier list. You can see us dragging things up. Um, at the very least that needs to be trimmed. I don't do any editing on it. It's basically unedited like a bootleg, but I trim off the beginning part where we're talking to each other in the end part after we're done. and me being me i want to do that in a lossless way and i had been using ffmpeg and asking gpt for the for the ffmpeg invocation because i could never figure it out i would just say you know i want to losslessly trim the ends off this thing here's the start point here's the end point give me the ffmpeg command line and i will copy and paste it but that got old and so i'm like okay this vibe coding error i'm going to make a little video lossless video trimming app that does the same thing i just give it a file i have two text boxes start and end and it just trims the ends off that's all it's supposed to do and i talked about that app i don't know several episodes ago and now i finally get to use it for the first time i get to use it for real we do a whole tier list thing we record it we get a video recording i go to the video recording i find the time stamps put them in hit trim check the video that it produced and it was messed up the video ended like 15 minutes before the audio so like the video was not as long as the audio which is not something that should happen because it was the full video and the full audio they were just massively different lengths probably because of variable bit rate or something or not variable frame i don't even know all i can say is it wasn't usable at all and when i made this vibe coded up i had mentioned this uh actual commercial app called lossless cut that does the same thing and way more it's like a way more sophisticated app but essentially does the same thing you want to losslessly trim audio and video this does it for you um and so i had to use lossless cut because my stupid vibe coded app didn't work but of course after i used lossless cut i went back to my vibe coded trimming app and i said i have a video file and i asked you to trim it and the resulting file looks like garbage fix it and it did so maybe next time my app will work cool making progress left and right i mean and i will i'll mention this as in another follow-up item if we get to it but like this is you know i see lots of discussion about like everyone will be making their own custom apps with vibe coding or whatever software development is software development even if you're literally not writing the code you like that phenomenon i've made an app and i think it works on all these test files and then you get a real file and you try and it doesn't work that happens with apps the humans code too all the time like that's just app development you're like oh i'm done with my app i made this little tool and it's great well have you how many bugs does your little tool have probably a ton whether you wrote it or whether whether an lm wrote it it's probably filled with bugs and so you're going to use it and then you're going to find more bugs and it's just like the the maintenance there is no avoiding the maintenance of writing something keeping it working it and finding all the bugs and all the dark corners where they're hiding there's no escape from that at least so far with current technology there's no escape from that with or without humans so here i am making these toy apps to make my life easier and all it's doing is reminding me that software development is software development even if you're not writing the code all right uh

12:15John Siracusa:I wanted to share with the listeners that I have, I think at this point, I can say I have leveled up my Sonos experience. And if you're not a Sonos or Unify person, maybe you can skip to the next chapter because this may or may not be relevant to you. But I wanted to tell everyone that, so when I first got my Sonos stuff, that was when I was still on Eero. And it all worked pretty much flawlessly. It was really, really reliable. Then I moved to Ubiquity. We talked about this several episodes ago. What episode was that? I think it was 693. And we talked about how what happens is, and my understanding of this is very faint, so forgive me if I get this factually incorrect, but the gist is Sonos stuff would create their own kind of ad hoc-ish Wi-Fi network that they call Sonosnet.

13:00John Siracusa:And what would happen is, particularly if you plugged one of your Sonos speakers into Ethernet, which most of the non-portable speakers have Ethernet jacks, if you plugged something into Ethernet and had Sonosnet running, it would try to bridge these two networks. And oftentimes, because of Spanning Tree Protocol or STP, there would be some sort of thing that would cause Ubiquity to crash out to use what the children use these days or basically just freak out. and it was annoying and so that led me to following ubiquity's official recommendations which is to either go all in on ethernet or not and so i took all my stuff off ethernet it was mostly fine but like latency particularly when starting or stopping something was kind of a pain in the butt and and occasionally i felt like i wasn't getting reception so to speak on a portable speaker where i really should you know i'm not that far away from a ubiquity wireless access point and yet it's still pausing or dropping the audio every 30, 40 seconds, something like that.

14:01John Siracusa:And then if you recall in 693, what we were talking about was that Ubiquiti put a shot across Sonos's bow because they had a marketing video where they talk about, oh, and if you have troublesome IoT equipment and they're focused on a Sonos speaker labeled as Sonos on their little Ubiquiti dashboard that Marco was talking about earlier, you can turn on the STP edge, which will prevent these sorts of issues, yada, yada, yada. And I tried that, and it seemed to help a smidge, but it didn't really help that much. Well, good news. Our long national nightmare is over. Our long international nightmare is over.

14:35John Siracusa:Because maybe a couple months ago, Sonos, this was about two months ago, came out with a new version of their apps that lets you turn off Sonosnet. So it will only use Ethernet or Wi-Fi. It will not create their own ad hoc thing. And so on Reddit's announcement post, there's kind of an FAQ. Should I disable SonosNet on my modern network? To which the answer is, for most modern high-speed routers, disabling SonosNet can actually help things run smoother. Newer routers use Wi-Fi technology that can sometimes conflict with SonosNet's mesh. So turning it off lets your system lean fully on your home network instead.

15:12So if you think about it, when Sonos was brand new, the quality of Wi-Fi,

15:17John Siracusa:I don't know when this was, but this was like 20-ish years ago, the quality of Wi-Fi in the home was frigging terrible. And they couldn't rely on your home or apartment's Wi-Fi in order to have a stable connection. And so SonosNet made sense 20 years ago, but I'm not so sure it makes a whole bunch of sense now. So there was a different post on Reddit where they have a full-on FAQ. One of the questions was, how does this affect surround stereo pairs in home theater? Which is a question I had because you want extremely low latency between like I have a Sonos Arc Ultra, which is wired into Ethernet.

15:47John Siracusa:and then I have two 1SLs in the back and a Sonos Subgen 3. And so how does that work? So the question was, how does this affect surround stereo pairs in home theater? No changes to your surrounds or sub. They still connect directly to your primary soundbar exactly like they do today with or without the setting. That low latency link is its own thing and isn't part of Sonosnet, so disabling Sonosnet doesn't touch it. The one nuance, if you had a soundbar connected over Ethernet, hello, with Sonosnet off, it'll stop bridging Sonosnet out to your other Sonos speakers. It'll still talk to its own surrounds and sub.

16:19John Siracusa:And heads up, disabling SonosNet is not the same as disable Wi-Fi in a single speaker. That one does cut off a surround or sub entirely. Different settings, different results. So again, the SonosNet thing was their own thing, and you can turn that off now, which is great. And then additional FAQ question, I'm on Unify. Can I finally mix wire and wireless without the network going crazy? I didn't write that question, but I could have. So the answer was with SonosNet running, it could create extra network paths back to your Unify gear. And if Spanning Tree wasn't enabled on all the ports with Sonos players with the right port cost, that's where a lot of the chaos came from.

16:55John Siracusa:Turning SonosNet off removes those alternate paths and should clear up that common scenario. I can't promise it fixes every Unify issue. That really depends on your topology and the specific problem. But for the wired plus wireless makes my network misbehave headache, this is the fix a lot of you have been waiting for. Hello, that's me. That's what I was waiting for. And then finally, I thought it was interesting. The official guidance on the Sonos knowledge base, if you will, has changed slightly. And so we'll put a link to the support article in the show notes. Wired setup without dedicated Sonos wireless.

17:24John Siracusa:I'm sorry, let me back up. This is to choose between wired or wireless Sonos setup. And then one of the options is wired setup without dedicated Sonos wireless network. With SonosNet disabled, you can connect one or more Sonos products to your router via Ethernet cable while the rest of your system stays connected to your home Wi-Fi. SonosNet can be re-enabled at any time. The setup is suitable for situations where your home Wi-Fi is strong, but you want to hardwire specific products for a more stable connection. That's me. You want the remaining wireless products to continue using your home Wi-Fi directly.

17:53John Siracusa:Also me. And you want to simplify your network setup and keep more control over your home network configuration. Yes, yes, and yes. so i apologize for the a lot of reading but i thought for the six of you that are in the same situation i am this is a very convenient and useful psa so if you two don't have any questions i'll move right along excellent uh friend of the show louis mantia wrote a very interesting blog post where louis was basically saying maybe squircle jail isn't so bad john can you tell us about this this is louis heel turn yeah he basically presents the argument for squircle

18:27Casey Liss:jail here's why it's good um and even though i think all the things he said are true i still hate it uh for the reasons he said so some of his arguments are um basically if you wanted to sum up the uh the gist of it is that uh squircle jail which as a reminder is the uh enforced by the operating system rule that mac os icons must fit within a particular shape squircle that apple defines and if your icon does not fit in that squircle apple will take whatever your icon is shrink it down and shove it inside a gray squircle so it's very draconian big sir recommended this way back when and some apps followed it and some apps didn't and apple decided you know what we're done recommending no choice you go in the squircle that's squircle jail uh and so louis basically says this effectively makes it so that the best icons have to be a little worse because the best icons were not always constrained by the squircle so they are necessarily going to be worse because now they have this new constraint that they didn't have before that makes them more samey and everything but it also makes the worst icons better because the worst icons were really bad like even in the the big sir recommendation the recommendation was you should be in a squircle that's like this and the squircle was not the size of the full icon the full icon would be like 1024 by 1024 or whatever but the squircle was inset from that but then things like games would ship and they'd be like oh our icon is the the character from our game and he fills the whole thing and so you'd have like on your dock you'd have a bunch of squircles and a bunch of other apps and then you'd have this big square that is taller and wider than all the other icons because they didn't get the memo about using margins because they don't care about the mac and you're just lucky this game runs on the mac at all you'll take what you can get so squircle jail makes it so no matter what garbage you put in your your app no matter what the developer of this app makes for their icon we will make sure that no matter what it is it is within the new designated squircle and that does make it so that you don't have these ugly third-party apps in your dock that just don't fit in and don't look right and look like they're garbagey shovelware because that's exactly what they are so it's kind of a leveling force um he also makes a mark and he's how like uh even though all the apps are the same shape now we can tell that a document is like a tall rectangle and an app is a squircle and blah blah blah and it's like that's great but like i don't need to know their apps if they're in the dock i can tell their i can tell their apps i don't get them confused with folders or whatever like that that apps are you know that basically is the argument the the iphone defined an app as a squircle and apps are squircles and all apps should be squircles because that's the app shape it's even used in like the sf symbols it's like app it's just a squircle uh i don't find that argument compelling because i still wish my apps could at the very least break the squircle if not be entirely non-squircle shaped um but yeah it is a leveling it is a uh make the bad icons better and make the greatest icons worse and the average goes up as a result of that but it makes this sameness that i do not find particularly pleasing and also i like people who can't make icons to have terrible icons that's an important signal like like that's it's like oh if only we could help them by programmatically trying to make their icon better it's like they should be punished for not making good icons if it's a bad part of a game and they've used the full icon area and their thing looks weird it should look weird let it work look weird i want the best to be able to flourish and like i don't i don't like this leveling so anyway you can read it and he's an icon designer and you know he goes through all like the the downsides of squirkle jail and he has plenty of complaints about uh this particulars of icon composer and how it works as we talked about last week and when we talked about our icon so i encourage you to read the post because it's very interesting and informed by literal decades of making icons professionally including i believe for apple so uh this is not just like some random opinion but i personally wish that that constraint did not exist i wish icons were free to look their

22:12Marco Arment:best and their worst yeah i i think i love that louis wrote this because it first of all i think it's a pretty uh bold stance to go to to write this in the sphere that we are in like in like the apple nerd sphere and you know you have people you know people like john gruber like you know holding up the the other end of the argument pretty strongly um and obviously you know john you but um to come out and say like hey here's actually some pretty good reasons why the squircles being enforced actually makes sense but i think what's you know you have to also kind of i don't know if louis came out and said this but i think the strong implication is that these are apple's reasons and i don't know you know because louis works very closely with in this world so like my guess is he has heard from people who have told him this is the reason and that I think is really interesting to hear that like the reason why Apple makes the design decisions they do is not always communicated to us it's almost never communicated to us and sometimes they do stuff like squircle jail where you know we can't see any upside of this like we we from the outside look at this we're like why would we do that why would they do that the only possible reason is that Alan Dye sucks but there actually was probably an argument internally about this particular thing.

23:31Marco Arment:And this was this is the reason they went with this behavior. And even though we don't necessarily agree with it, it was really interesting, I think, to be presented with that argument and to consider, oh, well, it isn't just because somebody's stupid. It's because they had a tradeoff that they were optimizing for and they chose differently than I would have chosen, probably. But they did have a reason.

Read the full transcript

23:53Casey Liss:Oh, I've always understood the reason, Like even when they did it in Big Sur, I totally understood why Apple wanted to do this. Like it was obvious. The thing that I hate so much about Squircle Jail is not so much their decision, which I disagree with, that they made in Big Sur that all Mac icons should be Squircles. I disagree with is that they entirely threw out the carrot and just brought stick. And they were like, because they tried in Big Sur. They said, this is what we want to do. And adoption wasn't terrible. Adoption wasn't terrible. Like we all saw a lot of Mac icons go squircle in Big Sur, which I think was Mac OS 11.

24:27Casey Liss:We're very bad at remembering these things, but it was a while ago. And obviously there's a discontinuity between 15 and 26. Anyway, Big Sur was a while ago. And when they did that, it's like, well, obviously they want to do this because apps on all their platforms are the same. And this is leveling and blah, blah, blah. And like it makes sense from a platform owner's perspective. And for someone who is in the mindset of like, look, I'm in charge of all Apple platforms and I want this uniformity. But, you know, the argument against it then was like that. We don't appreciate that uniformity on the Mac.

24:51Casey Liss:I don't care what icons look like. I don't care what app icons look like on the iPad and iPhone, blah, blah, blah. But they weren't happy with the adoption they were getting, which, again, I think was pretty good. I think most Mac apps post Big Sur in a couple of years went squirky. But some of them, like Rogarima, Rogamiba's app went squirky. But like, oh, the microphone and audio hijack is just poking out of the thing because it's just a fun way to do it. Makes for a nice composition. and the thing about squircle jail is not so much the decision that they made back in big sir but it was now if you're one pixel outside that thing we punish you we make your icons look worse your app that you already shipped that you didn't plan on updating the icon for like there was no sort of like oh you get a pass because you built this app before like we made this decision it was like nope i don't care when your app was compiled i don't care when it was built i don't care anything about it you are going in squircle jail that's the the thing about it that i think is so terrible obviously i disagree with the squirkeling decision too but the jail part the jail part adds an extra insult because it takes people's work and it makes it look worse and demand let me maintain his point it makes the terrible icons look better because they were so bad before that being put into squirkle jail improves them like they get a sheen on the top they're the right shape and the right size they fit in better in the dock but i don't care about garbage apps i don't help the garbage apps at the cost of punishing rogue amoeba for having a two pixel thing or what was the other one like like paste bot had like that a clipboard icon and the little metal, like the metal thing that you squeeze to, to put, you know, to hold paper in the clipboard, the top of that poked out above the clipboard by two pixels.

26:18Casey Liss:Squircle jail. So anyway, please, please check out the post. It comes from a, a, a very informed position and it's, it's balanced and very nuanced. It's not like he's coming out strongly in favor of squircle jail. He's like, it sucks in these ways and it's good in these ways. And in the end he is in favor of the leveling because I think because the appearance of one terrible icon in his doc bothers him so much that he's like yeah they deserve to be punished and shoved into a smaller spot but yeah opinions vary well and we've all seen those

26:46Marco Arment:apps like we've all we've all downloaded some like open source thing and you know it's and the icon is just like a big square that filled up the entire bounds and or a squircle that's the

26:55Casey Liss:wrong size i just did that recently with an app i don't want to blame this app it's not their fault because i think they were trying to do the right thing but their squircle was was like top to bottom that's right like there was no margins around the squircle so it was bigger than all

27:06Marco Arment:the other squircles it is a very modern appley thing to to to have the thought process of like let's make it impossible for people to do the wrong thing let's let's make it let's enforce with technical measures the design that we think is right and harmonious and whatever you know bs terms you want to use it's a very modern apple thing to enforce by like technical restrictions doing what they think is right.

27:35Casey Liss:Rather than competing on the merits. You're right, that is an Apple thing.

27:39Marco Arment:Well, yes, but yeah, totally. But like, you can see how they get there. Again, like, I don't think this was the right decision, but you can totally see how modern Apple with their modern sensibilities would easily be able to convince themselves that the right solution to a problem is to make it technically impossible to break out of the bounds that they want to enforce.

28:01Casey Liss:Yeah, and they were so mean about it too. like they're just two degrees off of a perfectly valid solution to give just two quick examples one they made it so that you get the squircle icon even on you can't have like like i was trying to do for all those times i had these amazing hacks to be like on the old os i get the old icon like on a mac os 15 and earlier i get my old icon and then 26 and later i get the new one apple's like nope we're not gonna bother letting you do that oh you figured out a way to do that we're gonna close that hole we don't want that to happen which is so dumb because it goes against their uniformity thing because a tahoe icon a mac os 15 does not fit in it looks really weird right so they did that and like if they wanted to make it essentially like it's hard to do the wrong thing make it hard but not impossible that's the power of the defaults people who don't care about their apps like oh by default they build with the new sdk we squircle jail them well but

28:48Marco Arment:then you then you have the power user problem the world is filled with people who are like i know what i'm doing so i'll check the power user box and then they shoot themselves in the foot

28:57Casey Liss:but it's not the shovelware game developers they're not going to dig out the checkbox buried 20 letters deep to make their icon uglier like you just gotta the defaults go a long way like there's a continuum which is like lead by example which they tried to do in big sir it's a recommendation and lead by example and i think that got them really far but they're like no it's not far enough there's still people who aren't following our lead so then i would have said make it default make it actually hard to do it the other way and then allow people's old icons to still be viewed on the old OS and new icon on the new OS, like have this transition period, right?

29:28Casey Liss:And then see what your adoption is like. And if it's in the mid 90s, you'd be like, we're probably done. We don't have to really, but they just went straight to draconian punishing everybody. And by the way, you can't use your old icon, old OS's either, which really sucked. Anyway, this is not the end of the world. It's just, like I said, when I posted about this, literally the worst thing Apple has ever done related to design on the Mac, not because it's so bad, but because it's so severe. It is a minor thing it's not that big of a deal but they have never like punished app developers in the same way when it comes to design when it comes to pricing and stuff and and uh you know business practices but they've done much worse things but design specifically never have they done anything this bad i think even like brushed metal and stuff that was not this bad what about all the icons of the menus no not as bad oh brave bold stance john syracuse you can turn that off real

30:17Marco Arment:easily one thing to consider also is you know you're like you are characterizing this as a punishment and i'm saying it's technically enforced but i think that also kind of kind of casts a an attitude or an intent that they that might not have been that considered it's very possible that like as they were writing the you know whatever parts of like the dock whatever parts of the system render icons they had to change a lot of that when they brought in liquid glass and icon composer icons because now these icons are not just bitmaps that they're rendering they now have this dynamic rendering that we talked about last episode of like they were you're defining layers and effects and it's rendering them in real time and so they had to probably rewrite a bunch of parts of the system that render icons in different places and maybe when they were doing that they were just like oh it's a lot easier to do this or it avoids some complexity or it avoids some possible bugs if we just draw in a squircle shape like and if we draw every icon in the same canvas shape it helps us somehow implement this feature and so it could have just been as simple as they were doing that and they were like oh oh yeah we got to figure out what to do with old icons and they might have just like crapped out a solution because they only had two seconds because mac os is under resourced all the time and they might have just like had to rush together some kind of solution for compatibility with old icons and that's just what they came up with as quickly as they could and maybe it wasn't that much thought about of like oh we should we should enforce this technically maybe it was just as simple as like this is a quick hack and then it passed some some review and they didn't have time to do anything better or no one really thought

31:48Casey Liss:about it well they have plenty of time to change it since then but i don't think your idea about the circle simplification is true because all of the icon runtime icon setting apis let you set it to anything you want and if you're not in the mac app store you can set it to anything you want too so obviously their icon rendering system still 100 supports all the old icon sizes formats and everything they just don't let you ship an app that does that by default through the mac app store right again if you're outside the mac app store there's apis where you can get around this but just it's it's punitive because it's punitive to your icon you did an icon design and they are going to make your icon design look uglier because nobody's icon looks good shrunken 70 and put on a background that is a that is a punishment that is a stick to make you do what you that you weren't doing before you didn't follow their lead in big sir and you've got a tiny thing poking out of the rim of your icon like they could have even made the maybe like you can have some things poking out and it's okay we won't put you into squircle jail like nothing and again tahoe's been out for a while and they haven't made any adjustments and i also think they haven't made any adjustments in 27 so it seems like they're pretty dedicated to this and i'll just have to live with it but it will go down in infamy if i could think of a shirt to yell them about it i would but i don't think this idea fits on a shirt very well yeah i see what you're doing there i don't know it doesn't work now i need to workshop that one we are sponsored this week by lisa let me tell you

33:07John Siracusa:a true story. When I do ad reads for ATP, I typically try to record them around seven o 'clock at night, which is just a little while before we start at eight. And I do that because I want my voice to sound kind of fresh, but also similar to how it'll sound on the show. Well, if I sound a little different or gravelly right now, it's because it's 10 51 PM and I'm tired. I am looking forward to bed. If you also want to look forward to bed, then let me tell you about Lisa. Lisa has a lineup of beautifully crafted mattresses tailored to how you sleep. Each mattress is design specific sleep positions and feel preferences in mind.

33:40John Siracusa:They have all sorts of different mattresses for all sorts of different body sizes, including, you know, kids. They have mattresses for your pet, or I should say pet beds. They have a wide variety of different mattresses, including they sent us a natural mattress, excuse me, actually a natural hybrid mattress. And they did this for Michaela's room because she was the one most in need of a new mattress. However, I wanted to be able to honestly talk about it on the show in moments like this one. And so before we moved it to her room, we put it in the guest room for a few nights, and I slept on it for a few nights.

34:14John Siracusa:And I got to tell you, it's genuinely very good. It's something that I kind of wish that I had done our mattress instead, because I really like the one that Michaela now has. From night one, you'll feel the difference. Premium materials that deliver serious comfort and full body support. No matter how you sleep, just take the Lisa Sleep Quiz, and you'll find your perfect match in two minutes or less. Lisa mattresses are meticulously designed and assembled in the USA for exceptional quality. And they back it all up with free shipping, easy returns and a 120 night sleep trial. So go to Lisa.com for 25 % off select mattresses, plus an extra$50 off with promo code ATP exclusive for our listeners.

34:51John Siracusa:That's L E E S A.com promo code ATP for 25 % off select mattresses, plus an extra 50 bucks off. And when you use our code, of course you're supporting our show Lisa.com promo code ATP. Thank you to Lisa for sponsoring the show.

35:08John Siracusa:Okay, so last week on Overtime, we discussed how Sony is no longer shipping PlayStation discs or game discs after 2028. Glenn Jarvis writes, one more aspect of Sony's decision to no longer sell PlayStation discs is its impact on competition. In making games digital only, Sony will run the only store where you can buy PlayStation games. Thus far, Sony has always had to compete with physical and online retailers, GameStop, Amazon, etc even if you only ever purchased digitally you still you have still benefited from this competition and the impact it has on prices and sales and the effect that there's a resale market

35:42Casey Liss:for reselling discs that's going to drive down the prices of stores but yeah that's why companies love it hey now we control all sales because we don't have to sell to retailers we don't have to you know ship inventory we don't have to give the retailers a cut of the sales it's you know It's wonderful for Sony, not so great for gamers.

36:00Marco Arment:Yeah. You know what every company loves? Hey, can we make more money? Yes, then we're going to do it. Let's do that. Let's make more money.

36:08John Siracusa:That sounds excellent. All right. And also within the same context, CD sales are up, as it turns out. Jay Peters at The Verge writes, CD sales are apparently going up, reportedly thanks to fans realizing they're an affordable way to support their favorite artists. according to a new report from research firm Luminate 16.3 million CDs were sold in the first half of 2026 in the US a 16 % increase year over year the growth in CD sales was driven by collection building, price accessibility massive albums such as BTS BTS's Ararang I probably pronounced that wrong, I'm sorry and strong K-pop release schedule, Luminate says

36:41Casey Liss:I feel like this is just a natural progression of the weirdos who were into vinyl but now there's like a slightly younger generation who are like okay well what is the equivalent you know my age minus whatever equals vinyl and then younger people come along they say my age minus whatever equals cds it's retro and and i've seen it in my daughter she suddenly got into cds and wanted to see all my cds which i did not want to let her touch but she was going to destroy them but she just took them anyway and she bought herself a little portable cd player and it's just like i remember this the new the new retro thing that instead of vinyl it's cds the good thing about cds is they have better sound quality at least and they don't uh get worse every time you play them so i endorse kids looking at and in fact they probably a better sound quality than most of the streaming they're listening to probably another mastered blah blah blah yeah but uh i still think it's kind of silly but it's less silly than vinyl so that's fine and i don't actually think people are buying them to support artists but it is true that artists surely get tons more money from you buying one of their cds than you streaming their song literally thousands or millions of times yeah i mean let's just let's just be honest cds

37:41Marco Arment:are cool that's it like and the reason why kids of various generations keep getting into you know old media first of all it is novel to them and it has the appeal of retro stuff like that's why do you know how many people i see walking around all the time wearing nirvana shirts who are you know like 15 years old it's like yeah weren't born when yeah not even close to born like you know like kirk cobain died probably 15 years before they were born like you know it's not not even close um but that's you know look like when we it's like when we were in high school and you know know somebody of our generation was around like a pink floyd shirt yeah yeah i don't know what the

38:20Casey Liss:offset is but like i said whatever the number of years is that you have to subtract because we weren't going back to like wax cylinders right like we weren't like oh rack cylinders are retro

38:28Marco Arment:it wasn't that we weren't that old but yeah but we weren't going back to eight tracks right so

38:33Casey Liss:exactly so there is there is some offset of like this number of years before you were born is when things are retro cool things before that nah don't bother things after that oh that's just normal so whatever that offset is i think now that the generation that is of age to listen to music the offset lands them in cds well and it's you know it's similar to like you know it's

38:53Marco Arment:that same generation also thinks old crappy digital cameras are really cool yes my daughter

38:59Casey Liss:was into that and borrowed my old crappy digital i told her the camera on your phone is better than

39:02Marco Arment:this she's like i don't care yeah they they know it's fine i mean yeah just like yeah everything sounds better than vinyl but no one cares right like that's people like novelty and i think the reason we didn't go back to like eight tracks when we were young is that we didn't have like ubiquitous streaming music that we that would that would be our alternative like we were living in a music scarce world still it was just scarce on different media and if we were going to go back and try to buy an eight track player and eight tracks they wouldn't have been nearly as cheap as cds are today like cds cost nothing like if you if you want to go buy a used cd you can go on you know whatever amazon or whatever or your local whatever store and get used cds for nothing like they they just cost nothing and i'm sure you can get cd people still make cd players so that they're not that hard to come by so for a young person today to get into that kind of stuff like it's more different than their experience of music today because it's it works so differently than streaming in so many ways and it's super cheap and your and their parents have a million of them

40:04Casey Liss:so yeah and music scarcity is novel like you said music scarcity was not novel to us when we were kids because it was just the situation but once you get into a place where music is no longer scarce the scarcity itself is a novel aspect of that draws into it and as for the cd players like the one my daughter bought is so clearly made for people her age like it is it was like a so you're a teen in 2024 this whatever you know ten dollar cd player is specifically for you like it's not if you look at it and like i think i told you about it once like the door to the place where you put the CD didn't open all the way.

40:37Casey Liss:So like you had to kind of like slide the CD, like you couldn't just put it vertically into it. You had to slide the edge in and put it down. It's like, don't they know how to, they lost the, they lost the knowledge about how to make CD players non-stupid, but she liked it.

40:49John Siracusa:That's all that matters. You know, your, your discussion about, you know, what's novel and what isn't reminds me of the Douglas Adams quote, anything that is in the world when you're born is normal and ordinary. And it's just a natural part of the way the world works. Anything that's invented between when you're 15 and 35 is new and exciting and revolutionary. and you could probably get a career in it. Hello, computers, for at least Marco and me. Or not computers, you know, like internet, et cetera. Anyway, the quote continues, anything invented after you're 35 is against the natural order of things.

41:16Casey Liss:Yeah, that quote doesn't go backwards, though. They should have said anything, you know, in heavy use 15 years before you were born is retro cool, whatever the offset is.

41:26John Siracusa:All right, then let's go to price increase corner, my favorite corner. Chance Miller, 9to5Mac writes, Apple has raised the price of Apple Music, which I saw and was like, yeah, whatever. I mean, I'm not a fan of them raising prices, but I mean, I have Apple One. It's fine, right? So Chance writes, Apple Music individual goes up to$12. I believe this is a month. Family goes up, I'm sorry, it's$12 from$11, so a difference of$1. Family goes from$17 to$20, difference of$3. Student goes from$6 to$7, which is a difference of$1. But then I kept reading and it says Apple One, uh-oh, individual,$19.95, which is remaining unchanged.

42:07John Siracusa:Family goes up$2 to$27.95 and Premiere goes up$2 to$39.95. And then one of you, I presume John, found a chart by BasicAppleGuy, which points out that when Apple One Premiere launched, which was roughly 2020 or thereabouts, it was$30 a month, which I thought was actually a legitimately decent to good deal. Now, six years later, it's$40 a month, and I'm not sure how much I feel about how happy I am about that.

42:35Casey Liss:Does that even track inflation, though? Like, I put this in here. I wish they had drawn an inflation line. What this shows is basically the prices of everything is going up. But that's just that happens with inflation and we've had some difficult inflation times here. And you can see there's kind of an increase in the plateau than other increase. So it's not steady. But I didn't do the math to figure out what would be the inflation adjusted price. But Apple One is still a good deal at 40 bucks. But just FYI, the price and everything is going up. And I'm pretty sure Apple Music and Apple One do not include any RAM or storage.

43:02John Siracusa:No, I thought Apple One did include storage.

43:05Casey Liss:You know what I mean? Like the actual service is not like it's not RAM prices are not going up. You know, what's driving these price increases? is it because ssds and ram cost more no this is just a service but yeah the price of things goes up over time and i think this kind of shows that apple services are are i don't know what the economic term for it is but like changing the price is probably not going to appreciably affect how many people continue to be subscribed to it because people who have the service like the service and are not that sensitive to a one or two dollar adjustment because like i said these all these adjustments are they're per month still but it's like one dollar three dollar one dollar two dollar two dollar two dollar those are the increases so even though percentage wise they might be high.

43:42Casey Liss:Absolute value wise, this is like one extra cup of coffee a month for these services. So I think Apple's going to get away with it just fine. But FYI, it's not just things that have RAM and storage that are increasing in price.

43:53John Siracusa:John, that was a much better joke than I gave you credit for. I was not picking up what you were putting down and I apologize. The offending person will be sacked. Mark Ehrman writes, prices for monthly AppleCare Plus packages went up by 50 cents for Macs and iPads while the annual offering rose by$5. The increases are exclusively for new signups so users of older plans will retain the current prices.

44:13Casey Liss:That's nice. It's too bad they don't do that with the other services, right? Can you imagine? If you're already subscribed, you keep your old price. I guess that's probably not tenable, but yeah. Pricing's increased corner. I'm not sure when it will go away, but it just keeps coming.

44:26John Siracusa:I asked, it was either last episode of the episode before about FileMaker, and what the heck is it? I think it was two episodes ago, or no, it was last episode, then we got some feedback about it. Some of that feedback includes Anonymous, who writes basically answering the question, why are people still using FileMaker? Anonymous says, FileMaker developers would probably admit that they are not the most technically knowledgeable, but they understand the business problem they are trying to solve better than most quote-unquote developers in any language. Over the years, many businesses have tried and failed to move off of their old FileMaker systems.

44:58John Siracusa:But now with agentic coding, more and more people are thinking it's possible. Claris is making a big deal about their upcoming agentic development support because it's the only way FileMaker will survive. If it were actually possible to quote-unquote vibe code a FileMaker app versus a web app, I still think FileMaker has some advantages. I promise this next part is not meant to be a sales pitch, but you asked what it's like to be a FileMaker developer, and these are some of the key reasons people still use FileMaker today. FileMaker doesn't scale to thousands of users well, but it's more than powerful enough for small businesses to run their internal systems.

45:32John Siracusa:Therefore, if FileMaker works, why spin up all this extra web technology that you don't understand as well. FileMaker is just an app that can work with no internet connection. It's an easy way to deploy an offline app on iPhones or iPads to collect data in the field, especially in low coverage areas or on the ocean. FileMaker apps aren't subject to all the increased CVEs due to AI, making it a safer choice for businesses to deploy internal tools. I first discovered FileMaker before I even knew how to code, and now, even though I do know how to code and agents are helping me code much faster, there's still something special about FileMaker being this all-inclusive app and platform that your app can be deployed into securely that is intriguing.

46:09John Siracusa:For those of us who know how to write code, it's easy to miss the value in an app like this. I think FileMaker's biggest hurdle is their licensing doesn't allow someone to try and use it for free for more than 45 days, which allows more modern apps like Monday.com to gain more of the market.

46:24Casey Liss:I feel like this is a list of things that are, you know, things that are still nice about FileMaker. and the largest argument for most file maker things is hey is it currently working and it supports your business then you don't need to change it so you enjoy the advantages and it's working so it's fine but it is so clearly just technology that i don't think anyone should start off using in this day and age even for a totally local thing it's possible to make entirely locally running apps that are not based on web technology or like you know just a regular app or you know even if it's like a local web app like we have so much better technologies and we'll see as we go on like filemaker these you know the the apps and and forms and everything they make they never looked good i remember back in the classic mac os days they always looked weird and they continued to look super weird uh and as for them being no cves due to ai finding flaws i think that's just a security through obscurity thing if you point if you pointed any kind of security researcher or human or otherwise at filemaker and it's uh various apps it would rip it to shreds because they're not made with security in mind now i granted they're not network connected so that helps them avoid things like this but anyway um i do not endorse starting any new projects in filemaker but the people who like it and use it there's no reason that they shouldn't continue to use it because if it's solving your problem if it ain't broke don't fix it right and i think that is the largest argument and that's why like i said the people who have a business running a filemaker as long as filemaker keeps going and stays as a viable application and they keep updating it and your thing keeps working and they add little handprints over time there's no reason to look elsewhere uh but yeah if you're starting something fresh i would still recommend maybe not

48:01John Siracusa:filemaker in 2026 point of order you said that they're not network connected i think that can be true but it's not exclusively true because i believe a few people wrote in and said that they have a way to deploy filemaker stuff to the web so before we get a bunch of feedback yeah it's this

48:15Casey Liss:the argument this person was making is like you can use them entirely offline and so on and so forth so but yeah i i would not vouch for the security strength of filemaker at this point

48:25Marco Arment:yeah because you could also use like a sqlite database and any app like there's a lot of ways to solve this problem with today's software and tools especially as we're about to get to

48:35John Siracusa:yep so uh robert thompson writes marco got me off filemaker after 18 years which of course definitely piqued my interest, and I presume Marco and John's as well. Robert writes, for the last 18 years, I used it to run the back end of my wife's wedding business. It handled leads, contracts, staffing, timelines, payments, and automated tasks like email reminders. It was also slow, looked terrible, never worked on mobile, and required basic plugins that cost hundreds of dollars every few years. Worst of all, building anything, and it felt like coding in Scratch with commands dragged and dropped into a script editor.

49:07John Siracusa:John nailed it. It's software anyone can use to create a database one step above Excel. This past May, when our renewal came up, we were quoted$4 ,100 for a three-year server license. That same week, HP 690 was released, and I heard Marco talk about using Claude Code to build his Saunter Map app in under an hour. I thought, if Marco can submit to the App Store in under an hour, I can do this in a month. So I said, screw it. Installed Claude Code on a local Linux box, hit the Fable window, and ran it. We should disclose once again that we have been sponsored by Claude, We will be sponsored by Claude.

49:39John Siracusa:This obviously does not, we didn't write any of this stuff, but just so you know, they are a former and future sponsor. Anyway, continuing with what Robert said, within three days, I had recreated and greatly improved the system we had used for 18 years. Everything is better. It's fast, it's organized in a proper web app. I can process a detailed lead with fancy timelines on my phone in under a minute standing in line. It's amazing. The entire rebuild cost me$100. We've now been using it for two months, and it's been remarkably solid, and I've been having a blast adding new features. I hadn't caught that episode, I would have paid$4 ,100 and been stuck in FileMaker hell for another three years.

50:13John Siracusa:Instead, I got out for$100.

50:15Casey Liss:Yeah, so this is the other thing about people who are using FileMaker. Like if it's working for your business or whatever, there's no need to change it. But when you're in a market like that, where it's like you are using some very old technology and your business relies on it, someone out there knows that your business relies on it. And maybe they also know that you're not particularly in a position to change that, which is why you've been using FileMaker for 18 years to run your wedding business. but lo and behold now suddenly you're not stuck in filemaker there's this new alternative that is not open to everybody but if you're the type of person who's basically a filemaker expert it's not that far of a leap to to go to like i bet you could do some vibe coding stuff to make a decent app now how will this turn out again see my caveats about vibe coded apps it's really easy to get the first one up and you're amazed that it works and everything where are all the bugs is it going to be as problem free as the thing you've been using for 18 years so on and so forth but on the flip side of that is the excitement of building something new with modern technologies and so on.

51:05Casey Liss:So I'm not saying everybody who's on FileMaker should vibe code their replacements, but the fact that that's now a viable alternative, I would imagine is going to further accelerate the decline of FileMaker usage. And I'm not sure if it's going to raise or lower the prices of these weird server plugin licenses that people are charging. But yeah, when you're in a technological backwater, things get weird. Yep.

51:25John Siracusa:And then both of, or all three of us, I should say, really wanted to shout out Adam1, who wrote us, John apparently computed a nearly 3 ,000 word email with 15 screenshots explaining the history and functionality of FileMaker. It was fascinating. I don't think it's published anywhere public, so I can't share it. And I wouldn't even begin to try to summarize it, but we wanted to shout out to Adam because that was really incredible. So thank you for taking the time.

51:50Casey Liss:You asked like, what is FileMaker and why do people still use it? I feel like that could have just been like a blog post somewhere and that answers every one of your questions with copious screenshots and history, like the background, where did it come from? How do people use it? Like if you were interested in FileMaker that was the answer. It's a shame that he just emailed us and did not publish it on the web, but maybe he'll make a FileMaker database. People can view it through their cloud integration.

52:13Marco Arment:Also, for an email to stand out among the FileMaker emails that we got...

52:20Casey Liss:Which we continue to get. One just came in while we were talking, by the way.

52:22Marco Arment:Yes. For one to stand out as being long really is saying something because I think the average word count of the FileMaker emails we've gotten has got to be like a thousand at least like yeah it seems like everybody who uses filemaker not only is emailing us but also has a lot to say about it i'm going to shout out to adam too i think

52:42Casey Liss:it was his email i think he's actually a pretty good writer i think it's maybe because i'm reading too much german stuff and i was reading this like look at these sentences they make sense they logically connect they're just it's never confusion about what uh what uh you know something refers to it's like wow i don't know if adam is a professional writer or something but it was very well written

52:57Marco Arment:I think at this point everyone who uses FileMaker is a professional writer like that's what our emails are showing but yes Adams really did stand out as being especially good and thorough

53:06Casey Liss:yeah maybe it's because they're all older and actually learned how to write in school or something I don't know it's just ageism I'm thinking all the old people are writing about FileMaker but anyway FileMaker fans are out there and they're writing us and I salute them but I think

53:22Casey Liss:FileMaker is not a sentence let's say

53:25Marco Arment:we are sponsored this episode by atp membership please consider becoming an atp member today so for just eight bucks a month or a little bit less if you do the whole yearly plan thing and we have a couple of local currencies as well you can support the show and what that gets you is first of all ad free episodes you don't have to hear ads like this or our regular ads in the show at all second of all gets you exclusive member bonus content every show has the overtime segment which is exclusive content just for our members a whole extra topic every show we also every month do member special episodes that are on things like movie reviews tier lists and other fun stuff like that you get access to the bootleg feed which if you want you can listen to the show faster than everyone else and totally unedited it's right after we finish recording usually on wednesday nights you get the bootleg version of the show which is totally raw you get to hear casey swearing and all the weird stuff we go through with all that stuff it's fun but mainly membership supports the show you know in times like this it's the summertime right now summer has been historically a pretty rough time to sell ads especially late july august like a lot of people are on vacation the ad sales go down and this is the nature of ad sales they go up and down and you know just in terms of broadly across the whole year like we can have like a whole kind of low year or we can have low months or low weeks and with atp membership we are very fortunate that our members support us directly.

54:44Marco Arment:And so ever since we started that, the pain of those dips of the ad sales cycle, it hurts a lot less because the members really support the show. So please consider becoming a member today. We really appreciate your consideration. We really appreciate all the members out there. For some reason, you're hearing this somehow, but you shouldn't be in your ad free feed. Anyway, please support the show. You can join us at ATP.FM slash join. Check that out today. Again, eight bucks a month to start ATP.FM slash join. Thank you so much. And please give it a shot.

55:16John Siracusa:All right, let's do some topics. We have, speaking of Gurman, we have a Gurman sandwich. We have two Gurman topics with something else in the middle. And we start with Gurman writing a few days ago as we record, early in the development of Apple Silicon, Apple was working on a processor that would be even more powerful than its Ultra chips, offering twice as many graphics and main processing cores. These chips are often referred to as the extreme chips. I'm told Apple had developed both M2 Extreme and M3 Extreme chips before winding down those projects due to cost and demand concerns.

55:50Casey Liss:All right, pause here for a second. I'm paraphrasing, as I often do with Gurman, and I'm editing him on the fly to try to make it sensible. But I think what he said was that he refers to them as Extreme chips. I'm not going to say we're the first people to refer to the quad chip as an Extreme chip. I think it was an obvious kind of like suffix with the suffixes that we knew to call, you know, they got the Pro, the Max, the Ultra. And imagine if they had two Ultra stuck together. that would be extreme um so he has to establish up front hey they were making the m1 extreme which i think there was like leaked like diagrams of how the four things were going to interconnect like the inter you know the interposer on it on the ultra is just like and and they just connect to each other we know what that looks like uh and then it was like well they never came out with anything bigger than that but here was the plan for the m1 extreme or the m1 quad or whatever and here's how it was going to connect and stuff and obviously they didn't ship that this post here which was like i don't know a few days ago was the first time i've seen a source non-frivolous like you know a information from what we has historically been a reliable source which is german even mentioning an m2 and m3 extreme maybe i've forgotten the history and those were like road mapped out or whatever but it's saying his his wording here which i didn't change is apple had developed both uh i'm told apple had developed both m2 extreme and m3 extreme chips before winding down those projects due to cost and demand concerns what does developed mean does that mean they were they were uh on a roadmap as in like we had a bulleted list of the chips we're going to make and those were on the bulleted list does that mean some people were at the point where they were designing that does that mean they had like how far did they get in development before they canned that because if they canned the m1 extreme for cost and demand and blah blah reasons like as far as i can recall that's like the last we heard of extreme which like they were gonna do the m1 one it was too big and they just really never revisit that topic and like i said in a past episode at some point during this when i was talking about this topic we got that other completely unsourced unreliable who knows rumor that like oh yeah don't expect anything extreme size until after the m7 or the m7 or later because the current roadmap includes none of them but maybe he just only has vague information so that's all i can give but saying they developed them before winding down those projects doesn't tell me how far along they got so it's not actually that interesting it could be that when they laid out apple silicon before a single chip had been designed they just laid out all the chips they're going to make and these run the list and then just when they canned the m1 extreme because we know about the m1 extreme because like we saw they got at least this far because here this diagram this is apparently supposedly an officially leaked apple thing so they got as far as this drawing did they get further than that i don't know but i've seen nothing about the m2 m2 extreme but just fyi german now says those were apparently more than

58:29John Siracusa:a twinkle in someone's eye they were certainly more than a twinkle in your eye but no i hear

58:35Marco Arment:what you know they were only a twinkle in my i cannot make them myself it sounds like it lines up with you know roughly what we have heard already they had these extreme chips that were like in development or being considered or being worked on and they decided for whatever reason probably cost not to ship them that i don't think is new um or not to make them like not just not

58:58Casey Liss:to ship them but like we're not going to even make this because i i don't remember if i talked about this on the show but i had some information back in the day of like that they they got to the point with the unwarm extreme where they were asking vendors if we asked you to do this for us how much would it cost and uh they would give back numbers that were basically like the equivalent of saying go away as in they would give a number that they know no sane person would ever accept like that we just want we won't do this for you here's here's a made-up number that is so much higher than anyone will pay and of course apple said yeah we're not paying that and that basically

59:28John Siracusa:killed the the project the m1 extreme yeah so uh mark german continues as for new mac pro hardware there were two additional models developed by apple that never saw the light of day j170 and j190 the former was a new intel corporation mac pro so a new mac pro with the

59:45Casey Liss:intel cpu sorry i didn't edit out the corp one i usually usually edit his thing when he says apple inc until that's a bloomberg style thing not german's fault but yeah a new intel mac pro

59:53John Siracusa:thank you uh the latter was an m3 ultra version that was supposed to launch alongside the m3 ultra mac studio in early 2025 which were you at the point that you would have bought that john no

1:00:03Casey Liss:i would not have bought that um so this is the you know the code names so again that uh there were two additional models developed by apple what does developed mean well apparently they were developed far enough to get code names i don't know how when that happens in the process do you give them the code name before you start work on it maybe um a new intel one is interesting just because like i guess they could have put newer zeons in it or whatever like i'm not sure what kind of revision that would be because it's not like the intel one was like super dated and in desperate need of an update um but okay so there was another intel one and it makes sense to can that one because apple silicon's coming along the m3 ultra version is what everybody thought when they came out with the m3 ultra mac studio whereas the mac pro that does the same thing not that it would have been great because we think it's stupid the mac pro would be exactly as powerful as the Mac studio and it's like well you did it once before it's no stupider when you do it the second time but they didn't even do that so I maybe that one got farther along maybe they got to the point where they actually did work on that one and then just decided we're gonna can this whole thing let's not bother but again this is the first information I've seen certainly about oh there was going to be a second Intel Mac Pro somewhere in the timeline I mean I guess that was maybe there is a hedge like we don't know when Apple Silicon's gonna be ready so we better to get another Intel Mac Pro ready, just in case this takes longer than we think, and it didn't.

1:01:16Casey Liss:And then the M3 Ultra Studio, or Ultra Mac Pro to go with the M3 Ultra Studio. And the M-Drylter Mac Studio was already pretty weird because it was a generation older chip introduced. But still, it was Apple's most powerful chip. And if they were going to keep the Mac Pro on life support for a little bit longer, this is what they should have made. Again, I don't know how far along it got, but oh well. More details for the sad death of the Mac Pro.

1:01:42John Siracusa:Was it sad? Anyway, EU drops the battery removal requirement for Apple Watch and AirPods. Hartley Charlton over at MacRumors writes, The European Commission on July 14th adopted new exemptions to its batteries regulation that free the Apple Watch and AirPods from having to offer user-removable and replaceable batteries. The commission has expanded its list of exempt product types to six additional categories, folding and wearables like smartwatches and fitness trackers. The reasoning largely mirrors why devices like electronic toothbrushes were already exempt. Opening a compact, sealed enclosure and failing to reseal it properly could let water in and create a safety risk.

1:02:18John Siracusa:Products can also qualify for exemption if their construction makes battery removal inherently dangerous or if there is no realistic way to build in user access given current manufacturing methods. The iPhone was already exempt from the removability ruled under the original regulation thanks to its battery cycle life and water resistance rating.

1:02:36Casey Liss:Yeah, I've seen a lot of people speculating that this was like the evidence of some kind of lobbying effort because like the electric toothbrush exception. Well, there are electric toothbrushes with replaceable batteries. It's not technologically impossible. So do the electric toothbrush manufacturers just lobby better somehow? And, you know, seeing a regulation like this where it's like, well, we don't want to require things that are, you know, as the thing said, no realistic way to build it with current technology. that seems to not be a thing they worry so much about when making software regulations not so much because you know that's the problem with software anything's possible in software but there is a feasibility curve to software like what are you asking people to do and how long would it realistically take and what kind of impact would have and how much would it cost whereas with hardware they're like ah people probably can't make an electric toothbrush with a replaceable battery that's safe even though those products already exist it doesn't quite make that much sense but i'm sure apple is super excited to not have to worry about it for the watch and especially the airpods and i feel like the airpods is the only product where it's like okay well you could do it there are earbuds with replaceable batteries like airpods themselves use those little coin cell whatever that's smaller than a coin but the little disc batteries like you could see how it could work but exempting it is just like do we want to fight this fight it's not that big a deal and we did talk about this ages ago that the iphone was exempt simply because the battery in it lasts so long like they have some threshold that's like how many cycles does the battery have to survive to and still maintain 80 of its life or whatever and because the iphones use expensive batteries or good batteries that it gets exempted as well and so when this when the rumors about this was first swirling was like in europe everything's gonna have replaceable batteries the phones are gonna go back like the old days where you could when you drop your phone the battery cover falls off and the battery flies out remember those days it's the thing that used to happen kids if we would drop our phones and the battery cover would fall off and your battery would come out and we didn't care because it was uphill both ways in the snow yeah put the battery back in and anyway and you know air air pods will have replaceable batteries and apple watches and everything will have replaceable batteries and they've just that's not the reality of these regulations your regulations are much more pragmatic it almost makes me wonder like okay well what does this apply to what benefit is this giving to customers so we'll see there was an item that i didn't end up putting in the notes for yet another um over ear headphone that had replaceable batteries and its innovation was not the fact that it had replaceable batteries but that it had replaceable batteries that were even easier to change it looked kind of like they're like kind of like cartridges think of like a big thick sd card yeah i saw that in the ears yeah like and i i love that i love seeing manufacturers compete with each other to uh not just you know say okay well our thing is like has replaceable batteries is more reusable just to make that experience better i think they were gaming headsets like and that makes kind of sense like if you're in this really long gaming session and your your uh your what do you call headphones die that's really bad you're like oh well i can plug it in is there a cord that can reach while i'm still playing the game because you know if it's the built-in battery you need to charge them first of all you better hope your your headphones can charge while they're in use because some of the worst ones are like okay when you're charging it doesn't work like you can't use it while it's charging because it will outrun the charger or something because the you know the charging wattage isn't enough but even if you can plug them in like the sony ones like oh i can kind of snake the cable from my playstation and it barely reaches my ear if i don't move my head too much and it's just but what you can do in two seconds oh my headphones are dying swap swap and that's what these things are fighting on which uh you know like competing on it's like not only can you change the battery you do it without even taking your headphones off because they're it's easy they're like on the outside they slide right down you shove a new battery in and you're good to go love to see it but apparently none of this removable battery stuff is coming from any of the apple stuff at all as far as i can tell including i guess the airpods max if they ever revise them again so i guess uh chalk one up for apple in the battle versus european regulations and i think also like

1:06:32Marco Arment:it's it's worth looking at like the actual problem they're trying to solve here making regulations is hard because when you when you make a regulation like this okay the it's well-meaning and there's a lot of reasons why you would want replaceable batteries first of all it makes things last longer so it reduces e-waste it also makes things more repairable which helps people afford them and use them longer and so like it helps people who are need to be on a budget it helps reduce waste it helps just make devices better and you know just more functional to be able to have you know better batteries easily replaceable for longer so like there's always having like long

1:07:14Casey Liss:use sessions without pausing to recharge you just swap your battery we used to do it on plane flights when when laptop batteries didn't last as long or in a long gaming session you just swap your

1:07:23Marco Arment:battery and keep going right like so there's there's a lot of reasons why a regulator would have the idea or we be presented with the idea we should have we should require that all these gadgets that currently have these like sealed up batteries we should make them all replaceable there's lots of good reasons for that yes there are but see the problem with designing regulations is that a lot of times the implementation details are like actually going to be impossible or will work against your goal or work against other important goals or will be we'll have to have so many holes poked in them that the regulation is effectively useless and i think this is you see exactly what happens here like okay you know with this particular regulation well how do you make how do you take a device like an apple watch or an airpod and how do you make the battery replaceable well first of all you have just sheer size issues of like okay batteries in a device like this, if you're going to open up an AirPod, say, what are you opening it with?

1:08:28Marco Arment:Like, what tools? These are sealed up devices for water resistance as well. So there's a lot of sealing and gaskets and glues and stuff to seal up water. Well, how are you going to get to the battery? And then how are you going to close the device back up after you get to the battery? And that's a really tricky thing, the smaller and more intricate and more precisely manufactured these devices get. generally what that means is you have to have some kind of like removable panel or some kind of maybe if you have it can be like screwed on or something but you have to have then screws and you have to be able to open those screws or like some way to pop the thing open or something and then you have to have some way for it to like assemble back together snapping or you know tabs or panels you know all those gaskets for waterproofing things that are currently glued have to become gaskets and what that does is that adds a bunch of size and mechanical complexity It adds points for mechanical failure, and it adds ways that the water resistance can fail either before or after the battery is replaced.

1:09:31Marco Arment:So by saying you want the batteries to be replaceable to extend the life of the device, what you're also then mandating, if you don't make any exceptions, you're also then mandating, oh, well, the device also is going to have to kind of get bigger and bulkier and, you know, maybe less reliable and maybe less water resistant when the battery is healthy. so that's that's problem that's full of problems right there then the batteries that are in these devices that are not user serviceable can also not have a bunch of safety um like guards against them like you don't want if someone's going to be opening up your device and swapping out a battery you don't want that battery to just be like a raw lithium ion cell with two wires coming out of it you want some kind of you know look at look at what replaceable batteries are they usually are themselves encased in plastic or metal and they have you know custom connectors that are like safe so like if a baby licks it you know it's not a problem like there's all these different you have to have all these different safety considerations around well what is the the battery part that you were inserting and removing how do you get those over time is it a standard or is it custom like all these different considerations you do and then what that does when you when you have like the additional layer of plastic or metal around the battery and whatever is keeping the battery safe when you are removing it and inserting it, that also adds bulk and size and weight and material and waste to those products.

1:11:00Marco Arment:So it really does start to become pretty cumbersome in a product that's very small, like a watch and like an earbud. But even phones, like, you know, the phones these days don't have a lot of free space, you know, like, I don't know if you've seen the inside of a modern iphone or you know it's it's pretty crammed in there so you can start to see like okay this regulation it's very well meaning but the actual implementation of something like this in the modern world has all sorts of complicating factors that either make it nearly impossible to do in a lot of devices or make it such that if you did it you wouldn't really want that resulting product or it would make the product substantially worse so it's a hard problem and like i think it's worth considering like when you hear about a regulation like this instead of doing some kind of knee-jerk reaction going back to the squircle jail with louis post earlier like it's worth kind of learning the other side of the argument like why wouldn't apple want airpods batteries to be replaceable by users is it only because they want to sell you more airpods like that's certainly a reason, but that's probably not the only reason.

1:12:09Marco Arment:And there might be really good reasons that you might not be considering or you might not know about.

1:12:14Casey Liss:Interestingly, at this point in technology, I actually think replaceable batteries in AirPods is easier than replaceable batteries in phones. Obviously, there are already phones and earbuds that have replaceable batteries. But if you look at the battery in an iPhone, that's not that changes from you. That's not a standard battery. It's custom. And not only is it custom, It's custom usually for the specific model of the specific phone. It's not the same even within a generation between the Pro and the Pro Max. Totally different batteries. Next year, Pro is going to have a different battery. There's no standardization there.

1:12:44Casey Liss:So it's very difficult. And just to reiterate, this regulation, according to this MacRumors article, is about user replaceable. It's not just about like you'd send it to a shop and they replace it. Whereas earbuds, they just use coin batteries from watches or whatever. You know, watch style batteries. Like even the AirPods, if you crack them open, all is in there is a standard coin cell battery thing. or air tags air tags that you know how are you going to figure out to have a door and do the extra stuff and blah blah and is it a standard the answer is yeah they're standard batteries that already fit in there in fact the entire air target tag is defined by the battery that is inside it which is very standard that that defines the entire product and they find a clever way to make a little door and it doesn't matter about waterproofness because whatever and by the way i don't even know what the waterproofness uh if that's that's not really worth i don't really know how waterproof the air but uh airpods are because they have like you know these speaker grills where the moving air from the speaker driver makes sound that goes into your ear i don't know if you immersed it and that filled with water would it be fine i don't want to find out i know people

1:13:43Marco Arment:have sent generally the the guideline is they're okay with like sweat and rain but not immersion

1:13:48Casey Liss:i think is although some people have sent their airpods through the washing machine and said they work fine so you know your mileage may vary and they've gotten better over time and it varies between models also but the point is we have a large menagerie of uh differently sized standardized extremely inexpensive coin cell batteries available for devices that size whereas for the phone we do not have a large selection of adequately powerful standardized easily to buy batteries it's all 100 custom with literally little wires and a press connector like it's not not friendly at all so if i had to pick like what device do i want this regulation to land on i would pick the smaller ones and not the phones because the phones everything you talked about the you know the size the bulk the the the phones would get hit by it so bad because phones really are pretty darn waterproof these days and phones are totally non-standard and there is no room in phones and every millimeter counts and it would just be really bad so i'm kind of surprised that they didn't end up applying it to uh airpods but you know just like the toothbrushes the electric toothbrushes that have replaceable batteries that have been on the market for years and their fine there's earbuds that have been on the pocket for replaceable coin cell batteries and they're fine like it's technologically solvable but when you're talking about like it's difficult to make regulations i was kind of thinking like um uh you guys know the monkey's paw i have to ask because

1:15:06John Siracusa:this is not a movie but it's a pop culture thing yeah i it's it wasn't a poe story or something

1:15:11Casey Liss:like that yeah and it's been parodied in a million things you may have seen the simpsons yeah yeah the idea is that there's the monkey's paw which is this magic thing i mean you see it with like a making a wish on a genie you know you rub the bottle the genie comes out and you make a wish lots of make-a-wish stories do the same thing and same thing with the monkey's paw which this will grant you your wish and so you say i wish for whatever and it always grants it in like the worst possible way it's like i wish for a million dollars and then it like it kills your mom because she had a million dollar uh insurance policy so you got a million dollars but now your mom's dead how do you feel right that's the monkey's paw the monkey's paw curls right you whatever you ask for it's gonna the worst possible thing is gonna happen making regulations is like making a wish on a monkey's paw no matter how you phrase it okay i want a million dollars i don't want anything bad to happen i don't want anyone to lose a million dollars i just want to have it and you have to go through all these caveats and no matter how you say it it's like okay and it's not it's not the magic of the monkey's paw that causes this to happen it's the adversarial companies that don't want to do what you're asking that you're asking them to do so they will find a way you know malicious compliance or whatever to not do it or just leave the market or whatever so it actually is a difficult problem and it is not a friendly relationship between regulators and the products but uh that's not a reason not to do regulation like that is entirely a reason to just get better at it uh you know so we'll see how uh this goes in the eu but it seems like the eu is being very pragmatic and careful and maybe bowing to lobbying efforts so that they can sort of and i think this is a good approach you know like a ratcheting approach don't do it all at once do it a little bit at a time allow give people time to adjust don't do it in the really hard cases off that if the electric toothbrush people complain fine you don't have to do it but little by little bit by bit you know as the years pass you keep changing the regulation making a little bit hard we get solid state batteries and we can put way more power in less volume and weight oh we ratchet up a little more uh that's what the whole world should be doing so i endorse the eu's efforts here

1:17:03John Siracusa:uh real-time follow-up it was not edgar allen poe it was ww jacobs again we regret the error

1:17:10Casey Liss:someone in the chat room said they wouldn't have come up with a monkey's paw example that was that bleak i think that's literally from one of the stories like i don't think i came up with that on my own and it may have been from the original story but that that's the monkey's paw for you don't make a wish on the monkey's paw it's bad news gracious all right uh and then finally tonight

1:17:28John Siracusa:uh there is an oled imac rumor that has just sprung up literally earlier today mark german writes apple's working on a new imac with an oled screen that's about it that is the entirety of

1:17:38Marco Arment:the rumor that's it the whole thing i mean that's incredible i think that's exactly the level of reporting that most people care about the imac and it's not to say the imac is like a bad computer it's just not an exciting computer it it serves its role very well and people like it for its role and for the most part it's pretty unexciting yeah so that the non-interest in the imac is why this

1:18:04Casey Liss:rumor is interesting even in its totally sort of like detail-free form and there was like no timeline given although again trying to parse what German wrote because he wrote about other iMacs coming with an m5 or an m6 chip or those two machines is that one presumably that's coming before the oled one or whatever so don't expect the next iMac to have an oled screen but the rumor is Apple is working on a new iMac that has an oled screen timeline and everything else unknown okay and presumably like i said there will be a refresh of the current version iMac with a different chip at some point before the OLED one the reason i think this is interesting is because the iMac is like not a top priority let's say from apple's mac division and macs themselves are not really a top priority it's it's kind of at the bottom of the barrel in terms of macs that need to be updated and unlike the mac pro and the mac studio with you know for reasonable reasons that make sense people who buy an iMac don't need a newly refreshed computer every single year they're not on the cutting edge running the the hardest uh you know lms on their system they just need most power no they're fine with it for years so it's fine for this to be in a slow update cycle which is why i think it is fascinating that there even is a rumor a glimmer of a rumor of a plan of anything about an oled iMac because it's like whoa whoa whoa why is the iMac going to get an oled screen i understand that like the laptops are going to get they're their most popular line of Macs and it just so happens that laptops have smallish screens because they're portable and so it's a lot easier to make a 12 a 13 inch and a 15 inch or 16 inch whatever OLED screen it's cheaper it's smaller it's like it's great iMacs are what are they 24 inches now i believe that's right that's getting pretty big apple just came out with a bunch of 27 inch screens and they don't have OLEDs and they cost a bazillion dollars including the one that they didn't even really change and again there are no timelines involved here but i would really not expect to see an oled mac rumor until oled had spread to every or at least oled rumors had spread to every nook and cranny of the mac line and i would think the oled oled iMac would be the last one because again a 24 inch oled screen maybe it's not hdr i guess maybe you can get it cheaper because it's not hdr and still the studio display xdr will be better because it's brighter and blah, blah, blah.

1:20:20Casey Liss:But what a weird place for OLED to appear. Not weird for it to be in laptops first. And we'll see if those make it into this year or early next year. That makes perfect sense. But the next place I would have expected it before the we just have the new monitors come out, which resets the clock for six years or whatever, is on the super expensive big Mac monitors. That's where you'd expect to see OLED next. They're already three grand or whatever. Throw an OLED on there. But they were like, nope, mini LED, special custom, blah, blah, blah. And so I guess the iMac gets the OLED before essentially before the Mac studio because from Apple anyway obviously there's third-party OLED monitors and believe it or not I know we talked about these ages ago I'm still waiting for some of the stuff that was announced to CES to come out because that's just how CES works uh so still we're halfway through the year and a bunch of stuff that they announced is not out so we don't have any great recommendations for Mac-ish Mac compatible OLED monitors but eventually they'll be out but yeah you get a fancy Mac studio and your choices are uh Apple super expensive mini LEDs, third-party OLED or mini LEDs.

1:21:22Casey Liss:And then the iMac is going to be out there at some point with its own OLED and the laptops will have OLEDs. Maybe the MacBook Air will have OLEDs and the next studio will be like, where's my Apple OLED? I don't know when that's going to happen.

1:21:32John Siracusa:All right, let's do some Ask AGP. And Jack Johnson writes, as someone who has never considered having a 3D printer, I was very interested to hear how Marco and the family have found it to be useful. What are the sorts of problems it solves? How often do you experience a situation where a 3D printer is useful? I have an 11 year old and imagine it would be easy for her to find fun things to use it for. But I'm more curious about non toy applications.

1:21:52Marco Arment:This has been something that I've really enjoyed. The toy applications are very fun. You know, you can go on the different sites like Maker World and Thingiverse and all this. You know, you can find lots of fun toys and they delight people. Like one of the things that like I found like a little tiny milk crate, like somebody like replicated the design of like plastic milk crates. And you can scale it to whatever size you want. If you make them really tiny, they're really cute. And it's really fun to have them on my desk. And I printed one or two on my desk. And then instantly Adam and Tiff spotted them and each demanded their own.

1:22:23Marco Arment:So now I have a few more. And they're all over the house now. And they're adorable. There's all sorts of fun stuff like that that you can do. Of course, everybody prints the dragons and all sorts of stuff that kids like to play with. But the toy applications, I would say don't discount those. But on the functional side, what I enjoy about a 3D printer functionally is, and I think this is maybe like the satisfaction that a lot of people have gotten when they get into woodworking that I've never done. But if there is a problem that I spot around the house, I can have a pretty good chance of being able to solve it.

1:23:01Marco Arment:That also applies to other people's problems around their houses. and so for instance having the restaurant there's like there's a bar it has like this wood ledge around the bottom of the bar like you could rest your feet on and that piece of wood cracked um along like where people if you like stand up and put all your weight on it it can crack in like the part over the ledge and we needed something that would hold it up that was strong um while you know for like a day while we went and you know went off island and got got some other stuff but like we need like a fast solution like what can we do to hold this up today so i printed a little bracket i quickly designed it in uh on shape and i printed a little bracket with like you know little ridges like triangular ridges ridges to support it and i printed it in pet cf so it's super strong and resists moisture because that's important and where we are and um and i made a really really strong corner bracket to help hold it up until we get some metal in there and you know what it's still there because it's actually stronger than most metal brackets you can find around here and it's strong enough to last the rest of the season and then you know in the in the off season we're going to replace that piece of wood so like that's one example another example is um i go to a physical therapist sometimes and um she was saying that that she had um one of those uh vaso machines that like it's basically like inflatable pants like you wear these like airbag things around your legs and they inflate and squeeze different parts of your legs and it can help you know a lot of leg stuff they have some connector that attaches air hoses to a machine and it's a plastic connector and she said she has a whole machine out of commission because the plastic connector on it broke so she can't attach the hoses correctly and the company's out of business and nobody sells these parts so i said all right give me the part and in one day i quickly replicated the part and printed one out in a pretty hard tpu so it's it's very very strong because tpu is a little bit flex like it's like it's bamboo's tpu for ams so it's a pretty firm tpu as they go but because it is flexible and has really good layer adhesion it's really strong in a lot of ways um and i was able to print that out and the whole thing took me including printing time an hour and a half like it was like a half hour to model it and an hour to print one and i brought it to her and it fixed the machine like just having this part now like you have that kind of thing enabled um we have wagons here at the beach like you know you like haul like a farm it's like a big farm wagon you like haul it up to the to town to pick up your freight and you haul it back and the hardware store in town sells these wagons so everyone in town has the same wagon and they all look the same also when people come over here they don't know that these wagons are actually purchased and privately owned so we've had people just kind of grab our wagon thinking oh these are just public for anyone to just take wherever they need so as a result many wagons have name signs on them and many wagons have air tags attached to them somehow.

1:25:54Marco Arment:I solved both of those problems. I made a custom name sign out of ASA because it has very high UV resistance with two colors. It has our last name on it and has a big cavity behind it to mount an air tag in one of Elevation Lab's 10-year air tag battery enclosures. So I'm going to screw that onto the wagon tomorrow with some security screws and now I have a name badge with a 10-year air tag for my wagon in town all again all custom modeled stuff i've made um to solve very specific problems um in town there's the chamber of commerce in town has these like animals around town that like and they had these qr codes under them and the way they put the qr codes on them was basically by cutting out like pieces of like cardboard with you know a taped on sign like laminated to resist the rain so and they look terrible and the head of the chamber is a friend of mine and she learned how to 3d printing and she's like can you make us qr codes and oh my god siri is talking in my pocket it's it's going great guys anyway so i again use the asa because it's very high uv resistance and i have black and white asa and so i made little qr badges for and they're now all over town so i improved my town by like being able to do this tiff had uh she has like a square ice cube maker and like for like cocktails that She makes fancy cocktails and she brought it over from the mainland, John.

1:27:22Marco Arment:And it has like a special storage box to store the square ice cubes in the freezer. So they don't touch each other and stick together, you know, melt together and ruin like the clear awesomeness of them. Well, she left the tray back and back on the mainland, John. And so, yeah. So, so it took me five minutes to model a quick little like plastic divider that she could then put four of them in this plastic divider and just put it in a plastic bag. and that replaced the need for this box in our freezer. I have a coffee filter holder that I made where like next to my coffee station where I like make coffee every day.

1:27:58Marco Arment:I make coffee at the beach two different ways depending on how many people I'm making it for. Either a regular coffee pot, which has those cone-shaped paper filters or AeroPresses, if it's just me or just me and Tiff, which has small circular paper filters. I made a coffee filter holder that mounts on the wall fits exactly like snaps on to like the gap between like the tile backsplash and the wall so like it fits perfectly with no adhesives no anything just like snaps on there and it has nice like wave shapes on it because it's the beach and holds both of those coffee filters my flagpole mount here at the beach um i i have a flagpole off the back of my house um sometimes it's american sometimes it's fish sometimes it's the progress pride flag it depends on my mood and um it's metal and it's the beach so it all corroded and rusted like crazy so i printed a flagpole mount that somebody else modeled but i modified the model to to account for a couple of screw holes i wanted to add to it and i printed that also out of pet cf because it's super strong and very moisture resistant because super strong carbon fiber reinforced plastic at the beach will probably not corrode as quickly as metal it probably will just be there and last for a very long time so this is just like a subset of you know i've mentioned in the past i made like a custom shower drain filter back in the other on the mainland john um i made a

1:29:18Casey Liss:custom shower drain filter and say in new jersey this much you're overwhelming me see i'm breaking the system galloping me with the mainland yep ddos uh if only you could fix my uh my breaking oxo cheese grater thing but unfortunately the part that breaks is essentially paper thin and And I'm not sure if there's any 3D printable material that can withstand that.

1:29:38Marco Arment:If you're willing to be a little bit flexible on the design details, try me. But no, like, can you do things like paper thin? It depends on where the strength has to be and in what direction and how, like, if can I add some like support?

1:29:49Casey Liss:It has to be paper thin because it slides into a paper thin slot, unfortunately. Well, look, I could try like it. I could also maybe the next the next broken one I have on mail to you and you can see what you can do about it because I have one that's in the process of breaking right now and it's probably going to be totally broken soon. Yeah.

1:30:04Marco Arment:So like what I would say is if you are interested in like the functional side of 3D printers, they are definitely very satisfying for that. You almost definitely want an enclosed one. Now, fortunately, there's a lot of really good enclosed printers now. Like, you know, I would definitely go bamboo because they're easy. And I know open source people, I please don't make me fight that fight. Bamboo printers are easy. They're awesome. I would say right now the ideal sweet spot of functionality to price in their lineup is the X2D. It's a relatively new model. It's the one I have here at the beach, and it is awesome because it's dual nozzle, which helps with a lot of support tasks and things like that.

1:30:44Marco Arment:And it only costs – it costs like$100 more than the P2S or something like that. And it also has a heated enclosed chamber. It has dual part fans. It has a really good filter, like an air filter. so if you have it in your office it's not going to kill you hopefully maybe um and so with that you can print basically any material you'd want to print on a home 3d printer including all these cool carbon fiber ones like the the x2d can print all of those um if you want to go a little bit higher end you can use you can get the one i have on the mainland john um the h2s which is um basically all the same capabilities but single nozzle and larger or the x2d if you want dual parts there but for most people's needs i think the x2d is awesome and uh it's about with all the accessories you're looking at about 900 um but it's i can strongly recommend it um anyway having that having that like creative outlet for me i've loved every bit of it it's absolutely fun it's absolutely great and it's super functional and and you can kind of you kind of have a superpower like if you can just fix your problem or someone else's problem like when i fixed my pt's vaso machine like she's thrilled about that because it was literally like this probably i don't know probably a thousand dollar machine um that was useless and she had no other way to fix it and i spent an hour and a half and probably i don't know two dollars worth of tpu and fixed it so to have that ability is really fun and i've never had that in any other way um i really enjoy this um and so i would say definitely get get an enclosed printer try it out it's super fun and you can start like the first custom parts i made i was using using tinkercad which is this web app the autocad makes for children to learn how to 3d model in school and it's very simple it's you know just drag and drop modeling um i did move to i did migrate to on shape for almost everything i do now because on shape is parametric which is wonderful because like you know once you are especially if you're a programmer like if you're listening to this show it's likely you'll enjoy on shape there's a bit of a learning curve you know i watched a couple of youtube videos to learn you know the basics of how to navigate it and what to make and stuff like that but what i like about on shape is you can basically define things as variables so you can you know instead of drawing the circle and saying this hole is five millimeters across you can say this hole is hole width across and then later on you can change hole width and so if you if you're as you're like iterating a design that can be very handy i love on shape for that and on shape allows you to do things that are impossible or difficult in tinkercad like if you want to have like a nice uh you know a fillet or i hope it's not pronounced fillet i don't think it is but if you want to have like a fillet like a curve on an edge or if you want to chamfer an edge like johnny i've like you can do stuff like that a lot like very easily in on shape whereas in tinkercad you really can't and that makes your parts generally look and feel and often work a lot better than you can do in Tinkercad, but Tinkercad is really fast and easy to learn.

1:33:45Marco Arment:You can do a lot with it. As you're learning, that's a great place to start. If the price is okay with you, I'd say start with the Bamboo X2D and Tinkercad. Thank you to our sponsor of this episode, Lisa, and to our ATP members who support us directly. You can join us at atp.fm slash join. One of the many perks of membership is ATP Overtime, our weekly bonus topic. This week on Overtime, we're going to be talking about the internal roles system in modern LLMs and how that relates to prompt injection vulnerabilities. This is going to be a deep dive. It's going to be kind of fun. So, listen atp.fm slash join to get overtime and all of our other membership perks.

1:34:23Marco Arment:Thank you so much for listening, everybody, and we'll talk to you next week. Now the show is over They didn't even mean to begin, cause it was accidental Oh, it was accidental John didn't do any research Margo and Casey wouldn't let him Cause it was accidental It was accidental And you can find the show notes at atp.fm

1:34:55Casey Liss:And if you're into Mastodon You can follow them

1:35:00Marco Arment:At C-A-S-E-Y-L-I-S-S So that's Casey Liss, M-A-R-C-O-A-R-M-N-T, Marco Arman, S-I-R-A-C-U-S-A-C-R-A-C-U-S-A. It's accidental. Accidental. They didn't mean to. Accidental. Accidental. Tech Podcast. So long.

1:35:29John Siracusa:So, Marco, you apparently have fixed another long-standing bug in Overcast. I can't say that I've hit this bug to the best of my recollection, but I believe this is not uncommon. So tell me about this.

1:35:41Casey Liss:Well, this doesn't say that it's a bug. It says you fixed AirPlay. Was AirPlay broken? Well, that depends.

1:35:48Marco Arment:All right, tell me more. It was kind of broken. when I first made AirPlay 2 support in Overcast which I think was around iOS 13 or something like that like whenever AirPlay 2 came out, iOS 12 or 13 I think when I first did that it was a huge undertaking because the way AirPlay 1 works is it just has like a fixed, usually a 2 second long buffer and you just play audio in your app the way you're playing audio anyway and AirPlay just buffers it up for 2 seconds and routes it to the speaker and that's fine but it's you know everything has a two second delay including like you're pausing and everything and that that's kind of an inconvenient you know the the ui showing two seconds off on the time stamps like it's not great airplay 2 works totally differently and the main behavior that you need to know for apps is that airplay 2 buffers up to usually about a minute and a half of audio so the way the entire way the app deals with it and supplies audio to it has to be aware of that it It has to work in a few different ways to be able to like render ahead all of that content without the UI showing that you're a minute and a half ahead, you know, etc.

1:36:56Marco Arment:So there's there's an API to do that. And if you just use AV player, like the default media playing API for Apple, Apple handles all that for you. You don't even have to know. And most podcast apps, that's just what they do because they what they what they need to do fits within AV players capabilities. but av players capabilities do not allow for a way for me to do smart speed so i've never used it with the exception of a few contacts on the watch which i'll get to so airplay 2 it was it was a pretty big undertaking to use like the the other api app provider that gives you like more raw access it's this whole like sample buffer renderer thing and that's for apps that that like play basically raw custom audio streams to work with airplay 2 and there's like this one sample project that they published back when they launched the API.

1:37:43Marco Arment:And that's basically the only documentation for this API. It's very limited. Almost no one uses it. So the information available out there on using it is pretty slim. It's pretty difficult and cumbersome to use. It seems easy until you try to do anything and then you run into problems. And you're like, oh, this is a weird bug. Okay, anyway, before the Overcast rewrite, which I guess is now launched about two years ago, maybe, what I had written for the AirPlay 2 system basically worked. It wasn't perfect, but it worked enough. Now, when I did the rewrite, that required me to change a whole bunch of the code around the AirPlay audio stream player that I wrote.

1:38:25Marco Arment:I didn't actually change anything about the AirPlay stream player, but I changed all of the code around it that manages open files, that sets up the audio stream and manages what file is playing and how times are reported and everything like that. Like all of that changed with the new architecture. But the actual AirPlay component I had written didn't. But for some reason, after the rewrite, everyone was complaining about AirPlay bugs. And I would occasionally see them myself too. Like bugs I'd never seen, like playback would sometimes just stop randomly or it would glitch. And like, what's going on here?

1:38:59Marco Arment:What changed around it that would cause that kind of problem? And I couldn't figure it out. And it's been one of the most common complaints about Overcast. so much so that i actually added a checkbox um saying uh basically force airplay one usage yeah

1:39:15Casey Liss:when i saw that option appear in overcast i'm like oh there must be some real problems with

1:39:19Marco Arment:uh airplay too because when that option appears i could not figure it out so a few days ago like over the weekend i was playing an episode of a podcast that i noticed was making audible glitches over airplay like every about every second or two it would have like a little like static blip in in the stream as if it was garbage audio data being played in the sample buffer and it was happening every time on this one file so i'm like oh this is interesting i have a reproducible case now

1:39:48Casey Liss:i just want to pause here for a second just to say that the excitement that programmers feel when they get a 100 reproducible instance of a bug they've been chasing for years it's like i don't know i'm sure there's like a an equivalent of scientists like when they do an experiment and something happens that they can't explain because when you're a scientist you're doing experiments you kind of know what's going to happen or whatever but if something happens that you can't explain you get so excited if you're a scientist you're like was it a mistake am i just mistaken let me just look at it again maybe it's just like an experimental error blah blah it's the most exciting thing ever for a programmer it's like it reproduces 100 of the time because once you get something that reproduces 100 of the time you will fix it because like now you can't there's nowhere to hide exactly so you know first i'm like i'm looking through i'm like

1:40:29Marco Arment:i don't i still don't see why this file is doing this and why other files aren't either always or never doing this there's nothing special about the file but whatever reason that file was doing great so i bring up fable by the way claude is a past and future sponsor of our show um so i bring a fable and clog code and i'm like this file that i'm playing is glitching about once a second maybe a little little more but you know about once every second or two it's making these audible glitches every time when I play through AirPlay. Audit the AirPlay code, try to figure out why, and fix it. And Fable goes, you know, churns away for, you know, 10 minutes, whatever.

1:41:06Marco Arment:You know, it's not a fast model, but I'm like, it's churning away for, you know, 10 minutes or so, looking at everything, and yes, I'm approving all those things. Do you want to run grep? Yeah, fine, go ahead and run grep. You know, do you want to fetch the sample project on the web? Yes, please, fetch the sample project on the web. Like, do whatever you want to do to try to solve this problem. Well, not...

1:41:22Casey Liss:Shift-Tab for auto mode, Marco.

1:41:24Marco Arment:Oh, really? Yes.

1:41:26John Siracusa:Yes, yes. I've discovered this recently as well. And I stand by that I think, especially as you're getting used to how this whole flow works, I wanted it to come to me after basically every step and say, is this cool, is this cool, is this cool, is this cool? But as I got more and more confident with the tool and instructing it and so on, yeah, shift-tab to go into auto mode and then basically just walk away for 20 minutes and come back.

1:41:51Casey Liss:That's great until it deletes everything. Yeah, exactly. Did you hear that story from Ranz? he was working with one of the another company's models and he put it into automotor and just deleted all of his code oh that's fine i fixed all the bugs it's still a danger so just keep that in mind but uh but anyway yeah uh you know make your estimation about the risk versus reward right i

1:42:10Marco Arment:had to go through and it's like all right i found a couple of edge cases like this apple's documentation of this feature says you should be calling it this way you miss this other chase this other chance here and if this thing ends up failing then this thing is null and this thing fails out and like it found a bunch of like you know kind of minor bugs like that all around my code and it's like all right i fixed all those bugs try it now so i i try it again and it's still glitching and i went back i said that didn't fix it keep looking so it churns away you know another 10-15 minutes like it's churning away i'm improving everything all of its grep commands and uh you know eventually it's like all right i i made regression tests and i tracked the entire audio path and i looked through every single possible way that your audio engine generates audio for this and i found everything was sample perfect there wasn't a single sample drop so it's not garbage data and so but i found this some other little edge case over here and i fixed that so try it now okay so i build it try it now it's not fixed so i'm back i'm saying it's still not fixed sorry sorry keep going keep looking so it's like okay it goes you know 10 minutes later it's like all right i found i dug through i replicate i made sure that your structure of your code matches the structure of the sample project exactly so you're calling it exactly the same way apple is calling it build and run apple's code on your device and see if that has the problem so i build and run apple's code and no it doesn't have the problem okay so let's go back okay so eventually fable's like all right try playing it with smart speed and voice boost turned off try playing it at 1x and 2x and i'm like oh actually the glitches are faster at 2x and then i eventually learned the reason why apple's code wasn't glitching was that apple's code doesn't have a speed control and when i played it at 1x with or without smart speed at 1x it didn't glitch so i was like okay all right so fable's like all right i got you so i went over you know audited the whole like speed control thing and it's like all right well if you call the speed it's like it's probably some kind of like difference in device clocks like in not like the time of day but like the audio refresh clock or the audio sample rate clocks so it added all this logging it's like all right run on your device and airplay at this and see what this log says and it's logging like which clock is being used to determine what the time is and everything and it's like you know is it is it the sample rate of the hardware is the sample rate of the file if those things are different and you know what is airplay using a different uh you know media time clock and you know for its algorithms and everything all this logging didn't fix it okay then it's like okay well that you should file this as a bug with apple this is a platform bug and i said i'm not doing

1:44:49Casey Liss:that by the way i think i talked about this before but earlier than you've gotten to now in your story i would have turned to the thing i've done a few times which sometimes works and sometimes doesn't but at least i enjoy doing which is hey this apple framework that doesn't have good documentation that uh is obviously involved in this problem that i'm seeing decompile that sucker tell figure out how it works i can't do this i literally can't i don't know how to decompile i don't understand the assembly code the i don't understand like machine code or like but these things like i have hopper installed i have a license for it it's got an mcp thing decompile the airplay 2 library figure out how i'm supposed to be using it because i don't trust the apple sample code from two years ago i'm not saying this would have solved your problem but i've done this multiple times which is again why i said like i said in an earlier episode i love having the source code for darwin because that's where all the low-level file system stuff is and i just go to the source code man like it's so glorious but yeah airplay 2 source code not available to you but you know who it is available to an llm with a decompiler that's i never thought of doing that

1:45:50Marco Arment:i've heard i've heard of people having success with decompiling with llms but i i've never thought

1:45:54Casey Liss:to do that i mean it's only it's only if that's the source of your problem it sounds like you've already narrowed it down to smart speed which is not like really airplay 2's fault you know well

1:46:00Marco Arment:No, it's narrowed down to speed. So the way Overcast has done the audio engine so far, I write the whole pipeline that deals with smart speed, voice boost, and the sample decoding from the file and everything. But I use Apple's audio unit for speed up and slow down. The base speed changes. So when you're playing at 1.2x, smart speed is engaged if you have it, but like i'm using apple's audio unit to to change like the actual media time of the file because apple has a really good algorithm for that it's really power efficient it sounds really good i set it on its highest quality mode and i just go and that served me very very well there's a couple of problems with that algorithm one of them is that the highest it goes you can set it to whatever speed you want but if you set it higher than 2.9x the way it speeds up audio after that is just by cutting out chunks.

1:46:55Marco Arment:So it's not really speeding it up after that. It kind of maxes out at 2.9 and just starts cutting words and cutting subsections of the audio after that, which is not helpful, I don't think. And that's why in Overcast, the Overcast speed label that says 3X is not 3X. It's 2.9X. And 2.9 and 2.8 and 2.7, I shifted them a little bit to kind of ramp up to that fake 3X that wasn't really 3x from what i understand almost every other app does something similar or it just does 3x and lets it cut out those chunks which i find unacceptable um and that's at that point smart speed is getting you more than that anyway so it doesn't really matter um but anyway so i'm like how how can you work around this so fable churns away and churns away and it's like okay the problem is that apple's airplay system library when you use the playback speed on this file.

1:47:53Marco Arment:Like when you tell Apple's code to play this file at 1.2x or whatever, anything other than 1x, it glitches. So it has a problem. So you should file this with Apple. And I'm like, well I'm not doing that. Work around it. What Fable then did was go and write its own speedup algorithm. Now the main algorithm to do this, it's called WSOLA. I'll link to an obviously ChatGPT written blog post about WSOLA, I have known for years that if I ever wanted to write my own speedup algorithm, I need to write my own WSOLA implementation. And the algorithms I've written for VoiceBoost, especially, I know that by having written those, I know enough about signal processing now that if I needed to write my own WSOLA, I don't know how it's pronounced, I could do it.

1:48:44Marco Arment:But it would take me probably like two weeks to get it right. And I don't know if mine would sound as good as Apple's or if it would be as power efficient as Apple's. You know, I use all the Accelerate VDSP code as much as I can, but I'm sure they're doing lots of tricks, and, you know, they've had this forever. And theirs sounds good, so I've never touched it. The other problem with relying on Apple for this, though, is that on the Apple Watch, I've never been able to run SmartSpeed and Voice Boost because the Apple Watch does not expose that API in AV Audio Engine. and the av audio engine on the apple watch is the only way to achieve smart speed so like i have to use av audio engine on the watch to get smart speed but if i do i don't have access to the speed up component so i would i wouldn't be able to play it at beyond 1x if i had smart speed enabled or voice boost so on the apple watch i've never had smart speed or voice boost i've just used av player because it allowed me to at least do the basic speed ups so as fable's trying to fix this airplay problem with apple's speed up algorithm it just wrote its own wsola algorithm and just tried that and it's like okay i wrote my i wrote my own with solar or whatever see if this has the problem and so after i picked my job off the floor and looked at the code i was like wow it's full of vdsp accelerate code and it's like this is a really good implementation it looks like has all the right windowing functions i'm like oh wow this is huh isn't that interesting so i pick up my jaw off the floor and i go test it and it works perfectly the airplay problem is solved and so i was like okay uh use that algorithm whenever you're using airplay then huh and then it took me all of about five minutes before i realized well now that i have my own wasola algorithm also add speeds above 3x that's where that came from because why not add speeds now that i have this like this was holding me back all this time and now listeners if you are saying who would possibly ever want speeds above 3x i assure you there are people who want them not least of which blind people because blind people especially blind people who like podcasts and are nerdy like us tend to ramp up the speeds of things like voiceover on their phone and everything as as they get used to it.

1:51:07Marco Arment:And once you start ramping up your speed of processing speech, you'd be surprised what you can understand once you are accustomed to it. So there are lots of people out there who actually want podcasts faster than 3x. So I said, why don't you make speeds up to 5x? First, I said up to 4x. And it's like, all right, I did it. And by the way, I didn't tell Fable about my little, not really 3x, 3x, but when it was adding all those speeds between three and four it said i also fixed your fudge about the 3x speed earlier and all the speeds leading up to it so i was like okay touche like you that's pretty good all right i would have asked to do it anyway but i didn't have to so it added all the speed and then one one beta build later i'm like why stop at four let's go to five i tested it out like does five work yeah does it strain the cpu no okay go to five why not so in the current pending version of overcast that's waiting for app store review as we speak it goes to 5x and it's a real those are real speeds that don't cut any audio out because now i have my own wasola algorithm that was stolen and so i'm also like okay while i have this why not bring smart speed and voice boost to the watch now because now that i have this part i can then use that to bring the entire av audio engine so i had fable do that and it literally took one prompt and about 15 minutes of churning and it brought the entire audio engine to the Apple Watch.

1:52:42Marco Arment:I had to modify a couple of things about the UI because it made some crappy UI to toggle on and off in the watch screen. I made it less crappy and I was looking at all the time like all I really needed was that WSOLA algorithm. Everything else I had all the other components I had written worked fine on the watch. Like I had written them all across platform. Everything else was there except that and for me to do this ever in the past again this would have been at least two weeks for me to write my own one and it probably wouldn't have sounded as good or or been as performant i even did a stress i'm like all right what if i play this on the watch at 5x with smart speed and voice boost is this going to crush the watch's cpu no five percent wow oh my god like this is amazing so i am blown away by first of all how much work fable did to help me find the airplay bug and it was really really good and it made the entire rest of the airplay subsystem much better and more stable and it accounts for more conditions it complies better with what Apple's sample code wants you to do.

1:53:50Marco Arment:It fixed all of those problems. It audited the entire rest of the audio stream that I build to see, is anything generating garbage data and make sure nothing possibly can? It did amazing work. It found tons of small stuff and a couple of big things. And then it wrote me this algorithm that, again, it would have taken me a while if I could even do it well at all. and that that enabled me to do these other two major things that people have been asking for for years and yesterday i went for a run and i had overcast on my watch and i played a podcast with smart speed and voice boost and it was amazing and i just i'm so blown away by how well these modern high-end models are able to find and fix bugs in code and the code they write I'm still not letting it write that much code.

1:54:47Marco Arment:But its ability to find and fix bugs, it's shocking. It really shows me how flawed humans are. Because I'm not a bad programmer. I'm not an amazing programmer, but I'm not a bad programmer. I'd say I'm a decent to good programmer. A lot of this code that I've written over the years, like this audio processing code or the low-level database code, A lot of that I've written very carefully, much of which I actually have done tests for. By the way, there's tons of tests now in Overcast, thanks to all of this work from Fable.

1:55:21John Siracusa:Look at that.

1:55:22Marco Arment:I wrote almost none of them, but there's a lot of tests now. And all that code that I've written over the years, I went through all my emails recently. I went through one-star reviews on the App Store. I went through the Reddit comments. and anytime I found a new bug for the last like you know couple of weeks I've just had fable like hey uh some users report that like the it's showing an infinite spinner on the purchase screen for some users I don't know why figure it out and it'll churn away for a while and it'll usually come up with something that I didn't come up with of like oh well if the connection fails over here because this variable is set to nil it never retries the connection I'm like huh look at that so i'll say all right fix it and i'll look over the fixes it makes and it's usually only you know a couple of lines difference it's usually not that much code that it's writing to fix these bugs it's just changing a couple of lines and i look and i'm like holy crap it's right i missed that and so i've been going through and the last couple of weeks again i've been fixing so many bugs because i'll tell it like i'll think about some part of the app that people complain about i'll say hey uh people complain about the reliability of streaming playback under low network connections go look at the streaming code and fix anything you find and it'll go it'll churn away in the streaming code and i'll say actually yes here's in this case you know you didn't account for this and over here if the server returns return to 200 here it would corrupt the data here it's like oh my god and again like you go look you're like it's right oh my god so fix that or you know people people have reported for years like hey uh ever since the rewrite it doesn't show up very well after the show it doesn't show up very well as the suggested app when you plug in airpods or when you connect airpods and it used to so i i told fable that i said hey this used to work works worse now figure out why and fix it and it'll churn away on my like upcoming media intent code that handles all that and sure enough it found something like oh well actually if there's a race condition during launch and if this database loaded before that database then this one would fail here and it would cause this problem and also you messed up you messed up here here and here oh my god and sure enough you go look yep i sure did and now it's fixed this is such a game changer for so much because as it turns out as i said humans are pretty imperfect and there's and so many of these of the bugs that it's finding for me are bugs that have been keeping me down for years and and i've you know i've kind of alluded to this recently but like i've been i've had a really hard time staying motivated to work on overcast recently because it seems like i'm just bogged down by just like old bugs that i can't find or fix because i've looked for months or years and i can't fix them and i can't figure out why and i've never been able to make tests that actually show them or whatever it is and now i can just ask this ai thing to fix it and like this is the kind of thing there are lots of programmers out there who you can like you can hire they can contract hire somebody who's like a real expert in this kind of thing who's really good at debugging who can look at the kind of stuff like this has existed for a long time but to hire somebody like that would cost a lot of money and i usually have either not been able to afford that or if i could afford it it's like i could just barely afford it with overcast profits and i'm like you know what i don't really want to like that's not really worth the cost of maybe getting someone to maybe fix it if it's going to cost you know tens of thousands of dollars you know for the to have them here it's like this is going to be like a lot it's like that's that's going to be you know pretty difficult for me to justify or or do whereas like now for like i'm on i'm on the 200 a month plan now only because i've been hammering fable like crazy but i did most of this on the 100 a month plan for 100 a month now yes claude's a past informer and future sponsor um for a hundred dollars a month now like i can have this amazing bug fixer available to me whenever i want so i am so literally every single time a bug comes to mind that people have reported or that i see myself i'll just go tell fable hey this audit the code paths around this see if you find anything and fix it and it has fixed so many bugs and overcast so if app review ever actually reviews this i'm on day three right now waiting but if after you ever actually reviews this the the like the 2026.7.1 version of overcast contains major bug fixes major improvements because it's all like stuff that was on my my bug hit list like in the last version the the 0.7 that one i included things like priority playlist priorities were messed up after the rewrite yes they were it fixed it and it was so subtle it was like you know like there was like a reverse greater than less than thing and like a couple other like minor things like that that like i would never have caught it fixed it and then it made a bunch of tests to make sure it fixed it like in this one i fixed storage usage bugs airplay bugs sync bugs i had it i had to look at the server sync code and the app sync code and say hey this is how this syncs figure out any possible way that the sync might have a race condition or lose data or anything like that and it found some and it fixed them so this has sync fixes this has airplay fixes storage usage fixes so background execution fixes like there's so many fixes in this build and i did it for almost almost for free and in almost no time so i do think again as i was as i say every time i talk about ai coding these days uh obviously this does not come with no downsides or no cost to society or no cost to developers like this all has some costs and some downsides and and depending on how you how you stand on certain things those may be you know significant to you or or not but i think if if you put these tools in the hands of conscientious people who know how to make good software and who know what to ask for the implications of to software quality are massive and again like i don't i don't want people to gloss over the fact that apple is themselves using claude heavily and when you see that giant list of fixes that apple's delivering in the 27 os's those obviously came with tremendous help from ai like if you think otherwise i think you're you're deluding yourself ask anybody who's working in this business today if you know anybody in Apple, ask them.

2:02:03Marco Arment:They're using it heavily. And so to see in one year a huge list of performance improvements, bug fixes, like little nice to haves being finally added, that's because they're getting assistance. Again, this is a bumpy road and it's not without its problems, but the implications here for the possible improvements in software quality are just so immense. and granted it is also allowing people to make more crappy software every time programming has been made easier it enables people who are not very good programmers to make software you see you know php visual basic you know like this i've i've been a user of all every language has done this and every time the accusation is always like oh look at all this crap written in php and like yeah but php also like enabled a bunch of first of all enabled people who couldn't program before to program at all.

2:03:02Marco Arment:And people who were good programmers were perfectly able to write good software with PHP. AI is kind of that same type of step, but in a much greater step all at once. It is enabling a lot more people to write software. And in good hands, it is also an amazing tool to enable people who know what they're doing to make even better software than they've ever made before. so I wanted to tell this story in part because I'm just so blown away by it and also in part to maybe if you're a programmer out there and you're inspired have it tackle whatever bug is on your hit list that you haven't been able to find or reproduce just say hey go look at this part of the code I'm pretty sure something is able to cause this behavior and I can't figure it out give it a shot you'll be shocked what it can find

2:03:55John Siracusa:you know it's funny if it's funny you bring this up because over the last few days, especially when Fable was, I don't know if it was unlimited or if it was just more flexible usage, you know, or whatever. I forget exactly what the situation was, but I know that I am now paying for Fable in a way that I wasn't just a few days prior. But anyhow, I had John, you know, the little disembodied head of John Syracuse in front of me saying, you know, every time there's a new model, especially with Fable, you should just point it at your app or your repository and say, find bugs. And I did that. And for the handful of you that are on the call sheet test flight, I can assure you, and you've seen, there were a couple of builds that were just like, boom, boom, boom, boom, boom, boom, boom, boom, boom, boom, boom, boom, boom, boom.

2:04:48John Siracusa:Like all of these different, admittedly, fairly esoteric bugs. Like it's not the same win that you're talking about, although I have had those in the past. We've talked about it on the show. but a lot of little things that were like the edgiest of edge cases that I wouldn't in a million years have found. And were they really affecting that many people? I don't think they really were, but is it better that they're now affecting hypothetically no people? Yes. Yes, it is better. I prefer it that way. And I've been really pleased with how this has gone and I have run it in auto mode, but when it's done churning and whatever, I will go through line by line, like it's some fresh college graduate and make sure I understand everything it's done and what it's doing and why it's doing it.

2:05:35John Siracusa:And there are times that I will go back and forth with it and be like, I don't think that's right. Or why did it justify to me why you made choice A, B or C? And usually I end up realizing, oh no, no, I see what it did and it's right. But I want to go through that process for me anyway. I want to go through that process of understanding what it's done, because otherwise I feel like I now have some phantom coworker whose code I don't understand. And that's not good when you're a team of one. And so I like spending the time to go through and understand top to bottom, inside and out, what's being done to what I still consider to be my code.

2:06:11John Siracusa:But the point I'm really trying to make is I have fixed a plethora of bugs that I wasn't even aware of, much less things that, you know, have been latently bothering me for years. I think your story is more impressive and more interesting because this is a problem that certainly affects way more people than any of my silly bugs do. But nevertheless, it's still useful to get rid of these bugs, no matter how big or small.

2:06:36Casey Liss:Yeah, I've been doing this for a while, and I wanted to give you the same caution I gave you last time I was talking about my LLM bug fix releases, which is, you know, there's some error rate per line. How many defects per line of code does any individual create and humans are flawed so that that is the non-zero value it's also true of lms the more code they write the more defects they're going to add and me being the person who was always uh pitting my speaking cylinders against each other to see who's better well i've been doing the same thing with all the various coding models uh and in particular a fun thing to do is to you know you know as as gacy said i'm always saying find bugs find bugs like bugs and by the way on that topic i was trying google's anti-gravity recently uh their like coding agent thing and i said find bugs in my program and it was like nope can't do that that would be insecure i don't find oh interesting i i you know i figured what his message was was like i can't find i can't take an open-ended request to find bugs because it's a security problem i tried to phrase it 17 different ways and every time it would come back maybe it's some snarky thing it's like even if x y and z i still can't do it i'm like all right well you suck anyway uh and i and also i did have i did have fable dropped down to opus 4.8 at one point because i did something that pissed it off and it was like nope i can't do that yeah i had that too like it's all my own source code man give me a break but anyway since i have multiple models going against each other what i will often do is first of all i'll always ask them all to find bugs and see if they find the same bugs and they don't they don't find the same bugs which is not which is like a little bit disconcerting because like how many bugs are there that they can find entirely disjoint sets and why are they not finding each other like why it is this bug that once it's pointed out like margo said like greater than her last end flip once it's pointed out to you're like oh yeah that was dumb must have been a typo or i was just stupid fat fingered or just wasn't thinking right but why did the other model not find that and it's like i guess they're just not looking at the whole code base or whatever but anyway so they find different bugs and on occasion i will ask them you know i'll usually phrase it as propose a fix for this book because there's seven ways this can be fixed and i know which way i would go but you propose a fix to it and often they will propose something and i've been doing this thing where i would where i will have multiple checkouts and i will say okay model one you fix the bug model two you fix the bug and then i will show each other the the fixes made by the other models i'll actually do this across three models and i will say and i will be like here's a fix for this bug are there bugs in this fix for this bug and then they will find bugs in each other's fix for the bug and all that has told me that basically they're other than their ability to you know inexhaustively like search forever and ever and do things so much faster than a human they still create some number of defects per line of code they generate and i don't know if it's higher or lower than than any particular human would be but either way it's non-zero so keep that in mind when you're there messing with things when you find that bug that's like oh that's never going to happen like that's such an edge of an edge of an edge case but i feel better with it being fixed if the fix has lots of lines of code think twice about it because you may have introduced a much worse bug than the one that you were fixing and you're like well i'll solve that problem by having another model or even the same model look at the fix and find bugs in that it's not a um it doesn't uh converge as far as i can tell so far again with current technology this changes every freaking week we say this all the time like just because we're saying this now doesn't mean it's going to be true literally next week which is why we keep going back to this well and trying it again.

2:09:59Casey Liss:But right now, in my experience with the top end models, they do not converge on perfection, just like humans don't converge on perfection. But they do, like as we've all pointed out a million times, they are such an incredibly powerful tool for helping that you absolutely should use them if you can stomach it. And they do amazing things that were not possible before, but they do not yet converge on perfection, which is kind of a shame. And what's also a shame is that we don't use curse words on the show or in our titles most of the time. And so we can't call this episode F everything. We're doing 5X.

2:10:31Casey Liss:youngsters don't get the 2004 that this the onion article is uh they don't know about the gillette razor because 2004 i guess we're probably still watching ads on tv maybe people who didn't have tiwo so we knew about the gillette five-bladed razor or actually was it just three blades then and five blades came later anyway i guess to go to know your meme or something but we'll pull a link in the show it's to the original onion article yeah all those kids walking around on their Nirvana shirts. I've never seen it.

From the publisher

Sponsored by:

  • Leesa: A mattress for every body and budget

Become a member for ATP Overtime, ad-free episodes, member specials, and our early-release, unedited “bootleg” feed!

More from Accidental Tech Podcast

All 184 episodes
701: I Improved My TownAccidental Tech Podcast · 2 h 11 min
Listen in VO