ATP

00:00:00 ◼   ► I sound a little bit bad this week because I'm a little bit sick because everyone's a

00:00:04 ◼   ► little bit sick because oh my god everyone is sick.

00:00:06 ◼   ► Like this might be, you know, anecdotally speaking, the most people I've ever heard

00:00:11 ◼   ► of being sick at the same time.

00:00:14 ◼   ► Everyone is sick right now.

00:00:16 ◼   ► Every one.

00:00:17 ◼   ► It's not all, you know, most of it's not COVID fortunately.

00:00:19 ◼   ► A lot of it is just like, you know, mild colds here and there, occasional flu is here and

00:00:22 ◼   ► there, but just everyone is sick.

00:00:25 ◼   ► - Yeah.

00:00:26 ◼   ► Michaela has been fighting a cough. She's been COVID tested, you know, once or twice,

00:00:29 ◼   ► and it doesn't seem to be that. She woke up today complaining of an earache. And so fast forward

00:00:34 ◼   ► three hours, she's getting treated for an ear infection. You know, our kids were sick over

00:00:39 ◼   ► Thanksgiving. Again, not COVID as far as tests and as far as we know. Yeah, I agree with you that it

00:00:44 ◼   ► seems like everyone is making up for lost time, unlike general illnesses. It's not, like you said,

00:00:50 ◼   ► it's not necessarily COVID, although it's certainly plenty of that going around. But just

00:00:54 ◼   ► even general illnesses, there seems to be plenty of it at the moment.

00:00:57 ◼   ► - Yeah, and the good news is that it seems like the most common thing I hear about is

00:01:01 ◼   ► people who just have like mild cold symptoms for a week or two at the most, which is really

00:01:06 ◼   ► annoying. In fact, I'm currently on, I think, my second one of those for the season. It's

00:01:11 ◼   ► annoying, but, you know, it could be a lot worse. It has been a lot worse. So, you know,

00:01:16 ◼   ► it's good that it's all mild. So everyone is just kind of like mildly annoyed with sickness

00:01:21 ◼   ► for this whole winter, which is, you know, again,

00:01:24 ◼   ► could be worse, but that doesn't make it

00:01:25 ◼   ► any less mildly annoying.

00:01:27 ◼   ► - Yeah.

00:01:28 ◼   ► What are you gonna do?

00:01:29 ◼   ► By the way, I don't think it made it

00:01:31 ◼   ► into the release recording, maybe it did,

00:01:33 ◼   ► but as we were hanging up, so not as the,

00:01:36 ◼   ► I don't think it was when the live broadcast was ending,

00:01:38 ◼   ► but as the three of us were hanging up on Zoom,

00:01:41 ◼   ► Marco sniped me and wished John a happy birthday in advance,

00:01:45 ◼   ► and it had completely slipped my mind last recording,

00:01:48 ◼   ► but I will now take credit for this one

00:01:51 ◼   ► and say happy birthday John, I hope your birthday was wonderful and your carrot cake was delicious.

00:01:56 ◼   ► Thank you, we're still working on the carrot cake, still some left.

00:01:59 ◼   ► Oh, I thought you meant that it hadn't arrived yet, I was like what are you talking about?

00:02:03 ◼   ► I thought Tina and Kate made that for you every year, I misunderstood.

00:02:06 ◼   ► Those are the questions, should we make the full-size cake? Yeah, make the full-size cake.

00:02:10 ◼   ► Would you ever not make the full-size cake? I mean-

00:02:12 ◼   ► I mean for just four people it's a lot of cake, but I think we're up to the task.

00:02:16 ◼   ► I mean carrot cake, so this is one of those things, you know, I hate to admit that you're

00:02:20 ◼   ► carrot cake is awesome, like I love carrot cake.

00:02:23 ◼   ► I'm going to your pizza place tomorrow,

00:02:27 ◼   ► like it sucks when you're right this much.

00:02:29 ◼   ► - No, the carrot cake is a relatively new birthday thing.

00:02:32 ◼   ► I mean, I think Tina made it when we were at Underscores.

00:02:36 ◼   ► - Yeah, that's right.

00:02:38 ◼   ► - Right, so that's, I mean, that's not the first time

00:02:39 ◼   ► we had it, but that was when the tradition

00:02:41 ◼   ► was relatively new, so that's about the age of this,

00:02:43 ◼   ► like, I don't know, like 10 years, 12 years, 15 years,

00:02:46 ◼   ► I guess, that's relatively new when you're an old person.

00:02:48 ◼   ► (laughing)

00:02:49 ◼   ► It's not just carrot cake in general, it's that specific recipe.

00:02:52 ◼   ► I forget who it is, maybe it's Ina Gardner or whatever.

00:02:55 ◼   ► It's carrot cake and it's got things in it that people may find controversial.

00:02:58 ◼   ► One, it's got raisins in it.

00:02:59 ◼   ► I know a lot of carrot cakes have raisins in it.

00:03:01 ◼   ► I think they're good.

00:03:02 ◼   ► And two, it's got pieces of pineapple in it, which maybe you haven't seen in a carrot cake

00:03:06 ◼   ► before, but I think works really well.

00:03:08 ◼   ► And then three is just cream cheese frosting.

00:03:10 ◼   ► Carrot cake with cream cheese frosting.

00:03:11 ◼   ► It's a classic combo.

00:03:12 ◼   ► Whether you want the raisins or the pineapples, you can adjust that.

00:03:16 ◼   ► I feel like that combo is overlooked.

00:03:20 ◼   ► And this is literally the one time a year I have it.

00:03:22 ◼   ► I would get sick of it if we had it all the time,

00:03:24 ◼   ► but most of the time people are making cakes or cupcakes

00:03:25 ◼   ► or whatever.

00:03:26 ◼   ► They're making chocolate or vanilla cake or whatever.

00:03:28 ◼   ► It's just for my birthday, carrot cake.

00:03:31 ◼   ► That's-- the pineapples are really good,

00:03:32 ◼   ► and a really interesting idea.

00:03:33 ◼   ► The raisins, I could see why that would be controversial.

00:03:35 ◼   ► Pineapple, though?

00:03:36 ◼   ► That could be really nice.

00:03:38 ◼   ► It keeps the cake moist, I think.

00:03:39 ◼   ► I mean, the first time she made it,

00:03:40 ◼   ► I'm like, are these dried pineapples?

00:03:41 ◼   ► But they're not.

00:03:42 ◼   ► It's fresh cut pineapple, bits of fresh cut pineapple.

00:03:45 ◼   ► and they cook in the cake and the pineapple gets dry

00:03:48 ◼   ► but the cake gets moist so they even out.

00:03:50 ◼   ► So it's not like they're hard pieces of pineapple in there

00:03:54 ◼   ► but the cake is very moist as well, it's good.

00:03:57 ◼   ► - Okay, yeah, 'cause I've recently,

00:03:59 ◼   ► I think I've talked about this before,

00:04:00 ◼   ► I've recently honed the skill of choosing good pineapples

00:04:03 ◼   ► at the grocery store and so now I'm able to fairly reliably

00:04:08 ◼   ► buy pretty decent to great pineapples.

00:04:11 ◼   ► So of course I'm always looking for ways to use them

00:04:13 ◼   ► and that's a good one, I should try that.

00:04:15 ◼   ► - I think this is it.

00:04:16 ◼   ► I mean, I think we've passed around the recipe before.

00:04:18 ◼   ► I'll double check for the show notes,

00:04:20 ◼   ► but Ena Garten, not Ena Gartner,

00:04:22 ◼   ► sorry, get her name wrong all the time.

00:04:23 ◼   ► And we don't put pineapples on top like this shows.

00:04:26 ◼   ► - No, well that's a different thing.

00:04:27 ◼   ► That's an upside down cake kind of thing.

00:04:29 ◼   ► That's a whole different-- - No, no, look at the picture.

00:04:30 ◼   ► Like it's cream cheese frosting,

00:04:32 ◼   ► but then they put bits of pineapple on top and we don't--

00:04:33 ◼   ► - Oh, that's interesting.

00:04:35 ◼   ► - Yeah, that actually could just be like an indicator.

00:04:37 ◼   ► It's like this, warning, this cake contains pineapple

00:04:39 ◼   ► in case for some reason you have terrible taste

00:04:41 ◼   ► and don't want that.

00:04:43 ◼   ► - I am actually not a pineapple fan.

00:04:44 ◼   ► I wish I liked it, but it is not for me.

00:04:46 ◼   ► - Like I said, easy to leave the pineapple

00:04:48 ◼   ► and the raisins out of this.

00:04:50 ◼   ► Then it just becomes a more boring carrot cake.

00:04:52 ◼   ► But you know, plain old boring carrot cake

00:04:53 ◼   ► with cream cheese frosting.

00:04:55 ◼   ► It's one of the classics.

00:04:56 ◼   ► - So what was your cake preference

00:04:58 ◼   ► before this thought technology entered your world?

00:05:01 ◼   ► - I don't think I had one.

00:05:02 ◼   ► - No cake, no presents, no party.

00:05:04 ◼   ► - Yeah, that remains.

00:05:07 ◼   ► That remains.

00:05:07 ◼   ► - Fair enough.

00:05:10 ◼   ► Now I am a fan of like just boring,

00:05:12 ◼   ► boxed like chocolate cake with--

00:05:15 ◼   ► - You're a cheap date, Casey.

00:05:16 ◼   ► - Yes, I am.

00:05:17 ◼   ► I know everyone's surprised by this.

00:05:18 ◼   ► But just chocolate cake with vanilla icing on the top.

00:05:22 ◼   ► And the best thing, it's like lasagna,

00:05:26 ◼   ► after you have a little bit, you refrigerate it,

00:05:28 ◼   ► come back to it the next day,

00:05:29 ◼   ► and the icing is just a little bit crunchy.

00:05:32 ◼   ► - I don't think I've ever heard anyone say that

00:05:33 ◼   ► about boxed cake mix, but okay.

00:05:35 ◼   ► - I'm saying, it's true, try it out

00:05:38 ◼   ► after you try your Velveeta.

00:05:40 ◼   ► - I mean, I've had leftover box cake mix,

00:05:42 ◼   ► not like I just feel like it's--

00:05:43 ◼   ► - I think it's more about the icing

00:05:45 ◼   ► than it is the cake, if I'm honest.

00:05:46 ◼   ► The icing gets a little-- - Do you like it

00:05:47 ◼   ► to get hard and crusty? - Yeah, yeah, yeah, yeah.

00:05:49 ◼   ► (laughing)

00:05:50 ◼   ► Yes, oh yes.

00:05:51 ◼   ► - Have you ever tried combining your loves

00:05:52 ◼   ► and maybe putting a layer of Alveda Mac and Cheese

00:05:56 ◼   ► between the cake-- - No, no, stop.

00:05:57 ◼   ► - Sponge layers?

00:05:58 ◼   ► - Look, okay, we gotta move on.

00:06:01 ◼   ► - I'm sure it could be done.

00:06:02 ◼   ► (electronic beeping)

00:06:04 ◼   ► - All right, let's do some follow-up.

00:06:06 ◼   ► Amy Lee writes, "I loved the recent conversation

00:06:08 ◼   ► around the copyright implications of AI-generated art and there was some hypothetical discussion

00:06:11 ◼   ► of how you'd feel if you were an artist and your art was used to train these models. I

00:06:15 ◼   ► was surprised that something like GitHub Copilot wasn't in the conversation, as Copilot is

00:06:19 ◼   ► probably the closest coding equivalent we have to stable diffusion. If you have any

00:06:23 ◼   ► open source code on GitHub, it's likely been used by Copilot and in a way that doesn't

00:06:27 ◼   ► necessarily respect the license applied to the code. OpenAI used all the public code

00:06:31 ◼   ► they could find regardless of license and used it to train the model. This is quite

00:06:35 ◼   ► similar to their approach with training on art.

00:06:37 ◼   ► Hypothetical for this case would be, what if it was also allowed to train on closed

00:06:40 ◼   ► source code on GitHub?

00:06:42 ◼   ► How would you feel if your code was used in this model even though you never publicly

00:06:46 ◼   ► released it or gave it a license to others?

00:06:48 ◼   ► That's potentially just a contract away between GitHub and OpenAI, no fancy decompilation

00:06:52 ◼   ► needed.

00:06:53 ◼   ► This was, I think, a very, very interesting point.

00:06:55 ◼   ► I really was flummoxed once I read it.

00:06:59 ◼   ► I don't think I have any problem with my open source stuff being sucked into, like, you

00:07:04 ◼   ► Copilot or Stable Diffusion or what have you, but I would not feel good about

00:07:08 ◼   ► closed source stuff ending up there. Just on principle.

00:07:13 ◼   ► Like that's my private stuff. Yes, it exists on GitHub, but that's not

00:07:18 ◼   ► GitHub stuff. That's my stuff, and I would not feel good about that at all.

00:07:22 ◼   ► To me, that's the line. It's like, if you have private repos, those should

00:07:27 ◼   ► remain private to you in all ways, from not only exposure to the files

00:07:32 ◼   ► of you know for people to see but also training to me that's that's an obvious

00:07:36 ◼   ► line yeah I completely agree with you I think the weird thing about copilot and

00:07:40 ◼   ► code stuff well there's two weird things about it one is my impression based on

00:07:45 ◼   ► seeing many examples of copilot and trying myself is that the odds of it

00:07:49 ◼   ► producing a snippet of code verbatim from its source is much higher than with

00:07:54 ◼   ► any of the AI image stuff yeah that's a good point you know what I mean like

00:07:57 ◼   ► sometimes people I've seen people say hey this is like literally my paragraph

00:08:00 ◼   ► of code for doing this thing, right?

00:08:04 ◼   ► And I think that does make a difference because that's not a transformative any of this, literally

00:08:08 ◼   ► just you copy and paste it, right?

00:08:10 ◼   ► And on the open source thing, like, open source doesn't mean you can do whatever you want,

00:08:14 ◼   ► that's why open source licenses exist.

00:08:16 ◼   ► So I feel like open source licenses should address this case explicitly.

00:08:22 ◼   ► Again, arguably all existing open source licenses do address it in some way that has yet to

00:08:27 ◼   ► be tested in a court, right?

00:08:28 ◼   ► Because you can say, well, I can look at the existing rules

00:08:30 ◼   ► and decide whether I think this falls within them or not.

00:08:33 ◼   ► But it would be much simpler if you could just

00:08:35 ◼   ► explicitly put it in there.

00:08:36 ◼   ► And same thing with the copyrighted stuff.

00:08:38 ◼   ► I feel like this should be something that--

00:08:42 ◼   ► if you're training an AI model on code,

00:08:45 ◼   ► you should be using code that you're

00:08:46 ◼   ► pretty sure you're allowed to use for the purposes of training

00:08:49 ◼   ► an AI model.

00:08:49 ◼   ► And so obviously, a very, very open, open source license

00:08:52 ◼   ► may do that.

00:08:53 ◼   ► Because it may say, hey, you can use this code

00:08:54 ◼   ► for whatever you want.

00:08:55 ◼   ► You don't have to credit us.

00:08:56 ◼   ► You don't have to pay us.

00:08:57 ◼   ► You don't have to do anything like the broadest

00:08:58 ◼   ► of broadest open source licenses.

00:08:59 ◼   ► That's probably game right now.

00:09:01 ◼   ► But the more restrictive ones, you know,

00:09:03 ◼   ► that requires credit or if it's a copyleft thing

00:09:05 ◼   ► or whatever, using that to trade in AI models

00:09:08 ◼   ► is still questionable.

00:09:09 ◼   ► And the second thing is, unlike with image training things

00:09:12 ◼   ► or whatever, the result you get from Copilot

00:09:15 ◼   ► is not immediately useful, let's say.

00:09:17 ◼   ► We talked about this before.

00:09:18 ◼   ► You can't just take code generated by an AI thing

00:09:20 ◼   ► and says, I assume this works.

00:09:22 ◼   ► You have to actually figure out whether it does

00:09:24 ◼   ► what you want it to do.

00:09:26 ◼   ► It's a really important part of programming.

00:09:27 ◼   ► not just writing the code the first time,

00:09:29 ◼   ► whether you read it yourself or an AI thing writes it,

00:09:31 ◼   ► then you have to figure out if what you wrote

00:09:33 ◼   ► does the thing that you wanted it to do.

00:09:35 ◼   ► And that's the hard part of programming, right?

00:09:38 ◼   ► Arguably, it's harder when some other thing

00:09:39 ◼   ► wrote the code for you, especially if it's complicated

00:09:42 ◼   ► and you don't understand it, but either way, it's difficult.

00:09:45 ◼   ► And so it's not like people are going to co-pilot up a thing

00:09:48 ◼   ► and say, "And I'm done."

00:09:49 ◼   ► No, you're not even close to done.

00:09:51 ◼   ► You have to now say, "Okay, I co-pilot up

00:09:53 ◼   ► "10,000 lines of code."

00:09:54 ◼   ► Figure out if those 10,000 lines of code

00:09:56 ◼   ► come close to doing what you wanted them to do.

00:09:59 ◼   ► And that's, you know, at the micro level,

00:10:01 ◼   ► does this paragraph do what I want it to do?

00:10:02 ◼   ► And at the macro level,

00:10:03 ◼   ► does the program do what I want it to do?

00:10:05 ◼   ► That is incredibly difficult.

00:10:06 ◼   ► Whereas if an AI art thing comes out and you say,

00:10:08 ◼   ► I like it fine, it's done.

00:10:09 ◼   ► Like it's not functional in that way.

00:10:11 ◼   ► It's not like the people are gonna look at the image

00:10:14 ◼   ► and then their eyes are gonna explode, right?

00:10:15 ◼   ► But that's the equivalent of that could happen

00:10:17 ◼   ► if you have code that doesn't do what you want

00:10:18 ◼   ► and you run it and just absolutely doesn't do

00:10:21 ◼   ► what your program is or crashes or does something like that.

00:10:23 ◼   ► So it is interesting and there are similar issues,

00:10:27 ◼   ► but code is so fundamentally different than images

00:10:31 ◼   ► or something that tells another machine what to do

00:10:36 ◼   ► and that presumably has some kind of correctness.

00:10:40 ◼   ► I don't know, you can say whether it does

00:10:41 ◼   ► what I intended it or not, is a little bit different

00:10:44 ◼   ► than something that just someone has to look at

00:10:46 ◼   ► and say, I find that acceptable, therefore I'm done.

00:10:48 ◼   ► - Yeah, I agree with you that the open source license

00:10:52 ◼   ► should prevent this, or may prevent this sort of thing

00:10:55 ◼   ► from happening.

00:10:56 ◼   ► I'm just saying I'm personally not particularly bothered

00:10:58 ◼   ► if that was subverted or ignored or what have you

00:11:01 ◼   ► for open source stuff.

00:11:03 ◼   ► But yeah, I find it really gross for closed source stuff,

00:11:07 ◼   ► even leaving aside what you're saying, Jon,

00:11:08 ◼   ► that it's not exactly an apples to apples comparison.

00:11:11 ◼   ► But yeah, I thought this was a really interesting point

00:11:14 ◼   ► for me, nevertheless.

00:11:15 ◼   ► - The closed source thing of like,

00:11:17 ◼   ► have any of us read the user agreement for GitHub?

00:11:19 ◼   ► It could be already in the user agreement for GitHub

00:11:22 ◼   ► that we have allowed GitHub by agreeing to that thing

00:11:25 ◼   ► and putting our code on there,

00:11:26 ◼   ► that they're actually allowed to make a contract

00:11:28 ◼   ► with OpenAI and say,

00:11:29 ◼   ► "Hey, we're allowing them to use your close with them."

00:11:30 ◼   ► Of course, if they did that, they could say,

00:11:32 ◼   ► "Well, you agreed to these terms,

00:11:34 ◼   ► then we just take our code off GitHub," right?

00:11:35 ◼   ► I think that would be a bad move for them,

00:11:37 ◼   ► but I'm assuming none of us have actually ever read,

00:11:40 ◼   ► in full, the terms for GitHub.

00:11:42 ◼   ► So that's just centralization

00:11:45 ◼   ► and having large, powerful players in the market.

00:11:48 ◼   ► That's why that's bad.

00:11:49 ◼   ► And to be clear, I don't think GitHub is some kind of monopoly

00:11:52 ◼   ► on Git or whatever.

00:11:53 ◼   ► If GitHub did that, people would take their closed source stuff

00:11:56 ◼   ► off GitHub.

00:11:58 ◼   ► There is enough of a competitive market.

00:12:00 ◼   ► They are not the only place that can host Git stuff.

00:12:03 ◼   ► And Git being decentralized, you don't need any one single place

00:12:05 ◼   ► to host anything.

00:12:06 ◼   ► So we're pretty well protected technologically from this.

00:12:09 ◼   ► Hopefully, GitHub would be smart with this.

00:12:11 ◼   ► But the counterexample is, well, GitHub

00:12:12 ◼   ► wasn't particularly smart with the rollout of Copilot

00:12:15 ◼   ► in terms of what they used to train it

00:12:18 ◼   ► and compliance with the licenses.

00:12:19 ◼   ► I think when they first rolled it out,

00:12:21 ◼   ► they were just like, everything's fine, right?

00:12:22 ◼   ► And they're like, well, wait a second.

00:12:23 ◼   ► My open source license says X, Y, and Z,

00:12:25 ◼   ► and it definitely says you can't use it

00:12:27 ◼   ► for something like this.

00:12:28 ◼   ► Did you use my code?

00:12:29 ◼   ► It seems like you did.

00:12:30 ◼   ► And I think since then, maybe they've adjusted and said,

00:12:33 ◼   ► okay, we will honor the license set on your repo

00:12:36 ◼   ► and only use the ones that we think we're allowed to use

00:12:38 ◼   ► according to the open source license or whatever.

00:12:41 ◼   ► - Yeah, I should have been more explicit, by the way.

00:12:43 ◼   ► GitHub Copilot is a thing where you can start typing code

00:12:47 ◼   ► in one of the many supported languages.

00:12:50 ◼   ► And based on what you type,

00:12:51 ◼   ► it will try to figure out what it is you really want,

00:12:54 ◼   ► and then go ahead and fill out your code for you.

00:12:58 ◼   ► So for basic operations,

00:13:00 ◼   ► like a database like CRUD, for example,

00:13:02 ◼   ► it can get stunningly close from what I've understood.

00:13:05 ◼   ► I haven't really played with this myself,

00:13:06 ◼   ► but my understanding is it can get just bananas close

00:13:09 ◼   ► to what you would actually write

00:13:11 ◼   ► if you were to write this out by hand.

00:13:12 ◼   ► And for tedious, like busy work,

00:13:15 ◼   ► of which there's a fair bit in programming,

00:13:17 ◼   ► that's actually super duper cool, but it's certainly--

00:13:21 ◼   ► - It's not cool, but then you have to read a paragraph

00:13:24 ◼   ► of code you didn't write and say,

00:13:25 ◼   ► does this really reverse a string?

00:13:26 ◼   ► Like whatever trivial thing you have.

00:13:28 ◼   ► - That's the thing, it seems like a subtle bug factory.

00:13:31 ◼   ► Like are you actually saving time?

00:13:34 ◼   ► - I don't know, maybe not.

00:13:35 ◼   ► Again, I haven't really played with this.

00:13:37 ◼   ► - That's why I feel like the thing that people do

00:13:40 ◼   ► that is different than this,

00:13:42 ◼   ► like having a computer do it for you

00:13:44 ◼   ► when you're a programmer is not that much more helpful

00:13:47 ◼   ► than you going and looking up something, whatever it is.

00:13:50 ◼   ► Look it up in Stack Overflow and people say,

00:13:51 ◼   ► "Oh, I just caught," people just copy and paste

00:13:53 ◼   ► everything from Stack Overflow.

00:13:54 ◼   ► But Stack Overflow has someone explaining the code to you

00:13:57 ◼   ► and people may copy and paste it,

00:13:58 ◼   ► but a lot of time what they do is like rewrite it

00:14:01 ◼   ► by looking at Stack Overflow

00:14:02 ◼   ► and that's a process of understanding, hopefully.

00:14:04 ◼   ► Hopefully you're not just blindly like,

00:14:05 ◼   ► "Well, I don't understand this at all,

00:14:07 ◼   ► "but I'm just gonna copy and paste this paragraph of code

00:14:09 ◼   ► "from Stack Overflow."

00:14:11 ◼   ► But even that, copying, pasting code from Stack Overflow

00:14:14 ◼   ► is I feel like has a higher chance of success

00:14:16 ◼   ► if it's a highly rated answer on a well-trafficked question.

00:14:21 ◼   ► That code probably has a better chance of being correct

00:14:24 ◼   ► than the AI-generated equivalent,

00:14:26 ◼   ► unless the AI-generated equivalent is verbatim

00:14:28 ◼   ► from somebody else's code,

00:14:29 ◼   ► because there's not a lot of,

00:14:31 ◼   ► especially for trivial code, mixing and matching,

00:14:33 ◼   ► like what are they gonna do,

00:14:34 ◼   ► rename the variables or something,

00:14:35 ◼   ► like half of what people do

00:14:37 ◼   ► when they copy from Stack Overflow.

00:14:38 ◼   ► But if it takes any sort of deviance from,

00:14:41 ◼   ► like there's not a lot of room for riffing, right?

00:14:44 ◼   ► Like, it either has an off by one error in the loop

00:14:47 ◼   ► or it doesn't, right?

00:14:48 ◼   ► It either correctly reverses strings of all possible links

00:14:50 ◼   ► or it doesn't, right?

00:14:51 ◼   ► It handles the zero length case or it doesn't.

00:14:53 ◼   ► You know, it's kind of an open shot on the micro level.

00:14:56 ◼   ► On the macro level, things get hairier and, you know,

00:14:59 ◼   ► then it's not like you can ask it

00:15:00 ◼   ► to write an entire program for you.

00:15:02 ◼   ► But you kind of can.

00:15:03 ◼   ► You can say, "Space Invaders!"

00:15:04 ◼   ► And it's like, here's an entire Space Invaders game

00:15:06 ◼   ► in JavaScript.

00:15:06 ◼   ► It's not really Space Invaders.

00:15:08 ◼   ► It kind of crashes sometimes, but it mostly works.

00:15:11 ◼   ► So, yeah, it's, again, it's different with code.

00:15:14 ◼   ► because it's not just like, I look at the code

00:15:16 ◼   ► and I'm happy, you have to actually run it,

00:15:18 ◼   ► and it's in the running where you find out

00:15:19 ◼   ► does this thing actually work.

00:15:22 ◼   ► - Fair enough.

00:15:23 ◼   ► John, tell me about ChickBit.

00:15:25 ◼   ► - Yeah, we were talking about data,

00:15:27 ◼   ► a bit rot last episode, someone had a question about it,

00:15:31 ◼   ► and there are lots of programs that can do

00:15:33 ◼   ► one of the steps that we talked about,

00:15:34 ◼   ► which is detect whether there are other errors.

00:15:37 ◼   ► Some file systems do it for you,

00:15:38 ◼   ► some operating systems do it for you.

00:15:40 ◼   ► Sometimes you can use a program outside

00:15:42 ◼   ► of the operating system to do it for you,

00:15:43 ◼   ► Here's one I don't think I had heard of last time we discussed this so I thought I would

00:15:47 ◼   ► Mention it and link it disclaimer. I have not tried this program. I don't know if it actually does what it says it does, haha

00:15:54 ◼   ► But if you want to check it out, it's written in Python

00:15:57 ◼   ► It's called CHK bit which I'm assuming is short for check bit

00:16:00 ◼   ► And as the website says a lightweight tool to check the data integrity of your files

00:16:04 ◼   ► It allows you to verify the data has not changed since you put it there and that is still the same when you move it

00:16:09 ◼   ► Somewhere else it just basically makes dot directories and writes about to check something in it

00:16:12 ◼   ► This is actually kind of a fun project

00:16:13 ◼   ► for a beginning programmer to become

00:16:17 ◼   ► a little bit more seasoned, because you're like,

00:16:19 ◼   ► oh, this is easy.

00:16:20 ◼   ► I know how to read a file and get a checksum.

00:16:22 ◼   ► I know how to make a directory.

00:16:23 ◼   ► And I know how to read and write files and check that,

00:16:26 ◼   ► you know, it seems like, oh, I can do all those things.

00:16:28 ◼   ► I'm a beginning programmer.

00:16:29 ◼   ► I have all the tools I need to do this.

00:16:32 ◼   ► If you then go and try to implement a program like this,

00:16:35 ◼   ► even though you think you know how to do,

00:16:37 ◼   ► you can co-pilot up every single one

00:16:39 ◼   ► of those individual steps,

00:16:40 ◼   ► you will very, very quickly become a not so beginning programmer as you realize

00:16:44 ◼   ► all the weird things you didn't consider about real people's discs and what are

00:16:49 ◼   ► the actual performance bottlenecks to make this unfeasible and how much of a

00:16:53 ◼   ► pain is it and what kind of weird errors do you get from the operating systems

00:16:56 ◼   ► and what are sparse files and what are hard links and what are SIM links and

00:16:59 ◼   ► what are permissions and what is sandboxing and you will learn so especially

00:17:02 ◼   ► if you try to do this on a Mac, you will learn so much so fast probably that

00:17:07 ◼   ► you'd never wanted to know.

00:17:07 ◼   ► And if you don't find that exhilarating, you're probably don't want to be a

00:17:10 ◼   ► programmer but presumably the people who wrote checkbit have done a bunch of this

00:17:13 ◼   ► work for you so if you just want to do something to check some of your files

00:17:16 ◼   ► this is one of many many options it's just one I hadn't heard of before so I

00:17:20 ◼   ► thought it was worth linking you are clearly not a child of the 90s like I am

00:17:25 ◼   ► you think you know but you have no idea so it was that MTV diary something like

00:17:29 ◼   ► that I forget man it's true life I don't know some MTV show you still love

00:17:32 ◼   ► probably true life we are sponsored this week by green chef a CCOF certified

00:17:40 ◼   ► meal kit company. Green Chef makes eating well easy with plans to fit every

00:17:44 ◼   ► lifestyle. Whether you're keto, paleo, vegan, vegetarian, gluten-free, or just

00:17:49 ◼   ► looking to eat more balanced meals, Green Chef offers a range of recipes to suit

00:17:52 ◼   ► your preferences. And in fact, you know, we previously were sponsored by HelloFresh.

00:17:56 ◼   ► Green Chef is now owned by HelloFresh. And the wider variety of meal

00:18:00 ◼   ► plans to choose from, there's something for everyone. So if you're looking to

00:18:03 ◼   ► start a keto diet in particular in this new year, Green Chef's making it easier

00:18:07 ◼   ► than ever to keep keto with their brand new limited time keto kickoff 2023 with

00:18:12 ◼   ► keto recipes for breakfast lunch and dinner plus meal bundles at the green

00:18:15 ◼   ► market keep keto the easy and delicious way and of course they also just have

00:18:20 ◼   ► tons of other variety they've expanded their menu recently you can now choose

00:18:23 ◼   ► from 30 weekly recipes with the option to mix and match meals from different

00:18:27 ◼   ► dietary preferences all in the same box without changing your plans so if you

00:18:31 ◼   ► want to do vegan one day keto the next you can do that they're offering more

00:18:35 ◼   ► customization than ever before. You can also swap protein in any meal that

00:18:39 ◼   ► features chicken, beef, or salmon. For USDA certified organic ground beef, USDA

00:18:44 ◼   ► certified organic chicken, or wild-caught sockeye salmon. You can also now add

00:18:48 ◼   ► chicken or fish to select vegan and veggie recipes each week for an added

00:18:51 ◼   ► protein boost. There are just so many great options with Green Chef. They're

00:18:55 ◼   ► also now offering 10-minute lunches. Each week's menu includes two convenient low

00:18:59 ◼   ► prep and nutritious lunch recipes ready in just 10 minutes, no cooking required.

00:19:03 ◼   ► So it's perfect for when you are on the go or pressed for time at the office so you can eat well at lunchtime too.

00:19:09 ◼   ► So check it out at greenchef.com/ATP60 and use code ATP60 to get 60% off. That's a ton! Plus free shipping.

00:19:19 ◼   ► Once again, greenchef.com/ATP60. Use promo code ATP60 for 60% off plus free shipping.

00:19:27 ◼   ► shipping. Thank you so much to Green Chef, the number one meal kit for eating well, for

00:19:31 ◼   ► sponsoring our show.

00:19:34 ◼   ► Marco, on a recent episode of Under the Radar, which is on a very good run, which is not

00:19:42 ◼   ► surprising because they are excellent episodes. Usually, Marco, usually not longer than 30

00:19:49 ◼   ► minutes. But anyways, on Under the Radar, you were talking about a couple of things.

00:19:55 ◼   ► You were talking about how you're kind of pulling the plug on using CloudKit for Overcast,

00:20:01 ◼   ► but you are allegedly, theoretically, maybe, going to be keeping the web player for Overcast.

00:20:09 ◼   ► So I would like to quickly celebrate the Overcast web player being--surviving for another day,

00:20:14 ◼   ► which is great news, even though I don't use it that much anymore.

00:20:17 ◼   ► I do think it's cool that you're keeping it around.

00:20:20 ◼   ► But I'm very interested in this CloudKit stuff.

00:20:23 ◼   ► So I have been working on something new that the boys don't really know about yet because

00:20:26 ◼   ► it's way too early for that.

00:20:28 ◼   ► But this new thing I'm working on—

00:20:29 ◼   ► Is that us?

00:20:30 ◼   ► Are we the boys?

00:20:31 ◼   ► Yes, yes.

00:20:32 ◼   ► You are the boys.

00:20:33 ◼   ► This thing I'm working on, it is currently written against CloudKit, and I was using

00:20:41 ◼   ► CloudKit as a data store for what it does.

00:20:44 ◼   ► And hearing you have a less rosy opinion of CloudKit than I expected is very selfishly

00:20:50 ◼   ► interesting to me.

00:20:51 ◼   ► So can you give the 10 second version

00:20:55 ◼   ► of your 29 minute podcast, if you don't mind?

00:20:58 ◼   ► And then I'd love to dig in a little more

00:20:59 ◼   ► about how you came to this conclusion

00:21:02 ◼   ► and what you think the implications are

00:21:03 ◼   ► for not only people like me

00:21:05 ◼   ► that are trying to write something new,

00:21:07 ◼   ► but people like you that are maybe moving something old

00:21:09 ◼   ► into this new world.

00:21:11 ◼   ► So I don't know, take it whatever direction you want,

00:21:12 ◼   ► but I thought we could tug on that string a little bit.

00:21:14 ◼   ► - Yeah, sure.

00:21:15 ◼   ► So over the last, I'll be brief,

00:21:18 ◼   ► 'cause this is mostly covered in another radar,

00:21:20 ◼   ► so we're mainly doing it here for any additional elaboration

00:21:23 ◼   ► and for you guys to yell at me.

00:21:25 ◼   ► But, you know, mainly I was,

00:21:28 ◼   ► I was facing a couple problems with Overcast

00:21:30 ◼   ► over the past year.

00:21:32 ◼   ► One was that I'm getting increasingly burdened

00:21:36 ◼   ► by my massive amount of legacy code,

00:21:39 ◼   ► and I'm trying to move myself forward in some way

00:21:42 ◼   ► so that I can, you know, basically adopt new things,

00:21:47 ◼   ► new languages, new frameworks, new techniques,

00:21:49 ◼   ► you know, Swift, Swift UI, Swift async stuff

00:21:52 ◼   ► and all that kind of stuff.

00:21:54 ◼   ► So I'm looking at a bunch of rewriting basically.

00:21:58 ◼   ► Not everything has to be rewritten,

00:21:59 ◼   ► like the audio engine I'm gonna keep what it is,

00:22:02 ◼   ► which is mostly C, and that's gonna be fine.

00:22:06 ◼   ► But a lot of the UI I wanna rewrite,

00:22:09 ◼   ► a lot of the data layer I wanna modernize,

00:22:12 ◼   ► in part to make it easier to change the UI

00:22:15 ◼   ► and to keep it up to date and to keep it fresh

00:22:17 ◼   ► and to add features and in part to fix certain,

00:22:20 ◼   ► you know, just certain like shortcomings

00:22:22 ◼   ► and reliability problems,

00:22:24 ◼   ► certain like weird obscure crash conditions

00:22:26 ◼   ► that could happen.

00:22:27 ◼   ► You know, some of those require

00:22:29 ◼   ► just re-architecting certain things.

00:22:31 ◼   ► So anyway, that's the motivating factor.

00:22:33 ◼   ► And the second big problem I was having

00:22:34 ◼   ► besides my massive legacy code bloat

00:22:37 ◼   ► is that my server load was really,

00:22:39 ◼   ► my server maintenance load,

00:22:41 ◼   ► not like the actual load average on a particular server,

00:22:43 ◼   ► but like the amount of my time

00:22:46 ◼   ► it was taking to maintain and deal with my servers

00:22:50 ◼   ► and their workload was getting to be a little too much.

00:22:53 ◼   ► And so over the year, over the past year,

00:22:55 ◼   ► I really have done a lot of optimization,

00:22:59 ◼   ► a lot of server experiments,

00:23:01 ◼   ► moving to certain techniques or services

00:23:04 ◼   ► that are supposed to try to improve things

00:23:06 ◼   ► and some of them did, most of them didn't.

00:23:09 ◼   ► I've had to deal with a lot of regressions,

00:23:12 ◼   ► a lot of bugs.

00:23:16 ◼   ► We talked on the show a few times about how

00:23:18 ◼   ► I would experiment with doing something like with S3

00:23:20 ◼   ► or with Cloudflare as a CDN in front of various things

00:23:24 ◼   ► and almost all of those have actually failed.

00:23:29 ◼   ► Almost all of those ideas I had either didn't make it better

00:23:33 ◼   ► or only made it better if I burned a ridiculous amount

00:23:37 ◼   ► of money way more than I could actually justify

00:23:40 ◼   ► or and even then it wasn't that much better.

00:23:44 ◼   ► So in many cases, they introduced a bunch of weird bugs.

00:23:47 ◼   ► I learned a lot.

00:23:50 ◼   ► One of the things I learned is that

00:23:52 ◼   ► all of those web services,

00:23:53 ◼   ► when you, for instance, write data to them,

00:23:57 ◼   ► and you try to read it back, sometimes it's there.

00:24:00 ◼   ► Sometimes it's not.

00:24:01 ◼   ► Sometimes it says it's updated and it actually hasn't.

00:24:04 ◼   ► It's sometimes the write just fails.

00:24:07 ◼   ► Why? Who knows?

00:24:08 ◼   ► Sometimes Cloudflare just caches things when it shouldn't.

00:24:10 ◼   ► Why? Who knows?

00:24:13 ◼   ► There's a lot of like, this works 98% of the time.

00:24:17 ◼   ► Well that's great, but when I write things to MySQL,

00:24:20 ◼   ► it works 100% of the time.

00:24:22 ◼   ► And then when I read them back,

00:24:23 ◼   ► that works 100% of the time too.

00:24:25 ◼   ► And so many of these optimizations

00:24:29 ◼   ► would introduce weird subtle bugs and problems

00:24:32 ◼   ► that eventually made them not worth it.

00:24:34 ◼   ► And so anyway, so I've just gone through all this crap

00:24:36 ◼   ► with my servers and so I devised this plan in the fall.

00:24:40 ◼   ► You know, and I've been thinking about this for a while,

00:24:43 ◼   ► you know, I should probably just move

00:24:45 ◼   ► as much user data as possible to CloudKit,

00:24:48 ◼   ► because my servers process two kinds of data.

00:24:52 ◼   ► You know, one is like global data.

00:24:53 ◼   ► It's like, well, this is the list of feeds I know about,

00:24:56 ◼   ► of RSS feeds I know about for different podcasts,

00:24:59 ◼   ► and here's the episodes in those feeds,

00:25:01 ◼   ► and you know, and so on,

00:25:03 ◼   ► and then the user-specific part of the data is,

00:25:07 ◼   ► all right, I have these users,

00:25:08 ◼   ► user number 123 is subscribed to feeds A, B, and C,

00:25:13 ◼   ► and in feed B, they've listened to episodes

00:25:16 ◼   ► one, two, and three, to progress 60 seconds completion

00:25:20 ◼   ► and two seconds, you know.

00:25:21 ◼   ► That's what I'm dealing with here, a lot of that data.

00:25:24 ◼   ► Huge volumes of very simple data.

00:25:27 ◼   ► And I know from running my previous web services,

00:25:30 ◼   ► I know the amount of server resources I was spending

00:25:34 ◼   ► per user was not a very good ratio.

00:25:38 ◼   ► I was spending too much per user.

00:25:41 ◼   ► That the number of active users I have

00:25:43 ◼   ► does not justify the amount of server resources

00:25:47 ◼   ► that I was needing to keep the system going.

00:25:50 ◼   ► And the amount of time it was taking me to manage it.

00:25:54 ◼   ► So anyway, so again, I had this idea,

00:25:57 ◼   ► let me go to CloudKit.

00:25:59 ◼   ► I'll move user data there, and I'll keep the public data,

00:26:01 ◼   ► like all the feed crawling and everything,

00:26:02 ◼   ► I'll keep that on my servers,

00:26:04 ◼   ► but I'll move user data to CloudKit,

00:26:06 ◼   ► thereby not only being a huge win in terms of like

00:26:10 ◼   ► getting possibly privacy sensitive data out of my hands,

00:26:14 ◼   ► which is always a great feature there.

00:26:16 ◼   ► It's like nuclear waste, like I don't want this.

00:26:18 ◼   ► Like I don't want people's data.

00:26:20 ◼   ► And then secondarily, that would presumably result

00:26:23 ◼   ► in a huge reduction in the server costs

00:26:27 ◼   ► and server complexity and server maintenance needs

00:26:29 ◼   ► that I'd have to deal with.

00:26:31 ◼   ► Over the course of the fall,

00:26:33 ◼   ► I decided in one of my fits of rage

00:26:36 ◼   ► in undoing one of my very bad ideas

00:26:40 ◼   ► for one of the various CDN hosted S3 bucket,

00:26:44 ◼   ► one of those various things,

00:26:46 ◼   ► I'm like, you know what, let me just buy some time

00:26:49 ◼   ► with the good old fashioned MySQL setup

00:26:51 ◼   ► that's just serving stuff directly.

00:26:53 ◼   ► Let me buy some time and see if I can optimize this

00:26:56 ◼   ► a little bit more.

00:26:58 ◼   ► And I took like a week,

00:27:00 ◼   ► And the one big table that was getting hit the most often

00:27:04 ◼   ► had like three indexes on it.

00:27:06 ◼   ► It's a huge table, hundreds of gigs.

00:27:08 ◼   ► It had like three indexes on it

00:27:10 ◼   ► and I ended up consolidating it down to one.

00:27:13 ◼   ► Both by a little bit combining two that were already there

00:27:16 ◼   ► and then rewriting the code to just not use the other one.

00:27:19 ◼   ► So a combination of basic database stuff

00:27:22 ◼   ► and of code changes made it so I could only use one index

00:27:25 ◼   ► if I had to.

00:27:27 ◼   ► And that single change made a way bigger gain

00:27:32 ◼   ► than any of the stupid stuff I did

00:27:34 ◼   ► for the whole rest of the year.

00:27:35 ◼   ► It was infuriating.

00:27:37 ◼   ► - That is both beautiful and awful all at once.

00:27:40 ◼   ► - Yeah, so I did a couple of other things,

00:27:43 ◼   ► like during that week I optimized a couple other things

00:27:46 ◼   ► like the feed crawling,

00:27:47 ◼   ► it was hitting one of the big tables twice

00:27:50 ◼   ► and I rewrote the code a little bit

00:27:52 ◼   ► so it only had to hit it once, stuff like that.

00:27:54 ◼   ► I did a couple of optimizations like that,

00:27:57 ◼   ► and it just made a huge difference.

00:27:59 ◼   ► And I was like, damn it.

00:28:01 ◼   ► So basically during that time,

00:28:05 ◼   ► oh, sorry, going back a second.

00:28:07 ◼   ► Before I had done that, when I was deciding,

00:28:09 ◼   ► okay, my future is probably gonna be CloudKit,

00:28:12 ◼   ► I put two things on my to-do list.

00:28:14 ◼   ► Number one, I'm like, all right,

00:28:15 ◼   ► I have to issue an update to the app

00:28:16 ◼   ► that will measure the percentage of people

00:28:20 ◼   ► who have CloudKit enabled.

00:28:24 ◼   ► Because I use CloudKit so far, I have used CloudKit

00:28:27 ◼   ► for a number of years now to do something very,

00:28:29 ◼   ► very, very basic, which is if you are logged out,

00:28:33 ◼   ► so if you restored a phone or you got a new phone

00:28:36 ◼   ► or whatever, if you're logged out of Overcast,

00:28:38 ◼   ► it shows you a list of your accounts

00:28:40 ◼   ► and you can tap it to just log in.

00:28:42 ◼   ► And that's just a list of basically login tokens

00:28:47 ◼   ► that are stored in CloudKit.

00:28:49 ◼   ► So I knew that CloudKit was available

00:28:52 ◼   ► on most of my customers' devices

00:28:53 ◼   ► because the system really has not resulted

00:28:55 ◼   ► in any big problems for me,

00:28:56 ◼   ► having this be tied to iCloud.

00:28:58 ◼   ► And then I also use CloudKit on the web

00:29:01 ◼   ► to help people log into the website

00:29:03 ◼   ► who have anonymous accounts

00:29:04 ◼   ► that don't have emails and passwords, which is the default.

00:29:07 ◼   ► And Apple does have a CloudKit.js web interface,

00:29:11 ◼   ► and it is horrific.

00:29:13 ◼   ► It is really hard to use. - I was gonna add.

00:29:16 ◼   ► - It is old and seemingly unmaintained.

00:29:20 ◼   ► It has a very high error rate,

00:29:22 ◼   ► and just the URLs you're hitting just error out a lot

00:29:25 ◼   ► and you just have nothing, you can't do anything about it.

00:29:27 ◼   ► The documentation is horrendous, using it is awkward.

00:29:29 ◼   ► - No, the documentation stinks?

00:29:31 ◼   ► - Yeah, but yeah, using it is incredibly cumbersome

00:29:35 ◼   ► and awkward, it was clearly designed by nobody

00:29:38 ◼   ► who has ever made a web anything ever.

00:29:41 ◼   ► It is designed only for JavaScript use

00:29:43 ◼   ► and it's definitely not designed for any kind of like

00:29:44 ◼   ► tying in with the backend, like it was just,

00:29:46 ◼   ► it's really extraordinarily cumbersome and bad.

00:29:51 ◼   ► So I already knew that if I was going in the direction

00:29:54 ◼   ► of CloudKit, the website, it was not something

00:29:57 ◼   ► I wanted to really offer, like with the web playback.

00:30:01 ◼   ► And I was thinking, I guess I'll still have

00:30:04 ◼   ► the directory on the web where you can search

00:30:05 ◼   ► and browse stuff, 'cause if nothing else,

00:30:08 ◼   ► I use it when I'm debugging feed problems.

00:30:10 ◼   ► I'll search for a podcast, browse this episode list

00:30:13 ◼   ► and see all that, so I was like,

00:30:14 ◼   ► I might as well keep the website.

00:30:15 ◼   ► But I was gonna lose all user functionality on the website

00:30:19 ◼   ► if I went with CloudKit.

00:30:20 ◼   ► Because I'm like, I'm not gonna have access

00:30:23 ◼   ► to what are your subscriptions,

00:30:25 ◼   ► what is your, what's your progress in this episode,

00:30:29 ◼   ► stuff like that.

00:30:29 ◼   ► That was all gonna be in CloudKit

00:30:31 ◼   ► and not on my servers anymore.

00:30:32 ◼   ► So this move that I was planning on doing to CloudKit

00:30:35 ◼   ► was going to kill the website for any kind of logged in use.

00:30:39 ◼   ► And so anyway, so I decided, you know what,

00:30:44 ◼   ► my servers are just killing me.

00:30:46 ◼   ► I'm just gonna finally rip the bandaid off

00:30:49 ◼   ► and just tell people, you know what,

00:30:52 ◼   ► the website's gonna have to go

00:30:53 ◼   ► because I wanna make this move.

00:30:55 ◼   ► And so I did that in, I think, November,

00:30:59 ◼   ► something like that.

00:31:00 ◼   ► I posted a thing to the website,

00:31:02 ◼   ► a little banner on top saying, sorry,

00:31:04 ◼   ► sometime in 2023, this website's gonna be discontinued,

00:31:07 ◼   ► mostly with the user functionality and stuff,

00:31:09 ◼   ► because I wanna leave all my options open

00:31:11 ◼   ► with my server stuff going forward.

00:31:14 ◼   ► And also because hardly anybody uses it.

00:31:16 ◼   ► So to give you some idea of what I mean

00:31:20 ◼   ► by hardly anybody uses it,

00:31:22 ◼   ► I'm looking at a few hundred people who use the website.

00:31:26 ◼   ► And that is not a large portion of the user base.

00:31:30 ◼   ► And this is like per day.

00:31:32 ◼   ► But it's under a thousand people.

00:31:34 ◼   ► And that's well under 1% or whatever.

00:31:39 ◼   ► It's a very small portion of the user base.

00:31:40 ◼   ► And so I'm like, why am I jumping over backwards here

00:31:44 ◼   ► for people, for under a thousand people?

00:31:47 ◼   ► I've made decisions to cut off support for an old OS

00:31:51 ◼   ► that still had thousands of people using it.

00:31:53 ◼   ► But the trade-offs outweigh that or whatever,

00:31:55 ◼   ► and it was still a small enough percentage

00:31:58 ◼   ► that I was comfortable with that or whatever.

00:31:59 ◼   ► So I'm like, this is okay.

00:32:02 ◼   ► I can lose the web player, it's not that big of a deal.

00:32:04 ◼   ► Most people won't care.

00:32:06 ◼   ► Boy was I wrong.

00:32:08 ◼   ► (laughing)

00:32:11 ◼   ► I heard from so many people.

00:32:15 ◼   ► And I think part of my problem with this decision,

00:32:19 ◼   ► part of what I think surprised me,

00:32:21 ◼   ► is that my analytics are not super granular.

00:32:24 ◼   ► This is just like a home built analytics system

00:32:26 ◼   ► that I report a couple of parameters to my server

00:32:30 ◼   ► that's all anonymized with every sync request

00:32:32 ◼   ► and I count 'em up and you know.

00:32:34 ◼   ► So I can tell things like how many distinct users

00:32:38 ◼   ► used feature X yesterday.

00:32:40 ◼   ► I can't tell you how many distinct users

00:32:44 ◼   ► used feature X in the past month.

00:32:46 ◼   ► So if you don't use something every day,

00:32:50 ◼   ► you're not counted every day.

00:32:52 ◼   ► Does that make sense?

00:32:53 ◼   ► So like if a different 200 people use this feature

00:32:57 ◼   ► on Monday, Tuesday, Wednesday, Thursday, and Friday,

00:32:58 ◼   ► if those are actually five different sets of 200 people,

00:33:01 ◼   ► that's a thousand people.

00:33:03 ◼   ► But my analytics will only show 200

00:33:05 ◼   ► 'cause it's only measuring it like per day.

00:33:07 ◼   ► - What kind of analytics do you have

00:33:08 ◼   ► where you can tell per day uniqueness

00:33:10 ◼   ► but not cross day uniqueness?

00:33:11 ◼   ► Is this like a one day cookie expiration or something?

00:33:14 ◼   ► - Sort of.

00:33:15 ◼   ► I've wrote the system years ago

00:33:17 ◼   ► so I forget the exact details.

00:33:18 ◼   ► But just as it is, there's two tables.

00:33:20 ◼   ► There's analytics today and analytics archive.

00:33:24 ◼   ► And analytics today is just some random numbers

00:33:27 ◼   ► and everything.

00:33:28 ◼   ► And then analytics archive is like the sum

00:33:32 ◼   ► of how many of those people were for each metric each day.

00:33:35 ◼   ► So I'm only really keeping the sum.

00:33:37 ◼   ► Also, the random number that the app uses

00:33:40 ◼   ► to associate people per day does rotate pretty quickly.

00:33:44 ◼   ► I forget what it is.

00:33:45 ◼   ► It's something like a random expiration

00:33:47 ◼   ► between 48 hours and 72 hours, something like that.

00:33:51 ◼   ► So that number, the random ID people are using

00:33:54 ◼   ► to report to me, it changes so often.

00:33:56 ◼   ► Anything more than a day granularity

00:33:58 ◼   ► wouldn't be useful anyway.

00:33:59 ◼   ► But I've never needed it really.

00:34:01 ◼   ► For the most part, I'm not measuring things

00:34:05 ◼   ► that are so critical that I would need

00:34:06 ◼   ► that kind of advanced measurement.

00:34:08 ◼   ► - So everyone is upset that you're killing the website.

00:34:13 ◼   ► - Yeah, and so I'm starting to get the feeling

00:34:16 ◼   ► that again, it's way more people

00:34:19 ◼   ► than the daily analytics suggest

00:34:20 ◼   ► because what people are telling me is,

00:34:23 ◼   ► I'm hearing the same stories over and over again basically.

00:34:25 ◼   ► It's like, oh I use this occasionally

00:34:27 ◼   ► when looking something up.

00:34:28 ◼   ► Or I use this on certain days when I'm at work,

00:34:32 ◼   ► on my Windows work PC.

00:34:34 ◼   ► And there were so many stories like that.

00:34:39 ◼   ► A lot of the use cases were people who were using

00:34:41 ◼   ► the file upload feature, which is part of Overcast Premium.

00:34:44 ◼   ► And I said, I'll find some way to offer file upload still

00:34:48 ◼   ► if I cancel the web player.

00:34:49 ◼   ► So that was less of a concern, but it was still,

00:34:52 ◼   ► it was a common enough thing, and I noticed

00:34:54 ◼   ► that occasionally I would use the features too.

00:34:57 ◼   ► Like if I hit a link to a website that had a podcast episode

00:35:02 ◼   ► and I would follow the overcast link,

00:35:03 ◼   ► and then I would add it to my account

00:35:05 ◼   ► with the web interface.

00:35:06 ◼   ► I'm like, "Damn it, I want people to do this

00:35:07 ◼   ► "after I make this change." (laughs)

00:35:09 ◼   ► And I started realizing,

00:35:11 ◼   ► there's a lot more uses for this thing.

00:35:13 ◼   ► It's kind of like the Mac Mini,

00:35:15 ◼   ► which is like, most people don't need a Mac Mini.

00:35:18 ◼   ► However, there are certain use cases

00:35:21 ◼   ► that only the Mac Mini can cover.

00:35:23 ◼   ► And it's a large number of very small use cases, effectively.

00:35:28 ◼   ► And that's how my web player was.

00:35:30 ◼   ► it covers a large number of very small use cases.

00:35:33 ◼   ► A lot of people end up using those every once in a while.

00:35:37 ◼   ► That was problem number one.

00:35:39 ◼   ► Problem number two happened when I shipped

00:35:42 ◼   ► the version of Overcast that captured the analytics

00:35:45 ◼   ► of how many people have logged in iCloud accounts.

00:35:47 ◼   ► And I thought that the number of people

00:35:51 ◼   ► who did not have CloudKit available,

00:35:54 ◼   ► which I thought meant they didn't have an iCloud account

00:35:56 ◼   ► on their device, would be very, very small.

00:36:00 ◼   ► I was wrong about two things.

00:36:02 ◼   ► Number one, not being able to use CloudKit

00:36:05 ◼   ► is not just caused by not having an iCloud account

00:36:10 ◼   ► signed into your device.

00:36:11 ◼   ► It is also caused by having iCloud Drive disabled.

00:36:16 ◼   ► That's a very different thing.

00:36:18 ◼   ► - That's so weird to me that the two things

00:36:22 ◼   ► are codependent, for lack of a better word.

00:36:24 ◼   ► I would not have expected that.

00:36:25 ◼   ► - Well, that's where CloudKit stars all the stuff, though,

00:36:27 ◼   ► right? - Yes, that's the thing.

00:36:29 ◼   ► So when people say iCloud, there's a bunch of different

00:36:33 ◼   ► services that actually is part of iCloud.

00:36:35 ◼   ► One that I've used before, which I think I might move

00:36:38 ◼   ► the account system back to, that little login token list

00:36:41 ◼   ► thing, is called Key Value Store.

00:36:44 ◼   ► And this is one of the very first things,

00:36:45 ◼   ► like when iCloud first launched, it was like two or three

00:36:48 ◼   ► services, this was one of them.

00:36:49 ◼   ► It's been there since the beginning.

00:36:51 ◼   ► iCloud Key Value Store is literally a key value store

00:36:53 ◼   ► for very small bits of data, it's limited,

00:36:55 ◼   ► it's very limited.

00:36:57 ◼   ► I forget the exact limits, but it's something like

00:36:58 ◼   ► you can have like a thousand keys and values stored

00:37:01 ◼   ► and each one can be up to like four kilobytes or something.

00:37:03 ◼   ► So you can store decent data there

00:37:07 ◼   ► but it's more for like preferences

00:37:09 ◼   ► and you know short bits of things.

00:37:11 ◼   ► It's not really for arbitrary user data

00:37:13 ◼   ► because the limits are so small,

00:37:16 ◼   ► I wouldn't be able to use KeyValueStore

00:37:18 ◼   ► because like way too many people

00:37:20 ◼   ► would bump into those limits

00:37:21 ◼   ► if you know if you had a lot of subscriptions or whatever.

00:37:24 ◼   ► But that service as far as I can tell

00:37:26 ◼   ► doesn't really count against anyone's data at all.

00:37:29 ◼   ► Like it doesn't count against your iCloud storage

00:37:32 ◼   ► in either in any way or in any meaningful way.

00:37:35 ◼   ► That data is also not user browsable at all.

00:37:39 ◼   ► Like there is no place I can go as an iCloud customer

00:37:42 ◼   ► and see the key value data for my apps for iCloud

00:37:46 ◼   ► as far as I know.

00:37:46 ◼   ► Maybe, I mean, you know, maybe who knows?

00:37:48 ◼   ► Maybe on the Mac it's in some like library

00:37:50 ◼   ► hidden folder somewhere.

00:37:51 ◼   ► But I don't know where that is.

00:37:52 ◼   ► And it's for the most part it's hidden from users.

00:37:55 ◼   ► So users can't go in and delete it or manage it really.

00:37:58 ◼   ► So that data's kept very small.

00:38:00 ◼   ► And as far as I know, that's not affected

00:38:01 ◼   ► by the iCloud Drive preference,

00:38:03 ◼   ► but I don't know that for sure yet.

00:38:04 ◼   ► Anyway, iCloud Drive, the amount of storage

00:38:08 ◼   ► that you're allowed to use is just whatever the user has

00:38:10 ◼   ► in their iCloud account.

00:38:11 ◼   ► So whatever amount of storage they have for free or pay for,

00:38:14 ◼   ► that's what you can use.

00:38:15 ◼   ► So it's, as far as the app's concerned,

00:38:18 ◼   ► kind of unlimited in the sense that you don't have

00:38:21 ◼   ► these arbitrary limits of four kilobytes or whatever else.

00:38:24 ◼   ► So that does make sense to have controllable

00:38:28 ◼   ► in some way by the user, to have visible to the user.

00:38:31 ◼   ► So the way Apple has chosen to do that

00:38:33 ◼   ► is to tie it into iCloud Drive.

00:38:34 ◼   ► That's where that data is stored.

00:38:37 ◼   ► So if you have iCloud Drive disabled,

00:38:39 ◼   ► apps, just as far as apps can tell,

00:38:42 ◼   ► you can't use CloudKit.

00:38:43 ◼   ► And it doesn't tell you necessarily

00:38:46 ◼   ► why you can't use CloudKit.

00:38:48 ◼   ► I think, I could be wrong about this,

00:38:51 ◼   ► I think it returns the no account status,

00:38:56 ◼   ► whether you have no account for real,

00:38:58 ◼   ► or whether you are logged in but iCloud Drive is disabled.

00:39:00 ◼   ► So you can't, there's not an easy message

00:39:02 ◼   ► you could show the user necessarily

00:39:04 ◼   ► that would be very accurate or whatever else.

00:39:06 ◼   ► So anyway, so I was shocked to find out,

00:39:10 ◼   ► not only that, but the percentage of not signed in

00:39:15 ◼   ► to iCloud or iCloud Drive disabled,

00:39:18 ◼   ► The percentage of my users who can't use CloudKit

00:39:23 ◼   ► is around 12%.

00:39:27 ◼   ► And it keeps going up, I keep following this metric.

00:39:30 ◼   ► Now granted, most of my data is from the end of December

00:39:34 ◼   ► and the first day or two of January,

00:39:36 ◼   ► so it is holiday break time for a lot of people.

00:39:40 ◼   ► A lot of people aren't at work during this time

00:39:42 ◼   ► where they take vacations or whatever else,

00:39:44 ◼   ► so it's not an extremely representative time.

00:39:46 ◼   ► Also just my overall usage is down by something like 20%

00:39:49 ◼   ► or something like that for during this time,

00:39:52 ◼   ► just because fewer people listen to podcasts

00:39:53 ◼   ► when they're not in their work routine

00:39:55 ◼   ► and they're like, you know, commuting and everything else.

00:39:57 ◼   ► So this data is not super strong yet,

00:39:59 ◼   ► but it's been staying, since this update shipped

00:40:01 ◼   ► a couple of weeks ago, it's been staying between 10 and 12%

00:40:06 ◼   ► of people who cannot use CloudKit in my user base.

00:40:09 ◼   ► This is way higher than I thought.

00:40:13 ◼   ► I was thinking it would be like 1% maybe.

00:40:16 ◼   ► Nope, it's 10 times that.

00:40:18 ◼   ► So it's way higher than I thought.

00:40:22 ◼   ► And if I can't use CloudKit for these customers,

00:40:25 ◼   ► it's different from if I move the required OS version

00:40:31 ◼   ► forward, I still have, like all those people

00:40:35 ◼   ► who I'm leaving behind still can use the old version.

00:40:38 ◼   ► Like it's fine.

00:40:40 ◼   ► But if I move all my user storage and sync to CloudKit,

00:40:45 ◼   ► I'm gonna cut off 12% of my existing users,

00:40:49 ◼   ► like not even just new users, existing users.

00:40:52 ◼   ► That's a problem.

00:40:53 ◼   ► Now, many of you out there might be thinking,

00:40:56 ◼   ► well who cares, why don't you just not offer sync

00:40:58 ◼   ► to people who don't have iCloud enabled or whatever.

00:41:02 ◼   ► And that's okay, that's a reasonable theory.

00:41:05 ◼   ► The problem is on an iOS device, in practice,

00:41:09 ◼   ► sync means backup.

00:41:11 ◼   ► If you don't offer sync,

00:41:13 ◼   ► then when somebody loses their phone

00:41:16 ◼   ► or it falls off of Casey's car or it--

00:41:19 ◼   ► - Oh, that was my watch, not my phone.

00:41:21 ◼   ► (laughing)

00:41:23 ◼   ► - Or they get a new phone and they don't quite hit

00:41:26 ◼   ► the restore in the correct order

00:41:28 ◼   ► and they have to reset everything

00:41:29 ◼   ► or they lose their synced key chain or whatever,

00:41:32 ◼   ► they just lose all their data.

00:41:35 ◼   ► That sucks.

00:41:37 ◼   ► And I know how people feel about that

00:41:39 ◼   ► from the handful of requests or the handful of support,

00:41:41 ◼   ► emails and stuff I do get about that problem

00:41:43 ◼   ► when my system doesn't work for somebody

00:41:46 ◼   ► with the whole CloudKit account sync thing.

00:41:49 ◼   ► People expect their data to be there.

00:41:51 ◼   ► They expect that no matter what happens,

00:41:54 ◼   ► when they get a new phone or they replace

00:41:55 ◼   ► their run over phone, they expect that everything

00:42:00 ◼   ► will just be there.

00:42:01 ◼   ► You can't just tell them, well hey look,

00:42:03 ◼   ► hey if you don't enable your iCloud account,

00:42:05 ◼   ► you know, this is it, no sync, no backup.

00:42:08 ◼   ► That's not really an acceptable outcome

00:42:10 ◼   ► for modern expectations of software.

00:42:13 ◼   ► that's just not an option.

00:42:14 ◼   ► So you have to have some kind of account

00:42:16 ◼   ► or something to back stuff up,

00:42:17 ◼   ► and ideally it's automatic.

00:42:19 ◼   ► And so anyway, so that's a no-go.

00:42:21 ◼   ► I can't just say, well, sync is optional now.

00:42:24 ◼   ► Like that doesn't fly in this day and age.

00:42:27 ◼   ► So I have to offer this.

00:42:29 ◼   ► And when I first, I first thought

00:42:32 ◼   ► that the reason people wouldn't have an iCloud account

00:42:34 ◼   ► would be really obscure situations like,

00:42:39 ◼   ► I don't even know what.

00:42:40 ◼   ► Like you're using some kind of weird test device,

00:42:42 ◼   ► 'cause it's actually somewhat challenging

00:42:44 ◼   ► to use an iOS device without ever signing in

00:42:47 ◼   ► an iCloud account.

00:42:48 ◼   ► Lots of stuff doesn't really work very well without that.

00:42:50 ◼   ► But I learned when I was, I guess, not tweeting about this,

00:42:54 ◼   ► when I was mastadoning about this,

00:42:57 ◼   ► I learned, I'm not gonna say tooting.

00:42:59 ◼   ► That's a different thing.

00:43:01 ◼   ► - You should say tooting.

00:43:02 ◼   ► Embrace the tooting.

00:43:03 ◼   ► - No, when I was masta-posting about this,

00:43:06 ◼   ► that sounds gross too.

00:43:07 ◼   ► - No, it's just making it worse.

00:43:09 ◼   ► Tude is great because it's so close to tweet.

00:43:11 ◼   ► When I was posting on Mastodon about this,

00:43:13 ◼   ► I learned that it's actually fairly common

00:43:19 ◼   ► for corporate issued and corporate locked down devices,

00:43:23 ◼   ► either to not have an iCloud account at all

00:43:26 ◼   ► and to disallow you from logging into one,

00:43:28 ◼   ► or apparently a lot of corporate IT departments

00:43:30 ◼   ► for whatever compliance or security reasons

00:43:33 ◼   ► or whatever it is, won't allow third party

00:43:36 ◼   ► cloud storage solutions.

00:43:37 ◼   ► So they will possibly allow your iCloud account

00:43:40 ◼   ► to be signed in, but they will specifically

00:43:41 ◼   ► disallow iCloud Drive from being enabled.

00:43:44 ◼   ► So that, which kills the whole thing for CloudKit.

00:43:47 ◼   ► So I think what I'm seeing here is way more people

00:43:52 ◼   ► than I expected use Overcast on a corporate managed iPhone.

00:43:57 ◼   ► That makes sense, people listen to podcasts

00:44:00 ◼   ► as part of their workday sometimes,

00:44:01 ◼   ► like they'll have it on in the background

00:44:02 ◼   ► or if you bring that phone to work,

00:44:05 ◼   ► you'll use that on your commute or whatever else,

00:44:08 ◼   ► like there's all sorts of conditions.

00:44:09 ◼   ► So what I've learned here is CloudKit is really a no-go

00:44:14 ◼   ► for things that might be run on a decent number

00:44:17 ◼   ► of corporate issued phones.

00:44:18 ◼   ► And so I decided, like right at the end of December,

00:44:23 ◼   ► I decided, you know what, now that I have this data,

00:44:26 ◼   ► this is not a good idea anymore.

00:44:27 ◼   ► And since I had in the meantime optimized my servers

00:44:31 ◼   ► a little bit better and made those gains,

00:44:33 ◼   ► I'm like, actually, is continuing to run my servers

00:44:37 ◼   ► the way they are really that big of a burden

00:44:39 ◼   ► is it really worth this loss of this many people?

00:44:43 ◼   ► And I decided, you know what, no it's not.

00:44:46 ◼   ► It's not worth, like it isn't worth this loss.

00:44:48 ◼   ► It's not, this whole plan I had to go to CloudKit

00:44:51 ◼   ► and everything, this is not gonna work for me.

00:44:54 ◼   ► And I'm not saying nobody can use CloudKit,

00:44:57 ◼   ► but I'm saying it's, I thought it was gonna be a good fit

00:44:59 ◼   ► for this particular app and for my particular priorities

00:45:02 ◼   ► and needs and I got more information and it's not.

00:45:06 ◼   ► So the website will stay and the Mac Mini remains a product in my lineup and

00:45:11 ◼   ► I'm going to, you know,

00:45:13 ◼   ► keep running my servers and I'll try to keep making it easier on myself.

00:45:16 ◼   ► In the meantime,

00:45:18 ◼   ► I'll get to go do anything else and not spend a year rewriting my whole sync

00:45:21 ◼   ► engine.

00:45:22 ◼   ► How swayed were you,

00:45:23 ◼   ► assuming this was not the case and it really was a fraction of a percent who

00:45:26 ◼   ► couldn't use it,

00:45:27 ◼   ► would you have been swayed by the horror stories from the people who say,

00:45:31 ◼   ► Oh, if you're going to use CloudKit,

00:45:32 ◼   ► you have to have something that's not CloudKit to serve as your backup for in

00:45:35 ◼   ► CloudKit craps out?

00:45:38 ◼   ► - Sync for a podcast app, I just finished telling you

00:45:40 ◼   ► why it's required.

00:45:42 ◼   ► But it doesn't have to be fast,

00:45:44 ◼   ► and it doesn't have to be super aggressive or super recent

00:45:47 ◼   ► because most of my customers are only using it

00:45:50 ◼   ► on one device.

00:45:52 ◼   ► So if CloudKit was having a bad day,

00:45:55 ◼   ► and you were still able to use the podcast app,

00:45:58 ◼   ► and you'd still, in my imagined architecture

00:46:00 ◼   ► with my server still existing to do feed crawling and stuff,

00:46:03 ◼   ► in that architecture, you would still even be getting

00:46:05 ◼   ► new episodes, like that wouldn't stop if CloudKit was down.

00:46:10 ◼   ► All that would stop would be your data wouldn't sync

00:46:12 ◼   ► between different devices of yours.

00:46:14 ◼   ► And so I decided in that kind of scenario,

00:46:17 ◼   ► okay, it's a risk I'm willing to take.

00:46:19 ◼   ► Like if CloudKit's down for a few hours

00:46:21 ◼   ► and your progress, your overcast progress

00:46:24 ◼   ► between your phone and your iPad and your watch

00:46:27 ◼   ► doesn't sync for a few hours,

00:46:28 ◼   ► that's not the end of the world.

00:46:30 ◼   ► - What about the horror story of it just getting wedged?

00:46:32 ◼   ► Like whether it's up or down for some period of time,

00:46:35 ◼   ► at a certain point it just gets wedged

00:46:37 ◼   ► and will just never sink again without some heroic measure.

00:46:40 ◼   ► - That was a little more concerning, for sure.

00:46:42 ◼   ► And frankly, I just hadn't gotten there yet.

00:46:46 ◼   ► I mean, if I had gone down this path,

00:46:49 ◼   ► chances are I would find different downsides to it.

00:46:52 ◼   ► Downsides exist with every solution.

00:46:55 ◼   ► Everyone's like, oh well, why don't you just,

00:46:57 ◼   ► my favorite phrase, why don't you just move to,

00:47:01 ◼   ► insert Facebook or Google or Amazon hosted solution here.

00:47:05 ◼   ► Like, oh, just move to XYZ cloud server parsley.

00:47:08 ◼   ► And it's like, I don't, like you don't understand,

00:47:11 ◼   ► this is not my first day on the internet.

00:47:14 ◼   ► Like this is not the first time I've, you know,

00:47:17 ◼   ► used web services or created web services.

00:47:21 ◼   ► Like every one of these solutions has pitfalls

00:47:25 ◼   ► and limits and downsides.

00:47:27 ◼   ► If you don't know what they are yet,

00:47:28 ◼   ► you haven't used them enough.

00:47:30 ◼   ► Trust me, everything has downsides.

00:47:33 ◼   ► Nothing is hands off, nothing is automatic,

00:47:36 ◼   ► nothing is taking all the problems out of your hands.

00:47:40 ◼   ► Every service has a bad day, every service has downsides,

00:47:43 ◼   ► every service has BS you have to deal with

00:47:45 ◼   ► as the person running it, or using it I guess.

00:47:48 ◼   ► At least with my current service,

00:47:50 ◼   ► of running a whole bunch of Linode servers

00:47:52 ◼   ► with MySQL and PHP and stuff,

00:47:54 ◼   ► I know where all the pitfalls are.

00:47:57 ◼   ► I know how that breaks.

00:47:59 ◼   ► I know when I need to intervene.

00:48:01 ◼   ► I know the BS I have to deal with.

00:48:03 ◼   ► Any other service I would move to,

00:48:05 ◼   ► I knew that I would be signing up for the devil I don't know.

00:48:09 ◼   ► I'd be signing up for some kind of future BS

00:48:13 ◼   ► and some kind of future sleepless nights

00:48:15 ◼   ► and headaches and stress that I didn't yet know

00:48:18 ◼   ► what they were.

00:48:19 ◼   ► It's like Dick Cheney's Unknown Unknowns,

00:48:22 ◼   ► whatever that was, or was it a from himself, who cares?

00:48:24 ◼   ► Anyway, I knew there would be some kind of pitfalls

00:48:27 ◼   ► with any of these other services.

00:48:29 ◼   ► and CloudKit is of course no exception.

00:48:31 ◼   ► Of course CloudKit has weirdness and problems

00:48:34 ◼   ► and shortcomings and downsides and downtime and errors.

00:48:37 ◼   ► Everything does.

00:48:39 ◼   ► I'm not scared away by the possibility

00:48:41 ◼   ► of a service having them.

00:48:43 ◼   ► I'm scared away by me not knowing them

00:48:46 ◼   ► and knowing that if I move to something like this,

00:48:49 ◼   ► there will be some amount of learning curve there,

00:48:53 ◼   ► possibly things that I would do very wrong as I learned.

00:48:56 ◼   ► It'd be one thing if I was starting a whole new app

00:48:58 ◼   ► with no users to start with, but when I'm moving an app

00:49:02 ◼   ► from, you know, when I'm thinking about moving an existing

00:49:05 ◼   ► app with a whole bunch of existing users,

00:49:08 ◼   ► that is currently making up, you know, my income

00:49:12 ◼   ► as a person, like, I don't wanna move that willy nilly

00:49:15 ◼   ► to just anything, you know, I wanna do that pretty carefully

00:49:20 ◼   ► and that's quite an operation to move something like that.

00:49:22 ◼   ► And so, I was entertaining CloudKit because I thought

00:49:26 ◼   ► would have significant benefits to me.

00:49:29 ◼   ► Like I thought that the benefits would outweigh

00:49:31 ◼   ► the possible downsides, and I got more information,

00:49:34 ◼   ► and they didn't.

00:49:36 ◼   ► - It makes sense.

00:49:37 ◼   ► It's obviously a big bummer for you

00:49:40 ◼   ► that you're not getting out of the server business

00:49:42 ◼   ► no matter what happens, but I don't know.

00:49:45 ◼   ► It's, I thought that a lot of Apple's stuff

00:49:49 ◼   ► is using CloudKit now.

00:49:50 ◼   ► Like I thought shared, not shared photo library,

00:49:53 ◼   ► I thought the iCloud Photos was on CloudKit, so.

00:49:56 ◼   ► - It is, and notes and reminder, I don't know if I remember,

00:49:59 ◼   ► I think all of Apple's recent stuff is backed by CloudKit.

00:50:02 ◼   ► And what I heard actually, when I was browsing

00:50:07 ◼   ► my post-a-dons, what I heard was that people

00:50:11 ◼   ► who have these devices, who run devices for work

00:50:15 ◼   ► without iCloud Drive for whatever reason,

00:50:18 ◼   ► a lot of them replied to me and a lot of them said,

00:50:20 ◼   ► yeah, a bunch of apps break in random, annoying ways.

00:50:23 ◼   ► Basically, like a whole bunch of stuff

00:50:25 ◼   ► doesn't work quite right.

00:50:26 ◼   ► A whole bunch of third party apps don't sync

00:50:28 ◼   ► and just don't even tell you why.

00:50:30 ◼   ► A whole bunch of Apple's apps just don't work right.

00:50:33 ◼   ► You are correct, a lot of stuff uses CloudKit.

00:50:35 ◼   ► And when you have one of these devices,

00:50:37 ◼   ► a lot of that stuff just breaks.

00:50:38 ◼   ► - I just, I find that hard,

00:50:40 ◼   ► I'm not trying to say you're wrong,

00:50:41 ◼   ► I'm not trying to imply you're wrong,

00:50:42 ◼   ► I'm not, none of those things.

00:50:44 ◼   ► But I just find that hard to believe, right?

00:50:45 ◼   ► Like, you know what I mean?

00:50:46 ◼   ► Like, it's just, it's stunning to me

00:50:49 ◼   ► that so many of your users apparently don't get to use

00:50:53 ◼   ► like half the junk on their phone.

00:50:56 ◼   ► Obviously I'm being hyperbolic,

00:50:57 ◼   ► but it's just bananas to me

00:51:00 ◼   ► that that's just a no-fly zone for so many people.

00:51:03 ◼   ► And I understand the other corporate thing.

00:51:04 ◼   ► I am not far enough removed from real work

00:51:06 ◼   ► that I don't understand it

00:51:09 ◼   ► or it doesn't make sense or what have you,

00:51:10 ◼   ► but golly, that is bananas.

00:51:13 ◼   ► And I don't know, I guess there's a couple of different

00:51:17 ◼   ► things that I'm thinking about completely selfishly

00:51:19 ◼   ► as I'm wondering if I can create a new thing

00:51:21 ◼   ► that rides on CloudKit.

00:51:22 ◼   ► And one of the things that I was--

00:51:24 ◼   ► the thing I was most worried about, in fact,

00:51:25 ◼   ► was that, oh, CloudKit would be unreliable,

00:51:28 ◼   ► or the API would be crummy, or something like that.

00:51:31 ◼   ► The API seems reasonably OK so far.

00:51:33 ◼   ► I have gotten far enough in the pre-alpha version of this app

00:51:37 ◼   ► that I can save stuff to CloudKit

00:51:40 ◼   ► and pull it back out and so on.

00:51:41 ◼   ► So the API seems fine.

00:51:43 ◼   ► In my singular use, it seems reliable.

00:51:49 ◼   ► But I never thought that, oh, if I release this app,

00:51:52 ◼   ► I'm shutting off 10% of iPhone users,

00:51:54 ◼   ► which granted, I mean, I can choose not to care about that,

00:51:58 ◼   ► which is kind of what you were saying earlier.

00:51:59 ◼   ► You don't get to make that choice.

00:52:00 ◼   ► I can make the choice that I can just choose not to care,

00:52:03 ◼   ► but ugh, that's not what I expected,

00:52:06 ◼   ► and that's a real bummer.

00:52:07 ◼   ► - Well, and to be clear,

00:52:08 ◼   ► this might not be 10% of all users.

00:52:11 ◼   ► This might just be 10% of my users.

00:52:13 ◼   ► - That's fair.

00:52:14 ◼   ► - My user base is not representative of the entire world,

00:52:17 ◼   ► and part of my user base is very nerdy, of course.

00:52:20 ◼   ► That's part of it.

00:52:21 ◼   ► It's a podcast app at all, it's also my podcast app.

00:52:24 ◼   ► So it's a very nerdy leaning for sure.

00:52:27 ◼   ► And also, it depends what the app is used for.

00:52:30 ◼   ► Again, in my case, I think my theory here is like,

00:52:33 ◼   ► in my case, a lot of people use podcast apps at work.

00:52:37 ◼   ► And therefore, my percentage of work-issued device usage

00:52:41 ◼   ► is probably higher than the average.

00:52:43 ◼   ► So if your top secret app that you haven't told us about yet

00:52:46 ◼   ► not that I'm bitter, is, if your top secret app

00:52:51 ◼   ► is less likely to be used by people on their work devices,

00:52:54 ◼   ► you might have a very different number.

00:52:56 ◼   ► - Yeah, yeah, I'm not so sure that's the case, unfortunately.

00:52:59 ◼   ► Well, fortunately and unfortunately.

00:53:01 ◼   ► Ah, that's something else.

00:53:04 ◼   ► So you said you've gotten over the immediate hurdle

00:53:08 ◼   ► of the feeling like your database is going to explode.

00:53:12 ◼   ► I'm not saying that's actually what was happening,

00:53:13 ◼   ► but it seems like you kinda felt like,

00:53:15 ◼   ► oh, whoa, things are getting a little dodgy over here.

00:53:18 ◼   ► So you got through at least the first hurdle

00:53:22 ◼   ► in front of you with some changes on your end,

00:53:24 ◼   ► some changes on the database end.

00:53:26 ◼   ► So now, to the best that you're willing to share,

00:53:29 ◼   ► what is the long-term play then?

00:53:30 ◼   ► Because these servers aren't gonna go away.

00:53:32 ◼   ► This problem is gonna continue to be a problem.

00:53:34 ◼   ► And God willing, it's gonna get worse

00:53:36 ◼   ► 'cause you have more users who are doing more with your app.

00:53:39 ◼   ► So what's the plan, man?

00:53:43 ◼   ► - Good question.

00:53:44 ◼   ► So one of the concerns,

00:53:47 ◼   ► one of the things that was frustrating me

00:53:48 ◼   ► was that I was so close to my capacities before.

00:53:52 ◼   ► And it's not, you know, look,

00:53:54 ◼   ► there's lots of things I could do to add capacity.

00:53:59 ◼   ► One of them that I, what I relied on in the past,

00:54:02 ◼   ► most of the time, with my SQL-based things,

00:54:05 ◼   ► and you know, I'm mostly database-bound,

00:54:08 ◼   ► you know, it's your standard kind of, you know,

00:54:10 ◼   ► lamp design of an app, it's, you know,

00:54:12 ◼   ► a monolithic app with a whole bunch of web server,

00:54:15 ◼   ► not even, you know, eight web servers

00:54:16 ◼   ► and a handful of database servers.

00:54:18 ◼   ► And the way I had scaled it in the past

00:54:20 ◼   ► was what everyone does at this kind of

00:54:22 ◼   ► low to moderate scale database based app.

00:54:25 ◼   ► You split some high traffic tables onto their own servers.

00:54:28 ◼   ► Eventually you have replication using MySQL replication

00:54:31 ◼   ► and you have the primary secondary

00:54:33 ◼   ► and then you try to send read queries to the secondary

00:54:35 ◼   ► and write to the primary.

00:54:37 ◼   ► I've done all those things.

00:54:39 ◼   ► You can get pretty far with that.

00:54:42 ◼   ► Obviously you couldn't run a very large scale operation

00:54:44 ◼   ► with this kind of architecture,

00:54:45 ◼   ► but you can get pretty far with it.

00:54:48 ◼   ► And so that's what I've been doing.

00:54:50 ◼   ► I have found MySQL replication to be significantly worse

00:54:54 ◼   ► since MySQL 8 and I'm looking to move,

00:54:58 ◼   ► to reduce my reliance on it to only like a backup

00:55:01 ◼   ► or kind of hot standby kind of situation,

00:55:04 ◼   ► not actually sending any queries to the replicas.

00:55:07 ◼   ► And that's where I am now.

00:55:08 ◼   ► Right now I have everything going to the primary,

00:55:10 ◼   ► no queries are being sent to the replicas,

00:55:11 ◼   ► they're only being used as basically live backups

00:55:14 ◼   ► so I could failover to them.

00:55:16 ◼   ► And I use replication, well no I guess,

00:55:18 ◼   ► no I use hotbackup to actually copy them for backups,

00:55:20 ◼   ► so that's something else.

00:55:21 ◼   ► Anyway, so replication is only for, at this point,

00:55:24 ◼   ► only for like failover scenarios.

00:55:26 ◼   ► So one thing I could do, rather than using replication

00:55:29 ◼   ► to build stuff up later, is I could just split up

00:55:31 ◼   ► the user data, like what we used to call sharding,

00:55:33 ◼   ► I don't know if that still means that today.

00:55:36 ◼   ► 'Cause the other thing is like,

00:55:38 ◼   ► my knowledge of server backend stuff is super outdated.

00:55:42 ◼   ► I know the way things were built in 2006

00:55:45 ◼   ► and I'm really frickin' good at it.

00:55:48 ◼   ► And anything that's come out since then,

00:55:51 ◼   ► chances are I don't know about it.

00:55:53 ◼   ► Or if I do know about it, I'd rather not use it.

00:55:56 ◼   ► Because here's the thing, the stuff in 2006

00:55:59 ◼   ► solves my needs really well the vast majority of the time

00:56:03 ◼   ► and I know how to do it and I know how not to do it.

00:56:07 ◼   ► Like I know what mistakes not to make

00:56:11 ◼   ► with this kind of setup.

00:56:12 ◼   ► I know how to avoid getting woken up

00:56:15 ◼   ► in the middle of most nights.

00:56:16 ◼   ► I know how to set this up so I can actually like

00:56:18 ◼   ► have a family life and take vacations

00:56:21 ◼   ► and have work-life balance.

00:56:23 ◼   ► Like I know how to run this.

00:56:26 ◼   ► So that's why I stick with it.

00:56:27 ◼   ► And it's served me just fine.

00:56:29 ◼   ► Anyway, one thing I could do is shard user data.

00:56:31 ◼   ► This is very, very simple.

00:56:32 ◼   ► My user data does not depend on other user data.

00:56:35 ◼   ► So I could say, you know, all right,

00:56:37 ◼   ► let's take the modulus of the user ID

00:56:40 ◼   ► and just make eight different databases,

00:56:42 ◼   ► and if your ID, you know, module eight equals two,

00:56:46 ◼   ► you go to the server, whatever,

00:56:47 ◼   ► and split up the user data across eight different servers,

00:56:51 ◼   ► and then all of a sudden I have eight times the capacity.

00:56:54 ◼   ► I could do something like that.

00:56:55 ◼   ► And I'm fortunate, fortunate,

00:56:57 ◼   ► in the sense that my business

00:56:58 ◼   ► is not really growing very quickly,

00:56:59 ◼   ► so I can predict what I'm going to need

00:57:02 ◼   ► with pretty reasonable certainty.

00:57:04 ◼   ► Like, I'm not gonna need eight times

00:57:07 ◼   ► as much capacity next month.

00:57:09 ◼   ► You know, things are not that,

00:57:10 ◼   ► I don't have hockey stick growth on my podcast app.

00:57:12 ◼   ► It's fine.

00:57:14 ◼   ► So, you know, so, you know,

00:57:16 ◼   ► I have lots of options that I can do,

00:57:18 ◼   ► but one of the reasons why I was so stressed all year

00:57:22 ◼   ► is that I did feel like I was like really at that threshold

00:57:26 ◼   ► of like I was, I felt like I was at like 90% capacity

00:57:29 ◼   ► all the time, and you know, things were not going well.

00:57:33 ◼   ► Sometimes, you know, things would fall behind,

00:57:35 ◼   ► you know, cues would fall behind or whatever.

00:57:37 ◼   ► It was very high stress, certain things would go down,

00:57:39 ◼   ► it was a whole thing.

00:57:40 ◼   ► Well, I wanted to do things that required storing

00:57:43 ◼   ► more database data, like for instance,

00:57:47 ◼   ► the thing I now have tracking data,

00:57:49 ◼   ► but I haven't actually shown it anywhere yet,

00:57:51 ◼   ► which is the how much hour time you spend

00:57:55 ◼   ► listening to each podcast per month or whatever.

00:57:58 ◼   ► People want that kind of Spotify unwrapped feature,

00:58:01 ◼   ► where you tell them at the end of the year,

00:58:02 ◼   ► I hear this every single December,

00:58:04 ◼   ► where you tell people at the end of the year,

00:58:06 ◼   ► hey, this year you spent 400 hours listening to ATP.

00:58:11 ◼   ► Like that's, people want that kind of stuff.

00:58:13 ◼   ► It's a very commonly requested feature.

00:58:15 ◼   ► Not only the end of the year wrap ups,

00:58:16 ◼   ► but people just ask for stats much more often.

00:58:19 ◼   ► And most of my competitors at this point provide stats.

00:58:22 ◼   ► So it's a very, very, very common user request.

00:58:25 ◼   ► And so I wanted to offer that.

00:58:27 ◼   ► So I needed to track more data than I,

00:58:29 ◼   ► before this, all I tracked per episode was like,

00:58:33 ◼   ► all right, you're at this position.

00:58:35 ◼   ► or you've completed it or you've deleted it.

00:58:37 ◼   ► The smart speed total time saved is just a double.

00:58:42 ◼   ► It's just a number.

00:58:44 ◼   ► Like the app just accumulates how much time you're saving

00:58:46 ◼   ► and periodically writes it to the database

00:58:48 ◼   ► and that number is synced.

00:58:50 ◼   ► That's it.

00:58:51 ◼   ► Like that's not tracking every single listening session

00:58:54 ◼   ► you do and adding it up.

00:58:55 ◼   ► It's just tallying up on a number.

00:58:58 ◼   ► So in order for me to add any kind of stats,

00:59:01 ◼   ► well now I have more data to store.

00:59:03 ◼   ► And the idea of writing to my database with more data,

00:59:07 ◼   ► I'm like, I just can't do that feature.

00:59:08 ◼   ► Like, I'm running at 90% capacity here.

00:59:11 ◼   ► I'm having trouble keeping up sometimes.

00:59:13 ◼   ► I can't add load to my database.

00:59:16 ◼   ► And well, I could set up a whole new database cluster

00:59:18 ◼   ► just to do that, but that's a very expensive operation.

00:59:22 ◼   ► 'Cause it's gonna be a lot of data.

00:59:24 ◼   ► How big of a server am I gonna need for that?

00:59:26 ◼   ► And then I have to have a replica for that,

00:59:28 ◼   ► and I have to add it to the database backup system,

00:59:29 ◼   ► and it's a whole thing.

00:59:31 ◼   ► So I would just block, I would just say,

00:59:33 ◼   ► you know what, I can't do that right now.

00:59:36 ◼   ► And so ever since I did those optimizations,

00:59:39 ◼   ► I thought, you know what, I bought myself some headroom,

00:59:42 ◼   ► now I can actually see, let me launch this feature

00:59:45 ◼   ► and I'll just launch it in the background

00:59:48 ◼   ► and if it turns out to be way too heavy on the database,

00:59:52 ◼   ► I'll just stop writing the data.

00:59:54 ◼   ► Like I'll just edit the PHP file,

00:59:56 ◼   ► that method of the controller,

00:59:58 ◼   ► and I'll just stop writing it to the database, who cares?

01:00:00 ◼   ► then I just won't launch the feature publicly,

01:00:03 ◼   ► but I can test, I can launch the app

01:00:05 ◼   ► that starts writing the data

01:00:06 ◼   ► and starts submitting it to the servers

01:00:08 ◼   ► and I can see can I actually handle it or not.

01:00:11 ◼   ► So I did that and yeah,

01:00:13 ◼   ► it turns out it handles it just fine now.

01:00:15 ◼   ► It turns out that's really small data.

01:00:17 ◼   ► (laughs)

01:00:18 ◼   ► And so I added enough headroom.

01:00:21 ◼   ► So again, I'm able to do all this stuff now

01:00:25 ◼   ► which feels really good because I have just spent

01:00:29 ◼   ► So much of the past few years with Overcast,

01:00:31 ◼   ► I have spent wishing I could do something,

01:00:34 ◼   ► wishing I could add some feature

01:00:36 ◼   ► or change the way something works,

01:00:38 ◼   ► but then the server side of it was prohibitive in some way.

01:00:43 ◼   ► Either it actually did cause too much load

01:00:47 ◼   ► or I expected it to cause so much more load

01:00:50 ◼   ► that I didn't even try it.

01:00:51 ◼   ► And so now I bought myself a lot of headroom,

01:00:55 ◼   ► I'm learning that some of the ideas I have

01:00:57 ◼   ► actually are cheaper on the database and stuff

01:01:00 ◼   ► than what I thought.

01:01:02 ◼   ► And at some point, my actual plan,

01:01:04 ◼   ► to answer your actual question here,

01:01:05 ◼   ► my plan for the future is,

01:01:07 ◼   ► I guess I'll just keep doing what I've been doing.

01:01:09 ◼   ► And if I reach the point where I can't just

01:01:13 ◼   ► double the instance size on Linode

01:01:14 ◼   ► to make my database have more capacity or whatever,

01:01:17 ◼   ► then I'll split it up.

01:01:18 ◼   ► Then I'll start sharding, and I'll start

01:01:20 ◼   ► splitting more tables into their own servers,

01:01:22 ◼   ► and do more user sharding and stuff like that.

01:01:25 ◼   ► That's all stuff I can do.

01:01:26 ◼   ► It's all a trade-off of cost and complexity

01:01:29 ◼   ► and management time and code adjustment time.

01:01:34 ◼   ► But it's all stuff I can actually do.

01:01:36 ◼   ► And so at the end of the day,

01:01:38 ◼   ► it's probably way less time and complexity

01:01:42 ◼   ► than rewriting my entire sync engine

01:01:45 ◼   ► to use a whole different system.

01:01:47 ◼   ► So this is probably the best way forward.

01:01:50 ◼   ► - Now you can do the fun stuff,

01:01:51 ◼   ► which is using SwiftUI to make a cool animated

01:01:53 ◼   ► year-end wrap-up screen based on the data that you have,

01:01:55 ◼   ► because you have to have, if you've looked at all the ones

01:01:57 ◼   ► that other things do, you have to have some kind of like

01:01:59 ◼   ► nice presentation of the information.

01:02:02 ◼   ► And that's perfect for SwiftUI because the data is small

01:02:04 ◼   ► and SwiftUI is pretty good at doing animations

01:02:07 ◼   ► and arranging images and stuff.

01:02:09 ◼   ► - Yeah, exactly.

01:02:10 ◼   ► Yeah, I am very much on board with that plan.

01:02:13 ◼   ► - To go back a step just for a moment,

01:02:15 ◼   ► you had made a comment a little while ago

01:02:18 ◼   ► that your knowledge is like been frozen in 2006, 2008,

01:02:23 ◼   ► whatever year you said.

01:02:25 ◼   ► And that makes sense. And you've said many times over many shows, including "Build and Analyze,"

01:02:29 ◼   ► that you do not like to be, you know, the tip of the spear for these sorts of things. You prefer

01:02:33 ◼   ► to use the boring stuff because it's the reliable stuff, and that also makes sense. Have you thought

01:02:39 ◼   ► about, like, doing a deep dive in, okay, how does-- how do people do these sorts of things in 2023,

01:02:46 ◼   ► and how is that different than 2006, 2008, whatever? Like, does that appeal to you? Do you

01:02:51 ◼   ► Do you think that would be a worthwhile expenditure

01:02:53 ◼   ► of your time or are you just gonna do the bare,

01:02:55 ◼   ► and I don't mean this, this is gonna come across snarky,

01:02:57 ◼   ► I don't mean it that way,

01:02:58 ◼   ► but are you just gonna do the bare minimum you can

01:03:00 ◼   ► but to kick that can down the road

01:03:01 ◼   ► as long as you possibly can?

01:03:03 ◼   ► - It's a good question.

01:03:04 ◼   ► I think when you're looking at,

01:03:07 ◼   ► is what I'm doing too old

01:03:09 ◼   ► and is the new thing better than the old thing

01:03:10 ◼   ► and should I use the new thing,

01:03:12 ◼   ► it's important to first ask yourself,

01:03:14 ◼   ► well, what's wrong with the status quo?

01:03:16 ◼   ► What am I not getting with the current thing that I'm using?

01:03:20 ◼   ► what problem am I actually trying to solve here?

01:03:23 ◼   ► And again, like I was saying earlier,

01:03:25 ◼   ► anything I would move to,

01:03:27 ◼   ► not just web services-wise, but tools-wise,

01:03:30 ◼   ► any software or stack component I would move to

01:03:33 ◼   ► would also have pitfalls and limitations

01:03:36 ◼   ► and downsides and bugs and crashes and downtime.

01:03:39 ◼   ► And I at least know how the current stuff works.

01:03:43 ◼   ► I know how to run the current stuff

01:03:45 ◼   ► so that it won't be problematic for me

01:03:48 ◼   ► and be a heavy load on me.

01:03:50 ◼   ► So the new stuff, I'm sure it's mature by now.

01:03:53 ◼   ► I'm sure there's many different options of what people do.

01:03:56 ◼   ► I mean, I think part of the problem is that

01:04:00 ◼   ► whenever people recommend something to me

01:04:03 ◼   ► or when people talk about how things should be done,

01:04:06 ◼   ► they seem to either not realize

01:04:08 ◼   ► that things can be run inexpensively

01:04:11 ◼   ► or totally underestimate the scale of the data

01:04:16 ◼   ► that I'm trying to store.

01:04:17 ◼   ► People will suggest things that, you know,

01:04:19 ◼   ► some kind of hosted service that purports

01:04:22 ◼   ► to take all your problems away,

01:04:23 ◼   ► and then you end up paying per query or per byte

01:04:27 ◼   ► or whatever, and it's like, and you actually look,

01:04:29 ◼   ► okay, well, even if I rewrote this to be very efficient

01:04:32 ◼   ► for this particular service, I would still need to hit

01:04:35 ◼   ► 1,000 queries per unit of time or whatever,

01:04:40 ◼   ► and it's like, okay, well, that's gonna then cost

01:04:42 ◼   ► like 30 grand a month or something.

01:04:45 ◼   ► It's some absurd, it's like, when you actually look at

01:04:48 ◼   ► what it would take, people don't realize,

01:04:50 ◼   ► people who have grown up in the,

01:04:52 ◼   ► just put everything in AWS kind of age,

01:04:57 ◼   ► don't realize how cheap hosting is for doing things

01:05:01 ◼   ► that are low level that you write yourself.

01:05:05 ◼   ► Another example of this is,

01:05:06 ◼   ► there's all these services that exist

01:05:07 ◼   ► to send push notifications for your iPhone apps.

01:05:10 ◼   ► Sending push notifications is so comically cheap for me.

01:05:15 ◼   ► It costs nothing, not even close to, like it's nothing.

01:05:20 ◼   ► If you are paying anything to send push notifications,

01:05:25 ◼   ► you're overpaying.

01:05:26 ◼   ► Like if you have a server to do literally anything else,

01:05:30 ◼   ► you can have a background job that reads a queue

01:05:32 ◼   ► to send push notifications and you will never even notice it.

01:05:36 ◼   ► It costs nothing.

01:05:38 ◼   ► So I send a lot of push notifications.

01:05:41 ◼   ► I have never had to scale any part of that at all.

01:05:44 ◼   ► Like it has never been a thing I've noticed.

01:05:46 ◼   ► Receipt validation, I do all that myself

01:05:49 ◼   ► 'cause it's like one call.

01:05:51 ◼   ► There's stuff that if I used a new system,

01:05:56 ◼   ► a lot of the new ways to do a lot of these things,

01:05:59 ◼   ► whether they're low level components

01:06:01 ◼   ► like a database kind of thing

01:06:04 ◼   ► or whether they're high level services

01:06:06 ◼   ► like we'll send you notifications for you,

01:06:08 ◼   ► a lot of them, it's like one of those as seen on TV

01:06:12 ◼   ► like what's he do at commercials

01:06:14 ◼   ► where you have the black and white scene,

01:06:16 ◼   ► the person's fumbling with the old one,

01:06:18 ◼   ► how can you make a salad with this

01:06:19 ◼   ► and you're like fumbling and dropping all your tools

01:06:21 ◼   ► in the black and white, you need this new tool.

01:06:23 ◼   ► It's like many of those are overstating the problem

01:06:27 ◼   ► or they're making you think something's a huge inconvenience

01:06:30 ◼   ► when it really actually isn't.

01:06:31 ◼   ► And it's like this weird complex doodicky thing here,

01:06:36 ◼   ► whatever, this is solving a problem

01:06:39 ◼   ► that it wants me to think is a big problem for me.

01:06:42 ◼   ► But is it really a big problem for me?

01:06:45 ◼   ► Is this problem actually something I need

01:06:48 ◼   ► to pay someone a large amount of money to solve for me?

01:06:50 ◼   ► What if I just try to do it myself?

01:06:52 ◼   ► What would it take?

01:06:53 ◼   ► What does it cost?

01:06:54 ◼   ► How complex is it?

01:06:55 ◼   ► How hard of a problem is it?

01:06:57 ◼   ► I have found that with a relatively small amount

01:07:01 ◼   ► of virtual server resources, like standard VPS,

01:07:04 ◼   ► like you get with Linode or whatever else,

01:07:06 ◼   ► with a pretty small amount of computing power,

01:07:09 ◼   ► you can do a ton if you just have a Linux server

01:07:13 ◼   ► running a database in some kind of front-end language.

01:07:16 ◼   ► You can do so much more with that than anybody thinks.

01:07:20 ◼   ► I've made a career out of running

01:07:23 ◼   ► moderate-volume web services very, very cheaply

01:07:27 ◼   ► so that I don't need to take VC money

01:07:30 ◼   ► and I don't have to have some kind of creepy business model

01:07:32 ◼   ► and I don't have to have some kind of big company

01:07:33 ◼   ► behind all this stuff.

01:07:35 ◼   ► Overcast has more market share by a lot

01:07:39 ◼   ► than many other apps that have staffs or VC funding

01:07:44 ◼   ► or some kind of startup to like,

01:07:45 ◼   ► oh, we're gonna be the Netflix or podcast.

01:07:47 ◼   ► Or like, part of the reason I'm able to do it

01:07:50 ◼   ► as a single person here is 'cause I'm really good

01:07:53 ◼   ► at hosting stuff cheaply.

01:07:55 ◼   ► So when people recommend a lot of these new alternatives,

01:07:58 ◼   ► it's not only not solving a problem I have,

01:08:02 ◼   ► but it's not solving a problem I have very expensively.

01:08:05 ◼   ► And so to the point where like,

01:08:07 ◼   ► it's not even in the ballpark of what would work for me.

01:08:10 ◼   ► Anyway, that's a long way of saying,

01:08:12 ◼   ► I will move to new things if I have to,

01:08:15 ◼   ► or if there's really compelling reasons to,

01:08:17 ◼   ► and so far, I haven't found those compelling reasons

01:08:21 ◼   ► for any of the tools that I'm actually using.

01:08:24 ◼   ► Once I factor in the cost of rewriting everything,

01:08:28 ◼   ► the cost of learning a new tool,

01:08:29 ◼   ► and learning its pitfalls and downsides,

01:08:31 ◼   ► and downtime, and bugs, and everything else,

01:08:34 ◼   ► what I have is working pretty well most of the time

01:08:36 ◼   ► if I actually optimize it

01:08:38 ◼   ► and I'm a little bit careful about it.

01:08:40 ◼   ► - That makes sense.

01:08:41 ◼   ► So when are you going to Postgres?

01:08:43 ◼   ► - Postgres, again, I've never used Postgres.

01:08:45 ◼   ► I hear it's great, but it doesn't seem like

01:08:48 ◼   ► it's greater enough than MySQL

01:08:51 ◼   ► to be worth the changing in the learning curve.

01:08:54 ◼   ► So maybe, I'll move to it if I have to,

01:08:57 ◼   ► but I'm not rushing to do that.

01:09:00 ◼   ► - All right, well, best of luck.

01:09:02 ◼   ► I mean, it certainly makes me wonder

01:09:05 ◼   ► about my strategy for this thing that may never even see the light of day anyway. So

01:09:08 ◼   ► it's not like this is a business crushing decision or anything like that. It's not dramatically

01:09:14 ◼   ► altering my world. But it makes me think and makes me wonder, you know, I had initially

01:09:19 ◼   ► started writing this, again, you know, it's extremely early stages, but I initially started

01:09:24 ◼   ► writing this thing with like, well, I'll just treat CloudKit as the data source. And it's

01:09:28 ◼   ► quickly becoming apparent that for other reasons too, but that's not going to work. And, you

01:09:33 ◼   ► CloudKit may end up being like a sync destination,

01:09:36 ◼   ► but having it be the one true source of data

01:09:38 ◼   ► is probably unwise.

01:09:39 ◼   ► So, I don't know, it's a lot to think about,

01:09:41 ◼   ► and I appreciate you taking the time.

01:09:42 ◼   ► And like I said, listen under the radar 258,

01:09:44 ◼   ► if you haven't already, 'cause it was really good.

01:09:46 ◼   ► And under 30 minutes.

01:09:47 ◼   ► - It's a much shorter version of this whole thing.

01:09:49 ◼   ► We are brought to you this week by Trade Coffee.

01:09:54 ◼   ► If you love drinking coffee every morning,

01:09:57 ◼   ► I know I sure do, you have to check out Trade Coffee.

01:10:00 ◼   ► Trade makes it super easy to get great coffee delivered

01:10:04 ◼   ► fresh from the finest local roasters around the country.

01:10:08 ◼   ► Now look, if you've been getting the same grocery store

01:10:10 ◼   ► coffee for your whole life,

01:10:11 ◼   ► you don't know what you're missing.

01:10:12 ◼   ► 'Cause coffee is best when it's freshly roasted.

01:10:14 ◼   ► It goes stale within a couple of weeks at most of roasting.

01:10:17 ◼   ► And Trade makes it easy to get fresh roasts delivered

01:10:20 ◼   ► to your doorstep.

01:10:21 ◼   ► And they have them from all these local roasters

01:10:24 ◼   ► to choose from.

01:10:24 ◼   ► And so this is a coffee subscription service.

01:10:27 ◼   ► And what's great about it is they have such great variety,

01:10:29 ◼   ► You can discover new coffees so easily

01:10:32 ◼   ► and make your best cup of coffee at home every single day.

01:10:35 ◼   ► All of this from the nation's top rated independent roaster

01:10:38 ◼   ► so that Trade can send you coffee they know you'll love

01:10:41 ◼   ► fresh to your home and on whatever your preferred schedule is.

01:10:44 ◼   ► So whether you're a coffee nerd

01:10:45 ◼   ► and you already know what you like,

01:10:46 ◼   ► or whether you're new to specialty high-end coffee

01:10:49 ◼   ► and need some guidance through it,

01:10:50 ◼   ► Trade makes it easy and convenient to discover new coffees

01:10:53 ◼   ► that you're guaranteed to love.

01:10:55 ◼   ► I personally have used Trade for I think over a year now,

01:10:58 ◼   ► and I love the variety I get from them.

01:11:00 ◼   ► It's so good.

01:11:01 ◼   ► And my tastes changed a while ago,

01:11:03 ◼   ► and so I was able to adjust my preferences

01:11:05 ◼   ► really easily with them and give feedback.

01:11:07 ◼   ► Oh, I like this one, this one wasn't so good,

01:11:09 ◼   ► I like this one.

01:11:10 ◼   ► And they learned quickly,

01:11:11 ◼   ► and it just hit after hit after hit now.

01:11:14 ◼   ► It's just so, so good.

01:11:16 ◼   ► So treat yourself or the coffee lover in your life

01:11:19 ◼   ► with Trade Coffee.

01:11:20 ◼   ► Right now, Trade's offering our listeners

01:11:22 ◼   ► a total of $30 off a subscription

01:11:24 ◼   ► and access to limited time specials

01:11:26 ◼   ► at drinktrade.com/atp.

01:11:29 ◼   ► That's drinktrade.com/atp for $30 off.

01:11:34 ◼   ► Drinktrade.com/atp.

01:11:37 ◼   ► Thank you so much to Trade Coffee

01:11:39 ◼   ► for keeping me caffeinated and for sponsoring our show.

01:11:42 ◼   ► Breaking news.

01:11:46 ◼   ► Samsung has decided to make the studio display

01:11:51 ◼   ► look like garbage, maybe, possibly.

01:11:54 ◼   ► So CES is happening soon, I guess.

01:11:57 ◼   ► I don't even know.

01:11:57 ◼   ► I don't pay attention to CES.

01:11:58 ◼   ► But there's some CES-related announcements.

01:12:01 ◼   ► The Verge covered this, and we'll also put a link to the

01:12:02 ◼   ► Samsung PR article as well.

01:12:05 ◼   ► But Samsung has announced the Viewfinity S9.

01:12:10 ◼   ► And it appears to be a shot across the bow at the studio

01:12:15 ◼   ► display, almost at the studio display.

01:12:19 ◼   ► It's 5K, same resolution.

01:12:21 ◼   ► It's matte.

01:12:22 ◼   ► It's not glossy.

01:12:24 ◼   ► It gets almost perfect DCI-P3 color reproduction.

01:12:29 ◼   ► It has HDR 600, which we'll talk about

01:12:31 ◼   ► what the heck that means in a minute.

01:12:33 ◼   ► It has a 4K external webcam,

01:12:35 ◼   ► which I guess you can slam onto the top of the screen

01:12:39 ◼   ► when you need it and then make it go away when you don't.

01:12:41 ◼   ► Allegedly, this includes automated zoom control

01:12:44 ◼   ► that will track subjects in the frame.

01:12:46 ◼   ► Does that sound like center stage to anyone else?

01:12:48 ◼   ► - 'Cause it worked out so well for the studio display.

01:12:50 ◼   ► - Well, I actually, I am a center stage apologist.

01:12:52 ◼   ► I like center stage, but the studio display's camera

01:12:55 ◼   ► fidelity is pretty bad.

01:12:57 ◼   ► It still is.

01:12:59 ◼   ► It'll rotate to portrait in the included stand.

01:13:02 ◼   ► It has HDMI input, Thunderbolt 4, USB-C, and DisplayPort.

01:13:05 ◼   ► It'll charge downstream, whatever you want to call it,

01:13:07 ◼   ► up to 96 watts.

01:13:09 ◼   ► It does not have True Tone.

01:13:10 ◼   ► And apparently it runs Samsung's--

01:13:13 ◼   ► what is this?

01:13:13 ◼   ► Tizen?

01:13:14 ◼   ► Is that how you pronounce it?

01:13:16 ◼   ► TVOS, and it has apps on it and whatnot.

01:13:18 ◼   ► So you can use it even when it's not

01:13:20 ◼   ► connected to a computer.

01:13:22 ◼   ► That's about everything there is to know about this,

01:13:25 ◼   ► except the deep dive into HDR 600,

01:13:27 ◼   ► which I'll let Captain John take us down in a minute.

01:13:30 ◼   ► But at a glance, not knowing a price,

01:13:33 ◼   ► not knowing when it's coming out,

01:13:34 ◼   ► at a glance, this looks like a pretty compelling option.

01:13:38 ◼   ► I'm not sure it's better than the studio display.

01:13:41 ◼   ► It may be more money than the studio display,

01:13:43 ◼   ► for all I know, although I'd be surprised.

01:13:45 ◼   ► - Unlikely. - It is unlikely.

01:13:47 ◼   ► But hey, I am all in on there being any competition

01:13:52 ◼   ► to the LG Ultra Me and the studio display.

01:13:55 ◼   ► So I'm here for this, at least in principle.

01:13:58 ◼   ► I'm excited to see more about this over time.

01:14:00 ◼   ► - The nice thing about it is that it looks

01:14:02 ◼   ► like the studio display too.

01:14:03 ◼   ► So if you want a studio display,

01:14:04 ◼   ► but you want a cheaper version of it,

01:14:06 ◼   ► that kind of looks like studio display from the distance,

01:14:08 ◼   ► from what I can see in the pictures,

01:14:10 ◼   ► it looks very similar because you know Samsung

01:14:12 ◼   ► and making things that look similar to Apple,

01:14:13 ◼   ► that they're not above that.

01:14:15 ◼   ► - No.

01:14:16 ◼   ► like they have here.

01:14:17 ◼   ► So yeah, because this is a pre-CS announcement, they haven't actually announced the price

01:14:22 ◼   ► or availability, so it's hard to talk too much about it, but the feature set is just

01:14:25 ◼   ► what you would expect, which is, we're going to do a thing like Apple, we're going to make

01:14:29 ◼   ► it look like Apple's product, we'll copy their aesthetic, we'll use the same panel presumably

01:14:32 ◼   ► or a very similar panel, we'll get to that in a second, and we'll undercut them on price.

01:14:36 ◼   ► And you don't have to pay extra for an adjustable stand, and our webcam is 4K, which I think

01:14:41 ◼   ► gives it a shot of not being crap when they do center stage, because they just have more

01:14:45 ◼   ► pixels to work with, right? It's an external camera which is way bigger than the bezel.

01:14:49 ◼   ► Like it's not huge, but like if you look at the picture, it's a circle that is like two to three

01:14:54 ◼   ► times the thickness of a tiny bezel on here. So hey, we can get a normal sized camera without

01:14:58 ◼   ► worrying about having to fit it in, and the camera can be 4K, right? And, you know, presuming it

01:15:05 ◼   ► undercuts it on price, it looks like it does everything the Apple one does. Now, the real

01:15:09 ◼   ► The real question is, is it the same panel?

01:15:12 ◼   ► So the HDR 600 thing, there's a bunch of terrible

01:15:15 ◼   ► marketing terms, we'll put a link in the show notes

01:15:17 ◼   ► to the HDR 600 one, but there's the VISA display HDR

01:15:21 ◼   ► certification things, if you watch any sort of

01:15:25 ◼   ► review channels for gaming monitors, which I watch a lot of

01:15:28 ◼   ► 'cause I'm always looking for a good monitor for my PS5

01:15:31 ◼   ► that is HDR, every monitor says, supports HDR

01:15:36 ◼   ► than some number.

01:15:38 ◼   ► Pretty much all of those numbers,

01:15:39 ◼   ► or let's say half those numbers,

01:15:41 ◼   ► mean you don't have HDR.

01:15:43 ◼   ► HDR 400, not HDR.

01:15:45 ◼   ► HDR 500, not HDR.

01:15:47 ◼   ► HDR 600, probably not HDR, right?

01:15:50 ◼   ► Because regular monitors go to that brightness.

01:15:53 ◼   ► So witness the Apple Studio display.

01:15:56 ◼   ► It goes to 600 nits.

01:15:58 ◼   ► But it doesn't have local dimming,

01:16:00 ◼   ► and 600 nits is brighter than 500 nits,

01:16:04 ◼   ► which was the previous kind of standard Mac display thing,

01:16:07 ◼   ► but not super bright.

01:16:09 ◼   ► And not having local dimming means you don't really get

01:16:11 ◼   ► that contrast that you would want in high dynamic range,

01:16:14 ◼   ► because the brighter you support in the screen,

01:16:17 ◼   ► the brighter your blacks are gonna be too,

01:16:19 ◼   ► because the backlight is always on behind the thing, right?

01:16:21 ◼   ► So if you look at the HDR 600 specification

01:16:24 ◼   ► on the displayhdr.org website,

01:16:26 ◼   ► one of the things that it says is,

01:16:28 ◼   ► of the HDR 600 standard specifically is,

01:16:33 ◼   ► real-time contrast ratios with local dimming.

01:16:37 ◼   ► That's in the bullet points of this marketing material

01:16:40 ◼   ► for HDR 600.

01:16:42 ◼   ► Local dimming means you can turn off the backlight

01:16:45 ◼   ► behind some portions of the screen.

01:16:47 ◼   ► The Verge said-- whoever's writing this research--

01:16:49 ◼   ► says, "I couldn't get an answer at press time

01:16:51 ◼   ► as to whether Samsung's latest monitor includes local dimming.

01:16:54 ◼   ► I'm guessing if it did, they would

01:16:55 ◼   ► be highlighting that feature."

01:16:57 ◼   ► It's kind of weird that Samsung wouldn't

01:16:59 ◼   ► say in their press release if it did support local dimming,

01:17:01 ◼   ► but they do say it supports HDR 600

01:17:04 ◼   ► and the HDR 600 marking bullet points say local dimming.

01:17:07 ◼   ► I don't know what this means, we'll find out.

01:17:09 ◼   ► But if it does have local dimming

01:17:10 ◼   ► and the local dimming isn't garbage,

01:17:12 ◼   ► like doesn't have, you know,

01:17:13 ◼   ► it was four zones or something, or it's not one of those,

01:17:15 ◼   ► they do like vertical local dimming.

01:17:16 ◼   ► I don't know if you know about this,

01:17:17 ◼   ► but like one of the crappy,

01:17:19 ◼   ► cheap local dimming technologies

01:17:21 ◼   ► just cuts your screen into a bunch of vertical slices

01:17:24 ◼   ► and the backlight can be on in any of those slices,

01:17:26 ◼   ► which as you can imagine is terrible if you have,

01:17:28 ◼   ► you know, like something that isn't the full height

01:17:30 ◼   ► of the screen because you get blooming and a vertical stripe.

01:17:32 ◼   ► It's bad.

01:17:34 ◼   ► So we'll see.

01:17:35 ◼   ► Obviously, Apple has no local dimming at all.

01:17:38 ◼   ► This is an IPS panel, which is what Apple has,

01:17:41 ◼   ► but it claims to be HDR 600, which is, oh, it's just

01:17:44 ◼   ► like Apple because Apple says their thing is 600 nits,

01:17:46 ◼   ► but the local dimming thing is still a mystery.

01:17:48 ◼   ► And as for True Tone, which is Apple's thing where they have

01:17:51 ◼   ► like a light sensor, it adjusts the color temperature,

01:17:54 ◼   ► essentially, of your screen based on how yellow

01:17:56 ◼   ► the light is in your room.

01:17:57 ◼   ► This doesn't have that because True Tone

01:17:59 ◼   ► is an Apple technology, but it does have

01:18:01 ◼   ► a color calibration feature built into the monitor

01:18:04 ◼   ► where you can either use your smartphone

01:18:07 ◼   ► with the Samsung app on it to sort of like

01:18:08 ◼   ► point it at the screen like the Apple TV thing

01:18:10 ◼   ► where it calibrates, or you can also use

01:18:12 ◼   ► a more professional calibration tool to calibrate it.

01:18:15 ◼   ► That's not calibrating to your room lighting,

01:18:17 ◼   ► it's calibrating the monitor to some standard or whatever,

01:18:20 ◼   ► but that shows where they're kind of aiming this monitor,

01:18:22 ◼   ► so I expect it to be in the ballpark of Apple's price

01:18:25 ◼   ► just because you don't put color calibration

01:18:27 ◼   ► on a monitor that people are just gonna use

01:18:29 ◼   ► to look at webpages or whatever.

01:18:31 ◼   ► It's trying to get that same market

01:18:32 ◼   ► of people who care about color accuracy on their displays.

01:18:36 ◼   ► The Tizen OS, this is not Samsung's first monitor

01:18:40 ◼   ► that does this.

01:18:41 ◼   ► They have a bunch of monitors that's been selling

01:18:42 ◼   ► for a while.

01:18:43 ◼   ► They're like little smart TVs, right?

01:18:45 ◼   ► Only in a computer monitor, right?

01:18:46 ◼   ► So you connect it to nothing except for power.

01:18:49 ◼   ► You can turn it on and inside of it,

01:18:51 ◼   ► it has not an A13 obviously,

01:18:53 ◼   ► but some kind of CPU and memory and storage

01:18:56 ◼   ► so that you can run essentially TV streaming apps.

01:18:59 ◼   ► You can even run Google Hangouts

01:19:00 ◼   ► and do video conferencing and maybe Zoom calls

01:19:03 ◼   ► with just the monitor,

01:19:04 ◼   ► because it has enough smarts in it to run stuff,

01:19:06 ◼   ► just like your TV has enough smarts in it.

01:19:08 ◼   ► This is just something, it's kind of weird,

01:19:09 ◼   ► it took this long to come to computer monitors.

01:19:11 ◼   ► In many ways, televisions were ahead of monitors,

01:19:13 ◼   ► because we buy a monitor and we expect this thing is useless

01:19:16 ◼   ► until I plug it into something to send it an image.

01:19:19 ◼   ► And because these monitors have an increasing amount

01:19:22 ◼   ► of computing power in them just to do their monitor job,

01:19:24 ◼   ► It's not too much of a stretch to say, you know what,

01:19:27 ◼   ► why don't we just run our TV OS in here too?

01:19:29 ◼   ► And so now all you need to do is connect it to power

01:19:31 ◼   ► and you can watch Netflix or look at YouTube

01:19:34 ◼   ► or do video conferencing with the supplied camera

01:19:37 ◼   ► or whatever.

01:19:37 ◼   ► I think that's a great idea.

01:19:38 ◼   ► And I think it's kind of,

01:19:40 ◼   ► it's definitely a huge waste of computing resources

01:19:42 ◼   ► for the A13 in the studio display,

01:19:45 ◼   ► not to be acting like a little Apple TV in that mode,

01:19:48 ◼   ► but that's something Apple chose not to do,

01:19:50 ◼   ► at least in the first iteration of their product.

01:19:51 ◼   ► So we don't know a ship date for this,

01:19:54 ◼   ► Maybe by the time this comes out,

01:19:56 ◼   ► a new Apple Studio display with local dimming will come out

01:19:58 ◼   ► and we'll have a better comparison.

01:20:00 ◼   ► But not having to pay extra for an adjustable display.

01:20:03 ◼   ► And if you really don't like glossy,

01:20:05 ◼   ► not having to pay the huge amount of money

01:20:07 ◼   ► for Apple's nano texture,

01:20:08 ◼   ► but instead to accept a presumably somewhat inferior

01:20:11 ◼   ► but way cheaper default matte display on this thing

01:20:14 ◼   ► may be to your liking.

01:20:15 ◼   ► So I'm with Casey.

01:20:18 ◼   ► Anything to get beyond the ultra fine

01:20:21 ◼   ► as being the only other option.

01:20:22 ◼   ► - Vuefinity is an awful name, but I think this is

01:20:26 ◼   ► some decent competition for Apple's display.

01:20:28 ◼   ► - And by the way, it's not Vuefinity, it's Vuefinity S9.

01:20:31 ◼   ► - Yeah, S9 reminds me of an Audi, so.

01:20:33 ◼   ► - No, I'm happy to see this.

01:20:36 ◼   ► Even if it's not something any of us want,

01:20:38 ◼   ► it's not healthy in this ecosystem

01:20:42 ◼   ► for there to be only one monitor available

01:20:46 ◼   ► at a given size that has the right DPI for Apple users.

01:20:49 ◼   ► And that's only a fairly recent thing

01:20:52 ◼   ► that's really been a problem.

01:20:53 ◼   ► Like, you know, the XDR is the first Apple monitor

01:20:58 ◼   ► I've ever bought.

01:20:59 ◼   ► All of my desktop use before this, that wasn't an iMac,

01:21:02 ◼   ► you know, when I go back in a pre desktop retina time,

01:21:07 ◼   ► it was always with third party monitors.

01:21:08 ◼   ► I had some Dell monitors, I had an HP monitor,

01:21:12 ◼   ► I had, you know, before that I had like ViewSonic

01:21:14 ◼   ► and the other like kind of PC brands.

01:21:16 ◼   ► There's always been good monitor choices from third parties

01:21:20 ◼   ► until relatively recent in computing history

01:21:23 ◼   ► for Apple people to choose.

01:21:24 ◼   ► That would have the right DPI,

01:21:26 ◼   ► that would have the resolution that we're looking for

01:21:29 ◼   ► and the size we're looking for,

01:21:30 ◼   ► and pretty decent specs if you didn't care too much

01:21:33 ◼   ► about some of the stuff Apple cares about.

01:21:35 ◼   ► And for me, most of that was always fine.

01:21:39 ◼   ► I never needed super color accuracy for my programming work.

01:21:44 ◼   ► Like that's not something I really ever need.

01:21:47 ◼   ► So it's been fine.

01:21:50 ◼   ► I've greatly enjoyed those monitors

01:21:52 ◼   ► at the time they came out.

01:21:52 ◼   ► And it was only when we went to desktop Retina

01:21:55 ◼   ► that the Darth of Monitors started.

01:21:58 ◼   ► And even when we first got the 27 inch category of monitors,

01:22:03 ◼   ► there were a couple of 5K ones besides the LG Ultrafine.

01:22:10 ◼   ► Like, Adele had one briefly.

01:22:13 ◼   ► I think ViewSonic actually had one briefly

01:22:15 ◼   ► that maybe wasn't even available in the US.

01:22:17 ◼   ► But there were a couple others at launch

01:22:19 ◼   ► and then they just disappeared

01:22:20 ◼   ► and were just discontinued very quickly.

01:22:23 ◼   ► But I hope there are more options.

01:22:25 ◼   ► Like Apple should not be the only company

01:22:28 ◼   ► making a 5K 27 inch monitor.

01:22:32 ◼   ► Apple should not be the only company

01:22:33 ◼   ► making a 6K 30 whatever XDR is either.

01:22:37 ◼   ► Like there should be other options here

01:22:40 ◼   ► and ideally more than one

01:22:42 ◼   ► so we don't have the ultra fine problem.

01:22:45 ◼   ► But that's a sign of a healthy ecosystem.

01:22:48 ◼   ► And I know for various reasons, like PC people

01:22:52 ◼   ► kind of went in a different direction

01:22:53 ◼   ► with what they prioritize with monitor design,

01:22:55 ◼   ► and a different direction than Apple

01:22:59 ◼   ► with the whole high DPI, retina,

01:23:01 ◼   ► game frame rate kind of situation.

01:23:03 ◼   ► I get that, but it's really good to have more options here.

01:23:07 ◼   ► And so if Apple screws this up again,

01:23:11 ◼   ► there's somewhere else to go.

01:23:12 ◼   ► And if this actually ends up being better

01:23:15 ◼   ► than Apple's monitor in ways that are important to people,

01:23:17 ◼   ► like number one, I'm sure price is probably gonna be better.

01:23:20 ◼   ► Number two, that webcam is almost certain to be better

01:23:23 ◼   ► because it's so much larger, it has room for better optics,

01:23:27 ◼   ► better sensor and everything.

01:23:29 ◼   ► - And it's 4K.

01:23:31 ◼   ► - Yeah, right, and so I expect certain things

01:23:34 ◼   ► about this to be better.

01:23:35 ◼   ► I expect certain things about it to be worse,

01:23:38 ◼   ► but that's fine, it's another option,

01:23:40 ◼   ► and the more options we have, the better.

01:23:42 ◼   ► We are brought to you this week by Squarespace,

01:23:46 ◼   ► The all-in-one platform for building your brand

01:23:48 ◼   ► and growing your business online.

01:23:50 ◼   ► Stand out with a beautiful website,

01:23:52 ◼   ► engage with your audience, and sell anything.

01:23:54 ◼   ► Your products, content you create, even your time.

01:23:57 ◼   ► Squarespace makes it incredibly easy for anyone

01:24:00 ◼   ► to make amazing websites, especially with all

01:24:02 ◼   ► their great new business features.

01:24:04 ◼   ► So for instance, maybe your business

01:24:06 ◼   ► is selling content to members.

01:24:08 ◼   ► They have member areas now.

01:24:09 ◼   ► You can monetize your content and expertise

01:24:11 ◼   ► in a way that fits your brand.

01:24:13 ◼   ► Unlock new revenue streams for your business

01:24:14 ◼   ► and free up time in your schedule

01:24:16 ◼   ► by selling access to gated content like videos,

01:24:18 ◼   ► online courses, or newsletters.

01:24:21 ◼   ► Of course, you can also have online stores of Squarespace.

01:24:23 ◼   ► Sell any physical or digital products,

01:24:26 ◼   ► all with the tools Squarespace has to sell online.

01:24:29 ◼   ► You can have email campaigns to promote what you're doing,

01:24:32 ◼   ► apply your own brand ingredients like site colors and logo,

01:24:35 ◼   ► have built-in analytics to measure the impact of every

01:24:38 ◼   ► email that you send.

01:24:39 ◼   ► And all of this, of course, is backed

01:24:40 ◼   ► by awesome, powerful analytics in other areas, too.

01:24:43 ◼   ► You can use insights to grow your business,

01:24:45 ◼   ► Learn where your site visits and your sales are coming from.

01:24:48 ◼   ► Analyze which channels are most effective.

01:24:50 ◼   ► Build your marketing strategy based on your top keywords,

01:24:52 ◼   ► your most popular products and content.

01:24:54 ◼   ► All of that and so much more is available

01:24:57 ◼   ► with Squarespace sites.

01:24:59 ◼   ► They make it so easy.

01:25:00 ◼   ► You can see for yourself, you can build the entire site

01:25:03 ◼   ► with their free trial.