PodSearch

ATP

183: I Filed a Radar

 

00:00:00   It is so hot in this room and so humid the humidity has been brutal here

00:00:06   It's only 77 degrees and I'm dying in this room because it's so humid and of course

00:00:11   It's probably 90 degrees in this room. I feel great air conditionings on

00:00:14   I haven't put my slippers on mid show because my feet were getting a little bit cold

00:00:18   All right, we should probably actually get the show started huh? No, that's making it in the show none of it. No, come on

00:00:24   No, ah, so here, you know how I'm gonna make you put some of that in the show watch this

00:00:29   Hey, so speaking of cars, we should probably do some follow-ups, starting with more reasons

00:00:34   for QNX.

00:00:35   Hey Casey, do we have any follow-up?

00:00:37   I hate you so much.

00:00:41   So Casey, do we have any follow-up?

00:00:43   We should probably start with some follow-up, Marco.

00:00:46   You make an excellent point.

00:00:47   And we had an individual write in with more reasons for using QNX.

00:00:51   And these actually were the, to my ears and eyes, the best reasons I've heard yet.

00:00:58   And what it basically boiled down to was incredibly fast boot times were generally important for

00:01:06   cars because you don't want to, say, have to wait until you're five minutes down the

00:01:09   road before your air conditioning or your radio turns on or something like that.

00:01:13   And the other reason that this individual gave was, you know, if you think about it,

00:01:17   the instrument clusters on a lot of cars are often driven by some sort of display or perhaps

00:01:22   driven by a signal that's coming off some QNX-derived computer.

00:01:27   And those need to be real time.

00:01:29   I mean, that data needs to show you

00:01:31   exactly how fast you're going right freaking now.

00:01:33   And so having a real time OS like QNX

00:01:36   makes that sort of thing a lot easier.

00:01:39   - Like Tesla's system is not based on QNX.

00:01:41   I'm pretty sure it's just some kind of Linux.

00:01:43   And you can tell.

00:01:44   Like you can tell because it's,

00:01:45   because like, you know, some, like, again,

00:01:48   like I said last time, like most of the time it works fine.

00:01:51   But sometimes, like we were upstate and I heard,

00:01:56   We were driving around, I was following ad directions,

00:01:58   and the map was getting a little bit wonky,

00:02:00   like it was starting to like flick out

00:02:01   and slow down a little bit and like not update

00:02:03   quickly enough, and that's great too,

00:02:05   like when you're following a map doing turn by turn

00:02:07   and it just doesn't update, so you're looking at

00:02:09   like the last turn from 30 seconds ago instead, that's fun.

00:02:12   And then there was one point where it made

00:02:16   the turn signal clicking sound

00:02:19   when the turn signal wasn't on.

00:02:21   - What?

00:02:23   - There was, in one of the, like I described last episode

00:02:25   I've had to reboot it three times so far over the three or four months, whatever it is.

00:02:30   One of the times that I had to reboot it, immediately beforehand, the turn signal was

00:02:35   not clicking. Like it would turn on and the light would, the indicator would blink in

00:02:38   the dash, but you wouldn't hear the click noise. So obviously the clicking noise is

00:02:41   made by the center console computer. Like that, it's adding it to the sound system in

00:02:46   all likelihood. Anyway, so one of the symptoms that I knew something was wacky this time

00:02:53   was the turn signal noise started clicking

00:02:55   when the turn signal was not on.

00:02:57   I was driving straight down a road,

00:02:59   I hadn't touched it, like that was it.

00:03:01   So that I rebooted again a few days ago.

00:03:04   - Goodness.

00:03:06   - So yes, I would have pre,

00:03:07   and when it's not booted, the turn signal does not,

00:03:09   like when it's rebooting, you can turn,

00:03:11   and the indicator on the dashboard did blink

00:03:14   when the computer was rebooting,

00:03:16   like it was still blinking,

00:03:17   and so I assumed the signal was still on,

00:03:20   I hope the light was still blinking outside the car,

00:03:22   but there's no sound.

00:03:23   And again, it's like one of those things where

00:03:25   when you tie these things to software,

00:03:27   you are at the whim of the software's stability,

00:03:32   responsiveness, uptime.

00:03:35   It's similar to like, one of the things that I didn't like

00:03:39   so much when I wore the Apple Watch is

00:03:42   so many interactions are tied to force touch.

00:03:45   And when you push really hard on the watch face,

00:03:51   and you expect that button click,

00:03:52   even though it's not really a button,

00:03:54   even though it's all fake and simulated,

00:03:56   you expect that better click back immediately

00:03:59   as if it were a button.

00:04:00   And the same thing with all the force text track pads,

00:04:02   like they have the same requirement.

00:04:04   Like it has to respond like the physical object

00:04:07   that it is mimicking, right?

00:04:08   It has to be quick, it has to be immediate.

00:04:10   There can't be any delay.

00:04:12   Well, on the watch, you know,

00:04:13   the watch is very slow hardware-wise,

00:04:14   and sometimes the software gets gummed up a little bit,

00:04:17   and so you push and sometimes there's a delay

00:04:19   before it actually clicks back at you.

00:04:21   And it totally breaks the illusion,

00:04:23   it makes it feel broken or cheap or wrong

00:04:26   or whatever the case may be.

00:04:28   And that's actually one concern I have

00:04:30   with the Force Touch home button

00:04:32   that is rumored to be on the next iPhone.

00:04:34   That like, you know, if you push that home button

00:04:39   and the iPhone software is like a little bit overwhelmed

00:04:43   or buggy or whatever, if you push that

00:04:45   and you don't get a click feel back

00:04:47   for a few seconds or at all,

00:04:49   like that's gonna feel really broken really quickly.

00:04:51   So anyway, similar thing with like this car stuff,

00:04:54   like you know, back to this follow up saying

00:04:55   that QNX is really good for boot time

00:04:58   and response times of interactions.

00:05:00   Like I totally get that and I totally respect that

00:05:03   because when you're tying things to controls

00:05:07   of a physical object, that makes a big difference.

00:05:09   - My complaint about the ancient iPhone 6 that I'm using

00:05:12   is that I have a physical home button and I press it

00:05:14   and it always you know, goes in just like you said

00:05:16   and I feel that physical feedback immediately,

00:05:18   but you know what doesn't happen immediately?

00:05:20   Springboard doesn't appear immediately.

00:05:22   That's what doesn't happen.

00:05:23   I click it and then I look at my phone,

00:05:25   I'm like, what are you doing?

00:05:26   Well, I push the button.

00:05:27   I know I pushed it because it went in and then it went out,

00:05:29   and then, oh, okay, now the animation's starting.

00:05:31   I'm getting picky in my old age.

00:05:33   I don't know.

00:05:34   - Getting picky?

00:05:35   What you should really do,

00:05:37   to make it seem like your phone is even older,

00:05:39   instead of saying, my old iPhone 6,

00:05:40   you should say, my first iPhone, which I'm still using.

00:05:44   - That's true, yeah.

00:05:46   I don't think my iPhone 6 is getting slower, but I'm just getting less and less patient

00:05:50   for like when it doesn't respond.

00:05:52   And I know it's not just because it's slow.

00:05:53   Like my iPad Pro, the 9.7 inch iPad Pro, which is pretty darn fast, I still feel like when

00:05:57   I hit the home button on that sometimes I have to wait.

00:06:00   I want that to be instant.

00:06:01   Just wait for the new touch ID sensor.

00:06:04   It's pretty damn fast.

00:06:05   Yeah, I know.

00:06:06   I've used my wife's phone.

00:06:07   I know that the touch ID sensor is fast.

00:06:08   I'm just saying like I feel like sometimes it's not responding to me as immediately as

00:06:13   I feel like it should.

00:06:15   Who knows?

00:06:17   Anyway, the other thing that this person had to say about,

00:06:19   why do cars have crappy hardware?

00:06:21   We read some other feedback from someone

00:06:25   who works on these systems talking

00:06:27   about how the car manufacturers want to save money and pinch

00:06:29   pennies.

00:06:30   And I said that was silly.

00:06:31   A couple other factors that were also

00:06:33   mentioned by many other people but reiterated by this person

00:06:35   is that there's a long lead time on hardware that goes into cars.

00:06:38   The development cycles are really long,

00:06:40   so they have to pick hardware that's

00:06:42   available like three years before the car even comes out.

00:06:44   And maybe it's even longer, so maybe it's

00:06:45   four or five years by the time comes out. So even if you pick something that was current it'll be

00:06:49   like three, four, or five years old by the time it ships in the car. And finally, stuff in cars

00:06:54   has to operate at extremes of temperature. If you've ever left your phone on the dashboard,

00:06:58   or even in a sealed cubby inside your car, like not in the sun, but just try taking your phone

00:07:03   and putting it in the glove box or in the little console thing anywhere inside your car on a hot

00:07:08   day, when you come back, chances are good after being away for several hours in the middle of the

00:07:12   your device will refuse to function because the temperature is too high and

00:07:15   the chips in the car obviously can't do that they have to continue working even

00:07:20   when they're like 150 degrees or whatever and they have to continue

00:07:23   working on it's negative 40 these are all real temperatures not a crazy

00:07:27   Celsius yeah it's similar like like if you ever heard of like how slow the

00:07:33   computers are on satellites or the space shuttle because like you know like you

00:07:36   hear stories you know about like how they're still running something it's

00:07:39   about as fast as like a 486 or something like that.

00:07:42   And like the reason, one of the reasons why they have

00:07:44   to use such slow types of hardware is that they have

00:07:48   to operate in extreme conditions that would just kill any,

00:07:51   you know, like the chip that's in an iPhone.

00:07:53   And so it's similar but less severe in a car.

00:07:55   - Yeah, that's their excuse, but like still,

00:07:57   it's a little bit ridiculous.

00:07:58   They, you know, like I said, the washing machine chips

00:08:01   they're using, like they're slower than the ones

00:08:02   in the space probes.

00:08:03   The space probes have PowerPC 601s that are like

00:08:05   rad hardened in them.

00:08:07   and they're using chips that are like from a Game Boy Advance, so it would seem worse.

00:08:11   Or maybe Game Boy Advance is faster than that.

00:08:14   Anyway, bottom line, we have the technology, we can make the hardware in cars better, and

00:08:20   I think we probably want to shorten up the cycle time on that too.

00:08:23   I know it's sort of all developed as one big unit, but you just can't afford to do that.

00:08:28   If a manufacturer can shave even just a year off of the cycle time and they can have a

00:08:32   year newer electronics, that can make a big difference, I think.

00:08:36   like I said, as the infotainment systems and even the instrument clusters and all that

00:08:39   stuff becomes a more important differentiator for cars, as the software becomes like all

00:08:45   our other products, the software becomes a more important part of the whole product mixture,

00:08:49   the company that can figure that software out is going to have a big advantage.

00:08:52   All right.

00:08:54   Tell us about AudioSync and Quartz Crystals, which we've gotten a lot of feedback about.

00:08:58   Marco, do you want to kind of cover this?

00:09:00   Yeah, sure.

00:09:01   So basically, so, you know, the last couple episodes we've discussed.

00:09:04   A while back I discussed my podcast tool

00:09:07   to automatically sync up tracks that were recorded

00:09:10   locally for each person in a multi-person recording,

00:09:12   and then sync it up to the master track

00:09:15   because if you try to import those tracks manually,

00:09:18   you sync it up, the difference in the sense of time

00:09:23   that each person's audio interface had,

00:09:26   it caused a problem called drift,

00:09:28   where if you sync up our recordings

00:09:30   like right up front in the beginning of a show,

00:09:32   an hour in, we might be out of sync

00:09:34   by like half a second or a second. My microphone interface, the little converter in it that

00:09:39   samples the audio at 44,100 times per second, has a very, very, very slightly different

00:09:46   interpretation of what that means, of how long a second is or how many times it has

00:09:51   to be, than the one in cases of John's computers. And over time, that very, very tiny error

00:09:56   can add up to quite a lot. And so this causes the problem of drift, so we've gotten lots

00:10:00   the feedback about why this happens.

00:10:04   Last episode I speculated that you're making

00:10:06   these components, these physical components,

00:10:09   and any little tiny bit of imprecision in making them,

00:10:14   when you are taking 44,000 samples per second

00:10:18   over two hours, a very small variance

00:10:22   in the clock performance of two different physical devices

00:10:26   will cause them to disagree and to drift over time like this.

00:10:29   So my speculation was basically like,

00:10:31   it's not really possible to make something

00:10:33   that is cheap and in a computer like this

00:10:35   that is more accurate than that

00:10:37   and we just have to deal with it.

00:10:38   And we had lots of people write in from the pro audio world

00:10:42   and from some people even from like

00:10:44   the scientific equipment world, which is pretty cool.

00:10:46   We have a lot of pretty cool listeners.

00:10:47   Also confirming what I said last episode

00:10:49   that pro audio gear, and actually people wrote in to say

00:10:52   also pro film gear does similar things.

00:10:55   Rather than trying to like sync up

00:10:57   a bunch of different audio tracks afterwards

00:10:59   in software from different audio recording sources

00:11:02   that might be on a film set or in a recording studio,

00:11:05   they use the concept of a master clock.

00:11:07   And they have one device, whether it's a clock generator

00:11:10   or just a device that has an internal clock,

00:11:13   and ProGear has usually clock in and clock out ports

00:11:16   on the back of it.

00:11:17   And so they actually physically wire all to each other

00:11:21   and they coordinate the clock based on one master source

00:11:24   rather than each device keeping its own clock

00:11:26   and therefore introducing this drift.

00:11:27   So lots of people wrote in saying that was true.

00:11:29   And then the best feedback we got,

00:11:31   I very casually mentioned last episode

00:11:33   that I had just anecdotally found

00:11:35   that laptops generally have more drift

00:11:39   than like a Mac Pro or a desktop, or like an iMac.

00:11:43   And it turns out, there's some basis for this,

00:11:45   the quartz crystals that vibrate at particular frequencies

00:11:49   that create these clocks and these devices

00:11:51   that have to be so accurate,

00:11:53   they are very dependent on stable temperature.

00:11:55   And if they don't have a stable temperature

00:11:58   that they're operating at,

00:11:59   or if they're just two different ones

00:12:00   operating at two different temperatures,

00:12:02   that can cause these very slight differences.

00:12:06   And laptops, usually their internal components

00:12:09   usually operate hotter than desktops,

00:12:11   and they also fluctuate more.

00:12:13   And desktops have much bigger heat sink mechanisms,

00:12:17   they have usually larger fans that are spinning faster

00:12:19   and pushing more air over them,

00:12:20   so desktops tend to be cooled better.

00:12:23   And so I think that alone might explain

00:12:26   that little anecdotal thing that yes,

00:12:27   this is worse on laptops,

00:12:30   or when you have one desktop on one end

00:12:34   and the other person on the other end is using a laptop,

00:12:36   that might have more drift between them

00:12:37   than if both people were using desktops

00:12:39   or if both people were using laptops.

00:12:41   - Yeah, and this is related to the reason I was,

00:12:44   I put the original feedback in follow-up from like,

00:12:47   this seems weird to me.

00:12:48   You tell me we can't make quartz crystals that are accurate

00:12:50   I remember from you know the world of watches which I'm not that involved in but that you know the fancy mechanical watches obviously are

00:12:57   Terrible keeping time because they have a bunch of gears and stuff

00:12:59   But like you know a 10 cent quartz watch keeps amazing time for you know for an entire year

00:13:04   Maybe we lose a second

00:13:05   It's like so how can you make a stupid plastic quartz watch that keeps amazing time and it's very accurate over the course of an entire

00:13:11   Year, but we can't make a quartz crystal for our computer

00:13:14   That is equally accurate

00:13:15   And there are a bunch of factors that have to go into that having to do with the specifics of the quality of the crystals

00:13:21   for even for a 10 cent watch or whatever, but

00:13:23   one of our readers

00:13:25   One of our readers wrote in to tell us that the advantage a watch has is that it is essentially kept at an even temperature by

00:13:32   being next to your skin

00:13:33   So you are essentially

00:13:35   Temperature regulating the watch the entire time you're wearing it and that's the best thing for a quartz crystal actually in scenarios where they really want

00:13:41   Them to be stable they put them in a little everywhere

00:13:43   they called it, but they put them in a little device that keeps it at a stable temperature.

00:13:47   So having a watch and wearing it all the time keeps it accurate, much more accurate than

00:13:52   a laptop that the temperatures are going up and down and you're putting it to sleep and

00:13:56   you're playing a game and it's all over the map.

00:13:58   So this is all very explicable and now we should feel every time we learn anything about

00:14:04   the actual analog physical world of components inside your computer it makes you feel scared.

00:14:08   So now you can know that the things regulating the clocks are pieces of crap too and change

00:14:13   all the time, there's nothing you can do about it.

00:14:15   - Yeah, like whenever you learn anything

00:14:16   about the analog component world,

00:14:17   like you start, like we live in like this digital world

00:14:20   where we think everything is just,

00:14:22   like you know, a one is just on, and a zero is just off.

00:14:25   And of course, like when you get down to the analog level

00:14:27   of like the physical implementation of these chips

00:14:29   and these components, it isn't that simple,

00:14:32   and everything's kind of like tolerances and variances

00:14:35   and approximations and tricks, and it's like,

00:14:38   it's kind of amazing all the stuff we have works at all

00:14:41   as consistently as it does.

00:14:42   I mean, even just the clock, like you learned from like CPU design, even just propagating

00:14:46   the clock signal around the die of a large CPU is kind of sketchy at best.

00:14:50   And they all have all sorts of phase lock loops and other things to make sure that that

00:14:54   actually propagates everywhere and it's the same everywhere and you don't have delays

00:14:57   and yeah.

00:14:59   So anyway, I think we are now all completely satisfied that we understand why Marko's

00:15:03   tool is necessary in life.

00:15:05   Yep.

00:15:06   speaking of, in your quest to avoid doing actual work, you actually filed radar.

00:15:12   - Yes I did. - I'm very proud of you. So do you want to tell us about this?

00:15:17   - Okay, so last episode in the after show, I got my time in the sun or whatever the metaphor

00:15:26   is and whatever floats your bubble. And so I got to finally explain the details of the

00:15:34   MP3 file format and the efforts I was trying to do with variable bitrate or VBR encoding

00:15:41   and why podcasts couldn't practically use VBR encoding, which basically boils down to

00:15:47   the method to seek a VBR file that you don't have all of. So like if you are playing a

00:15:52   stream and the user jumps ahead to a timestamp that's way forward in the stream that you

00:15:57   haven't downloaded like the part of the file between those two points, you can't really

00:16:01   know exactly what byte position to jump to to get to timestamp one hour thirty without

00:16:07   using these lookup tables at the beginning of a VBR file that just don't have very

00:16:12   much precision in the current standards or in the old standard. They're very imprecise

00:16:16   and so over the course of like a two hour podcast, you only have in the most common

00:16:21   jump table format, you only have a hundred entries. So you have minute precision at best.

00:16:26   You know, it might even be less than a minute precision. I've learned a lot since then.

00:16:30   I did a lot of experimentation,

00:16:31   I talked to a bunch of people,

00:16:33   and I learned a lot about this.

00:16:36   So I was recommending the use of this,

00:16:39   there's an ID3 tag,

00:16:41   it's abbreviated MLLT for MPA location lookup table,

00:16:45   and this is basically an ID3 tag version

00:16:49   of the VBR offset jump table to tell you

00:16:52   which bytes map to which time stamp,

00:16:54   so you can jump between the file easily

00:16:55   without having to have download the whole thing

00:16:57   and just scan through manually.

00:16:59   The benefit of the ID3 tag is that it can be any size.

00:17:02   Like the VBR jump table thing is restricted to the size

00:17:06   of like the one MP3 frame that they shoved it into

00:17:09   for compatibility reasons, so it has to be like

00:17:12   basically below a kilobyte or so.

00:17:13   The ID3 tag can be any length you want it to be,

00:17:16   so the ID3 tag version of this would be great

00:17:19   because you can basically have arbitrary precision.

00:17:22   As much space as you're willing to devote to this,

00:17:24   and like for me, I was able to encode one second precision

00:17:28   of the time stamps with something like 16 kilobytes of a total jump table. So for like

00:17:33   a 45 megabyte podcast, 16K for the jump table is fine and by having VBR you're saving like

00:17:39   20 megs on the file size in that case so like it's totally worth the savings to embed this

00:17:45   little jump table. So I also heard from Devin Govitt. After recording last episode I discovered

00:17:51   a GitHub open source project called AudioCogs.

00:17:55   And it's a group of people who make JavaScript implementations

00:17:59   of decoders for MP3, AAC, a couple other audio formats,

00:18:03   and a whole audio player, all written in JavaScript,

00:18:06   that can decode and play these audio formats entirely

00:18:11   in JavaScript, and that support MLT tag, these proper VBR seek

00:18:16   methods, and if they didn't, it's

00:18:18   an easy way for me to add support to this.

00:18:20   'cause I could then have my site's web player

00:18:24   switch to this and then I would eliminate

00:18:26   a lot of this problem.

00:18:27   Anyway, so I heard from one of the authors of this,

00:18:30   Devin Govit, we went back and forth a few times.

00:18:33   Devin informed me that the Fraunhofer VBRI tag,

00:18:37   which is an alternative version of that stupid

00:18:40   100 entry jump table, is basically a more precise

00:18:44   version of that, it still has to fit within one MP3 frame,

00:18:47   but instead of being 100 bytes, it can be like,

00:18:50   1.3 kilobytes, so that's better.

00:18:54   You have more space, more precision.

00:18:57   So I did some experimentation with that.

00:18:59   Couple problems came up though.

00:19:00   First, I discovered in the experimentation

00:19:02   and going back and forth with Devin

00:19:04   that not only does Apple's decoder not support

00:19:08   the MLT or this VBRI tag that has the more precise version,

00:19:13   even that stupid little 100 byte version

00:19:16   in the Zing tag that we talked about last episode,

00:19:20   Apple doesn't use it.

00:19:21   Their decoder completely ignores it.

00:19:24   It does read these tags to get the duration of the files.

00:19:28   So it's parsing them,

00:19:30   'cause the duration's also one of the fields in these tags.

00:19:33   So it reads them for the duration,

00:19:34   and if you edit them in a hex editor

00:19:35   and you put any duration you want there,

00:19:37   it'll show up like in QuickLook

00:19:38   and everything has that duration.

00:19:38   So we know it's reading them,

00:19:40   but it completely ignores any of the entries

00:19:43   that are in these seek tables

00:19:44   that tell it which byte maps to which timestamp.

00:19:47   In any of these formats, it ignores all of them.

00:19:49   This is bad.

00:19:51   So I basically wrote up a bug report

00:19:53   and I emailed some people inside Apple to say like,

00:19:55   hey, here's this problem I'm having.

00:19:57   And if you guys would support really any of these formats,

00:20:02   except for maybe the stupid 100 byte one,

00:20:05   but if you'd support the Fraunhofer VBRI frame

00:20:08   or the awesome MLT ID3 tag,

00:20:12   either of those would provide usable precision

00:20:15   for a two hour podcast to be able to seek reliably within it

00:20:17   within one to 10 seconds of the desired point

00:20:20   as opposed to the stupid 100 byte one

00:20:22   which is like a minute off and their estimation

00:20:25   which could be any amount off really

00:20:27   and is frequently like 30 to 60 seconds off.

00:20:30   Anyway, so I emailed this around and I made a blog post.

00:20:32   I basically made as much noise as possible about this issue

00:20:36   because I've realized,

00:20:37   and sorry for the massive diversion here,

00:20:39   I like talking about like ranty Apple stuff on a podcast

00:20:43   where it doesn't get me in trouble

00:20:44   and where you guys can tame me a little bit and rebut me.

00:20:48   It's better for me to reserve my blog for issues

00:20:51   of maybe greater importance or more boring topics

00:20:56   or whatever else.

00:20:58   My blog is a great way to spread a message

00:21:01   to people who don't always follow me.

00:21:04   And that is often the problem.

00:21:07   Like when I get myself into hot water,

00:21:08   it's often because a whole bunch of people

00:21:10   who don't follow me and don't really get my context

00:21:14   are reading something that I didn't write very well

00:21:17   and where I assume people would get my context

00:21:19   and would understand me and they'd be the ones reading it

00:21:21   and that gets me into trouble.

00:21:23   But something like this, where I'm requesting

00:21:26   that Apple implement an esoteric standard

00:21:29   of the MP3 file format that's 20 years old,

00:21:32   that is really boring and doesn't spread

00:21:34   onto Business Insider or CNBC.

00:21:36   And there's not much in battle that's really controversial,

00:21:39   although Hacker News found some things,

00:21:40   but most people would not find this controversial.

00:21:43   So this is, I think, is a very good use of my blog,

00:21:46   and it kind of, it's helping inform me

00:21:48   like how I should use my various outlets going forward.

00:21:51   Anyway, I found a giant bug report with example files

00:21:54   and examples of why their way of just ignoring

00:21:57   these seek tables for long VBR MP3s was bad,

00:22:01   you know, why that's bad, how you can fix it.

00:22:03   That's it, nothing has happened yet on this front.

00:22:05   However, I have gotten rumblings here and there

00:22:08   that this bug report has traveled inside of Apple.

00:22:13   but that's all I know, and I don't know anything else

00:22:15   that's going on with it.

00:22:16   It is probably too late, even if Apple decided I was right

00:22:19   and they wanted to do this, it is almost certainly too late

00:22:22   to get it into iOS 10 or Mac OS Sierra,

00:22:27   but I would just love for this to happen sometime soon.

00:22:29   And the main reason why I can't do the audio cogs

00:22:34   and the Aurora JS, like the JavaScript version

00:22:38   that I was telling you about a minute ago,

00:22:39   the HTML5 audio element can fetch a file

00:22:42   from basically anywhere, but for JavaScript to fetch a file

00:22:45   you run into all these CORS issues.

00:22:47   And you also, you can't cross from HTTPS to HTTP,

00:22:52   and all these, like all these restrictions of like,

00:22:54   you know, for various web security purposes,

00:22:57   all these restrictions on what JavaScript

00:22:59   is allowed to fetch from.

00:23:01   And for me to be able to play podcasts

00:23:04   from arbitrary podcast hosts,

00:23:06   I basically can't use this thing.

00:23:07   I could run a proxy, but if I do that,

00:23:10   then the podcasters don't get unique hit information really,

00:23:14   so that's no good.

00:23:16   So there's a whole bunch of crappy reasons why,

00:23:20   in practice I can't use the JavaScript approach

00:23:22   of just decoding the whole file in JavaScript

00:23:24   and basically then running my own decoders.

00:23:27   So I'm reliant on Apple, I have to,

00:23:29   if podcast VBR is ever gonna happen,

00:23:32   Apple has to build it into their decoders, that's it.

00:23:35   - Well, but hold on, so a couple of questions here.

00:23:37   First of all, let's assume you did desire to proxy it,

00:23:42   because this was that important to you for the web client.

00:23:45   Why couldn't you proxy each request individually?

00:23:48   I mean, it's a crud load of bandwidth,

00:23:50   but is there any other reason why you couldn't do that?

00:23:53   So at least all of these hits are unique,

00:23:55   they're just coming from you

00:23:56   and not the actual person that's asking for them?

00:23:59   - The main problem with that is that

00:24:01   even if you architected the proxy such that

00:24:04   every inbound request equaled

00:24:06   one back end outbound request.

00:24:08   - Which would be silly in general,

00:24:09   but in this case may make sense.

00:24:11   - Right, so even if you did that,

00:24:13   the request would still appear to be all coming

00:24:15   from one IP address.

00:24:17   - Sure.

00:24:17   - So, and most, it's kind of a highly debated topic

00:24:20   in the podcast world of like what counts as a download.

00:24:24   You know, 'cause there's, as with most things,

00:24:26   it's complicated, and so you can't just say,

00:24:29   oh, well every hit you get, nope, doesn't work that way.

00:24:31   Because that's, yeah, a hit doesn't match up

00:24:34   to a listener necessarily.

00:24:35   So anyway, most podcast hosts have their own idea

00:24:39   of what a download should count as,

00:24:41   and most of them involve the IP address

00:24:43   of the source in some way.

00:24:45   So it might be like, you know,

00:24:47   a unique download is like one IP,

00:24:49   or like the IP has to be unique within a certain

00:24:51   time interval for it to count as a unique download

00:24:54   or something like that.

00:24:54   Like there's all these different tricks that people do,

00:24:57   but basically if all your requests came from my single IP

00:25:00   that was running the proxy, it would not,

00:25:03   people's stats would under count.

00:25:06   That being said, I actually already wrote

00:25:09   and run one such proxy.

00:25:11   I already have this, that I designed exactly this way

00:25:14   and that has exactly this problem.

00:25:16   And the only reason that it's not really a problem

00:25:20   that it hasn't gotten me into any hot water with anybody

00:25:22   is because it's hardly ever used.

00:25:25   And I designed it because Overcast has a Twitter card.

00:25:28   On any share link I have Twitter cards.

00:25:30   And so if you view a tweet with an Overcast link

00:25:33   on Twitter's website or in any other clients

00:25:35   that support their cards,

00:25:36   which is very few of them in practice.

00:25:38   I show a whole little overcast embedded player

00:25:40   and it works and it's great.

00:25:41   You can stream, you can do timestamps,

00:25:43   it works perfectly, it's just like the website but tiny.

00:25:45   Twitter's cards require that all assets loaded through them

00:25:49   must be served over HTTPS.

00:25:52   So I actually have kind of like a little function,

00:25:55   like a mapping that many big podcast hosts,

00:25:58   including Libsyn, our host,

00:26:00   Many of them have just a simple way

00:26:03   that you can transform their insecure URLs

00:26:07   into HTTPS URLs with a simple string replacement

00:26:10   on certain things and everything.

00:26:12   So some hosts I can just redirect,

00:26:14   I can do a straight redirect and it's fine,

00:26:16   but not all hosts.

00:26:17   So I actually run this proxy that does this,

00:26:21   that follows redirects,

00:26:22   and when it's a host that it knows about,

00:26:24   it can send you along,

00:26:26   and when it's not, it does that proxy

00:26:28   of one-to-one connection mapping exactly the way you'd think.

00:26:31   And the only reason it isn't a problem

00:26:32   for either my bandwidth cost or people getting mad at me

00:26:35   is that the cards get pretty low usage, relatively speaking.

00:26:39   Not a lot of people play podcasts that way,

00:26:41   as far as I know.

00:26:42   So that's the only reason that works.

00:26:44   And to answer Glass_ in the chat room,

00:26:48   who said send them something like the X forwarded for header,

00:26:53   that I do send X forwarded for,

00:26:55   that is a header that proxies use

00:26:57   to tell what they're fetching from the IP

00:27:00   of the person fetching it from them.

00:27:02   So it's kind of a way to forward the source IP.

00:27:05   Problem is you can't trust that.

00:27:07   So if you have a podcast metrics thing

00:27:09   that is trying to measure unique IPs in an honest fashion,

00:27:13   you can't really trust if somebody sends you

00:27:16   an X forwarded four header and you get a whole bunch

00:27:18   of requests from one real IP, but that IP says,

00:27:22   oh, I'm actually forwarding you these requests

00:27:23   from these other 10,000 IPs.

00:27:25   You really can't trust that for your purposes of stats that you tell sponsors.

00:27:30   Our first sponsor this week is Betterment.

00:27:33   Betterment is the largest independent automated investing service out there.

00:27:36   You've probably been hearing a lot about Betterment in TechCrunch, the Wall Street Journal, and

00:27:39   other major news outlets because Betterment makes it easier, more straightforward, and

00:27:43   less expensive to invest.

00:27:46   Betterment is built on smarter, cutting-edge technology to bring you sophisticated investing

00:27:49   and financial advice all at a much lower cost than more traditional financial services.

00:27:55   More than 175,000 customers choose Betterment's advanced advice algorithms and beautiful user

00:27:59   interface to manage over $5 billion of their customers' dollars.

00:28:03   This is the perfect time to get started with Betterment and to get started saving for your

00:28:06   retirement or your other financial goals.

00:28:08   ATP listeners can get up to six months of no fees.

00:28:12   Now their regular fees are already very low, way lower than what you'd think for most

00:28:15   financial advisors.

00:28:16   Up to six months of no fees by going to betterment.com/ATP.

00:28:21   That's betterment.com/ATP.

00:28:24   investing made better.

00:28:26   (upbeat music)

00:28:29   - So you had said before,

00:28:32   well if Apple just implements this,

00:28:33   then you can start using it on the web

00:28:37   because the HTML5 audio tag will support it,

00:28:40   et cetera, et cetera.

00:28:41   But what about your beloved Windows users?

00:28:44   Or next year's gonna be the year of Linux on the desktop,

00:28:48   so what about them?

00:28:49   - First of all, we don't have that many of them, honestly.

00:28:52   like not a lot of people use Windows and use my app

00:28:56   or follow the share links generated by my app

00:29:00   or anything else.

00:29:01   I would love to change that because there are a lot

00:29:02   of Windows users out there.

00:29:03   I would love for more Windows people to see the share links

00:29:06   from Overcast 'cause that would mean that the share links

00:29:08   are being spread far and wide and that's the whole point

00:29:10   of the share links.

00:29:12   But reality is different.

00:29:13   Reality is that they really don't get much Windows usage

00:29:15   at all.

00:29:17   So that's problem number one or rather dodge number one

00:29:20   where I can dodge this issue.

00:29:21   (laughing)

00:29:23   Same reason why I never test my websites in IE

00:29:25   because my audience and my customer bases

00:29:28   tend to have such incredibly low,

00:29:30   or whatever IE is called now, what is it called now?

00:29:33   - Edge.

00:29:34   - Edge.

00:29:35   (upbeat music)

00:29:36   - The edge.

00:29:37   (laughing)

00:29:39   - Anyway, the main reason why I don't really need

00:29:43   to care that much about this problem

00:29:45   for other browsers other than things

00:29:47   that use Apple's built-in decoders and my own app

00:29:50   is because it's only a problem for seeking to timestamps

00:29:55   that haven't been downloaded.

00:29:57   That's it.

00:29:58   In all other ways, VBR MP3s work great already.

00:30:01   - Don't you have the same problem

00:30:02   even if they added support for it?

00:30:04   That it doesn't matter because you have to wait

00:30:06   for everyone to upgrade to an operating system

00:30:08   that includes that support that Apple just added.

00:30:11   - In practice, yes, but the amount of time I have to wait

00:30:14   isn't that much, especially for podcast listeners

00:30:17   to a tech show.

00:30:18   Like, that's a pretty upgradey group.

00:30:22   We don't have to wait that long.

00:30:25   - Even on the Mac?

00:30:26   Like, iOS probably, like, their upgrade curves

00:30:28   are always pretty good, but the Mac upgrade curves

00:30:30   have not been great.

00:30:31   Like, you'd wait a year and it'd be like 50% adoption,

00:30:33   which is better than Windows, but still.

00:30:35   - For people who would see, 'cause again,

00:30:38   I don't have to worry about users of my app.

00:30:40   I control my app, I can update my app.

00:30:42   I have to worry about people who are gonna see

00:30:43   the share links.

00:30:44   You know, right now the share links

00:30:46   don't have big audiences.

00:30:47   They get used some, and some people see them,

00:30:49   and that's great.

00:30:50   I want to make them bigger,

00:30:51   because I think podcast sharing could use help,

00:30:54   and you know, I mean everybody in podcasting

00:30:55   thinks podcast sharing can use help.

00:30:57   But like, here I am, like actually,

00:30:59   I actually have something that does help,

00:31:00   and I want it to get bigger and better.

00:31:02   The reality is, even if, you know,

00:31:06   if Apple had support, suppose it makes it

00:31:07   into the next major versions of OS X and iOS, whatever,

00:31:10   you know, not the ones coming out in a few weeks or whatever,

00:31:12   but next year or in the spring or whatever else.

00:31:15   It'll probably be about a year after that

00:31:17   before I can reliably use it, because at that point,

00:31:22   you know, it's gonna be way more than half,

00:31:23   because first of all, the people who are seeing share links

00:31:28   from social media are gonna be way more likely

00:31:30   to be on mobile than on a desktop.

00:31:32   So the slower adoption curve on Mac OS

00:31:35   is not going to be as big of a problem

00:31:36   because it's gonna be offset by the large proportion

00:31:39   of those viewers who are gonna be on iOS and Android,

00:31:44   if anybody can tolerate me who's using Android,

00:31:47   which is a big ask.

00:31:48   So I think iOS will be updated faster,

00:31:53   especially as Apple has figured out how to make people

00:31:55   upgrade by adding emoji and messages tricks.

00:31:57   (laughing)

00:31:58   So again, I think give it a year after they add support,

00:32:02   and then I think almost anybody can responsibly do this,

00:32:05   and it would be fine.

00:32:08   Podcast listening, if anybody doesn't know,

00:32:11   Rob Walks, the CEO of Libsyn,

00:32:13   Libsyn is a massive podcast host.

00:32:16   They've been around forever and they host a ton of podcasts

00:32:19   including this one and if you listen to this,

00:32:22   they probably host a lot of other podcasts you listen to.

00:32:24   I'm pretty sure all of the Relay FM shows are hosted there.

00:32:27   It's a very big podcast host.

00:32:31   And they do a podcast, I forget what it's called.

00:32:36   I'll put it in the show notes.

00:32:38   They do a podcast where it's like tips for podcasters.

00:32:41   And then something like once a month,

00:32:42   the CEO of Libsyn goes on there and gives stats

00:32:46   of like podcast user agents, like what clients,

00:32:49   what platforms are downloading podcasts.

00:32:51   And Libsyn is a pretty good source of this

00:32:54   because they host so many podcasts across so many markets

00:32:58   that it isn't just like tech heavy.

00:32:59   Like any stats I can give you on like

00:33:01   how people use Overcast is like how Apple nerds

00:33:04   who know who I am mostly use Overcast.

00:33:06   like the top podcast and overcast are mostly tech shows.

00:33:10   They're not, but if you look at the top podcast

00:33:12   in the world, it's mostly not tech shows.

00:33:14   So obviously, overcast usage is not representative

00:33:17   of all podcasters out there.

00:33:19   But Libsyn's stats are really close, I think.

00:33:21   I would say Libsyn's stats are probably

00:33:23   the best representation we have outside of Apple,

00:33:26   and they're not talking of how podcasters behave in mass,

00:33:30   what the overall market looks like.

00:33:32   And the stats they give, I listen to this

00:33:35   take notes every month.

00:33:37   And so here, let me just pull this up here.

00:33:39   So the most recent ones were stats for June.

00:33:41   77% of listens were on mobile devices.

00:33:45   And every month that number increases.

00:33:47   So already, very mobile heavy.

00:33:50   iOS to Android is a little over three to one.

00:33:54   And that ratio's going down.

00:33:55   Android is becoming more popular now

00:33:57   because people are finally building in,

00:34:00   like Android vendors like Samsung

00:34:01   are finally building in podcast clients.

00:34:03   They weren't for a long time.

00:34:04   and so that's adding a lot to the Android side.

00:34:08   But so basically, you still have like more than three times

00:34:12   of any people listening on iOS than Android.

00:34:15   And only like 22% of listeners listening on computers.

00:34:20   So this is a very mobile heavy market

00:34:23   and it's a very iOS heavy market.

00:34:25   So that's why I think it's fairly responsible

00:34:28   for a lot of podcasters, especially if you are in like

00:34:31   the Apple tech world like we are,

00:34:33   where your audience is gonna be even more skewed

00:34:35   towards recent Apple platforms.

00:34:38   I think one year after they add support to this,

00:34:40   it's totally safe to do, and I might even do it sooner.

00:34:43   - So do you feel optimistic about them

00:34:45   actually adding support, or do you feel like

00:34:47   it's gonna be two years before this bug is closed

00:34:49   as it behaves correctly?

00:34:50   - When I filed this bug in my bug reporter,

00:34:54   it was next to two other bugs that have been open forever.

00:34:58   I'm lucky if I get a response.

00:35:02   I don't file that many bugs with Apple

00:35:04   because the effort on my side to feedback

00:35:07   and potential benefit I will get from that

00:35:11   ratio is so terrible.

00:35:13   Like, you know, this took the better part of two days

00:35:16   to do research properly to like figure out

00:35:19   that Apple didn't even support these tags at all.

00:35:20   You know, I thought they were supporting

00:35:21   at least the crappy one.

00:35:22   They supported none of them.

00:35:24   And you know, to modify my, these test files,

00:35:27   to like, you know, write these tags properly

00:35:29   and everything, took a lot of work.

00:35:30   to write a bug report, give proper test coverage

00:35:33   and everything, it takes a lot of time, a lot of work.

00:35:36   And usually I file bugs and they go nowhere.

00:35:39   If I'm lucky, they might be closed as a duplicate.

00:35:43   And when they're closed as a duplicate,

00:35:44   I believe I then lose any visibility on that.

00:35:48   I forget, I think they changed it recently, anyway.

00:35:50   So the feedback loop is terrible for filing bugs for Apple.

00:35:55   People who do it are good people.

00:35:58   they probably floss, they're probably very good people.

00:36:02   I have a hard time doing it most of the time

00:36:04   because again, I see what happens with the few bugs

00:36:07   I do file and they sit around forever

00:36:10   and they don't even get closed.

00:36:12   And the few that do get closed are often closed

00:36:16   in a way that I consider invalid.

00:36:18   They'll do something like, we're gonna close this

00:36:23   unless you tell us really soon that this is still happening

00:36:25   on the newest build of iOS.

00:36:27   for a bug that I filed like six months ago

00:36:29   that is easily testable.

00:36:30   Like, they clearly have a wide variety of quality of people

00:36:35   who go through the bug reports,

00:36:38   and many of them clearly just want to close

00:36:39   as many as possible without actually doing any work.

00:36:41   Like, the incentives are wrong there.

00:36:43   - Are you saying you don't floss?

00:36:45   (laughing)

00:36:47   - So anyway, I usually don't file bugs

00:36:50   because I've had a poor history of any response

00:36:54   from Apple on bugs, but this time I filed it

00:36:56   because I figured basically there's nothing else I can do.

00:36:58   I can't work around this.

00:36:59   Like, I mentioned the JavaScript thing.

00:37:01   I tried doing the JavaScript thing,

00:37:02   but because of all the CORS restrictions,

00:37:05   it makes it pretty much impossible to use

00:37:08   for an arbitrary set of hosts that you don't control,

00:37:12   most of which don't send CORS permission headers already.

00:37:15   So that's a non-starter.

00:37:17   I could write my own decoder in my app

00:37:20   that calculates these offsets, but that's only one app.

00:37:24   And so to make something, like even our podcast,

00:37:28   that you would think, you know, what percentage of people

00:37:31   who download our podcast do you think listen in Overcast?

00:37:34   And a lot of people would probably guess

00:37:36   it's a pretty high ratio, and it is,

00:37:37   compared to Overcast's global ratio.

00:37:39   I haven't looked recently,

00:37:40   but I think it's something like 60%.

00:37:42   That's still 40% of our listeners

00:37:44   who don't listen in Overcast who wouldn't have,

00:37:47   who would need compatibility in this way,

00:37:49   you know, for seeking in streams.

00:37:51   I basically, the reason I'm following this bug

00:37:53   is that it's my last hope to get the VR MP3s to be a thing.

00:37:56   I think they'd be great if they were a thing.

00:37:57   I think it's ridiculous that they aren't a thing yet,

00:38:00   but if there's any hope of them being a thing,

00:38:02   it's basically on Apple to do.

00:38:04   'Cause if Apple doesn't do it,

00:38:05   nobody else can and nobody else will.

00:38:08   - Did you make a strong recommendation

00:38:09   that they implement a specific one of these?

00:38:11   'Cause if you just said,

00:38:12   "Oh, here are all these things you don't support,"

00:38:13   I can imagine them supporting that terrible one

00:38:15   that only gives you 100 points,

00:38:17   and then being like, "Done, closed, fixed."

00:38:19   - Yeah, no, my bug report actually reads

00:38:21   a lot like the blog post.

00:38:22   I wrote the bug report first and then edited it

00:38:24   to be a blog post, but basically it's a similar format.

00:38:28   And I recommended, I gave the three options and I said,

00:38:33   you don't parse any of these right now.

00:38:34   This old one is not precise enough, don't use this.

00:38:37   So I said basically the MLT tag would be the ideal one to do

00:38:42   because it can be arbitrary length and arbitrary precision.

00:38:44   So that's the ideal one.

00:38:46   If you only pick one, pick that one.

00:38:49   The Fraunhofer VBRI tag, that's like 1.3 kilobytes

00:38:52   worth of stuff, like that's a good second choice.

00:38:54   Like if you do that, I'll be happy,

00:38:56   but the best one to do would be the MLT tag.

00:38:59   - I wonder if they don't support it,

00:39:01   not out of laziness, because like you said,

00:39:02   they are actually parsing the durations out of them,

00:39:04   but kind of for the same reason related to something else

00:39:07   you said, you know, why web hosts can't trust X forwarded for.

00:39:10   Maybe they're afraid it's gonna be filled with garbage data,

00:39:13   and then they're afraid to like expose exploits

00:39:15   in their like a buffer overflow or something,

00:39:17   they put crazy offsets in there and then trigger a bug. Who knows? I don't know. The type of thing

00:39:22   where it is, or they just don't trust the encoders to put good data in there, they don't want to try

00:39:26   to read garbage data. These are all solvable problems, like, you know, make your thing not

00:39:31   have silly buffer overflows and sanity check the offsets in the map and make sure they seem

00:39:38   reasonable before blindly following them and maybe disregard if it looks like line noise. But

00:39:43   I do wonder, because someone did have to write the code

00:39:46   to pull these durations out of there.

00:39:47   Why did they do that?

00:39:48   And they're like, "Well, I'm in there.

00:39:49   "Why don't I just actually parse this whole format

00:39:52   "and just implement it?"

00:39:53   - So, I mean, your concerns are totally valid.

00:39:56   And I can completely understand how,

00:39:59   it wouldn't surprise me at all

00:40:00   if there was an engineering meeting at Apple

00:40:02   some ridiculous amount of time ago.

00:40:05   Somebody was like, "Yeah, you know what?

00:40:06   "We could parse these things out,

00:40:08   "but the data might be wrong,

00:40:10   "and if we just do this percentage of time

00:40:13   to byte offset thing that we,

00:40:15   kind of like a dumb approximation,

00:40:16   like that'll be close enough and it'll be consistent.

00:40:20   And so there is a school of thought that says

00:40:24   you should override the stupid MP3 encoders

00:40:27   'cause who knows what garbage you're gonna get there

00:40:29   and you should just do your thing

00:40:31   and just make it so that it works.

00:40:34   It's kind of close for short songs

00:40:36   and then it doesn't matter anymore.

00:40:39   - It's good enough for scrubber work.

00:40:40   Like they want the scrubber,

00:40:41   'cause when you're moving a scrubber around anyway,

00:40:43   it's probably like an inch on some webpage

00:40:44   and you can't tell if it's like what pixel it's on

00:40:46   or whatever, but for your specific use case,

00:40:48   which is no, no, no, someone's not dragging a scrubber here,

00:40:51   I'm putting an offset in the URL down to the second

00:40:53   and I want that precision out of it.

00:40:55   Say, hey, YouTube does it, you should do it too.

00:40:57   - Exactly, yeah.

00:40:59   So there is an argument to be made there,

00:41:02   but A, they're already reading the duration of these tags.

00:41:05   Even if they have the whole file,

00:41:07   they still read the duration from that header.

00:41:09   And so if you put a garbage duration,

00:41:10   as I did during testing, it says, all right, yeah, sure,

00:41:12   this file is nine minutes long instead of an hour.

00:41:15   It's happy to use that value,

00:41:17   so it's already trusting it on some level.

00:41:19   Also, MP3 encoders don't change that much.

00:41:22   There's very few that are actually in active use today,

00:41:25   and they all write good data.

00:41:27   This isn't 1997 anymore.

00:41:29   We have solved the MP3 encoder problem.

00:41:32   MP3 encoders work, and they work well,

00:41:34   and no one else besides me is going in with a hex editor

00:41:38   and messing with these values.

00:41:40   you're not gonna get total garbage on a regular basis.

00:41:43   So I think what's more likely to have happened

00:41:47   is that maybe these decoders were written

00:41:49   a very long time ago, maybe around that timeframe,

00:41:52   when the world of MP3 was still very much in flux

00:41:55   and you had crappy encoders in the late '90s,

00:41:59   and maybe they just haven't revisited it since then

00:42:00   because there wasn't a reason to.

00:42:02   That is the way more likely explanation for this,

00:42:04   is like, this is very old code

00:42:07   that no one has had any justification to touch

00:42:08   for a long time, it works well enough, so fine.

00:42:12   And that's why I think ultimately

00:42:14   it probably won't get done, because I would be surprised

00:42:17   if anybody was really motivated to devote,

00:42:20   people have limited time, so who's gonna devote time to this

00:42:25   out of their engineering time budget inside Apple

00:42:28   when they might have more pressing things to worry about

00:42:31   with supporting the new Bluetooth headphones

00:42:33   on the next iPhone?

00:42:34   If anything else these people could be working on

00:42:38   is probably more important than this

00:42:39   to Apple's overall corporate goals.

00:42:41   - You're just supposed to be using AAC anyway,

00:42:44   it's the future.

00:42:44   Just use a different container format

00:42:46   and those offset things can be all at the front of the file.

00:42:50   - Yeah.

00:42:51   - I love that Hacker News thread though.

00:42:55   Just use a different container format,

00:42:56   doesn't that solve all your problems?

00:42:58   - Yeah, that was the best, yeah.

00:42:59   - All the existing container formats are bad,

00:43:00   but just invent your own,

00:43:01   doesn't that solve all your problems?

00:43:02   And then I guess make every encoder on the planet

00:43:05   understand your new container format, it'll be fine.

00:43:08   Yeah, this has been a lot of time on this topic.

00:43:09   I'm sorry to everybody who doesn't care.

00:43:11   You probably tuned out long ago,

00:43:13   but we actually got a decent response on it.

00:43:15   People actually enjoyed hearing about all this crap.

00:43:17   I'm very surprised by that, actually.

00:43:19   But I guess our listeners are both cooler and geekier

00:43:23   than I would've assumed.

00:43:25   - They are definitely cooler.

00:43:27   Geekier, I'm not sure.

00:43:28   - We're also sponsored this week by Tracker.

00:43:33   Tracker makes losing things a thing of the past.

00:43:35   Go to thetracker.com right now.

00:43:36   that's thetracker.com and enter promo code ATP

00:43:39   for 30% off your entire order.

00:43:41   Now, Tracker is a coin-sized device

00:43:44   that locates misplaced keys, wallets,

00:43:46   bad computers, anything you want.

00:43:47   It's this little thing you just attach it to anything

00:43:49   in seconds.

00:43:50   You just pair Tracker to your phone

00:43:52   and you attach it to anything.

00:43:53   It communicates over Bluetooth

00:43:55   and you can find its precise location

00:43:57   with a tap of a button.

00:43:58   So here's how this works.

00:43:59   The Tracker itself, this coin-shaped thing,

00:44:00   it has on it a button.

00:44:02   And so when you push that button,

00:44:04   you can locate your phone.

00:44:06   So it works that direction.

00:44:08   You can also launch the app on your phone

00:44:11   to locate whatever you have attached the tracker to.

00:44:14   So it's great for anything you lose,

00:44:16   your keys, you know, keys are a big one obviously.

00:44:18   You know, your backpack maybe, like your purse,

00:44:21   bags, computers, whatever else, things you have to find,

00:44:24   objects you have to find in your home

00:44:26   or around the workplace, whatever else,

00:44:28   attach a tracker to it, you can find it very, very easily

00:44:30   with your phone.

00:44:31   And then you can also, as I said,

00:44:32   use the tracker in reverse, push the button on the tracker,

00:44:35   and your phone makes a beep sound,

00:44:37   even if the speaker is switched to silent mode.

00:44:39   So pretty great.

00:44:41   With over 1.5 million devices,

00:44:43   Tracker has the largest crowd GPS network in the world.

00:44:46   So your lost item actually shows up on a map,

00:44:48   even if it's miles away.

00:44:49   Never lose anything again with Tracker.

00:44:51   Listeners of our show, get a special discount

00:44:53   of 30% off their entire order.

00:44:55   Go to the, that's T-H-E, tracker.com, thetracker.com,

00:45:00   enter promo code ATP.

00:45:02   The hardest thing you'll ever have to find is their website.

00:45:04   then you'll be done.

00:45:05   Thetracker.com/ATP for 30% off your entire order.

00:45:09   Thanks a lot to Tracker for sponsoring our show.

00:45:11   (upbeat music)

00:45:14   - So John, if the cell signal at your house is so crappy,

00:45:18   why not enable Wi-Fi calling and solve all your problems?

00:45:21   - The first reason I didn't enable this is

00:45:23   I didn't think my carrier supported it,

00:45:25   but I was just looking at the wrong place in settings.

00:45:28   For some reason I had it in my head

00:45:29   that only AT&T supported this and not Verizon,

00:45:31   but now Verizon supports it too.

00:45:33   Of course I've now forgotten where the setting is, but anyway it's in there somewhere, you

00:45:36   can find it.

00:45:38   But the second thing is when I turned it on I'm like, "Oh okay, I'll turn this on and

00:45:41   give it a try."

00:45:43   It makes you enter your physical address because once you're on Wi-Fi calling, if you call

00:45:48   911, 911 can immediately tell where you are because you're coming from the internet essentially

00:45:53   and they can't get a location for you.

00:45:55   Which is kind of weird because you can get locations from Wi-Fi base stations and everything,

00:46:00   a lot of things use that big map of base station MAC addresses to physical locations and stuff

00:46:06   to help aid location awareness.

00:46:08   But anyway, what iOS says is please enter a physical location so when you call 911 if

00:46:14   you're too injured just speaking to the phone or whatever, don't worry, they will come to

00:46:18   this address.

00:46:20   But of course that means that if you go someplace else and you're on wifi, say you're at work

00:46:24   and you're on wifi, and you call 911 and don't get a chance to tell them where you are, they're

00:46:28   going to go to your house.

00:46:29   my understanding anyway of what this message is telling me.

00:46:32   Enter this address here, if you call 911 this is where people will go to.

00:46:37   But I didn't want to enable it because I'm like well then I have to remember to turn

00:46:41   off Wi-Fi calling when I leave my house because then if I call 911 people will go to my house

00:46:46   instead of where I am.

00:46:48   So that's one reason why I didn't leave it on.

00:46:51   And the second reason is, this is the stupidest reason, that reason I think is only vaguely

00:46:55   silly.

00:46:56   this reason is really dumb but nevertheless is a reason everyone has

00:47:00   the reasons this is mine it changes the thing that appears in the status bar to

00:47:04   make some ugly thing that says like VZW whatever it doesn't say like the Verizon

00:47:09   a little fit Wi-Fi fan symbol it says different words like in all caps and it

00:47:13   looks ugly I don't like it that's a reason like I can understand even though

00:47:19   I think it's a little ridiculous the whole address thing fine whatever but

00:47:23   But because you don't want to look at VZW in your status bar, that's your reason?

00:47:28   You try it, turn it on and see what it does to your status bar.

00:47:31   I use AT&T, like a gentleman, and so it says AT&T Wi-Fi.

00:47:34   But it gets rid of the fan thing, doesn't it?

00:47:36   Or maybe it doesn't.

00:47:37   Nope.

00:47:38   Anyway, it changes what's in the status bar.

00:47:39   I didn't like it.

00:47:40   The location one is the larger reason, but I was kind of glad that I had a legit reason,

00:47:44   so I didn't need to look at that ugly status bar anymore.

00:47:46   It would be nice if like an iOS enhancement would be, an iOS enhancement would be, "Only

00:47:51   use Wi-Fi calling when connected to this base station."

00:47:53   I would like that setting.

00:47:54   You know what I mean?

00:47:55   Because then I could say, "Use Wi-Fi calling when connected to my home base station," but

00:47:58   never anyplace else.

00:47:59   And then I don't have to worry about this.

00:48:00   Then I could set my home address to my home base station, but no.

00:48:04   And anyway, after I made the decision, I made a couple of successful voice calls from my

00:48:09   house where people could actually hear my voice.

00:48:11   So maybe Verizon loves me for coming back to them and not using this filthy internet

00:48:16   calling.

00:48:17   Wow.

00:48:18   Okay.

00:48:19   All right.

00:48:20   Well, I don't even know what to make of that.

00:48:23   - Yeah, that's. (laughs)

00:48:26   - Moving on.

00:48:28   Graham Spencer has indicated that Google and Facebook

00:48:31   and others also do a charity match on bug bounties.

00:48:34   So just FYI.

00:48:36   - And did it before Apple.

00:48:38   Most people didn't write it into Gloat

00:48:40   about how all these other companies did it before Apple,

00:48:42   but they totally did.

00:48:43   - Do you have any actual topics tonight

00:48:47   or is this all follow-up?

00:48:48   - We're getting there.

00:48:49   - There's topics down there, believe me.

00:48:51   Don't worry, we got another podcast

00:48:53   two days. I know. Yeah, that's true too. And somebody had to go all deep on the MP3 stuff,

00:48:58   but actually it's very entertaining and I enjoyed it, so I shouldn't give you a hard time anyway.

00:49:02   Imagine how much dead space we'd have in like the topics and everything this week if I didn't do

00:49:06   that. Fair. You don't know what's lurking down there in topics. There's all sorts of stuff.

00:49:10   Is there more about Teemo? Yes, there is actually. Really? Yes, I'm so excited. Oh, that's fantastic.

00:49:16   I'm so overjoyed. All right, well, hopefully we'll run long enough that we won't get there. I mean,

00:49:21   I mean, anyway, let me tell you about my iMac.

00:49:24   It has been rebooted.

00:49:26   I did it on purpose.

00:49:29   It ran 21 days on the stock RAM.

00:49:33   - And without a UPS.

00:49:34   - And without a UPS.

00:49:35   I've gotten in contact with Mac sales, OWC,

00:49:38   whatever they call themselves,

00:49:41   and have requested and am told that I'm receiving a RMA,

00:49:45   a Return Merchandise Authorization,

00:49:47   as I think that that stands for.

00:49:48   So I think what they're doing is sending me a box

00:49:50   send the RAM back in and they said they will replace it post-haste. I have plugged it into

00:49:56   the UPS, I have pulled the UPS power and seen the iMac stay on for at least long enough to make me

00:50:03   feel better about myself. I did not leave it disconnected long enough to see whether or not

00:50:09   the Synology would be smart enough to shut itself down, but I believe I have those settings squared

00:50:13   away in the Synology so it should shut itself down gracefully. The iMac won't, but you know I'm used

00:50:18   I'm used to that at this point anyway.

00:50:20   Hey-o!

00:50:21   So anyway, I am running still on the eight gigs

00:50:25   of stock RAM until I receive my new batch of OWC RAM.

00:50:29   I will try that.

00:50:32   If that doesn't work, then I will probably,

00:50:34   very politely but very sternly, ask for my money back

00:50:36   and get Crucial RAM, which the entire internet

00:50:39   has written to tell me is the only RAM

00:50:40   I should ever really buy.

00:50:42   - I don't think Crucial buys any different RAM than OWC.

00:50:45   I forget, someone at one point sent me a long email

00:50:47   me about the different bins of who buys the good chips versus the cheaper chips and i think crucial

00:50:52   was in the same bin with owc but anyway i've bought owc ram for years and like i said i think

00:50:57   the last time i had owc thing go bad the chip because my computer is ancient the chip the the

00:51:03   the ram dim was like four years old five years old maybe it was six years old like whatever it is i

00:51:09   just i just assume like well it's so old whatever warranty or whatever they had it must be completely

00:51:14   gone by now but I just called them up and they said I will send you a new one.

00:51:17   That's it. Like I don't know if they have like a forever lifetime if this RAM

00:51:20   ever goes bad they'll just replace it forever and ever and I know it's a

00:51:23   hassle to return if you have to keep returning it it's a pain but I've had

00:51:26   them last for years and years and then go bad years later and they just send me

00:51:30   a new one. I think it's like miraculous like try doing that with a

00:51:34   hard drive for example hey my hard drive died after six years can I get a new one?

00:51:37   Ha! No I mean like almost all RAM has a lifetime guarantee like almost all RAM

00:51:42   and that you would buy separately from a computer yourself.

00:51:44   They almost all have lifetime guarantees,

00:51:47   but it's up to the retailer or the manufacturer

00:51:49   to make that a good or bad experience

00:51:52   if you actually have to claim it.

00:51:53   And I too have had only good experiences with OWC RAM.

00:51:57   I don't use it anymore, now I just buy my computers

00:51:59   with enough RAM from Apple,

00:52:00   'cause I'd rather just not deal with it anymore.

00:52:02   And the OWC RAM, when I was using it,

00:52:05   which was up until like two years ago, was always great.

00:52:07   I had to return it one time,

00:52:10   And that was one time in something like eight years

00:52:14   of using it, I think it's pretty good.

00:52:16   - Yeah, and I mean, they've been really good about it

00:52:18   so far, I didn't expressly reach out to them,

00:52:20   they reached out to me via Twitter,

00:52:22   I'm not entirely clear how they caught wind

00:52:24   of the fact that I was having issues,

00:52:25   but they just basically said, "Hey, can you send us

00:52:29   "an email and we'll talk about it?"

00:52:30   And I said, "Hey, here's the situation."

00:52:32   And I spelled it out and said,

00:52:36   it had been rebooting every week, look at this,

00:52:38   now running for 20 some days at this point. I feel like it's the RAM. So they said, "Yep,

00:52:44   you're probably right. I will get you an RMA and we will replace it immediately."

00:52:49   I don't know how they heard about this. I just told 80,000 people about it for a month

00:52:54   and somebody told them. Well, but what I mean is I don't recall having seen anyone mention

00:53:00   me and them on Twitter. Maybe that did happen and I just missed it, but it seemed like they

00:53:05   came out of the woodwork as opposed to somebody tagging them to kind of wave the flag in their

00:53:10   face and say, "Hey, you should pay attention over here." So I'm not sure how that came to be. But,

00:53:15   you know, hey, I'm happy that they reached out. I'm happy that they seem to be more than happy

00:53:20   to replace the RAM. Because, I mean, like you were saying earlier, I bought this RAM in January. I

00:53:25   looked up my order number and I forget when it was in January, but it was still January. I mean,

00:53:29   that was somewhat long ago. And I guess, you know, most things have a year warranty, but

00:53:34   I don't know, it's nice of them to not fight me on it because I'm used to most retailers being like,

00:53:40   "Well, are you sure it's the RAM? Why don't you take it to Apple first?" Yeah, etc, etc, etc.

00:53:46   And they were like, "No, no, no, send it back. We'll get you RAM."

00:53:48   So like I said, you didn't have to wait for someone to hear it and tweet you. You could

00:53:51   have just called them on the phone. All you got to do is tell them you have bad RAM and notice that,

00:53:54   like, last time I called from mine, like I said, it was years after I bought it, I said,

00:53:57   "I've got a bad DIMM." They said, "Okay, read me the serial number. Okay, we'll send you a box."

00:54:01   Like that was it. There was no I didn't even give them a reason it was dead

00:54:04   Yeah, and I truly hand on heart was planning on calling sometime this week and then right as I decided no, it's been long enough

00:54:11   I'm happy. Well, I'm not happy but I am satisfied with my my scientific experiment

00:54:16   That around that time when I had finally concluded that was the case. That's when they reached out. We're like, hey what's happening?

00:54:22   So that was my plan

00:54:24   Marco I am deeply sorry, but I have to ask John

00:54:28   Tell me about TiVo lifetime service, please

00:54:31   We are sponsored this week by Automatic, the small adapter that turns your clunker into

00:54:36   a smarter connected car.

00:54:38   Go to automatic.com/ATP for more info.

00:54:41   Automatic has just launched the Automatic Pro, their new unlimited 3G car adapter with

00:54:46   no monthly fees and no subscription and no service charges.

00:54:50   You just pay once upfront for the Automatic Pro and you don't pay anything per month to

00:54:55   use their services and their abilities.

00:54:57   Now, the Automatic Pro includes always-there 3G connectivity

00:55:01   to let you know where your vehicle is parked at any time

00:55:04   and let you track your vehicle in case of theft.

00:55:06   And of course, this builds on what Automatic

00:55:08   has been doing for years.

00:55:09   It works with If This Then That for endless customization,

00:55:13   connecting your car to the rest of your life.

00:55:15   You can do things like automatically turn on

00:55:17   your Nest thermostats when you're nearing your home,

00:55:19   and of course, tons of other abilities that you can do

00:55:22   using If This Then That and the Automatic API.

00:55:25   In addition, of course, Automatic provides

00:55:26   built-in functions they always have. Things like gas mileage tracking, reaching your goals

00:55:31   for fuel economy, tracking your expenses for businesses. So it integrates with apps such

00:55:36   as Concur and Expensify to be able to track costs associated with your driving. There

00:55:41   is so much you can do with Automatic and of course now with Automatic Pro. It has always

00:55:46   on 3G connectivity which makes so many of these things even better and possible. Automatic

00:55:50   can also call for help in case of a crash. It detects severe accidents and if you aren't

00:55:55   able to respond, a trained responder will call for help because you can't.

00:56:00   Now Automattic is normally $129.95, but when you use our exclusive offer code ATP, you'll

00:56:05   save $20 bucks, so it makes it about $110.

00:56:08   Great deal, no monthly fees after that.

00:56:10   Again, you buy it up front and that's it.

00:56:12   You don't pay per month to use these services.

00:56:14   So go to automatic.com/ATP for more information and use offer code ATP to save $20 off the

00:56:21   regular purchase price.

00:56:23   Thanks a lot to Automattic, the smart car adapter, for sponsoring our show once again.

00:56:28   [Music]

00:56:29   >> John, go ahead and tell us about TiVo.

00:56:33   >> You're saved by real-time follow-up.

00:56:35   Someone just tweeted at me that from the Verizon Wi-Fi calling fact, "When using Wi-Fi calling,

00:56:41   911 calls will always try cellular service in the local market first, even when the device

00:56:45   is in airplane mode or cellular service is off."

00:56:48   So I don't know.

00:56:50   Now I'm back to just the status bar, I suppose.

00:56:52   Which is a silly reason.

00:56:54   I'll think about it.

00:56:55   I'll think about it.

00:56:56   I don't like getting phone calls at home anyway.

00:56:57   And I have landlines.

00:56:58   Oh my god.

00:56:59   I should take a screenshot of that status bar.

00:57:02   We can mark it and switch to the show art to it so you can just see what a horror it

00:57:07   actually is.

00:57:08   All right, TiVo.

00:57:10   The news today about TiVo?

00:57:12   Not really that big of news, but apparently TiVo is discontinuing service for the TiVo

00:57:19   Series 1.

00:57:20   TiVo series 1 that was, what, introduced in 1999?

00:57:26   And what they mean by discontinuing is it will no longer get any guide updates or anything.

00:57:32   So you can still watch shows that are recorded on it, but you can't record any new ones because

00:57:36   it won't know when anything is on.

00:57:39   And it doesn't work like a VCR, we can do it manually anyway.

00:57:43   And it's because they're changing the format of the guide data, probably modernizing it

00:57:46   in some way that we would be horrified to learn about.

00:57:50   we knew exactly what they were feeding because of these TiVo Series 1.

00:57:53   TiVo Series 1 was an analog device.

00:57:55   It recorded video from analog sources.

00:57:58   It was pre-digital.

00:57:59   And of course standard def because it was all analog.

00:58:02   I didn't know that you could still use a TiVo Series 1.

00:58:06   I thought they were already non-functional entirely, but apparently...

00:58:09   Did you make it with like semaphore flags?

00:58:11   Like how does that even work?

00:58:13   If you have a TiVo Series 1 and your 16 year old hard drive, or maybe a hard drive you

00:58:17   replaced it because you can do third-party hard drive replacements or

00:58:19   whatever. We're still using it to record analog standard FTV.

00:58:24   They're terminating your service. This is even if you bought quote-unquote

00:58:28   "lifetime service." That's what the story about this the tweet is like, "Oh I bought

00:58:31   lifetime service, but now you know I'm still alive and the box is still alive

00:58:37   but somehow the service is ending." So the tweet was "Tivo series 1 lifetime"

00:58:41   and scarequote service lasted about 16 years. And for this inconvenience Tivo is

00:58:46   offering you a $75 prepaid visa card so you get 75 bucks if you if you held on

00:58:50   and kept using your TiVo series 1 for this long and they say there are 3,500

00:58:54   of these series 1s still in use which is a small number but you know anyway I

00:58:59   can't believe that there wasn't some fine print written into the lifetime

00:59:03   service that says after 16 years we reserved the right to turn off the

00:59:06   service maybe you should buy a new DVR but I salute the people who are out

00:59:10   there still using it I never expected like lifetime service I have lifetime

00:59:14   was on all my Tevo boxes but I replaced the boxes long before I get to the lifetime service thing.

00:59:18   People used to be better about letting you transfer lifetime service for some nominal fee

00:59:22   or for free to another device to encourage you to upgrade or whatever but those days are mostly gone

00:59:26   and the deals have been getting worse. Anyway, I just thought this was funny that there are

00:59:30   actually people out there still using a series one DVR. I don't think I have any piece of electronics

00:59:35   in my house that is 16 years old and still working including smoke detectors, telephones,

00:59:43   Microwave maybe I think my as discussed on a recent reconciled with em, so I think my microwave

00:59:47   Maybe more than 16 years old, but I should look at the date on the back of it

00:59:50   I should see exactly how old it is, but other than that

00:59:52   No DVR. So that's for sure

00:59:55   fair enough

00:59:57   We had spoken a few episodes ago about Twitter verification because all three of us are now verified and

01:00:04   a friend of the show Brianna Wu has been verified now in the last week or so and

01:00:12   Apparently it's made everything immensely better.

01:00:17   Brianna tweeted a couple of days ago as we recorded this fact, "I have not seen a single

01:00:21   death threat or rape threat since being verified and getting the 'quality filter' quote.

01:00:26   Everyone should have this."

01:00:28   Funny how that works.

01:00:29   Yeah, that's exactly it.

01:00:30   I mean, like, we talked last time, like I joked, like, we have computers now, like,

01:00:35   any computational overhead of offering this to everybody, like, that's totally worth it.

01:00:40   And we've since heard a little bit more discussion from various other places about

01:00:43   it.

01:00:44   It seems like the gist of the problem here is not that turning this on would be computationally

01:00:49   expensive for everybody.

01:00:51   It's that Twitter, fundamentally Twitter is very, very tied to the idea internally,

01:00:58   politically that there should be this kind of open platform and that there should be

01:01:03   no filters by default.

01:01:05   And that's a really nice theory.

01:01:07   So is communism.

01:01:08   But in practice these things don't work quite that well.

01:01:12   It is really ideal and that would be nice and it worked that way for a while and that's

01:01:16   cool, but the reality is the harassment problem is very big and very real and that wonderful

01:01:23   world where anybody could tweet anybody and they would see it by default, I think that's

01:01:28   a nice idea.

01:01:29   Like blog comments, that's a nice idea but in practice it doesn't work.

01:01:34   In practice there's lots of problems with that.

01:01:36   So I maintain that the quality filter feature that is available to verified Twitter accounts

01:01:43   not only should be available to everybody, but I'd even say it should be on by default.

01:01:47   Yeah, I would agree with that.

01:01:49   Yeah, I don't know why you wouldn't have it on by default.

01:01:51   I guess the idea is that maybe the filter is bad and things will accidentally get filtered

01:01:56   out, but it's not like Gmail has spam filtering off by default.

01:02:00   Of course it's not by default.

01:02:02   Sometimes things end up in your spam folder when they shouldn't be in there, but no one

01:02:07   would say the correct default is off.

01:02:09   Yep, I don't get it.

01:02:12   Finally, in follow-up, over an hour in, a few people have taken me to task for saying

01:02:19   in the last episode, or maybe it was one of the recent episodes, that I was not satisfied

01:02:24   with Apple's diversity numbers, particularly with ethnicities and how many white employees

01:02:32   Apple has as opposed to other races. And what I'd said was, "Hey, this is really unfortunate,"

01:02:41   and it looks like as I look up these numbers again, 56% of Apple is white as they self-report

01:02:46   today. And a lot of people said in various degrees of obnoxiousness, "Well, what's the

01:02:50   appropriate amount of diversity because the country is..." And I don't have those numbers

01:02:54   in front of me, but something like 60 or 70 percent white. So if they're less white than

01:03:00   America, then what's the problem? And I don't have any simple answer for that. It's something

01:03:06   I hadn't considered, and it's a fair point, but I would point out that not all of Apple

01:03:12   is in America, and so that probably should be factored in as well. But I think the biggest

01:03:20   thing to me was that Apple seems to have made not a lot of change from just a

01:03:27   couple of years ago when they started reporting this, which I admire. They

01:03:31   didn't make that much change, yet it seems like the thing that was completely

01:03:36   unacceptable a couple a couple of years ago is now, "Look at us and how wonderful

01:03:40   we are." And that just seems kind of weird to me. I don't know if you guys had any

01:03:46   thoughts on this, but that's basically the thing that I still stand by, which is

01:03:50   which is it hasn't gotten that much better.

01:03:52   Why are we really celebrating it

01:03:54   like it's completely turned around?

01:03:56   - So before we let Jon tell us the truth

01:03:58   and why this is, you know.

01:03:59   (laughing)

01:04:00   Before we let Jon basically rock this

01:04:02   and do everything correctly,

01:04:04   I will just kind of speak for myself

01:04:06   and for I think a lot of people

01:04:08   who listen to this kind of stuff

01:04:09   and who see these kind of things.

01:04:11   I know, having been just in the area

01:04:16   of people discussing this for a while now,

01:04:19   I know now that it is never that simple.

01:04:22   And so when you throw out a stat like,

01:04:27   "Well, you know, America is X percent white,

01:04:29   "and so that's fine," it is usually not that simple.

01:04:33   And so I know to keep my mouth shut with things like that,

01:04:37   because I know, I'll hear something like that,

01:04:38   and I'll be like, "Yeah, that might make logical sense

01:04:41   "when you read it in 140 characters,

01:04:43   "but I bet there's more to the issue than that."

01:04:45   And so it's important,

01:04:48   When you are responding to people, not you Casey,

01:04:50   but you the public, and I'm talking to myself too,

01:04:54   don't just jump to a conclusion like that

01:04:55   because chances are the issue

01:04:56   is way more complicated than that.

01:04:58   So John, tell us why.

01:04:59   - I was gonna start getting to the thing,

01:05:02   but I'm basically on what you just said.

01:05:04   That's something I always notice on Twitter,

01:05:08   and we all notice it to some degree or another,

01:05:10   whether we're doing it ourselves

01:05:11   or we see other people doing it,

01:05:14   where people will start from a premise

01:05:18   and then start throwing out facts in support of that.

01:05:21   And that's fine, right?

01:05:23   Because why wouldn't you?

01:05:24   You know, this is what you think.

01:05:25   And so you will find facts that support what you think

01:05:27   and then you will list those facts.

01:05:30   But it's always important to kind of ask yourself,

01:05:32   why am I seeking out the facts that I can find

01:05:38   that align with this thing that I already think?

01:05:40   Like, am I actually thinking about the situation

01:05:42   I'm merely just trying to find something that will support my release.

01:05:46   So that maybe, I mean, this is not universally true, but perhaps someone who was telling

01:05:51   Casey, "Hey, you said this percentage was bad, but it's about the ratio of the people

01:05:57   in the United States."

01:05:59   You'd have to go look up the ratio of the people in the United States.

01:06:01   You could do it.

01:06:02   This is a bad example because they're probably too intellectually honest in this.

01:06:04   But a lot of times, you see an issue come up and people immediately go looking for something

01:06:10   they can throw out that supports their thing.

01:06:12   really asking why am I so desperate to find things that support this rather than looking

01:06:18   at the issue from first principles or understanding the wider context.

01:06:22   Because there will always be things, especially statistics, that can help support one side

01:06:27   or the other.

01:06:28   But if you're only ever looking for and saying repeatedly to anyone who mentions anything

01:06:31   about the topic the few facts you have in your back pocket that support the one thing

01:06:35   that you care about, you probably are missing the larger issue.

01:06:37   And what you're really missing more is why is it so important to you that your current

01:06:41   view of this be exactly correct. So anyway as for these specific stats I

01:06:46   don't know I thought the stats were just US only so I think there are percentages

01:06:51   for you know white employees that are actually accurate to the United States

01:06:56   the United States has that percentage of the population and then that's there you

01:06:58   know so I thought that was an apt comparison but like Margot said it's

01:07:02   never quite that simple because you know especially with statistics you can

01:07:05   they're taking like all their employees right and all US employees Apple

01:07:10   employees a lot of people and not all those jobs are the same. If you start

01:07:15   looking at different sections of employees the percentages, I think they

01:07:18   do this don't they break it down on the page, the percentages will change you

01:07:22   know pretty dramatically by five to ten percent maybe even more. If you look at

01:07:27   like the top level of the org chart director level C level executives they

01:07:33   are not 56% white right so and then all the way down to look at engineering

01:07:39   versus QA versus retail versus, you know, all the different categorizations. If you lump them all

01:07:46   together, it's nice to get a nice aggregate, but you don't know what the pieces are, right? And so

01:07:49   Apple's not trying to mislead anybody there. They're not going to break down their hiring and

01:07:55   their employees by individual job level and location and, well, what is the population in

01:08:01   the state of Maryland and what are the retail employees in the state of Maryland and what about,

01:08:04   you know, like it's just, they're not going to do that, right? But that is Apple's job internally

01:08:08   to work on this problem because if their overall numbers happen to work out to the same percentage

01:08:14   of the United States but it's only because like all the leaders of the company are white men

01:08:19   and all the rest of the company are not, that's not that you're not you're not achieving your

01:08:24   goal of diversity right your goal is not to make a bunch of numbers match a bunch of other numbers

01:08:28   you can always pick numbers to make you look better or worse and to some degree Apple may be

01:08:32   doing that in the high level numbers that are up there but that's not the goal of the thing you're

01:08:36   not — Apple is not an amorphous blob of people that combine to one Uber person. They are a bunch

01:08:40   of individual people, and you want to know if, you know, if I get a job at Apple, what are my chances

01:08:46   of, you know, being in charge of all software at Apple? What are my chances of rising to the level

01:08:52   of the org chart where I'm taking meetings with Tim Cook and stuff? What, you know, what are my

01:08:56   odds of becoming an engineering manager or leading whatever? And if you look around you at Apple,

01:09:01   and it's like, "Gee, everybody who's an engineering manager is a white dude, and everyone in the board

01:09:06   meetings is a white dude, and everybody who's on stage at WWDC is a white dude." To a first

01:09:12   approximation, that was the case not too many years ago. And like I said, even now, the high

01:09:17   levels of the company are like that. Maybe you feel like, "Oh, sure, I can get a job at an Apple

01:09:21   store. They're good at it, but do I feel like this company is giving equal opportunity to everybody?"

01:09:30   And that's what Apple is working towards, and I don't think they've yet achieved it,

01:09:33   which is why they keep putting out these numbers. So as Casey said, if these numbers that were given

01:09:41   are these big aggregates and they haven't changed that much from year to year, it just shows they're

01:09:45   making slow progress, right? And so it's not phenomenal progress, but it certainly doesn't

01:09:49   mean because you can say 56% is less than 60%, therefore they're done, it absolutely doesn't

01:09:54   mean that because it doesn't take much thinking to realize, well, that's just a big aggregate

01:09:58   number and it's not that's not the goal the goal is not to simply people are asking that like

01:10:02   honestly asking like what is the goal what numbers you're supposed to see here to let you know that

01:10:06   apple has achieved its goal of diversity and the answer is those numbers are never going to tell

01:10:11   you that right otherwise you know they wouldn't they wouldn't be continuing along this path they

01:10:15   would just say when we're done we've accomplished our goal isn't that great right um and that's

01:10:20   before you even get into details of like percentage of u.s population versus percentage of u.s

01:10:24   population of working age, like not babies, not children, not retired people, you know, so it's

01:10:29   numbers are complicated. But anyway, I'm confident that Apple understands the problem before them and

01:10:35   is working towards it, just as I am not particularly tied to the specific numbers they throw out on

01:10:42   these pages. All we're basically looking for is are the little graphs they put up going in an

01:10:46   upward direction? Do they have enough of them every year so they show they're making progress?

01:10:50   and how fast is that progress, what do the slopes look like.

01:10:54   Yeah, that's about it.

01:10:56   All right.

01:10:58   Well, we are out of follow-up.

01:10:59   Thanks a lot to our three sponsors this week, Betterment, Tracker, and Automattic, and we

01:11:04   will see you next week.

01:11:06   Now the show is over, they didn't even mean to begin, 'cause it was accidental.

01:11:15   Accidental.

01:11:16   Oh, it was accidental.

01:11:18   Accidental.

01:11:19   John didn't do any research, Marco and Casey wouldn't let him

01:11:24   'Cause it was accidental (accidental)

01:11:27   It was accidental (accidental)

01:11:30   And you can find the show notes at ATP.fm

01:11:35   And if you're into Twitter, you can follow them

01:11:40   @C-A-S-E-Y-L-I-S-S

01:11:44   So that's Casey List M-A-R-C-O-A-R-M

01:11:48   ♪ Anti-Marco armen ♪

01:11:51   ♪ S-I-R-A-C ♪

01:11:53   ♪ U-S-A-C-R-A-Q-S-A ♪

01:11:56   ♪ It's accidental ♪

01:11:57   ♪ It's accidental ♪

01:11:59   ♪ They didn't mean to ♪

01:12:01   ♪ Accidental ♪

01:12:02   ♪ Accidental ♪

01:12:04   ♪ Tech podcast ♪

01:12:06   ♪ So long ♪

01:12:08   - Well, I wanted to talk about the Intel Fab thing.

01:12:10   That's a topic.

01:12:11   - Can we make that quick?

01:12:15   - Hmm.

01:12:16   - Yeah, exactly.

01:12:18   Maybe. I don't know. Let's dig down what we have here. What do we have for an aftershow then?

01:12:22   We can do Instagram stories.

01:12:24   Instagram stories is a good idea because that's fluffy.

01:12:26   Yeah, it's this thing that appeared on top of Instagram recently that's confusing all the old

01:12:31   people like me. You're not that old.

01:12:33   Your wife's face is always up there. I know. She figured it out very quickly because she is

01:12:39   younger at heart and also biologically than I am. I am too old.

01:12:45   Am I what, three months?

01:12:47   - I think like six, yeah about six, yeah.

01:12:50   So I am too old to understand Snapchat,

01:12:53   so Instagram has basically cloned a major part of Snapchat

01:12:59   into the Instagram interface,

01:13:01   which has just confused Instagram for me,

01:13:03   and now I have this row of heads on top of Instagram

01:13:06   and I tap them and weird things happen,

01:13:09   and I don't really know what I'm supposed

01:13:10   to be doing with this.

01:13:11   Can you explain it to me?

01:13:12   - So John, I don't think I've seen any posts from you yet,

01:13:15   So you're not producing using this,

01:13:18   although I presume you're at least consuming it.

01:13:20   Is that fair?

01:13:21   - Yeah, I don't see myself ever making one of these videos,

01:13:23   but they're up there and they tap on the faces.

01:13:25   - Okay, so the idea here, from what I can understand,

01:13:29   as someone who's never used Snapchat

01:13:31   and didn't really have any interest in Snapchat,

01:13:33   is with Instagram Stories,

01:13:36   you can take a photo or a video that is semi-ephemeral.

01:13:43   So I believe it's after a day they will self-destruct and your phone explodes.

01:13:49   It's very inconvenient.

01:13:50   No, they'll self-destruct and then they won't be available anymore.

01:13:53   Not to you, not to anyone else.

01:13:55   And so the idea is if your Instagram profile and the photos that you post on Instagram

01:14:02   proper is the like super staged, super deliberate, super serious version of you, Instagram Stories,

01:14:10   which is this ephemeral thing, is more of the candid, fun-loving, like, "Hey, here's

01:14:17   the real me, take it or leave it" sort of thing.

01:14:20   And I actually wanted to call out the latest episode of Connected, which is episode 103.

01:14:26   The host did a really incredible job of talking about kind of what the motivations are behind

01:14:33   Instagram Stories and how it's different than regular Instagram.

01:14:36   But regardless, I've really been enjoying it a lot more than I thought I would as someone

01:14:43   who had not even the slightest interest in Snapchat.

01:14:47   And one of the common complaints about Snapchat that I've understood is that anyone who is

01:14:51   over the age of about 18 finds the interface completely inscrutable.

01:14:55   And everything is gesture-based, and as much as we all hate tutorials as you install an

01:15:00   app for the first time, this is one of those instances where a tutorial or walkthrough

01:15:07   or onboarding would have perhaps been useful in Snapchat, because everything's gesture-based,

01:15:12   nothing's obvious, everything's weird.

01:15:14   In Instagram, there's a little bit of that in Instagram Stories, but it's not nearly

01:15:18   as bad, and I was able to figure it out pretty quickly, and I'd classify myself as an old

01:15:23   man just as much as you two are.

01:15:26   And I've been enjoying both consuming and creating.

01:15:30   One of the things I've wondered as I consider whether or not to post a new entry in my Instagram

01:15:34   story which sounds super cheesy, but anyway, one of the things I've been considering is

01:15:39   why would anyone else give a crap about this?

01:15:42   So like, as an example, I was going to take a picture of my setup at work, and you know,

01:15:47   after having obscured the sensitive parts of what was on my computer screen, and then

01:15:51   it occurred to me, I don't think anyone really gives a crap.

01:15:55   And so while I was at the beach, you know, I'm on vacation, whee, that's exciting and

01:15:59   kind of cool or whatever. I had posted a handful of times then, but now that I'm back on my

01:16:05   normal grind, well it's not my normal high roller, you know, look at me, I'm so awesome

01:16:09   life, it's my normal grind. So I don't know if I'll be producing that much or posting

01:16:15   that often from now until the next time I do something interesting, but I really like

01:16:21   the idea and I echo what Mike had said on Connected, and he's talked about it and he

01:16:28   and I both have talked about this on analog as well, that Instagram is kind of my happy

01:16:32   social network. I get a lot out of Twitter and I enjoy Twitter, but Twitter is nevertheless

01:16:38   kind of a dumpster fire where I cannot remember a time that Instagram has made me anything but

01:16:46   happy. And even with stories, it makes me happy in a different way because it's kind of the

01:16:52   fun, ephemeral, you know, "Hey, it's not going to be perfect and that's okay because life isn't

01:16:56   perfect. And it's brought, and this is a super sleazy business-y term, but it's

01:17:01   brought my engagement with Instagram up even more. And I check Instagram a

01:17:05   handful times a day because I enjoy it so much. But now I found myself checking

01:17:08   it more often wondering which one of my friends, because I tend to follow mostly

01:17:12   friends on Instagram, which one of my friends has posted something new and

01:17:16   something worth looking at. And so I have really been enjoying it. Two thumbs up

01:17:20   from me. I think it's super, super interesting and it seems to have

01:17:24   I have traction so far.

01:17:26   Don't think it's a flash in the pan.

01:17:29   This isn't the next Peach, I don't think,

01:17:31   but remind me of that in six months.

01:17:32   This is future Clean Shoppers.

01:17:34   - Yeah, I mean like this, I have,

01:17:37   I've been using Instagram more and more recently

01:17:38   because I've been using the Explore View,

01:17:40   the little search thing at the bottom,

01:17:41   which Tiff introduced me to,

01:17:42   and I didn't even realize that's what, that was there.

01:17:45   I thought it was literally a search,

01:17:46   and I never wanted to search for anything,

01:17:48   so I never visited that tab.

01:17:50   And it turns out it's more of like a browser Explore View,

01:17:53   and you pick topics, so mine's full of watches

01:17:56   and puppies now, which is awesome.

01:17:58   And--

01:17:59   - I didn't know that either, actually.

01:18:01   - It is really nice to, so I'm using Instagram a lot now,

01:18:04   and I have the same feelings as you.

01:18:05   Instagram is just a feel-good place.

01:18:08   It is, both to browse and to post,

01:18:10   it is a just nice place to be.

01:18:13   People aren't universally terrible.

01:18:15   - Unlike Twitter?

01:18:16   - Well, in general, I'm saying,

01:18:18   including all services that we have,

01:18:21   people in general are not terrible.

01:18:24   The design of a social thing online,

01:18:27   a social construct, a social service,

01:18:29   the whole design of it, the structure,

01:18:31   what's allowed, what's possible,

01:18:33   how things work, who sees what,

01:18:36   that all affects how horrible people are enabled to be

01:18:41   and what kind of impact their horribleness can have.

01:18:46   So the same group of people can use two different services

01:18:51   and act very differently on them

01:18:52   and get very different things out of them

01:18:54   just because of differences in their design.

01:18:56   And so even if Twitter's entire user base

01:18:59   is also Instagram's exact user base,

01:19:03   you have better results on Instagram

01:19:05   because just the form and the structure

01:19:07   and the way everything works in a setup

01:19:09   and the expectations of it and the format and everything

01:19:12   just encourage better behavior

01:19:14   and limit the places people can be horrible to you.

01:19:17   So that's good.

01:19:18   What worries me about this,

01:19:20   And one of the reasons I never used Snapchat was,

01:19:23   and again, yeah, that conversation I connected was awesome

01:19:26   and we will link to it.

01:19:28   One of the reasons I never had to get into it

01:19:30   was because no one that I know really uses it.

01:19:33   Like all the people I would be interacting with

01:19:35   are on Twitter and Instagram.

01:19:37   So I don't need to use, first of all,

01:19:39   I don't need to use Facebook, which is great

01:19:41   because I don't want to use Facebook.

01:19:42   And I also, and yes, I know they own it

01:19:44   and that doesn't matter.

01:19:46   And I also never had to use Snapchat

01:19:47   because my people aren't there, really.

01:19:50   So to bring that to Instagram,

01:19:53   to basically bring a major feature of Snapchat to Instagram

01:19:58   is kind of okay on one level because it's like,

01:20:01   well, I wasn't gonna use Snapchat anyway

01:20:04   because my friends weren't there.

01:20:06   But on the other hand, it really shows,

01:20:09   Facebook has always been a very tasteless company.

01:20:12   Facebook is pragmatic and ruthless and tasteless,

01:20:15   just like Microsoft was in the 80s and 90s really.

01:20:18   Facebook now is this conglomerate really.

01:20:21   They have tons of properties that are major on the internet.

01:20:25   They have Facebook itself, which is probably

01:20:28   the biggest web property in the world

01:20:31   and probably will stay that way for quite some time.

01:20:34   Anything that gets really big that they think

01:20:36   might threaten that, they just buy it.

01:20:37   They have enough money to keep doing that indefinitely.

01:20:40   And they are clearly totally okay with like,

01:20:44   well if you won't let us buy you,

01:20:45   we're just gonna rip you off,

01:20:47   and we'll just kill you that way.

01:20:49   And that strategy doesn't always work.

01:20:53   Microsoft did the same thing.

01:20:55   Apple has done things on a smaller scale

01:20:57   with Sherlocking things,

01:20:58   but they don't usually do a huge scale job of it

01:21:00   with stuff like this.

01:21:02   Google does it sometimes,

01:21:03   and that strategy tends to work.

01:21:06   In general, when the big powerful companies

01:21:09   that already have all the users, and all the usage,

01:21:13   and all the attention going to them,

01:21:14   and all the time spent on their services,

01:21:16   if they clone some smaller service,

01:21:18   their clone often wins.

01:21:21   And if they can't clone it, they can buy it.

01:21:24   And this just kinda, it's a bit of a warning sign

01:21:28   of the power that Facebook has.

01:21:29   And I'm not saying Snapchat will be killed by this.

01:21:32   In all likelihood, Snapchat is too big now

01:21:34   and they'll probably be totally fine

01:21:36   because Snapchat's really big.

01:21:37   I think they're bigger than Twitter.

01:21:38   They're really big.

01:21:40   But it does kinda scare me a little bit

01:21:43   that Facebook is willing to so cavalierly

01:21:46   just like completely rip off the stories feature,

01:21:50   call it the same thing, do the same things,

01:21:52   use the same gestures, like completely rip it off.

01:21:55   It's so closely and so shamelessly

01:21:58   and they just couldn't possibly care less

01:22:00   because they are just that ruthless.

01:22:02   And we live in an age now where these handful

01:22:06   of big internet companies have a lot of control over us

01:22:11   and a lot of resources.

01:22:13   And the centralization of power here

01:22:16   is just getting more and more severe.

01:22:18   And that kinda worries me for the future

01:22:20   of internet things in general, things we do online

01:22:24   that use the internet or that involve the internet

01:22:26   or are the internet.

01:22:27   It is a little concerning.

01:22:29   - The thing that struck me about using this

01:22:31   and having used Snapchat a little bit

01:22:33   is what you often hear, and you guys said it before,

01:22:37   is like, oh, well, I don't understand Snapchat,

01:22:39   the kids understand it or whatever. And there is a demographic thing about what age groups use,

01:22:44   which applications, that's definitely true, and more older people use Instagram or whatever.

01:22:48   But Snapchat and also the Instagram stories have bad interfaces. They have bad user interfaces.

01:22:57   There is no sensible structure to the interface. They don't use OS conventions, that's for sure.

01:23:05   So the normal affordances and sort of the things you're used to in like generic UI kit,

01:23:10   you know, I don't know about another platform, I'm just talking about an iOS, but like it

01:23:13   doesn't look like iOS, that's for sure, right?

01:23:15   So it's kind of its own thing, and within its own thing, they don't really define any

01:23:19   particular spatial metaphor or something to hang your hat on where you can say, "Oh, I

01:23:25   see all up swipes to this."

01:23:27   I mean, like basically if you, you know, if someone's like, "Oh no, Snapchat is totally

01:23:30   sensible," what you'd basically be doing is listing, "This does that, this does that,

01:23:34   does that, just does that, there, now it's, that's not a system. That is just a list of

01:23:38   actions and the resulting, you know, what you're looking for a system is like, because I understand

01:23:42   the system, before you tell me the list of all things I can do, I can predict, oh, then obviously

01:23:47   the way to do x is going to be y. That's when you know you have a system. Snapchat doesn't have a

01:23:51   system or not a very good system. It's a lot of arbitrary crap in there. Not using native controls,

01:23:56   not using sort of unprecedented, right? So sort of setting their own standard. And why does,

01:24:03   Does that mean, is that why young people get Snapchat

01:24:06   and old people don't?

01:24:07   No, like there are other factors there

01:24:08   in terms of who it's advertised towards

01:24:10   and who the first users were and you know,

01:24:12   critical mass and social networks

01:24:13   and where my friends are and all those other things.

01:24:15   That all factors in as well.

01:24:16   But I think there is something to the idea

01:24:18   that young people will,

01:24:21   they do something because a bunch of their friends

01:24:24   are doing it, they'll just figure it out.

01:24:26   They have a high tolerance

01:24:28   for learning arbitrary stupid crap.

01:24:30   If I think about all the arbitrary stupid crap

01:24:31   that I learned, which basically is the entire early years

01:24:34   of computing where everything was terrible

01:24:36   and nothing made sense and the Mac was the only one

01:24:38   that had kind of an understandable system.

01:24:40   Think of all the things that I memorized,

01:24:41   things I learned how to do in video games.

01:24:43   There was no system, it was just arbitrary.

01:24:44   And kids have a lot of time and good memories

01:24:48   and they'll just plug away

01:24:50   until it becomes second nature to them.

01:24:51   So to kids, Snapchat makes perfect sense

01:24:53   because they memorized all the stupid gestures and commands,

01:24:56   even though those gestures and commands

01:24:57   are almost entirely arbitrary,

01:24:59   countered everything that every other UI paradigm

01:25:02   on the device they're holding has taught them

01:25:04   and are just plain bad, right?

01:25:06   They're just bad, this is a bad UI.

01:25:07   And so Instagram Stories didn't use all the same things

01:25:11   as Snapchat, but it's a similar thing

01:25:12   where Instagram was a very sensible,

01:25:14   straight up the middle iOS application

01:25:16   with a reasonable UI using metaphors and controls

01:25:19   and interfaces that we understood

01:25:21   so you could predict what would happen

01:25:22   when you did things for the most part.

01:25:23   There's always weird edge cases

01:25:24   and custom controls and stuff.

01:25:26   But then when they did Stories, it's into La La Land again,

01:25:29   where nothing makes sense and you can't you have no you have no nothing to hang your hat on except for like

01:25:34   Maybe it'll behave like snapchat

01:25:36   Which again maybe that's establishing a standard and if you're familiar with snapchat

01:25:40   That might be a good way to go if you're trying to do like what Margo said it like oh

01:25:42   We want to do a snapchat like thing then just copy a snapchat gestures because all the kids who do snapchat

01:25:47   It makes perfect sense to them and if they go to your thing and swipe or drag or tap and it doesn't do what they?

01:25:52   Expect from using snapchat. They think your thing is broken

01:25:55   But again, it's not a good UI just because they make it like snapchat for like or you know

01:26:00   Make things similar snapchat to try to get it's a little bar

01:26:03   They've got on the top with a little segmented timers and what happens when you swipe in different directions

01:26:07   and a little rotating effect versus the sliding versus up and down and tapping and it's just

01:26:12   These are bad interfaces. And so I maintain as the old cranky person that not only

01:26:18   actual reason like old people don't understand snapchat because they made a bad UI and

01:26:24   And because older people have less tolerance for figuring out a bad UI that doesn't look

01:26:29   like all the other ones they're going to do.

01:26:31   And I don't just think it's bad on many, many levels.

01:26:34   And again, this doesn't explain why old people don't use Snapchat.

01:26:37   There's so many other reasons, but this is a factor.

01:26:39   I'm saying this is a factor in that giant stew.

01:26:41   It's not the biggest factor, it's not even maybe a major factor, but it's there.

01:26:46   And that annoys me, because I feel like Snapchat could have been successful with a clever,

01:26:52   innovative UI.

01:26:53   games with this, like a UI that uses no native controls, totally custom, doesn't use anything

01:26:57   from UIKit, is the same on Android and iOS, but is delightful and understandable and fun

01:27:03   to use.

01:27:04   And someone can pick up and learn.

01:27:06   Again, this is my hobby horse, but it's really true.

01:27:09   If you make some kind of physical world-based metaphor like, "Oh, I have stacks of cards

01:27:14   and we can slide them from side to side and up and down and bring them together and collapse

01:27:17   them, or I can crunch them up."

01:27:19   Like if you make some sort of thing like, these aren't physical things, but in some

01:27:23   ways they behave in physical things or there's like a larger map that you can't see of this

01:27:27   application, once you learn the layout of that map, you know how to move around in it

01:27:31   and you know how to make things happen and you know what a button looks like and you

01:27:34   understand like any kind of metaphor like that, even if it uses totally non-native controls,

01:27:41   can become understandable where people start using it, learn one or two things that, oh,

01:27:45   I see how this works and then they can figure out how to do the other things without being

01:27:48   told explicitly, whereas Snapchat, it's like all mystery meat navigation, a bunch of inscrutable

01:27:53   icons and no real sense of where you are, what you're doing, where you're going.

01:27:57   You just memorize how it works because you're young and you have a lot of time or use the

01:28:01   application a lot and that becomes second nature to you and you just think that's normal.

01:28:04   So if you ask any kid, is Snapchat easy to use?

01:28:06   They'd be like, everyone knows how to use Snapchat.

01:28:08   It's like I know how to write my name, I know how to put one foot in front of the other

01:28:11   to walk and I know how to use Snapchat.

01:28:12   It's the most intuitive interface on my entire phone.

01:28:14   No, you just use it a lot.

01:28:16   That's what that means.

01:28:17   a good UI. And the reason they like using it is because it makes people like us that

01:28:22   angry. No, but that's not why we're angry about it.

01:28:25   I'm just saying it's a bad UI. You're angry about it because it's like you're making silly

01:28:28   videos of yourself and sending them to your friends and you're like, "Oh, these kids these

01:28:30   days shouldn't be taking video--" I don't care about that. Whatever, take videos. It's

01:28:34   the same thing when we take videos in our things and send it to our friends. It's fine.

01:28:38   Whatever. I have no anger towards people who use Snapchat. I have anger towards the developers

01:28:42   of Snapchat for making a bad UI.

01:28:44   [laughs]

01:28:46   Snapchat users, thumbs up.

01:28:48   Snapchat developers, thumbs down.

01:28:51   [beeping]