0:00

I am

0:01

back. Where

0:02

were you? I

0:03

was using homepod minis.

0:06

Okay. And

0:08

I am back. on big boys?

0:10

As long as they last. With

0:13

big home pods, like the cheese graders.

0:15

Oh my goodness. How did this happen? So

0:17

last time I went to Westchester,

0:20

there they were two perfectly

0:22

good full size home pods being

0:25

used in a kitchen by nobody. The fourth house

0:27

has just been gutted. So,

0:30

you know, I'm like, you know, these are just sitting

0:32

here rotting away with whatever

0:34

electrical flaw is what eventually kills them.

0:36

Those the capacitor or whatever that everybody says

0:38

is is bad or wrong or whatever. No one's

0:40

using them. And I'm sure they're gonna get replaced

0:43

at some point in the next decade by Apple. But

0:45

until that happens, let me

0:47

take them back. So I brought them

0:49

I have been using them and they are exactly

0:53

as glorious and annoying as

0:55

they always were and I'm so I'm

0:57

so happy I made the switch because

1:00

while they are playing music,

1:02

which you can eventually make them do most

1:05

of the time, eventually.

1:08

They do sound incredible. Like,

1:11

can you can you tell me something and I need you to really

1:13

and truly be honest. Are they white or are they black? they're

1:15

white. They happen to be white. So here's the

1:17

thing. I'm so glad you answered that question.

1:19

I feel like I I feel like Jack Ryan in hunt

1:21

for October about a crazy Ivan. We should do

1:24

that as movie thing, by the way. Anyway, I

1:26

I feel like your white home

1:28

pods are my white BMW. Hear

1:30

me out. When my white BMW

1:33

was working, it was amazing.

1:37

You know, it was only working for about a week every

1:39

month. But for that one week, oh,

1:41

it was good. It was real good for

1:43

that one week. Your white home pots, which by

1:45

the way, just happened to you -- Mhmm. -- knows when

1:47

they're working, I guess they're real good,

1:49

aren't they? No. And and by the way, I

1:52

chose the white because I

1:54

know I know this is unpopular for

1:56

a nerd like me to say that the black option is not

1:58

the best option on something, but

2:00

white home pods look better than black home pods.

2:02

They're not really white, they're gray. I mean, come

2:04

on.

2:05

they're pretty close

2:07

to I mean, you know, it's the outside is made

2:09

of something that's kind of resembles cloth, so you

2:11

can't get it like super super white. They're

2:13

not white. they're not even starlight. They're light

2:15

gray. Well And the black is

2:17

definitely a dark gray. Yeah. The black

2:19

is very very dark gray. But yeah. Anyway,

2:21

so the white looks better. And so I chose them and I

2:23

stand by that. I'm I'm unashamed

2:26

in in choosing the white home pods.

2:28

I also actually so my phone is temporarily

2:31

white again, but I'll tell you what. Some quick

2:33

follow-up on last week's case

2:35

discussion. I said I briefly

2:37

said last week I had tried the peak design, the Patoka,

2:40

and the Patoka, you know, I like the way

2:42

it felt, but it didn't I I hated the way it looked.

2:44

The peak design was a little bit bulky

2:46

and I didn't like how how

2:49

little how little relative, like, tackiness that,

2:51

like, the back kind of cloth like surface

2:53

head?

2:53

Well, I have

2:55

been using the peak design case all week.

2:58

because I just like

3:00

it. Like, and it's it it makes

3:02

relatively little sense. I can't

3:05

really justify it over the other cases

3:07

in any kind of, you know,

3:09

subjective or objective, I guess, objective

3:12

means. It is

3:14

more expensive than most of them. It is

3:16

thicker. and it's less grippy.

3:18

However, it

3:20

feels the nicest and it looks the

3:22

nicest. Therefore, been

3:24

sticking with it. I and I kinda like it. I kinda like

3:26

the the weird little square on the back. It's kinda like

3:28

a reverse pop socket. And so

3:30

it's definitely something that sticks out. It's some it's like a hole

3:32

that you can, like, put your finger on the inside of

3:34

the hole to lift it out of your out of your pocket or whatever.

3:37

Like, it's it's nice. I even

3:39

I I just now or just yesterday, just

3:41

order a couple. There are, like, mounts that go on the back. Just

3:43

see what I could do with that. Oh, here we go.

3:45

You're gonna put on your motorcycle? Yeah.

3:47

Yeah. For well, first, that's a buy a motor just for

3:49

this case, you know, I I like the case so much. I bought a

3:51

motorcycle as an then I bought them out, and

3:53

the motorcycle happens to be white. And I I bought

3:55

them out that that loops around. That way,

3:57

I'm I'm gonna start becoming a

3:59

what are those, like, the people who, like,

4:02

make GoPro, like, action videos on their vehicles?

4:04

Like, what what are those do we have a name for

4:06

those? Like, cloggedgers or something?

4:08

Oh my gosh. Yes. Let's go with cloggedgers.

4:10

That's perfect. So I'm gonna become a

4:12

cloggedger. This it's all it's and it's all thanks

4:15

to this case. I tried that Marco. Not as

4:17

easy as you think. Yeah.

4:20

There we go. Oh, yeah. So anyway,

4:22

Turns out, that's the case I'm going with for a while.

4:24

Okay. Update on my creaky clear case.

4:26

I've got a this is the first clear case I've

4:28

used for any pre show amount of time. I've got white

4:30

a collection of crumbs and

4:33

dust collecting and it visible

4:35

all around. If you look if you just circle

4:37

all around like the top edge

4:40

of the thing. It is so gross. I

4:42

cannot wait to get this thing off. One of my cases

4:44

shipped. I got a shipping notification. I think

4:46

possibly it'll be here by next week's show.

4:48

but I'll keep you updated. I cannot wait

4:50

to get this thing off my phone. You

4:52

know, an alternative is just is to just

4:54

not care about having a covered bottom And

4:56

then there's zillions of cases you can try.

4:58

This is I mean, options too. I mean, I picked to

5:00

two of them because they were expensive. There were there's more

5:02

than two options I could affect. I figure all

5:04

like one of these two, but I'll let you know. You

5:06

can try the paint design. It comes in

5:09

grayish black and a weirdly

5:11

light green color and that's it. Yeah.

5:13

I mean, obviously, if I if I needed a secure

5:16

mount or something, I would probably pick that, but

5:18

the the little square that you mentioned, like, oh, I can use

5:20

it to pull my phone out of pocket or whatever. I just I

5:22

don't want that to be there. I feel like my

5:24

fingers would find it and it would just be annoying.

5:26

You're

5:26

right. That is a thing. And and

5:28

the little square, the very first day I used

5:30

this case, I thought I felt sharp. Now it

5:32

doesn't push up many anymore. Maybe it just it just sanded

5:34

down my fingers. I don't know. But what I like about the Pizza

5:36

Hut, in addition to the fact that it looks

5:38

nice system, it feels nice the

5:40

buttons also feel nicest.

5:43

The the mute switch toggles a little

5:45

deep, like the whole the little recession to get to

5:47

the mute switch is a little bit deep. but the the

5:49

side buttons, you know, all the buttons are covered. The

5:51

side buttons feel great in

5:53

this case, better than most cases. And overall,

5:55

this case is the only it's the only

5:58

iPhone case I have ever used from anybody

6:00

except Apple, where it felt

6:02

like they went for the nice materials

6:04

instead of the cheap materials. And

6:06

it is by far the nicest

6:09

case I've ever seen, felt, or used,

6:11

that was not made of leather. So

6:13

for whatever that's worth, if you either wanna

6:15

go leather free or if you

6:17

just want something nice, it's just it's

6:19

really nice. Alright.

6:22

Let's do some follow-up. What

6:25

is going on with shared photo libraries

6:27

and what is shared with them, John? And last

6:29

week, I said some things about

6:31

what wasn't shared in photos and I was

6:33

wrong about a few of them. I

6:35

said that keywords, favorites, and location and

6:37

stuff were not shared? They are

6:39

shared. So those aren't attributes of the individual

6:41

photos. So keywords, and I tested this with my

6:43

little thing. If you put a keyword on a photo, get shared

6:45

with the other person. If you favorite it, it shows up

6:47

you know, that we're talking about photos that are

6:50

in the shared library. You can do that stuff

6:52

to them. You can assign keywords. You can set their favorite

6:54

status. You can add location information and everybody

6:56

sees that.

6:57

All

6:58

the other stuff I talked about, the the non

7:00

photo items remain unshared. So you

7:02

don't get albums, you don't get smart albums, you don't get

7:04

slide shows, you don't get book projects, all that stuff.

7:07

And also, someone asked about this when I was having

7:09

discussion on about it

7:11

on Twitter. What about duplicates?

7:14

What if two people have

7:16

the same photo, like, maybe like a person to

7:18

airdrop the same identical photo

7:20

to two different people at two different times,

7:22

and they both add it to

7:24

the shared library, what happens? The

7:26

answer, according to my experimentation, using

7:28

the Mac version of photos only and two different

7:31

accounts that had the exact same file,

7:33

I imported the file into both of their private

7:35

libraries and they're they're separate at that

7:37

point. And then I had both of them add

7:39

that photo to the shared library and

7:41

you end up with two copies of that photo in the shared library.

7:44

Just

7:44

two completely identical copies of the

7:46

photo in the shared library. I didn't edit them, so it's not like

7:48

one of them was edited and one of them was imported

7:50

into the individual things. then add one person

7:52

added to the shared library, and then add the second

7:54

person added to the shared library. Still in

7:56

beta, maybe it'll change that. And interestingly, I

7:58

was like, okay. Well, maybe they do that

8:00

for safety or, like, just to make sure photos

8:02

don't squish other photos with edits or whatever. But

8:04

isn't there a d duke feature that they've been talking

8:06

about in iOS sixteen and stuff?

8:09

they will find your duplicates for you, like built

8:11

into the iOS sixteen photos

8:13

thing. I believe so. If that exists

8:15

in Mac photos, I could not find it. So

8:17

it's kind of weird if that ends up being an

8:20

IOS only feature. There, obviously,

8:22

there are tons and tons of ways to do photos

8:24

in your photo library with Mac applications.

8:26

Lots party applications have been doing that for years,

8:28

but it's kinda weird that if they're bringing

8:30

that as a first party feature to the photos app

8:32

on the phone that they don't also bring it to the Mac.

8:34

So I hope that appears

8:36

in a future beta or something. We

8:38

were sponsored this

8:40

week by Khaled. IT admins

8:42

often feel like they've between their commitment to cybersecurity

8:45

and they're due to protect their employees'

8:47

privacy. Naturally, of course, you need to

8:49

safeguard company data against hacks and

8:51

breaches. you also don't want to turn your

8:53

workplace into nineteen eighty four. Traditional

8:55

MDMs give the IT team complete

8:57

access and control over company

8:59

devices to a fault because Since

9:01

employees are inevitably gonna use their work

9:03

laptops for some personal activities, those

9:05

tools can saddle you with surveillance capabilities

9:08

over personal stuff you never wanted

9:10

access like photos or browser

9:12

history. Before you know it, your end users

9:14

are complaining about security agents slowing down

9:16

their laptops. Developers are frustrated

9:18

by the lack of autonomy. and people

9:20

start secretly working on their personal devices just

9:22

to get things done. That's in that

9:24

case everyone loses. So it's easy to

9:26

fall into these traps of

9:28

top down security. but not

9:30

the only option. Collide

9:32

is an endpoint security solution built

9:34

around honest security. Their

9:36

philosophy is that employees aren't your biggest

9:38

security risk they're biggest

9:40

allies. Your relationship with them should be based on

9:42

transparency and informed

9:44

consent. Collide works by

9:46

notifying your employees of security issues

9:48

via Slack. edging them on why they

9:50

are important, and give them step by step

9:52

instructions on how to resolve them

9:54

themselves. For IT and security teams,

9:56

Kalai provides the right level of visibility

9:59

for Mac, Windows, and Linux

10:01

devices. And it addresses high risk

10:03

issues that can't be solved through brute force

10:05

or automation. and your end users can

10:07

see exactly why and how

10:09

every piece of data is being collected by a

10:11

collides user privacy center and

10:13

their open source code base. you

10:15

can meet your security goals without

10:17

compromising your values. Visit

10:19

cali dot com slash ATP to find

10:21

out how You follow that link to hook you up

10:23

with a goodie bag just fractionating a free

10:25

trial. That's collide, K0LIDE

10:29

collide dot com slash

10:31

atp. Thank you so much to collide for

10:33

sponsoring our show.

10:38

Alright. The European European

10:40

Union has mandated that

10:43

all phones in somewhere between

10:45

one and seventeen thousand years

10:47

will have USB C on them

10:49

as is for or as as

10:51

was foretold. Yeah. Well,

10:53

do we know I mean, the the

10:55

the europeologists here will tell us,

10:57

like, well, this actually they didn't

10:59

really mandate it yet. Now it has to go to this

11:01

committee, then it has to go to this board, then it has

11:03

to I bet there's something I'm gonna has the to

11:05

those questions. So let me read from the

11:08

notes. I didn't know

11:10

if we were going to make any initial opening

11:12

remarks, I guess not. here we go. That's why you shouldn't

11:14

have editorialized based on the

11:16

based on the title item. Mhmm. Because

11:18

notice what the title item says, would you like to read the

11:20

title item as written? EU

11:22

USB C mandate passes vote.

11:24

Passes vote. It does not so

11:26

it's mandate as a noun. Right? The

11:29

USB C mandate What happened to it? Did it

11:31

go into effect? Is it real? Is it whatever?

11:33

No. But it passed the vote. What the hell does passes the

11:35

vote? I mean, and here we get the text. I

11:37

mean,

11:37

look, hey, hands up who wants us to

11:39

pass? three of us. One, two, good. Okay. She

11:41

has to vote. Like, that could that doesn't mean anything.

11:43

Oh, I mean something. Anyway, gone. Alright. On

11:45

October fourth, European parliament voted

11:47

overwhelmingly. in favor of new legislation that

11:49

would eventually require all mobile

11:51

phones sold in the EU to use the to use

11:53

a USB c port for wired

11:55

charging. The EU's new rules

11:57

are yet to be formally approved. Although

11:59

they've been given the thumbs up by the bloc's

12:01

parliament, the common charger legislation still

12:03

needs to be signed off by the Council of

12:05

the EU and published in the EU

12:07

official journal. It would then enter

12:09

into force twenty days later.

12:11

So clear. But even

12:13

once that happens, just you

12:15

wait. But even once that

12:17

happens, companies like Apple will still

12:19

effectively have a two year grace period. That's designed to

12:21

ease the transition to a USB C future.

12:23

This means the rules are likely to come into force

12:25

by the end of twenty twenty four.

12:27

Devices already on the market what

12:29

won't need to be withdrawn. So if Apple

12:31

launches a Lightning port iPhone ahead of the

12:33

deadline, it can keep selling the

12:35

phone. Or what if they just kill

12:37

the ports entirely. They already did one. Why not

12:39

take the other? Yeah. So twenty twenty four

12:42

is a is a ways out if, you know,

12:44

things go according to what they seem like they're go to.

12:46

That's plenty of time for Apple to

12:48

transition to SBC, quote unquote, on

12:50

its own. You know what I mean? I love that it

12:52

says, like, companies like Apple

12:54

too. Like, this is all about Apple. Like, what other

12:56

what other companies are left? There's no comp as

12:58

as Tim Cook will tell you, there are no other companies

13:01

like Apple. Only apples would be stubborn enough to keep

13:03

lighting for this long. Nice.

13:05

I I don't know how I feel about

13:07

this. So I I'm gonna seal from

13:09

former guest, Christina Warren, who

13:11

tweeted about this, and I I agree with her.

13:13

I would like USB C on my

13:15

phone for several reasons which we can investigate if

13:17

we care, but I would like USB C on my

13:19

phone. I don't

13:21

love that it will arrive there

13:23

if at all. because of a

13:25

government mandate rather than either

13:27

market forces or I don't know.

13:29

And I feel very like, oh,

13:31

free market this. You know, no government. And I

13:33

feel about most things, but in this particular

13:35

case, I I don't love how we're

13:37

ending up on a position that I

13:39

think I'm going to love. I mean, it's gonna it's

13:41

gonna end up there not because of this mandate. Because

13:43

if the twenty twenty four date ends up being anything

13:45

close to correct, Apple essentially

13:48

has painted itself into a corner to

13:50

transition regardless of this law because they

13:52

keep making phones that can

13:54

create these massive files

13:56

like shooting the, you know, high resolution

13:58

video with high frame

13:59

rate and trying to get them off a

14:02

phone as a nightmare. And so they

14:04

either have to upgrade lightning to be much

14:06

much faster, which will be big pass hassle. I'm

14:08

not even sure if it's possible, or they have to go to

14:10

USB c. So there is technical

14:12

motivator that's going to make Apple

14:14

make some kind of change whether it's the

14:16

USB C or something else just because it's

14:18

it's

14:18

plain ridiculous how long it would take

14:20

to get

14:21

video off of a phone, you

14:23

know, that you can record at the at the highest quality.

14:25

It would it just takes too long. It's too

14:27

slow. So

14:28

that's that's my guess about why Apple

14:31

will be transitioning. I'm sure

14:33

this mandate doesn't hurt. the

14:36

schedule. Right? It can only help it.

14:38

But given how long these things take

14:40

effect and given how it's been not weakened,

14:42

but like, have even allowing people to ease

14:44

the transition, giving them a long time to do it,

14:46

not having to withdraw existing products, which really

14:48

helps them. Like Apple could release it

14:50

depending on the timing within twenty twenty

14:52

they get released their last lightning phone

14:54

to be the twenty twenty four phone.

14:56

Right?

14:56

And not have to worry about it until

14:58

a twenty twenty five phone. Right? That's how much time

15:00

they have to do this, but I think they'll probably

15:03

change before then. David Schwab

15:05

had some other ideas about things Apple

15:07

could do to skirt this. I think these are less likely

15:09

than the straightforward thing, which is let us go to

15:11

USB c. But here they are, he says the EU law

15:13

contains an exception for devices that only use

15:15

wireless charging. assuming Apple

15:17

really doesn't wanna switch the out front of lightning to USB c, do

15:19

you think it might just replace the lightning cable with

15:21

mag safe charger in the EU implement one of

15:23

these as a legal workaround? These

15:25

are probably not

15:26

increasingly silly, but some of them are silly, like, put a

15:28

service only sticker or plug over the lightning

15:30

port. I don't see that happen. No. You know,

15:32

it's kinda like the little, like, the diagnostic part

15:34

on the watch. disable charging

15:36

through the lighting port and software. Right.

15:39

You could still have the lighting port not USB

15:41

c as long as it doesn't charge through.

15:43

Yeah. only for EU phones. Right?

15:45

Check an EU phone without a lighting part just

15:47

like the US phone doesn't have a symmetry. I

15:49

guess they put a plastic spacer in there.

15:52

And then he says, think EU regulators may have

15:54

underestimated how much Apple hates being forced to modify

15:56

their designs to satisfy regional laws. But I really

15:58

don't think that's what's going to force

15:59

Apple to do. I think that the march of

16:02

progress of storage size and

16:04

the the size of video files.

16:07

And

16:07

from that matter of photo files, is that you're

16:09

taking you're shooting everything with forty eight megapixel raws,

16:11

each of those photos photos is eighty megabytes and you

16:13

got a one terabyte iPhone. Try transferring

16:15

that at lightning speeds at the USB

16:17

two point o speeds or whatever it does. It's it's

16:19

a little bit silly. Their quote unquote pro phones need

16:21

to be able to transfer data faster. And at

16:23

this point, trying to make

16:25

a new version of Lightning that is faster

16:28

seems very silly

16:29

in light of how much USB C is

16:32

spread throughout the rest of Apple's lines. You know, they did

16:34

it on the iPad. they're they're

16:36

either gonna do it on the phone or they're gonna

16:38

get rid of ports altogether, but I don't see

16:40

them like it's not it's not the EU that's

16:42

forcing them to put out a

16:44

nonlighting phone in twenty twenty five is just

16:46

sanity. I

16:47

want them to do it not because

16:49

a government is forcing them

16:51

to. but because it's the right thing to do.

16:54

Exactly. And and I think they, you

16:56

know, the rumors have been

16:58

fairly consistent that starting

17:00

with next year's iPhones that we are

17:02

apparently going to USB C. And that's been

17:04

consistent now for a number of years.

17:06

So I I kind of put

17:08

some weight behind it. And I think, you

17:10

know, Apple must have decided a couple years ago,

17:12

like, you know, to make this change. And

17:16

Changing over the iPhone in any kind

17:18

of major component change is

17:20

is not a small deal. They have to

17:22

worry about, like, first of all, can

17:24

we even get or create

17:26

enough USB c

17:28

connectors to to keep up with the

17:30

iPhone's volume? Like, that's actually a real concern that,

17:32

you know, something that's as high

17:34

volume and as,

17:36

I guess, high stakes

17:39

as the iPhone is. Like, because every single

17:42

thing that an iPhone has or

17:44

has to do has to

17:46

be nearly a hundred percent perfect,

17:48

nearly a hundred percent of time

17:50

because they just sell so many of

17:52

them. And it's so important to the company that

17:54

if they have a part that has like

17:56

a point 001 percent

17:58

failure rate, that's too high. They can't have that because that

18:00

could cause a scandal that could have a big

18:02

problem for the iPhone that year. Like, you know,

18:04

so they have to be so

18:06

careful and they have to make sure they can

18:08

create the volume and have the high

18:10

yields and have the good reliability of all these

18:12

different parts. You know, when you look at phone when you look

18:14

at the iPhone, as a

18:16

product. It's really amazing

18:18

when you compare it to not

18:20

only anything else that Apple makes,

18:22

but anything else that anybody makes, it's really

18:25

amazing how rarely anything

18:27

goes wrong with them. Like how like

18:29

manufacturing defect wise, like,

18:31

How often have you opened up a

18:33

new iPhone and something's been broken and you've had

18:35

to exchange it? It's almost unheard of like it

18:38

happened so rarely. compared to the number of

18:40

them that they sell. And so

18:42

again, they have to be super cautious. So I'm I'm sure

18:44

there were reasons like that.

18:47

that led them to take this long

18:49

to get here. But I do

18:51

think it looks like things are lining up that

18:53

they will be getting here. And And I

18:55

think, you know, part of the reason, as John

18:57

said, they transfer rates of having these

18:59

giant video captures. They're they're little they're

19:01

marketing the pro phones, and they're

19:04

these software features and hardware features to optimize

19:06

for things like like pro

19:08

res and and raw photos and everything

19:10

that generate these huge files that are comically

19:12

slow to get off the phone that

19:14

is one side of this. But the

19:16

reality is, and Apple knows this, most

19:18

people with most iPhones will never do

19:20

those things. And so that

19:22

doesn't necessarily need to be the reason.

19:25

That's a reason. But I

19:27

think the reason, the the much

19:29

bigger reason, is just that it's a

19:31

pain in the butt to have two different

19:33

phone chargers out there. And

19:35

when Apple went with lightning,

19:37

the Android world was not as unified

19:40

as it is now. Now,

19:42

everything is USB c and it has

19:44

been for a number of years now. And

19:46

it's spreading to all sorts of other devices that

19:49

aren't even phones. You know, the laptops now all

19:51

charge via a USB C. They can, at least. They

19:53

don't have to, but they can. And then you

19:55

look at every, like, all

19:57

hardware in the world, flashlights

19:59

charge of our USB c. Like, the other I

20:01

I saw I got an ad on Instagram for

20:03

a power drill that charges via USB

20:06

C, which by the way, well

20:08

targeted at. Like,

20:11

everything is USB C now. and

20:13

the the very few things that aren't are

20:16

every

20:16

iPhone and like our

20:18

AirPods cases or whatever. It's like it there's

20:20

not much else left on

20:22

the stupid Apple. my keyboard. Yeah. keyboards. Yeah. My

20:25

trackpad. But, like, the Apple

20:27

the Apple battery case that

20:29

isn't the case or whatever it's called. Mhmm. But, yeah, but, you know, you look

20:31

at the market and, like, everything else

20:34

is USB c now. It's a it's a

20:36

very, very different scenario now.

20:38

now than it was when lightning was introduced. You know,

20:41

when they had to decide to

20:43

go from the dock connectors to lightning,

20:45

again, that was a very different world wasn't

20:47

this consensus. There wasn't this one

20:49

amazing universal standard. There was a

20:51

bunch of miscellaneous crap and mostly

20:54

micro USB, which sucked. and

20:56

this is a totally different ballgame now. It's

20:58

a different time with different needs.

21:00

And the right thing to do now

21:03

whether the EU gets around to

21:05

mandating it for them or not is to use

21:07

USB C for everything. The rumors

21:09

again suggest we're going there and I hope

21:11

they're right because it is

21:13

such a pain in the butt, like,

21:15

to have, like, a family of mixed devices of

21:17

just, like, you know, we need USB c for

21:19

pretty much everything except

21:21

our iPhones. Like, that's that's so annoying when our

21:23

Apple watches again, whole separate thing there.

21:26

But the the world would be

21:28

better off if they made the iPhone

21:30

USB c. And I hope that's the

21:32

reason they're doing it. And

21:34

not because of government pressure

21:36

or not because of

21:38

only pro is needing it. No. Everyone

21:40

needs it. It'll benefit everyone. Now

21:42

speaking

21:42

of not having enough parts and stuff, I seem to recall

21:44

a story back when I first came out, that the

21:46

limiting factor on the phones they could manufacture

21:49

was the ability to get that little the

21:51

little lightning connector thing that you

21:53

guys saw obviously Apple was the only company in

21:55

the entire world that needed that thing made, and they

21:57

needed a lot of them, and they needed a lot of them fast, and

21:59

they needed to be up to Apple

22:01

standards of quality and everything, and that was a

22:03

a problem. But at at this

22:05

point, getting quality USB C

22:07

connectors should not be a problem for Apple.

22:09

One

22:09

would hope not. But, no, I mean, I haven't been

22:11

traveling too much, but I've been traveling more than zero, which

22:13

is, you know, more than I can say for the last couple

22:15

years. And having everything

22:17

USB c is extremely convenient. And,

22:20

yes, I did spend an absolutely

22:22

hilarious amount of money on my, you know,

22:24

travel magsafe situation.

22:26

But it would still be nice to

22:28

know that if I needed to plug in, all I

22:30

need is the cable that I can use for my laptop

22:32

or the switch or for any number

22:34

of other things. and that I don't need

22:36

a bespoke cable just for my iPhone. And, yeah, I

22:38

I don't begrudge Apple for having gone

22:41

lightning. I do kind of begrudge them

22:43

for not having gone to USB c

22:45

sometime last year or two. And I think that, you know, it

22:47

would have been better a year or two ago. The next best

22:49

time is as soon as possible.

22:51

And and and I just the

22:53

other day, recorded the kids were doing, like, a

22:55

little play for the two of us for Aaron and

22:57

me. And I recorded it and I

22:59

recorded it as, like, one ten minute,

23:01

you know, don't remember what my settings my

23:03

phone are, but I think it was a one ten minute sixty

23:05

sixty frames per second four k video.

23:07

And it's something like three gigs. I don't even

23:09

remember how big it is, but it's massive. Right? And

23:11

yeah. Getting that oh, I'm sorry. It's

23:13

actually nine gigs. Nine

23:16

gigs. Getting that off of my phone

23:18

via cable was effectively impossible.

23:20

Like, yes, it is it is literally

23:22

possible, but it was effectively impossible.

23:25

And so what I ended up doing was air

23:27

dropping, which took a couple of tries and was

23:29

not exactly reliable, but I eventually

23:31

got it from my phone to my

23:33

computer. And I'm going to eventually, you

23:35

know, put in I in fung fu

23:37

pro and do things with it, but it is

23:39

not fun to get, you know,

23:41

more than a minute of four

23:43

k sixty frames video off of your phone

23:45

using a lightning cable. It's just I know you guys

23:47

said this a minute ago, but it's so true. And this

23:49

is something that I ran into just in

23:51

the last three days. I

23:54

I cannot wait for for USB C

23:56

to be a thing. That

23:58

being said, I

24:00

wouldn't be in highly surprised if

24:03

Apple went no ports at all

24:05

or perhaps no ports on

24:07

non pro phones and ports on a

24:09

pro phone. And the reason I say that is,

24:11

what do they have to care about? They have to care about

24:13

developers who they don't really care about.

24:15

Or people who have

24:17

wired car And there are

24:19

solutions like I use. They're not great, but they

24:21

work to change wired car play into

24:23

wireless car play. So do do either

24:25

of you guys see them going

24:27

to a completely wireless world?

24:29

There

24:29

was that the rumor that, like, the magsafe

24:31

puck. Like, imagine the magsafe puck as they exist

24:33

now, but also with, like, thing the middle of

24:35

until it's data be transferred, then that would be

24:37

their solution essentially, like, MaxSafe. I don't know what

24:39

number they're on. MaxSafe two, three, and then I

24:41

guess the numbers are on the laptop. and I

24:43

believe they reset the timeline on that one. Yeah. It's just it's

24:46

yeah. As we said, like,

24:48

previously, not Magsafe, but Magsafe. Anyway,

24:50

That was they had patents filings on

24:52

that and everything, but it's hard to tell whether that's just a

24:54

thing they were considering for the Magsafe Talk

24:56

and just didn't do or if it's a thing

24:58

for the future. it doesn't really solve

25:00

the the car play problem at all. One of

25:02

the problems it does solve is Apple's ability to

25:05

charge peripheral

25:07

manufacturers money to sell things that sort of

25:09

made for iPhone, whatever. I'm

25:11

sure of any directly does anybody make an

25:13

a mag safe pocket besides Apple?

25:15

Like,

25:15

the actual Yeah. Well, sort

25:18

of, my travel thing that I

25:20

keep talking about over and over again,

25:22

that there, you know, there's three

25:24

pieces. There's a Qi charger

25:26

that's about the shape of an AirPods

25:28

case. There's an Apple Watch charger,

25:30

and then there's a honest to

25:32

goodness Magsafe. not not a puck, but, you know,

25:34

there's a max safe mat on on I

25:36

think it's the right most of the three

25:38

different parts of this charger. So, yeah,

25:40

it is on honest goodness,

25:42

magsafe, but it is not a

25:44

first party magsafe puck as far

25:46

as I'm aware. Well, and and to be

25:48

clear, like, I'm sure they I'm sure they make a

25:50

decent amount of money with the

25:52

licensing of everything, you know.

25:54

But have you

25:54

ever seen

25:55

anybody in the real world using officially

25:58

licensed MFI stuff that didn't come with their

26:00

phone. Everyone just buys the knock off crap in the

26:02

drugstore in Amazon. And

26:04

doubt they're getting any money from that. So III

26:06

wonder if we might be overestimating the

26:08

value of that. I think there's a lot

26:10

of it in that. I mean, I I

26:12

think, like, all of the manufacturers that you see

26:15

selling stuff on Apple Store, for example, like, Belkin

26:17

and stuff there, venue manufacturer. They send tons of

26:19

stuff. If you just do a random for

26:21

any kind of, like, wire peripheral thing. The

26:23

odds of you getting a Belkin match

26:25

are high, and I'm assuming everything Belkin

26:27

does is on the up and up because, again,

26:29

they're an Apple store, so they're probably, you know I don't know.

26:31

Like, it's it's someone's job to maximize

26:34

MFI income. Right? And and that person

26:36

is not in charge of the whole company. but

26:38

that is a factor in weighing this. And obviously, it's not

26:40

gonna stop them from going to USB C.

26:43

Like, they did it on the iPad. Right? No more no more

26:45

revenue from all those lightning cables that we were selling

26:47

to iPad owners. like they'll in the end, do

26:49

the right thing from a technical perspective,

26:51

but it remains to be seen what Apple

26:53

thinks is the right thing on the phone because

26:55

they in the same way they remove the

26:57

headphone port. although they remove that from the

26:59

iPad too. But anyway, they may say, well, on the

27:01

phone, we need every ounce of space we can

27:03

get. It's not like we have room for plastic spacers

27:05

in there. So we gotta get rid of the

27:07

port and everything is going to be

27:09

magnetic pucks from now on. I hope they don't do

27:11

that. It seems like a much more straightforward

27:13

and smarter thing to do go with you. You

27:15

see? And in the end, I think the iPhone for

27:17

all of Apple's sort of

27:20

punctuated moments of daring tends

27:22

to be a conservative product.

27:24

And, you know, change happens slowly.

27:26

So the iPhone ten was a big change.

27:29

going retina was a big change. The big iPhone six

27:31

was a big change and going,

27:33

you know, growing for thirty planes. The lighting was a

27:35

big change. I think, you know, but

27:38

those those events happen, but in

27:40

general, Apple's not

27:42

keen to rock a boat on the idea that a

27:44

phone is something that's a rectangle where

27:46

you plug in a thing at the bottom to charge

27:48

it. So I'm right now, my money

27:50

is still on a USB

27:52

DC port where the lightning port was.

27:54

I don't know. We'll

27:55

see. it is tempting, you know,

27:57

Apple famously frequently overdoes

27:59

their minimalism, especially on

28:02

hardware. And so I

28:04

I see why we would think

28:06

that, you know, oh, this is a big risk. And and

28:08

I think it's a it's a risk that

28:10

they might do this, but I think it's a small risk

28:12

because Ultimately,

28:16

wired is in many ways, of

28:18

course, it's in many ways simpler, but

28:20

it's also in many ways better.

28:22

wired charging, first of all, is way more

28:24

efficient. Now we've already seen, like, I believe there's a

28:26

feature in the sixteen point one beta that came out

28:28

yesterday or today. that there's now

28:30

an option to set your

28:33

iPhone to charge when

28:35

it has, like, environmentally

28:38

friendly energy generation in your

28:40

area or like so if there's like an at

28:42

a time of day in your area where

28:44

they use only solar or wind power

28:47

or something, it can have your phone, like, try to only charge during

28:49

those times. And, you know, that's the kind

28:51

of feature that the reason they do that kind of

28:53

feature is that it has a pretty

28:55

massive environmental impact when

28:57

you're talking about the the number of iPhones

28:59

that are out there. If you can make them

29:01

charge a little more efficiently or

29:03

using certain resources instead of others.

29:06

It's a small power draw. Yeah. But it's

29:08

it's it's like millions and millions and

29:10

millions of small small power draws.

29:12

And so to only

29:14

would make almost everyone

29:18

use an inefficient charging

29:20

method on their iPhone that

29:22

loses a good percentage. I mean, what

29:24

is what is the Chi MagSafe

29:27

efficiencies? Probably something like seventy or eighty

29:29

percent. I think that's a very optimistic Yeah.

29:31

Exactly. I would guess that it's less than

29:33

fifty percent, much less. Right. And especially if

29:36

you have a case on your phone, then you're

29:38

getting those coils further apart, and I bet it

29:40

makes the efficiency worse. So

29:42

if they're gonna put all the effort

29:44

environmentally to do other other good things

29:46

for like energy conservation and

29:48

and know, energy smart energy usage and everything.

29:50

It seems like a step backwards to require all

29:52

of a sudden everyone to

29:55

go you know, cheer or wireless only.

29:57

In addition to the fact that at first, I I just

29:59

I hope that

29:59

this rumor is not true

30:00

or that that this idea wouldn't happen. I

30:03

hope because I

30:05

know from a developer's point of view, like

30:07

the Apple Watch development situation,

30:09

not being able to hardwire to it,

30:11

is just so inferior. to the iPhone where

30:13

I can just hardwire in. So

30:15

I really hope they don't do that. Oh,

30:17

great. Agreed. But but yeah. But I I just I think

30:19

there's there's lots of reasons for this why

30:21

they probably wouldn't do it for any

30:24

iPhone, let alone for

30:26

all iPhones. Yeah.

30:29

I I don't know. I I don't want there to be an old

30:31

wireless feature, but I

30:32

I wouldn't put it past Apple

30:34

one way or the other.

30:36

We responded this week by

30:38

Linode, my favorite place to run servers.

30:41

Visit linode dot com slash

30:44

ATP. look, if you need to run servers and,

30:46

you know, if it was me to the show, I think you're

30:48

disproportionately likely because probably why the advertiser

30:50

does, you probably have needed to run

30:52

a server at some point or need to run one now. Linode

30:54

is by far my favorite place to do

30:56

that. I've run so many servers in my

30:58

career, and I'm actually trying to run fewer

31:00

of them, but for some reason to do that, I need

31:02

to keep adding more. But

31:05

I am I am so happy with Lenovo

31:07

as a customer. They have everything I need

31:09

from compute instances, to specialty plans,

31:12

like GPU compute, high memory,

31:14

dedicated CPU, stuff like that,

31:16

and they have all sorts of managed services now

31:18

as well. So in addition to things I've

31:20

used for a while like load

31:22

balancing and manage backups. They also now

31:24

have block storage, object

31:26

storage like s three, and

31:28

manage database services now launching. And

31:30

so it's really AAA huge

31:32

amount of value being created there. And they

31:35

really don't charge much for their

31:37

offerings. Like, they are an amazing

31:39

value and I have not found a better

31:41

value in the business. And that's why that's why I went

31:43

to them in the first place. And that's one of the reasons why

31:45

I've stayed there all this time because

31:47

I run a lot of servers. The the money would

31:49

really add up in any other place and and

31:51

Linode is by

31:53

far the best value I found in the

31:55

business. So it's just it's amazing. I got an amazing

31:57

support, amazing control panel. They have a full

31:59

API if you need it. sorts of capabilities. I

32:01

don't even understand because I don't need to. But

32:03

it's just it's a fantastic host. So see

32:05

for yourself at linode dot

32:07

com slash atp. and

32:09

let them make your your

32:11

cloud computing fast, simple, and

32:13

affordable. So you can focus on your project,

32:15

not your infrastructure. Once again, linnode dot com slash ATP

32:18

Create a free account there. You get a hundred bucks

32:20

in credit. linnode dot com slash

32:22

ATP new free accounts there get hundred dollars in

32:24

credit. Thank you so much to Lynode

32:27

for being an awesome host and for running

32:29

all my servers and, of course, for

32:31

sponsoring our show.

32:34

So I have a

32:36

couple of updates with regard to my weird

32:39

Apple Watch band return.

32:41

John, not Syracuse. has a

32:43

theory that I genuinely

32:45

wish I could have confirmed or denied,

32:47

but all this came in after I

32:49

had completed my return. John writes, here's why I

32:51

think you were asked to bring your watch bundle to

32:53

return the band. The bundled

32:55

band is not exactly the same

32:57

with as the retail version. as

32:59

you can see in pictures that John provided, which really

33:01

you can just use your mind painting

33:04

to figure this out. It has no bar

33:06

code and no specifications about its color.

33:09

From the case, I can only tell that it's in my case a sports

33:11

band. On the other hand, the retail version

33:13

has more stickers for the model in the barcodes.

33:15

See another picture. again, not gonna bother with

33:17

the pictures. And the show notes partially for John's opsec.

33:20

But anyway, I don't

33:22

think this is the case. I thought

33:24

on certainly on the exterior. You know

33:26

how on the Apple Watch, it's like a

33:28

piece of paper that wraps the watch

33:30

box and the band box. Well,

33:32

on that or it's not a piece of paper, like a that thin piece of

33:34

cardboard. But anyways, on that, it

33:37

says what band is within. Now,

33:39

maybe the band wasn't specific, but

33:41

I could swear. It had a picture

33:43

of the correct band on the outside. I could swear

33:45

it had like the size on it and so

33:47

on and so forth. So I don't know. Maybe

33:49

John is right, but I'm skeptical. But

33:51

it's a plausible theory. Now from an

33:53

anonymous Apple Watch employee or

33:55

an anonymous Apple employee, they

33:57

had the following to say,

34:00

with regard to the genius or or retail

34:02

person that I spoke to, they were

34:04

wrong. We don't need the watch. All we need

34:06

is a serial number of the watch. Here's what actually happens during

34:09

a band swap. systematically regardless of whether your watch is there or not, Apple

34:11

is returning the whole thing and reselling it to you with

34:14

a new band. It's just that they've rejiggered the

34:16

system to

34:18

hide most of the exchange from us as in the retail employees,

34:20

and the customer. The steps look like

34:22

this. One, the specialist chooses the

34:24

item swap option on

34:26

their Isaac which is the handheld device that carry. Two, the system

34:28

asks for a watch serial number. This can be

34:30

scanned using the barcode from

34:32

the box. be

34:34

manually entered if the box isn't there. We

34:36

can also scan the original receipt. Since every

34:38

serial number is unique, that number allows us

34:40

to pull up the original transaction. Quick aside,

34:43

I later asked this anonymous genius, hey, what if I had

34:46

the, you know, W12245679

34:48

online order number? And they weren't sure,

34:50

but they said, yeah, you know, I bet that would work.

34:53

Right? So back to the to the retail employee,

34:55

after we have a step three after we

34:57

have received or retrieve, excuse me,

34:59

the original transaction, The system asks for us to scan in the new

35:01

band. Four. If the new band is the same price as the old

35:04

band, the transaction is pretty much finished except for the

35:06

receipt which is printed

35:08

or emailed. If the new band is

35:10

a different price, the customer pays the difference or receives a refund for the difference depending.

35:12

Easy peasy, lickety split, So

35:16

this is mostly what

35:18

happened when I went back with the boxes and

35:20

so on and so forth. But the the key

35:22

here is that really they is

35:24

a way to get the serial

35:26

number. And once they got that, then they're

35:28

off to the races. And that makes

35:30

sense and kind of stands to reason

35:32

based on what I saw. but here's anonymous

35:34

Apple employee telling us exactly what the

35:36

truth is. Alright. So

35:38

let's talk about something

35:41

that is definitely brand new and definitely hasn't been talked to death

35:43

over the last two months. Let's talk about

35:45

AI art. And I feel like

35:47

we should start right

35:49

away by saying, you should really consider listening to

35:52

Cortex episode one thirty three, the

35:54

ethics of AI art. That

35:56

was a real I

35:58

I really that show. But that was a really, really great episode in which lot

36:00

of the ins and outs of all this was

36:02

discussed. How

36:04

do we want to approach this? I guess,

36:06

we should maybe kind of do the quick summary of what do I

36:09

mean by AIR? John or Marco jumping when

36:11

you're ready, but the general

36:14

just is there have been there's been a lot of work put into various

36:16

products, some open source, some not,

36:18

that allow you to do many different

36:20

things. One of the things they allow you to do is type

36:22

a prompt

36:24

type, picture of the three hosts of

36:26

the external tech podcast drawn as

36:29

pixel art. And these

36:32

these different products will use a

36:34

whole bunch of machine learning and and artificial

36:36

intelligence to try to,

36:39

figuratively speaking, draw whatever picture you've asked them to draw. Some of them

36:41

are better than others and we're gonna talk about that I think

36:43

a little bit. But it's very very

36:46

interesting and some of these products I

36:48

haven't played with all

36:50

of them In fact, I've only played with one of them. But the

36:52

one I played with, a lot of times it gave

36:54

me straight up garbage, but occasionally it would

36:56

come up with

36:58

something reasonable. And when it

37:00

did, it was kinda mind shattering

37:02

that I could ask a computer with plain text, like

37:04

describe a phantom picture I had in my

37:06

mind in plain text. and have the computer basically come up with

37:08

it, it it's really

37:10

wild. And so that's kind of what we're talking

37:12

about here.

37:14

that I don't know. John, I think you were most excited to talk about this. Where

37:16

how do you wanna proceed? Maybe we should list some of the

37:18

ones that are out there

37:19

that you wanna try them all of these links in the

37:21

show notes. There's dolly DALL

37:24

hyphen e as a play on the

37:26

Salvador dali and Wally,

37:28

the the robot from the Pixar movie. There's

37:30

stable diffusion,

37:32

mid journey, Google Imaging, a whole bunch of other ones.

37:34

Right? One of them is

37:36

available as like a standalone application that you could

37:38

run on your ARM

37:40

based Mac what is that one

37:42

called? Stable diffusion, I believe that's the one. Right. But is

37:44

it, like, diffusion b or something? It's

37:46

diffusion b. Yeah. We'll put a link to that. That's the one

37:48

I play with on my Mac as well. A lot of other

37:50

ones have interfaces. I think Dolly used to be invitation

37:52

only for a while, but now I think it's

37:54

open to everybody. So you can, you know, follow the links

37:56

and and try them out and

37:58

people post you know, the

38:00

interesting things they come up

38:02

with. And the the thing I wanna talk about with

38:04

this is that, you know, it's been a topic of

38:06

conversation because I mean, first, it went around, like, I don't know, a year or two

38:08

ago whenever the first one of these started coming out

38:10

as, like, a technical

38:12

curiosity. And started to get

38:14

mainstream enough that you'd see like articles about it and

38:16

just regular sort of tech websites and other

38:18

stuff like that. And at a

38:20

certain point, it started to be

38:22

so mainstream that people were using it, not just as a speculative

38:24

curiosity, but they were using

38:25

it to make a picture. that

38:28

they

38:28

would then use. I think there was some controversy because one person was actually writing. I

38:30

don't think it was an article about AIR, but they

38:33

or maybe it was. They

38:35

used AIR to They talked

38:37

about us in the cortex episode to

38:39

make an image that they included

38:41

with their article online rather

38:43

than rather than paying an artist and people were mad,

38:45

like, why didn't you pay an artist to do this

38:47

and that sort of gets into the the I think the most

38:49

interesting part of this debate is, given that people are

38:51

working on this tech, what

38:53

does it mean for the future

38:56

of all things related to

38:58

making pictures? And by the way, there are movie ones

39:00

now as well. Are you gonna ask it? Like, someone you know,

39:02

you're gonna ask it like, a

39:04

painting of an ice cream cone

39:06

melting in the sun, and it will do a

39:08

video of, you know, a video of a painterly

39:10

style ice cream cone melting in the

39:12

sun. Right? So it's not just the and still

39:14

images. And this stuff is developing

39:16

so quickly. It presents

39:19

a lot of very Thorny

39:22

questions. Obviously, if you

39:24

are if your profession is

39:27

drawing pictures for money, and

39:29

there's a program that lets people

39:31

draw pictures by typing what they wanna see

39:33

in the picture that probably doesn't make you

39:35

feel good about your

39:38

chosen profession. there is the predictable sort of

39:40

luddite versus check enthusiast,

39:42

you know, battle there. between

39:45

saying, oh, computer can never do what a human

39:48

does and, you know, like,

39:50

the anytime there's any

39:52

kind of technology that previously

39:54

does something could only be done by

39:56

humans. There is this battle saying that the new way to do it is

39:58

soulless and bad and evil

40:00

is going to corrupt youth

40:03

and so on and so forth. And then the other people who were excited about

40:05

the tech and just wanted to go forward. And many times throughout

40:07

history, there has been a technology

40:09

that has caused entire professions

40:11

and entire industry used to

40:14

basically disappear or shrink to the point

40:16

or transform in a way that's, you know, not

40:18

even recognizable, witness the entire

40:20

industry surrounding. having horses

40:22

pull things with people in them and the advent of

40:24

the automobile. Like, we don't have horses. Not like people

40:26

don't have jobs making saddles for horses and chewing

40:28

horses and taking care of horses. All those

40:30

jobs still exist. But boy, that industry looks a lot

40:32

different than it did before the end of the

40:34

automobile. A lot different. And so

40:37

this AI art thing brings all

40:39

those issues up and that people are going around around that

40:41

debate. But I think one of the most

40:43

interesting things the interesting aspects of

40:45

this debate is how these how these

40:47

things work? Like, how do you how do you make a program where

40:49

you type in words and it draws you pictures.

40:52

Right? And like most sort of machine learning

40:54

style models, they

40:56

are, I don't know, trained is the right word, but

40:58

they are they are given

41:00

a set of images

41:02

to to and you know, and

41:04

associated words and phrases and stuff

41:06

to say

41:06

to to feed into the model so

41:08

that they can do this. Right? So

41:11

I don't know. So like millions and millions of images,

41:13

I'm not sure how they're tagged. Maybe each one is just

41:15

tagged with a caption or something like that. And

41:17

they grind that up into a big soup. and,

41:19

you know, that's the big deal term. And then when

41:21

you say, you know, a picture of an ice cream cone, they give

41:23

you a picture of an ice cream cone because they have

41:26

enough images of enough things within enough of the

41:28

words associated with it that they

41:30

can synthesize a picture based on everything they've ever seen

41:32

and the association of those words of the

41:34

images to get used not

41:37

just one, but multiple pictures of ice cream cones

41:39

and, you know, trying various attempts

41:41

at it. And

41:42

and that

41:44

raises a lot of interesting questions in particular,

41:47

what images were

41:49

fed to this thing? I

41:51

most of the things that are done

41:53

for, you know, universities or, you know, this is

41:55

like sort of research type stuff. And the the

41:57

first versions of these, there are lots

42:00

of image sets that are used for

42:02

research purposes that are

42:03

presumably, like, millions and

42:06

millions of correctly

42:06

annotated royalty

42:09

free images that have been used in

42:11

lots of different computer vision studies for

42:13

years and years. Right? but

42:16

there are so many of these things and they're so popular and the

42:18

new ones come to every day. It's

42:20

not entirely clear within

42:21

all the image sets that they're using. So there was

42:23

a story about this on our

42:25

sector, oh, I guess, actually is fairly recent.

42:28

Hebine is artist finds private medical

42:30

record photos in popular AI

42:32

training dataset. Someone

42:33

who found

42:36

private medical record of photos taken

42:38

by her doctor in twenty thirteen referenced

42:40

in the LAION five b

42:42

image set. Right. Whoops. Because a

42:45

a lot of the ways they find images is like,

42:47

oh, I'll just scrape the web. And anything I find the

42:49

web, I'm sure it's fine to use. no. Because

42:51

there might be a Docker website that has poor security and has an image exposed to the web

42:53

and a and a web scraper comes along

42:55

and finds it. And low

42:58

and behold, your, like, medical images end up in a in a dataset that

43:00

is, you know and and to be clear,

43:02

is this is this also how my email

43:05

address ends up on so many people's

43:07

list. Oh, you must have both of them from us a nope.

43:09

I sure didn't. Well, I mean, it's

43:11

like bad website security in that, like, there's

43:13

a page that is detected,

43:15

you know, security through obscurity. Technically, if you know

43:17

the URL, you can get to it, but there's no

43:19

link to it. So how could people find it? Well, a lot of the

43:21

web scraping things can find links that are not

43:24

visible because they're hidden on a page or

43:26

they do scraping techniques that allow them to find things by iterating on IDs or stuff

43:28

like that. And that's not

43:30

great.

43:31

And on top

43:31

of that, there,

43:33

as we all know, if you are an

43:36

artist who, you know,

43:38

does any kind of art and puts it on the

43:40

Internet, you will find

43:42

that art all over the place. And you may think this is a show

43:44

with three software developers that

43:46

doesn't apply to us, but

43:48

we make a small amount of

43:50

art, you may have seen some of it and purchased some

43:52

of it on our t shirts. And let me tell you

43:54

that very simple art that is on our t shirts

43:56

is all over the freaking web because

43:58

we put it there but it's

43:59

everywhere. Right? If you put

44:01

an image on the web. Oh, when you go to Cotton Bureau

44:04

page and it shows you a picture of the search, there's that

44:06

image on the web and a web

44:08

scraper could find it because that page of Con Bureau, totally

44:10

unprotected. So if a web scrape defines it, that

44:12

image is probably in some image set somewhere.

44:14

Right? It's probably in all these

44:16

image sets. because, you know, it's the wild west out there in the Internet. You can find

44:18

the image, you can put it in. So there are you

44:20

know, and obviously, this is just a tiny, you know, we're

44:22

doing t shirt graphs, whatever. Imagine you are an

44:24

actual artist

44:26

by profession. And you do this artwork and maybe you have it on your website that

44:28

shows your portfolio of all your great artwork that

44:31

you're gonna do. And your

44:34

artwork probably also your name and the descriptions that stuff gets

44:36

shoved into one of these image sets. To

44:38

the point where, let's say, you're a famous artist and

44:40

you have a name that people know, you know, it

44:44

says, like, you know, picture of a toaster oven

44:46

in the style of Ralph Macquarie.

44:48

Right? That is knows

44:52

who Macquarie is based on the images they pulled, is they're all of the freaking Internet, and

44:54

his name is attached to all of them. And it's gonna

44:56

show you a toastroven drawn in his

44:58

style. I think he's dead now, so it's not. But

45:00

Napa, like,

45:02

tons of living artists, like, you'll type something over these things that

45:04

are like, that looks a lot like something

45:06

I did once. And you know why?

45:10

Because some of your artwork is probably fed into this machine, and it's

45:12

popping out. And these people are like, well, you

45:14

know, how can you do that? The are you allowed?

45:16

They're like, what we're It's

45:18

not really your art. made It's work. It's like, yeah,

45:21

but it's an original work informed

45:23

by work that

45:26

I did. And that doesn't seem like, should I get some kind of royalty for

45:28

this? Should you have my permission? Does that count as

45:30

a derivative work for copyright reasons?

45:32

Yeah. Or should should you at least get my permission to

45:34

include any

45:36

of my in there or should you clean your data set to make sure that the images you

45:38

have, you actually do have the rights to, and

45:40

that's almost impossible because you need millions of

45:42

images to do this or at least a very large number of

45:44

images. And having a human

45:46

vet each one for copyright is just, you

45:48

know, it's like everything that's true about the

45:50

Internet is true of these image

45:52

things. Like, the the

45:54

perfect world where you're like, oh, we have to make sure

45:56

every one of the images that contributes to this to

45:58

this input is free and clear and we need

45:59

all the rights to it and that's impossible

46:02

at scale. Like, that's just not how the Internet works. We, you know, we can't

46:04

even get all the movies and television shows made

46:06

in the pre Internet or on streaming services

46:08

because people can't figure out how

46:10

to do Right? So that is

46:12

a much smaller problem than millions of images and image sets. So

46:14

there's that whole debate

46:16

and rat hole about do

46:19

artists

46:19

deserve to get paid? You

46:22

know, should should they be allowed to do

46:24

this? Is there some kind of royalty

46:26

structure? Should should this stuff

46:28

be removed?

46:28

I think the

46:30

final interest model two

46:32

more interest things about this. One

46:36

is, given that that's the the way

46:38

these things work, Well,

46:40

actually,

46:40

before we run to that, I actually ask you too. Do you

46:42

have an opinion on the on the the

46:44

artist having their work sucked into the thing? Like,

46:46

what do you what do you think

46:48

about about

46:49

the validity of the artist complaining in that

46:52

scenario. I think it's

46:54

really still yet to be

46:56

proven like what are acceptable standards are for

46:58

this. So you know, my my

47:02

barometer for, like, what is

47:04

an unacceptable level of copying, you

47:06

know, just ethically. And and there's legal definitions as well, which I think

47:08

kind of can port with this. But

47:10

anyway, is it's based on,

47:12

like, are you to to make a

47:14

new work

47:16

are you pretty much lifting most of

47:18

your stuff from the same source

47:20

or the same very small number of

47:24

sources? then that's kind of over the line. Whereas if you

47:26

are taking bits of inspiration from

47:28

a diverse set of sources so

47:32

that the resulting work doesn't look like just a straight

47:34

up clone of like one other person's work, but

47:36

it looks like, okay, maybe you were inspired in this

47:38

way by this person. And in this way by this

47:42

work, and this way by this style, you know, but it it all comes together into

47:44

a more diverse soup of a

47:46

product. I think that's okay. And so

47:49

you can look at these AI generators and say, well, you know,

47:51

if you ask for, you

47:54

know, something that that is

47:56

in one particular

47:58

person's style, that could result

48:00

in something that is over the

48:02

line, whereas if you just ask

48:04

for an image of a, you know, a slice of

48:06

pepper any pizza on a table, Like, that's gonna be

48:08

probably drawing from so many different

48:10

data points and and input

48:12

sources that I don't think if

48:14

the texture on the

48:16

pepperoni slice happens to look like the

48:18

way you text or something in photoshop once,

48:20

I I think that's less of a concern.

48:22

But the problem is, you know, you can

48:24

use these tools the way whatever the

48:26

operator wants. And if the

48:28

operator says rip off this one

48:30

person's work or their or their

48:32

style, you're gonna have a problem.

48:35

But I don't necessarily know that that's

48:37

the fault of the technology. That's the

48:39

fault of the user. yeah. So

48:41

assigning blame on on on this, you

48:43

know, how is this different than, you know,

48:45

other scenarios? Signing blame

48:48

is always fun when it's a computer program, quote unquote, doing it,

48:50

but then the user is prompting it to do

48:52

it. And that kinda leads to

48:54

the question, that you

48:56

usually end up at in these habit rates

48:58

is, is this thing doing

49:00

anything different than one people do? If you

49:02

ask a human artist to

49:04

draw something, they

49:04

have a corpus of images they've seen through

49:07

their entire life. That contributes to

49:09

the output. Right?

49:11

the it's, you know, you

49:12

can say, like, well, if I give it to a person, they're gonna do

49:15

original work. But the original work of an

49:17

artist is necessarily informed by their entire

49:19

life experience of seeing

49:22

everything. I've

49:22

seen things in real life, obviously, but also I've seen other

49:24

pictures and works of art inevitably.

49:27

And some people would

49:28

say, well, this AI program what

49:31

it's doing is absolutely no different than what a human

49:33

does. It has a series of inputs and

49:35

that contributes to what it's going to

49:37

make. If you ask a human, to give

49:39

you a logo in the style of a Solvass logo,

49:41

they can probably do that because they know about those

49:43

logos because they're very famous and he's a very famous

49:45

logo designer. And if you do that, he's

49:47

not gonna keep hitting a dead deal. He's not gonna rise from

49:50

the grave and sue you

49:52

because you can't

49:54

sort of trademark a style. If

49:55

I tell you to draw something in the style of

49:57

any living artist, you can do that

49:59

and they can't say, oh, it's a

50:01

legal future that just copied my

50:03

style. Now they may look down on you and say you didn't come

50:05

on with your own original style, but

50:08

every style is a, you know, everything's

50:10

a remix. Every style is a that that we think of as new and

50:12

novel is itself informed by all the other styles

50:14

that came

50:16

before it. it So in one

50:18

sense, I agree that this program

50:20

is doing something

50:22

that if you squint, it looks very similar

50:25

q what to what people also do. But that

50:27

leads to the second question, which is,

50:29

can this program Not

50:33

not can't make anything new. Can't anyone of these

50:35

brothers make anything new. But,

50:37

like, if you if you sort of fast

50:39

forward this, you do the whatever it

50:41

is. It's not argument to add absurdum now. But if it's if you just it's

50:43

an infant time on argument. Okay? So No.

50:46

For instance, say the the

50:49

artists become the horse and

50:51

buggy salesman. Right? And they still exist, and they're

50:53

out there. But, boy, there are a few a lot fewer

50:55

of them because these AI programs get so good that

50:57

in the in the average working life of a person,

50:59

nobody actually pays an artist to do anything.

51:01

We just type words into a program when we get

51:03

an output. Right? If

51:05

all the

51:05

inputs to these programs are

51:08

images made before these

51:10

programs existed, then how does

51:12

that

51:12

sustain itself? Can you feed the output

51:14

of these things? back in is the So forget about

51:16

computers. You've just got humans making art. Humans

51:18

make art that new humans arrive and see the

51:21

art made by the previous humans and they in

51:23

turn make quote unquote new art that

51:25

then the future human see and it feeds back in. Right? So

51:27

you can see how the people are kind of like programs

51:29

in this scenario where they they see

51:31

existing art. They make quote

51:33

unquote new art. and then that cycle repeats itself. If you

51:35

took the humans out of

51:36

the equation, could the machines continue

51:38

to do the same to do

51:41

the same thing thing? taking as input all

51:43

the art ever made by humans, and then going forward, taking

51:45

as input all the art made

51:47

by AI programs. or would they

51:49

would they stagnate and feed in on themselves to

51:52

everything was just a giant gray mush

51:54

because, like, you know, mixing all the paint

51:56

colors together, or would they be just as a versus as human And

51:58

that I think makes me personally circle

52:00

back to, are they doing what people

52:04

do? And I think fundamentally, they are not doing

52:06

what people do. Big strokes seems like

52:08

they're doing, hey, they see pictures and they make new

52:10

pictures. That's exactly what

52:12

people do. but

52:13

it's not. Like, that's what Lloyd's AI

52:15

things, and that's why, you know, general

52:17

artificial intelligence or whatever you wanna call it is

52:19

so far away.

52:20

This Even if these these programs were operating in the exact

52:22

same way that the center of our brains

52:24

that makes pictures do and they're not.

52:26

But even if they were,

52:29

there's so much more to human

52:32

mind than the

52:32

part that makes pictures based on

52:35

work prompts. And what these

52:37

programs don't have and won't

52:38

have for a long, long, long time is

52:41

the life experience of a human,

52:43

all the sensory input they've ever had,

52:45

all the emotions they experience the

52:48

the way humans judge a

52:50

picture, whether it accomplishes the goal they set out

52:52

from us, the ability to set a goal for

52:54

themselves that the ability to experience art and make it and

52:56

feel what the art is meant to feel, thus

52:58

judging whether this art has achieved what you

53:00

wanted to

53:02

achieve or inspiring you to do something else based on how something you saw you feel. None

53:04

of these programs can do any of

53:06

that and it necessarily makes

53:10

their there the the funnel through

53:12

which they have to shove all of their creative

53:14

efforts so narrow. They they do not

53:17

have the wealth of experiences of a human.

53:19

All they have is visual input

53:21

and description. or just such, you know, they don't they don't have

53:23

an experience of the art. So the art that they

53:25

make can only be informed by that

53:27

those tiny little things because they literally

53:29

can't experience anything else they have

53:32

no memory, no life not

53:34

memory in that sense. Like, no memories, no

53:36

life experience, no sensory organs, no

53:39

emotions, no thoughts, no awareness like

53:41

they're not they're not

53:42

artificial intelligence in that sense. Do

53:44

you need that

53:45

to make a picture of repeats on a slice

53:47

on a table? No. But I think you

53:49

need that to continue the

53:52

cycle of creation

53:54

of art with the

53:56

quality level that we have come to expect from

53:58

humans. Because as we make each new generation of humans,

54:00

they have new experiences, their their life

54:02

experiences, and the art

54:04

that they see and the things that they feel inform, the things that they create,

54:06

and it is a rich tapestry as they

54:08

say. And it's great to be able to feed that into

54:10

an AI and have it chop

54:12

that down. But if you take the humans

54:14

out of that equation and leave the AI as stupid as they are now,

54:16

it would basically be like they were working

54:17

from the same set of data forever and they would

54:20

just grind it to

54:22

a pulp and it would just be this

54:24

incredible stagnation. Not that I think this is gonna happen because you can't stop humans from making

54:26

things unless the the machines kill us

54:28

all terminator style, we don't have to worry about

54:32

that. But just as an academic exercise, I think

54:34

AIR is a sustainable

54:37

thing without human creativity as

54:39

an important input. it would be

54:41

sad to think that the only purpose of

54:43

human creativity in our work would be defeated to AIs to do

54:45

most of the drudgery, and then, you know, again, they'd be like

54:48

the people who own horses now. They're out there. There's a

54:50

lot of them, but not nearly as many

54:52

as there were. And I don't think we have

54:54

any particular fear of that in

54:56

our lifetime. But

54:57

that's kind of where I I come down on this, setting

54:59

aside the the legalities and everything. These programs are

55:01

so dumb and so bad at what

55:03

they do. We're impressed

55:06

you know, it's this this one analogy like seeing it right now as a dance. You're impressed

55:08

that it could do it, but boy, the dancing isn't that

55:10

great. Right? And they're

55:12

never going to be adequate

55:16

to sustain a creative

55:18

timeline of works of

55:21

art like humans until their experience

55:23

of life is as rich as a human's experience life which point we have

55:25

lots of other problems. And

55:27

we're not even

55:30

close So don't worry about it. Don't let the people who are telling you that AI is gonna take over

55:32

and kill us all. If you're listening to this now, that

55:34

will not happen when you're alive, so don't worry

55:36

about kinda

55:38

like self driving cars. So Well,

55:40

but I think there's actually some overlap there,

55:42

you know. It because I don't think

55:44

that this is going to put artists, you

55:46

know, out of business as

55:49

a whole. It's more

55:51

like thinking about this is a

55:53

this is a new digital

55:56

tool that can save a lot of busy

55:58

work. It's gonna make certain types of

56:00

art more accessible than they

56:02

were before. to more people. And it's going

56:04

to save a bunch of time

56:06

on on work that previously

56:08

was more

56:10

manual. if you think about it kind of

56:12

like, you know, when when, you know, digital art came around, when Photoshop and

56:14

everything came around and and digital drawing

56:16

tools and and things like that,

56:19

there was a whole industry before that

56:22

of people who were doing a lot of this stuff by

56:24

hand. If photo retouching by

56:26

hand, you know, painting and

56:28

illustrating by hand, And when

56:30

you move to digital, a whole

56:32

bunch of things got easier. And

56:34

things became much more easily

56:36

possible that weren't easily

56:38

possible before. And so that did inevitably put out

56:40

of work like sign

56:42

painters and things like that, you know, to

56:44

some degree, but

56:46

most people who were artists in some way

56:48

embraced the new tools in

56:50

in some form and just became, you know,

56:52

their job just became a little bit different. But

56:56

it didn't kill the art. It just changed what was

56:58

out there and what was available and and how you how

57:00

you had to use it and and what was

57:02

possible. And a bunch again,

57:04

a bunch of new people were able to do it, who weren't able

57:06

to do it before, or or maybe it was, like, a

57:08

little bit too tedious before. And now people

57:10

were able to do these things who, like,

57:13

wouldn't have done the old tedious way, but we're we're willing to

57:15

do the new digital way. And so it just

57:18

it changes things. And

57:19

not everyone comes

57:20

along on those transitions. You know,

57:22

every time technology gets better.

57:25

Certain jobs aren't necessary anymore, you

57:27

know, the horse analogy. And not

57:29

every person who was keeping

57:31

horses became an auto mechanic. You know, that that it's it's and

57:33

that doesn't work that way. But a lot

57:35

of people, you know, a lot of people do become

57:37

auto mechanics when when that demand arises up.

57:40

In this case, you know, when digital

57:42

arts tools came around, a lot of

57:44

people became digital artists. They were

57:46

not everyone who was previously drawing stuff by hand,

57:48

which by the way still exists and is fine,

57:50

but like not everyone who did that

57:52

went to digital, but many people

57:54

did and many new people started

57:56

on digital. And so art is a thing

57:58

that's that's still it's still a major thing in

58:00

the world. It's just different than it used to

58:03

be. I think these AI tools using them

58:05

and figuring out how to how to make

58:07

these text prompts, how to control

58:09

them, what knobs and

58:11

dials to adjust, how you

58:13

word things, what you even think to

58:16

create. That's all art.

58:18

That's part of the process. These

58:20

are now just able to generate

58:23

things, you know, much more quickly than a human can,

58:25

but then humans are still directing them. Humans

58:27

are still tweaking them. Humans are still

58:30

deciding, okay, you know what, generate pictures

58:32

of this thing. And I'm gonna pick the

58:34

one that I like out of this hundred and

58:36

have you riff on that a little

58:38

little more.

58:40

And then go to that one. Generate a hundred riffs on that one.

58:42

Okay. I'm gonna pick these two. Let's

58:44

follow these through and do more with

58:46

these. Like, That's art.

58:48

That's humans doing art with a

58:50

different tool. And it doesn't have to be

58:52

entirely used for entire images

58:55

too. as the tooling and technology gets more,

58:57

you know, more mature and and

58:59

more established. These kind of tools can be used

59:01

for things like Okay. You know what? I'm

59:03

drawing this thing in Photoshop. I have a brick wall here. Can

59:06

you just put a brick texture on this wall that looks

59:08

good? That hasn't been used a million times by

59:10

everyone else has ever used Photoshop their life. and can generate

59:12

a brick texture. Fine. You know? Or, hey, you

59:14

know, this car that's in the background of this photo,

59:16

I don't want this car to be here. Can you delete

59:18

that in a way that's even smarter than

59:20

Constant Aware? fill and stuff like that.

59:22

Like, as the AI tools get

59:24

better, it adds a lot of

59:26

capabilities for

59:28

artists to eliminate busy work that used to exist or to

59:30

do things in a in a nicer way than they used to

59:32

be able to be done. And so

59:34

III see this

59:36

really as a mixed

59:38

bag. You know, there are, yes, some

59:40

downsides and some artists

59:42

will be put out of work by this, but it

59:44

also opens up so much potential

59:46

for artists to use these tools. The work

59:48

that that is going to be reduced

59:50

by this is gonna be

59:54

stuff like the crappy client saying, hey, can you show me fifty different

59:56

version of this of my of my logo?

59:58

You know, like, maybe they can skip that step and

1:00:00

move move on to more

1:00:02

interesting things. And again, that's

1:00:04

that's not gonna take everyone along with them.

1:00:06

But I don't see these tools as

1:00:08

like a universal bad or like

1:00:10

a a doomsday scenario for human

1:00:12

creative artwork. quite the opposite. It's just new tools for humans to

1:00:14

use. I think that the crappy state these things

1:00:16

are

1:00:16

now, like the relatively primitive state

1:00:19

think now is the time that is the

1:00:22

most rich for artists to

1:00:24

potentially have legal action against them

1:00:26

because it's very difficult to tell without sort

1:00:28

of knowing you know, I

1:00:30

I know this is not how they work, but imagine if

1:00:32

you could ask one of them. Okay. So you made an

1:00:34

image for me. Can you tell me what images

1:00:36

contributed to this image that you made? And that again, that's

1:00:38

not how that work. They don't take five images and

1:00:40

smush them together. But, like, big

1:00:42

picture wise in the abstract, lots of

1:00:44

millions of images of input and then

1:00:46

output. Right? And sometimes I can imagine that these more

1:00:48

primitive, very early versions of

1:00:50

this produce a a work

1:00:52

where you could overlay an actual

1:00:54

existing thing

1:00:56

from its corpus on a section of it and say, okay, this is literally

1:00:58

just lifted. Like, it's it's smooshed and

1:01:00

smoothed a little bit, but literally I I drew

1:01:02

the slice

1:01:04

of pizza. and you put it in the image and you rotated it and scaled it.

1:01:06

Right? And that, you know,

1:01:08

that's a no go. Like, you would get if you did that

1:01:10

with if you did like a,

1:01:12

you know, a

1:01:14

cover of a magazine and you did it by, like,

1:01:16

stealing the cover of a different magazine

1:01:18

and just, you know, cropping out everything except for

1:01:20

the slice of pizza. Like, draw your own slice

1:01:22

of pizza. Right? there is a line to

1:01:24

be drawn there like, oh, I was doing a collage or whatever,

1:01:26

but that's what these legal cases are about. And, you

1:01:28

know, if there's a whole other thing to be

1:01:30

said about the the sad state of legal cases on songs that

1:01:33

are identical. But for artwork, you could say, oh,

1:01:35

I was doing a collage. It's a derivative work, so

1:01:37

on and so forth versus I just straight up

1:01:39

lifted this pizza slice. from this other

1:01:41

artist thing. It didn't change it enough for it to be legally distinct. As these things

1:01:44

get better, they will be less of that, less

1:01:46

chance of that happening. That it really

1:01:48

will be

1:01:50

all new totally fresh work. But part of that relies on

1:01:53

a big sort of leveling up

1:01:55

of these things

1:01:57

in understanding

1:01:59

literally anything. What

1:02:01

they understand now is so limited.

1:02:03

You could say these things don't know what a slice of pizzas, but

1:02:05

they kind of do because of the pictures

1:02:07

of slices of pizza and the fact that they could say that triangle

1:02:09

thing is probably the pizza slice because I have a hundred

1:02:12

thousand examples of it. Right? I mean, in all

1:02:14

fairness, like most of America doesn't notice slices of

1:02:16

pizzas. Right. Exactly. but they don't

1:02:18

actually know. They may be able to

1:02:20

pick out the thing that corresponds to

1:02:22

pizza from an image, but they have

1:02:24

no idea what

1:02:26

pizza is. you know, could could they do something like draw heat

1:02:28

lines coming off of pizza? Only because they

1:02:30

have existing artwork that shows heat lines, but they don't

1:02:32

know pizza as hot. They don't know what hot is. They don't know

1:02:34

what pizza is. They

1:02:36

don't want food is. Like, they're again,

1:02:38

they're they're they're so incredibly dumb that even

1:02:40

even if they're synthesizing new

1:02:44

images in the same way that our brains synthesize new images, our brains

1:02:46

have so much other stuff that informs

1:02:48

the thing that we're making, which is why you're saying

1:02:50

murder, you need a human to guide this because these

1:02:52

things know nothing. And even

1:02:54

guide them to do things that require

1:02:56

them to have literally any understanding of

1:02:58

anything that they're doing. Right? You

1:03:02

know, like, Could you put more play settings at that table? Maybe

1:03:04

if they have images that say play setting and there's

1:03:06

different numbers of them, then I could down, but they don't know what a

1:03:08

table is. They don't know what a play setting is. They

1:03:10

don't know what people are, that

1:03:12

they sit at tables, and like future

1:03:14

versions of this will be better in that

1:03:16

regard, and then they will be much

1:03:18

better tools because they have to have some kind of

1:03:20

understanding. In fact, probably in very

1:03:22

specialized areas, they'll gain that

1:03:24

understanding. But getting computers

1:03:25

to understand what a person is, what a table is,

1:03:27

what a pizza is, and how they relate to each

1:03:29

other, we've been working on that for decades and decades. It's way

1:03:31

harder than you think it is. These things

1:03:33

look like

1:03:34

magic because they're like, why are we doing that?

1:03:36

It's like trying to make an airplane by

1:03:38

making a thing that flat to swings. That's the wrong way to do it. Even though that's how

1:03:40

birds fly, it's stupid for us to try

1:03:42

to make a mechanical bird. Instead,

1:03:45

But we a fixed wing thing and we put a lawn mower

1:03:48

engine on it and a propeller and that's

1:03:50

a way better way to make an airplane

1:03:52

even though it has nothing to do with how

1:03:54

birds fly or a little do with it, but don't

1:03:56

it's not an ornithop there. Right? These are like

1:03:58

that for image generation. You know

1:04:00

what? Because because we can't make we

1:04:02

can't make a thing that thinks yet.

1:04:05

but we can make some incredibly

1:04:07

dumb thing that we feed enough of our

1:04:09

intelligence into by saying, here's a bunch of images.

1:04:11

Here's descriptions of another words. You know, words are

1:04:14

well there. Now, do that.

1:04:16

And we've got just enough to do this magical

1:04:18

stuff, but like as a tool,

1:04:20

trying to herd this towards something that

1:04:22

you want is harder than trying to

1:04:24

hurt an artist towards what you want. because at least

1:04:26

you can tell the arcs to make the logo bigger. And if they

1:04:28

don't do it, it's because they think you're a jerk now because they

1:04:30

don't understand what you mean. orna doctor,

1:04:32

a machine designed to achieve flight by

1:04:34

means of flapping wings, today I

1:04:36

learned. Yeah. Same. I knew that I knew

1:04:38

the word existed. I could not in a million years have

1:04:40

told you what it Exactly. You should note

1:04:42

from the nineteen ninety four movie Doon where they all talk about, let's get

1:04:44

in the ornithopter and they get into these things that have

1:04:46

wings that do not flap. Come on.

1:04:48

Marco didn't see it. Neither

1:04:50

of us. Please don't make me junk.

1:04:52

Please don't. The new dune movie, those

1:04:54

wings flap, baby. Bye. Now

1:04:57

the thing that that really changed my opinion

1:05:00

about this well, and I didn't have a strong opinion

1:05:02

about it, but would really kind of blew my

1:05:04

mind, I guess. is the

1:05:06

three of us are in a are in a Slack together

1:05:08

and and another person in that Slack was saying,

1:05:10

oh, and I'm heavily paraphrasing

1:05:12

here, but oh, I was I was looking at designing,

1:05:14

like, an app icon or an image. I forget exactly what it was. And

1:05:16

I knew a vague direction of where

1:05:18

I wanted to go with it, but

1:05:22

I didn't really know specifically what I wanted to do. This wasn't

1:05:24

the case, but, like, let's say for the sake of discussion

1:05:26

they were trying to draw a settings icon

1:05:29

and they knew they wanted a gear. But they didn't know,

1:05:32

do I have a gear or, like, several

1:05:34

concentric gears? Do I have a series of gears

1:05:36

all touching each other on the

1:05:38

outside? Like, you know, what what exactly what what am I

1:05:40

looking for here? I just know I want something with gears.

1:05:42

And they said they they put basically

1:05:44

app icon with gears in it. There's something along

1:05:46

those lines. one

1:05:48

projects, and it spit out, you know, like,

1:05:50

fifteen different options. And

1:05:52

this person, like, didn't really

1:05:54

love any one of the options, but

1:05:57

they said that it did a really good job of kind of getting their

1:05:59

creative juices flowing and saying, okay, now I have something to

1:06:02

work with. Now I know kind of where I wanna go with this.

1:06:04

And this is what you were alluding to

1:06:06

earlier, Marco. I just think something like that having

1:06:08

this kind of a fascinating tool

1:06:10

in your tool belt is

1:06:12

extremely cool. And and

1:06:14

as someone who can

1:06:16

barely draw a stick figure. I

1:06:18

think being able especially as these

1:06:20

things get better, being able to I

1:06:22

don't know, like, make my own app icon

1:06:24

potentially or get close.

1:06:26

You know, not that I have any problems with

1:06:28

the app icons I have. I love them and I

1:06:30

got them, you know, from a dear friend of

1:06:32

mine, but

1:06:34

nevertheless, it would be neat if I was capable of even putting, like, a

1:06:36

placeholder icon there that

1:06:38

wasn't utter garbage. And I

1:06:41

just think having this tool available

1:06:43

to more people, especially non

1:06:46

artists. I think that's neat. And the

1:06:48

thing that gives me pause

1:06:50

about it is, well, what happens to artists? In the same way that, you

1:06:52

know, I I worry about, like, GitHub,

1:06:54

Copilot or whatever it is. And and

1:06:56

I worry,

1:06:56

well, what happens to us? what

1:06:58

happens to developers. Don't don't worry about that. No. I I know. But you get

1:07:01

my point. Well, in the same way, you know, like,

1:07:03

co pilot is basically fancy

1:07:06

auto complete. And I think we can

1:07:08

look like this is basically like

1:07:10

fancy bucket fill. Fair.

1:07:12

Well, but the difference is, not to go

1:07:14

off on the tangent and copilot. bucket

1:07:16

fill to determine whether it

1:07:18

has done its job adequately. You look at it and

1:07:20

go, is it okay for what I want it to

1:07:22

be fun?

1:07:23

code, not quite that

1:07:24

easy. because if we could look at code and

1:07:26

figure and know whether it was doing what we intended it,

1:07:29

we wouldn't have to use copilot. And so

1:07:31

copilot will generate some

1:07:33

code does the code do what you wanted to do? Why don't you look

1:07:35

at it? And tell me, that turns

1:07:37

out to be really, really hard

1:07:39

to do. So I don't think we have

1:07:42

because, you know, In the end, you can

1:07:44

use AI art things to

1:07:46

full stop substitute for a thing that a

1:07:48

person could do. But

1:07:50

co pilot, you you need a human to look at that

1:07:52

before you check it into the air traffic

1:07:54

control system once. Like -- Yeah. -- you know,

1:07:56

maybe for a game, you can get away with it or something

1:07:58

or non

1:08:00

encryption credit but a co pilot has no idea what it's

1:08:02

doing even more so than a person. So we

1:08:04

need people to look at it and

1:08:06

to check that what same

1:08:08

way with them when you use autocomplete. But John, wouldn't

1:08:10

your unit test catch any problem? Yeah.

1:08:12

If it If you auto complete and

1:08:14

installed Can it write my unit test

1:08:16

for me? you wish. Maybe. It probably can. But

1:08:18

are the unit tests. Right? Like, in the same way when you do auto completing, you think you're auto completing NS

1:08:20

string, but it auto completes. What was

1:08:22

the thing that used to do like

1:08:25

or whatever before I ex go to No. It

1:08:27

was it was much less common, like like, in a scanner or something like that. Yeah. Like, when whatever came

1:08:29

first, automatically. If you don't

1:08:31

notice, that's what out

1:08:34

of auto complete. Guess what? Your program's not gonna work. You always have to look at the code that is you're not sent. Like, it's a help. It's you

1:08:36

know, content aware fill

1:08:39

to really help you. especially

1:08:42

on program interviews. You should say, can I use

1:08:44

co pilot? Great. Now I'm gonna reverse this red black tree for you. But,

1:08:46

yeah, you gotta you gotta check its work. But for the AI things,

1:08:51

the checking of the work is much simpler. You look at it and you

1:08:53

decide, am I happy with what it has made?

1:08:55

Yeah. Yeah. I just III

1:08:59

fear and feel for artists that, you

1:09:01

know, maybe wouldn't be able to make a living as artists anymore. But I agree

1:09:03

with what you

1:09:06

were saying that that's quite a ways in the future and we're nowhere near

1:09:08

there yet. And eventually, you know, the

1:09:11

only thing that's inevitable has changed.

1:09:13

You know, if if you're

1:09:15

a developer that's getting your

1:09:17

job you served by artificial intelligence, then you're gonna have to find a new way to

1:09:19

make money and same thing with an artist. But

1:09:23

I don't know. my initial my initial reaction

1:09:25

was, you know, get off my lawn. This is barbaric. You know, we can't

1:09:27

take from from real

1:09:30

and true artists. But I I

1:09:32

worry that as I get older, that's my natural reaction. It's it's

1:09:34

to just yell get off my lawn, and so I'm trying very desperately to

1:09:38

fight that. And I think having this tool available, especially to

1:09:40

people like me who I I have

1:09:43

no artistic ability whatsoever, I think

1:09:45

it's exciting and I think it's very fascinating.

1:09:47

the problem with having well, first of all, just so you ask what you said there. I I

1:09:49

totally agree with Marco that this is yet

1:09:51

another tool that will

1:09:53

be used by artists. like, content aware fill is the greatest

1:09:56

example. They didn't use AI to to advertise that because

1:09:58

it wasn't in fad at the time, but and

1:09:59

it worked slightly differently.

1:10:02

But content aware fill powered by this

1:10:04

type of thing. It's even better. It's a

1:10:06

tool that artists will use and the mundane

1:10:09

tasks that artists do. For example, example

1:10:11

for my childhood painting cells in Disney

1:10:13

Animation, like coloring in, like the people's

1:10:15

dresses and making the sky blue

1:10:17

and the grass green and everything. that used to be

1:10:19

a job where you would paint it to fill those regions because how

1:10:22

else are you gonna make something filled with the color

1:10:24

green if

1:10:26

you don't fill it with the color green? computers made that real easy. You just click the bucket

1:10:28

tool and nugget just filled the whole area with

1:10:30

green. That put all the people who are painting

1:10:32

themselves out of a job. You're like,

1:10:34

oh, they were just doing a mundane test.

1:10:37

that was incredibly skilled work. It's only mundane for the computer

1:10:39

to do it because very often, the strengths of computers are

1:10:41

the exact opposite of the strengths

1:10:43

and weaknesses of humans. a

1:10:47

computer finds it really easy to fill a region, especially if that

1:10:49

region is correctly contained with a

1:10:51

solid color, whereas the human has to

1:10:53

carefully control a brush and, you know,

1:10:55

and so the you would say the person doing

1:10:57

that work is incredibly skilled and the computer doing that work is as dumb as rocks. And that

1:10:59

often is the case, but it doesn't change the fact that they're

1:11:02

out of work because now the computer just does the fill

1:11:04

on all

1:11:06

that stuff. And same thing for hand drawn animation in the

1:11:08

ages three d animation. 33D animation

1:11:10

is incredibly difficult. They're incredibly skilled artists

1:11:12

that do that. They have to have all the

1:11:15

skills of traditional a artist on top of computer skills. But if you are A2D

1:11:17

artist and you don't know how to use computers and

1:11:19

don't care to learn, you've got a

1:11:21

problem. In

1:11:22

fact, if you watch the Disney

1:11:24

plus on Disney Plus, there's an ILM

1:11:26

documentary. What is it called? I don't know. Just go to Disney Plus and search for ILM.

1:11:28

And part of

1:11:31

that documentary is seeing what

1:11:33

the advent of computer technology did to industrial light and magic to the people who

1:11:35

were there like the model makers and the the creature shop people

1:11:40

or whatever. Like, if technology comes slow enough, people

1:11:42

die and retire, and then the new generation does Newtek. But if it

1:11:44

comes fast enough, people actually end up getting

1:11:46

booted out of their job. You have to

1:11:48

learn skills, and

1:11:50

that's just part of the

1:11:51

world. But I think mostly this stuff, at the rate even at the rate it's people think, oh, it's going so fast.

1:11:53

By next week, you'll be able to make a feature

1:11:55

length movie by just Writing

1:11:59

a phrase like no, you won't. Because to get to Casey's early or

1:12:01

a later point, you like the fact that

1:12:03

you don't

1:12:04

have artistic skill, but you could just

1:12:06

ask this thing to make, you know, a

1:12:08

picture. Now I'm

1:12:09

gonna sort of show the counterpoint to my earlier

1:12:11

point

1:12:11

about it. It's

1:12:13

easier to

1:12:16

tell whether you're happy with the picture

1:12:18

than to tell whether the code copilot generated does what you wanted to do. Part of making

1:12:20

that decision, so let's

1:12:22

say, I have no artist

1:12:25

taste, but now I can just make it make

1:12:27

the image for you. If you have no artistic taste and no

1:12:29

artistic skill, your ability to judge whether what it generated is good or

1:12:31

not is also impaired. right

1:12:34

there. So, like, you know, yes,

1:12:37

being able to do it is one skill,

1:12:39

but also having the taste to know this is

1:12:41

the good icon with the gears versus this is

1:12:43

the bad com for the years is itself an

1:12:45

artistic skill. And just because a computer made you fifty

1:12:47

of them and you get to pick one, the picking

1:12:49

of the one is the skill, which is why you

1:12:52

need artists to use these tools. Right? Content aware

1:12:54

fill is available on all our copies of Photoshop. And yet if we use Photoshop, we can't do what a great artist

1:12:56

can do with Photoshop because

1:12:58

we're not great Photoshop artists. Right?

1:13:02

And I think picking, you know,

1:13:04

like, let's look at anything. Anything that requires

1:13:06

any kind of, you know, taste.

1:13:07

Like, even if you're presented with

1:13:09

a thousand options, If you don't know which one is actually

1:13:11

better or the one you pick is not the one that

1:13:14

the world thinks is good. And, you know, for reasons

1:13:16

that you don't under you don't understand

1:13:18

why it's not pleasing, but you don't like any of them or the one you

1:13:20

think is awesome. Everyone else thinks is ugly.

1:13:22

Like, there's there's always a place for that

1:13:24

because, again, the the

1:13:26

things that are generating this have

1:13:28

no awareness of anything. They're just being

1:13:30

led by us. And

1:13:31

so like any tool, the

1:13:34

result is going to

1:13:36

be heavily

1:13:37

informed by the person using a tool, even if the court using of

1:13:39

the tool is just pointing to a grid

1:13:41

of pictures and saying, I

1:13:43

like that one. And even

1:13:45

if you repeat that process a thousand times, give me give me fifty more. Give me fifty more. Give me fifty more. And just keep

1:13:47

pointing to the ones that you like. If you

1:13:51

have bad taste, you will end up

1:13:53

with a bad eye at the end of it, no matter what. Yeah. There's no no computer is gonna save

1:13:55

you. So, you know, and and that gets back to, like, what

1:13:58

is it? What is it that makes good

1:13:59

taste? Like, can

1:14:03

is this is this are the computers doing what we

1:14:05

do? And I I have to say that, you

1:14:07

know, like most things in AI, the

1:14:09

answer is no until they

1:14:12

until their experience of their

1:14:14

existence is something close to what our experiences,

1:14:15

which would allow them to

1:14:19

learn things and have

1:14:21

memories and experience life the same way we do. Something that

1:14:23

doesn't do that will never

1:14:27

be able to create or judge art

1:14:29

in the same way that we do, so we will always be a necessary ingredient in

1:14:31

that stew. We are brought

1:14:34

to you this week by

1:14:38

Squarespace, the all in one platform

1:14:40

for building your brand and growing

1:14:42

your business online. With Squarespace, you can

1:14:44

stand out with a beautiful website

1:14:46

engage with your audience whatever you need to and sell

1:14:49

whatever you wanna sell. Your products,

1:14:51

your content, even your

1:14:54

time. Squarespace makes it super easy to make

1:14:56

all kinds of websites from very

1:14:58

simple, you know, basic informational sites,

1:15:00

all the way up to full

1:15:02

blown storefront hosting. sell your products in

1:15:05

an online store where you sell physical

1:15:07

or digital products. Squarespace has all

1:15:09

the tools you need

1:15:11

to sell online. You

1:15:13

can add an email campaign to your

1:15:15

customer. Click email subscribers, convert them to loyal customers. Start template and customize it

1:15:18

by applying your brand ingredients

1:15:22

like your site colors or your logo, you have built

1:15:24

an analytics measuring the impact of every

1:15:26

email sent. All of this comes with

1:15:28

SEO tools so that your

1:15:30

website or online store has suite of

1:15:32

integrated features useful guides to help maximize

1:15:34

prominence among search results, everything you need to run a business, analytics, member

1:15:37

areas if you wanna

1:15:39

sell member content, so

1:15:41

much is available at Squarespace, and it is all backed by amazing support, amazing infrastructure,

1:15:43

and just a whole bunch of stuff

1:15:46

that you don't have to deal

1:15:48

with. you get you get

1:15:50

to do whatever it is you do with your business or your hobby or whatever your pleasure is for making your site. You to

1:15:52

focus on that unless

1:15:54

Squarespace handle all of the

1:15:57

you know, kinda boring and tedious hosting

1:16:00

and maintenance kind of stuff. So see for yourself at

1:16:02

square space dot com slash atp to start a free

1:16:04

trial. When

1:16:06

you're ready to sign up and launch your site, use offer code ATP

1:16:09

to save ten percent off your first purchase

1:16:11

of a website or domain. Once

1:16:13

again, squarespace dot com slash ATP

1:16:15

to start that free trial and use offer code ATP at launch to save ten percent your first purchase.

1:16:17

Thank you so much

1:16:20

to Squarespace. for

1:16:23

sponsoring our show. Alright.

1:16:24

Let's do some

1:16:28

Ask ATP and let's start

1:16:31

with Philip, who writes, I recently got

1:16:31

a MacBook, and I'm struggling to build a

1:16:33

solid mental model for window management on Mac

1:16:36

OS. I've been

1:16:37

using Linux with a tiling window manager and

1:16:39

things felt simpler. I'm not trying to replicate

1:16:41

this setup and wanna learn the quote unquote

1:16:43

Mac way, but I can't seem to grok it.

1:16:45

I'm not sure which features I should

1:16:47

be using between mission controls, spaces, applications which

1:16:49

are hide, minimize full screen with tiling, hot

1:16:52

corners, etcetera. And I often end up

1:16:54

with a lot of window clutter where I can't even seem to find the one window I need. Can you refer to a primer on window

1:16:56

management? How do you think about

1:16:58

and organize your applications and windows?

1:17:02

I've listened to the windows of Syracuse County, but can't

1:17:04

tell if John is messing about. No,

1:17:06

he was not messing about, which is

1:17:08

why Marco and I still 567

1:17:11

years later or gobsmacked by that episode.

1:17:13

I think it's episode ninety six. We're

1:17:15

not mistaken. That, like, half hour forty

1:17:17

five maintenance segment of ATP might, to this

1:17:19

day, be my favorite part or a fact favorite

1:17:21

segment we've ever done at ATP because we didn't know

1:17:23

it was coming. And I don't wanna speak

1:17:25

for Marco, but I'm a speak for

1:17:27

Marco in saying, it was flabbergasting. Like,

1:17:29

just stupefying the the absolute

1:17:32

bananas way in which John

1:17:34

Syracuse manages his seven trillion windows.

1:17:37

And, yeah, what have you learned since then? No. You Have you learned that your ways? No. You

1:17:39

have not. Because I'm not a monster that keeps telling me I'm

1:17:42

just still just a little baby.

1:17:46

Right. Exactly. So getting to

1:17:48

Philip's question, what is the Mac way to manage

1:17:50

your windows? I think you too, at this

1:17:53

point, are better equipped than me to answer this

1:17:55

because I think the answer is just do a windows people

1:17:57

to Zoom everything full screen because you have no freaking idea

1:17:59

how the old windows. No. No.

1:18:02

So the window. The great thing about Mac OS

1:18:04

window management is that Mac

1:18:06

OS ten started out trying to

1:18:09

get people who used Mac OS

1:18:11

nine and earlier, two like it. I don't know about that.

1:18:13

I feel like it did the opposite. Didn't say

1:18:15

it succeeded. It's I'm not sure it'd

1:18:17

really even tried, but I get what you're

1:18:19

saying. There was there was

1:18:21

some acknowledgement that there was something that existed before the Mac, but

1:18:23

it was a grudging acknowledgement. Right? So anyway, as it

1:18:27

went on, it later

1:18:29

took a larger role in trying to get Windows people to like it. And

1:18:31

then later on, it

1:18:34

took a larger role in

1:18:37

trying to get iOS people to like it. And now it's in this weird mishmash

1:18:39

where they're trying to move it

1:18:41

forward with iOS and the

1:18:44

iPad somehow And

1:18:47

and so the result of all of this is that there

1:18:49

are a million different ways to

1:18:51

manage Windows on Mac OS.

1:18:53

It it basically ended up

1:18:55

with this mishmash where it kinda supports

1:18:57

all of these different things you might wanna do. So what I would

1:19:00

recommend is

1:19:03

basically play with different options just see what works for you. Now,

1:19:05

I can tell you what I do, you

1:19:07

know, because I was

1:19:09

a Windows person until

1:19:12

two thousand four

1:19:14

ish. When III began AA2 year transition to Mac full time at that

1:19:16

point. But anyway, what I

1:19:18

do, first of all, you know,

1:19:22

hide versus minimize. You want

1:19:25

to hide. Command h is your

1:19:27

new best friend. As a new

1:19:29

person on Mac, you're gonna want that

1:19:31

all the keyboard shortcuts to to make us easier

1:19:33

and command h is gonna be one of the

1:19:35

things you use the most up there with command

1:19:37

q for quit. almost never hide windows. Almost

1:19:39

-- Oh, no. -- but anyway, it might be new

1:19:41

to you the fact that, you know, on the Mac,

1:19:44

an app can have no windows but

1:19:46

still be running. So when you're done with an

1:19:48

app, command queue. If you wanna close

1:19:50

a window or tab, command w,

1:19:52

and then command h for hide. These

1:19:55

are things Other other, you know, Windows has alt tab. We have

1:19:57

command tab that's in the same position on

1:19:59

the keyboard even. We

1:20:02

also have Since, you know, alt tab and windows go between different

1:20:05

windows, you know, all

1:20:08

individually, on the Mac,

1:20:10

command tab goes through apps. much to cases

1:20:12

the way it does this, but it goes

1:20:14

through apps and then command tilde, the button

1:20:16

right above the tab key on

1:20:18

the US keyboard at least, goes between

1:20:21

different windows of the app you're currently

1:20:24

in. So again, these are kind of like in the training wheels

1:20:26

and getting into Mac window management from other systems, most likely windows.

1:20:28

So that

1:20:31

those I think are the are the main entry points. And then,

1:20:34

you know, whether you, like,

1:20:36

maximize the full screen,

1:20:38

like John just accused Casey of

1:20:40

which macOS makes kind of difficult. Whether you can

1:20:42

use full screen mode or just make Windows big, that's up to

1:20:46

you. I on those things I I make windows need to

1:20:48

be, but not when they don't. I don't use full

1:20:50

screen mode on anything because it sucks in many

1:20:55

other ways. So And then, you know,

1:20:57

spaces well, that's virtual desktops. Every windowing system has that. If you use spaces

1:20:59

in in your previous systems, you

1:21:02

might need to hear, you might

1:21:04

not It's up

1:21:06

to you. You could try it. Feel free.

1:21:08

No one's gonna bother you. The things like mission control or what

1:21:10

used to be called expose, where you kinda like

1:21:14

Zoom all the windows out at once and do stuff with

1:21:16

them. I don't really do that. I I used

1:21:18

to when it like, when it was

1:21:21

expose back in the early days. I don't really

1:21:23

do that now. The one thing I that I do lot is the f eleven

1:21:25

to show desktop where you're hitting f hit

1:21:27

hit f eleven or whatever the

1:21:29

desktop key is on on

1:21:31

the Apple keyboard. and it shoves all the

1:21:33

windows out to the sides and exposes the desktop where I

1:21:35

keep, yes, files I'm working

1:21:38

on. That's a whole thing.

1:21:41

I do it. Everyone does it. Who

1:21:43

cares? So I oftentimes will like zoom all the windows out, grab a file at desktop, hit

1:21:45

f eleven again to bring all

1:21:47

the windows back in, and

1:21:51

then drop that file onto something I'm working on onto a

1:21:53

window of an app or whatever. Marco, you're

1:21:55

showing you're showing your keyboard

1:21:58

preferences here because

1:21:59

f eleven isn't

1:22:00

what you're describing. It's volume down. It is when you're

1:22:02

using a non Apple keyboard. Yes. But for everyone else in

1:22:04

the world that uses a

1:22:07

Mac, it's volume down. understand

1:22:09

what you're going for, but your your your preference for keyboards is is coming

1:22:11

through here. Yeah. Well, anyway, so

1:22:16

Yeah. So basically, that's that's the kind of

1:22:18

like the basics that I like to do. But whether you use all this

1:22:20

stuff like the, you know,

1:22:22

mission control and all this stuff,

1:22:25

It's really personal preference. And again, because of the history

1:22:27

of macOS and and the jumbled design leadership it has had

1:22:29

over time and it ended very

1:22:32

different dark It

1:22:35

has tried to attract people from over time. It it it

1:22:37

kind of offers all of these

1:22:39

things. Now we we're

1:22:41

even gonna have stage manager if it ships

1:22:43

in Ventura. And that's a that's its own

1:22:45

whole other thing that, frankly, III

1:22:47

don't think is working

1:22:49

very well so far.

1:22:52

But that's that's its own thing. Try it out.

1:22:54

See what works for you. If you ask any two Mac users, you're gonna get two different answers because

1:22:56

there there are

1:22:59

so many methods and MAC users are

1:23:01

largely at least they we used

1:23:03

to be power users. And so everyone has their own, you know, certain workflows

1:23:05

and quirks and habits

1:23:08

and preferences and

1:23:10

they're all gonna be a little bit different because

1:23:12

there are so many different ways to do it. Alright. IIII

1:23:14

don't think either of you was necessarily

1:23:15

wrong, but I think you gone

1:23:18

directly into the deep end. And I think

1:23:21

we need to to back up a bit.

1:23:23

So coming from Windows, I I have used

1:23:25

Linux, but it's been woof, it's

1:23:27

been a long time. Coming from Windows,

1:23:29

I think the most startling thing about using the

1:23:32

Mac is what you

1:23:34

had made brief mention of

1:23:36

earlier go is

1:23:38

that you can have an application that

1:23:40

is running even though it does not

1:23:42

have not a single window open. This

1:23:45

is very different than Microsoft Windows, or at least the last I used it ten years ago, where

1:23:47

if you close the final window of Outlook

1:23:50

for the sake of discussion,

1:23:54

you're suddenly not gonna get any new email. Again,

1:23:56

this may not be true anymore. It doesn't matter. But the way that's

1:23:58

the way it used to be. If you close the last Outlook window,

1:24:00

can email anymore you're not getting email anymore.

1:24:02

If you just close not even necessarily hide. Close the

1:24:04

last mail window. Yours did

1:24:06

that mail the mail

1:24:09

app is still open. And this

1:24:11

isn't universe clearly true, which has gotten even over time. But generally speaking, that's

1:24:13

true. If you close the last window, that

1:24:15

does not necessarily mean

1:24:18

that the app is quit. the app could still be running and mail is a

1:24:20

quintessential example of this or Safari if you

1:24:22

don't have any tabs open for example.

1:24:26

They're still running So what I would say and I think what Marco

1:24:28

got right is get used

1:24:30

to the keyboard because the

1:24:34

keyboard is your friend for doing windowing things on macOS.

1:24:36

You don't have to do it. You don't have to touch

1:24:38

the keyboard at all, but it is your friend. So

1:24:42

When

1:24:42

you're done looking at something, command

1:24:44

w closes that thing, be it

1:24:46

a tab or a window. Generally

1:24:49

speaking, you're not quitting the

1:24:51

app usually. Even if you close the

1:24:53

last window, you're just closing that window.

1:24:55

So as an example, if I'm

1:24:57

looking at my mail using, you

1:25:00

know, the standard Mac OS Mail app. After

1:25:02

I'm done reading and responding to mail, I command w. That closes mail,

1:25:04

but it does

1:25:07

not quit mail. it leaves it running to

1:25:09

get new mail if new mail comes in. If I want to play ignorant and

1:25:11

don't want to get new mail,

1:25:14

then I command q for quit.

1:25:18

and

1:25:18

that will quit mail such

1:25:20

that I don't receive new mail.

1:25:22

That

1:25:23

so the whole clothes

1:25:25

were quit thing, close being command w, quit

1:25:27

being being command q, that is something to get used to. And similarly, if you were to stoplights

1:25:30

in the upper left, The

1:25:33

red stoplight is not quit. It is closed. So

1:25:35

you are closing an entire window that maybe that

1:25:40

is a window with a bunch of tabs in it,

1:25:42

maybe it's just a single window like in mail, but you're closing it, you're not quitting it. And there is

1:25:44

an option buried somewhere in

1:25:46

system preferences. I forget where. that

1:25:49

you can have, like, little light bulbs that show under the app

1:25:51

icon in the dock to indicate what is actually running, which is the

1:25:53

things that have light bulb. They they

1:25:55

aren't light bulbs anymore. used

1:25:59

to be Whatever they are. Now they're just the dots. You're right. You're

1:26:01

right. But now I'm showing my age. Yeah.

1:26:03

And also Mac OS lies

1:26:05

about that. It's way more complicated I can do that.

1:26:07

You're right. You're right. But I'm I'm trying to do a

1:26:09

I'm I'm trying to ease into the shallow end here. Yeah.

1:26:11

The the problem with what we're

1:26:13

saying here is, like, like, everything that Casey has just

1:26:15

said has a bunch of asterisks on it now, again,

1:26:17

through the course of, like, history and different goals and

1:26:20

different emphasis. You're

1:26:22

right. you're right overall. Like, your your overall theme

1:26:24

is right, but it is I I think it's part

1:26:26

of the reason, like, it it's a little bit frustrating

1:26:28

that the Mac is is not as simple

1:26:31

as it once was because there's been all these different, like, ideas and directions and then mandates over

1:26:33

bad designs, like,

1:26:36

over time. I

1:26:38

mean, but the the thing is, like, what you're first of all, in case

1:26:41

you're going off on a tangent here, which is basically

1:26:43

about process management versus window management, and I see

1:26:45

how they're so relate They're interrelated.

1:26:47

Kind of. But, like, the the technical

1:26:49

details that we know about for, like, you know, I think I had

1:26:51

a big paragraph on it. One of the

1:26:53

macro center reviews that basically you

1:26:56

can have Applications

1:26:58

with no windows that are running, applications with windows that aren't running, applications with a dot under to the dock that aren't applications

1:27:00

without appearing in the

1:27:02

dock that are running, like

1:27:05

every combination of things that you think

1:27:07

shouldn't be possible or possible, but that's

1:27:10

just what we know from a technical

1:27:12

perspective. What's important is the user model,

1:27:14

the mental model. You're not supposed to know that an application

1:27:16

that doc that has a dot under it might

1:27:18

not actually be running because the OS is

1:27:21

working to provide the illusion that if it's got a dot, it's

1:27:24

running. And it does everything it can to

1:27:26

make that illusion true. So for example, if

1:27:29

the if the OS has quit an app,

1:27:31

but left the dot under it and the dot it did that because the app

1:27:33

supports whatever the hell the thing is called.

1:27:35

But when you click it again, behind the

1:27:37

scenes it relaunches it for you and

1:27:39

lets it auto restore the state.

1:27:41

So to you, it just looks like it just brought that app to the front, which is what it would do if it was running, but

1:27:43

it didn't. It actually

1:27:47

relaunched it. Right? And by the same

1:27:49

token, sometimes when you quit an app, the OS has the option to the, oh, yeah, I'm not actually gonna

1:27:52

quit it. I'm

1:27:54

actually gonna keep it So if

1:27:56

you quit an app and the dot disappears from running the dock,

1:27:58

and then you click that icon on the dock again, you're like, wow, that launch really fast.

1:28:01

You know why? because the OS

1:28:03

didn't freaking quit it. And that

1:28:05

only happens in cases where the application supports whatever API that Apple had

1:28:07

introduced in Mac OS ten point seven

1:28:10

point what you know,

1:28:12

like, read my macro as ten reviews

1:28:14

to see all this insanity. I'm not sure how much of it is still in play, but the point is those details don't matter

1:28:16

because if that

1:28:19

happens behind the scenes, it's

1:28:21

it's meant to provide the illusion that the model is dot means running. No dot means not running. And

1:28:23

what does running mean versus not

1:28:26

running mean? Well, running means

1:28:28

that If

1:28:30

you bring it to the front, it looks like it did be when

1:28:32

you saw it before. Like, it preserves state in the

1:28:34

window. It remembers your selection or whatever. And for

1:28:36

apps to support to correctly support those

1:28:38

APIs, that's what it's supposed to do. So you shouldn't

1:28:41

be able to tell that it's not running and

1:28:43

it relaunched. Because from your perspective, it looked exactly

1:28:45

the same as if it already

1:28:47

was running. How successful? how successful

1:28:50

are individual applications to those APIs

1:28:52

in achieving that debatable. Right? But that's the

1:28:54

goal of those APIs. So I think what people

1:28:56

need to understand is what is the

1:28:58

supposed mental model? What is the abstraction? How how is the OS trying to tell me

1:29:00

that it works? And then

1:29:02

there's all the cases where that

1:29:06

attraction falls apart. It's like, oh, it's you

1:29:08

know, it's it kinda seems like that app

1:29:10

wasn't running because even though it had the dot under

1:29:12

it, when I clicked on it and and I saw the

1:29:15

window again, it didn't like last left it. Why is

1:29:17

that? Oh, actually, relaunched and that app doesn't support state restoration for that

1:29:19

one particular thing and blah blah blah, you

1:29:21

know, it gets super complicated

1:29:23

really fast. But it's the same

1:29:25

thing on iOS where we complain that people are, you know, going to the application switch and flicking up

1:29:27

what are essentially images of applications

1:29:29

that haven't been running for

1:29:32

three weeks. Right?

1:29:34

because I think they're quote unquote quitting the apps. Right? It

1:29:36

provides the illusion that these are all the

1:29:38

apps that are running, but they're not

1:29:40

all running. Like, this this five hundred pictures there.

1:29:43

How could they all be running? They're not. It's just

1:29:45

it's just literally an image of what that thing looked

1:29:47

like the last time it was running. Right?

1:29:49

And so that's the illusion it's

1:29:51

trying to provide. but that illusion is not true and that

1:29:53

only matters when you care about the technical nuances or when you become an

1:29:56

obsessive, you know, force quitter because you think you're doing

1:29:58

something and all you're really doing is removing a

1:30:00

bunch images, which

1:30:02

in itself may be a goal that you wanna

1:30:04

achieve, so go for it. But anyway, I don't wanna

1:30:06

get into that debate again. So so I think

1:30:09

worrying about the nuances here are not as important as just getting

1:30:11

what the supposed mental model is. Because if you get the supposed mental

1:30:13

model that you

1:30:15

just described, Casey, then it's

1:30:17

easy to or easier to explain the exceptions. And unfortunately, as you both

1:30:20

noted, one of

1:30:22

the exceptions is there

1:30:24

are a certain class

1:30:26

of application that when you close their last window, they quit themselves. How do I know what those applications are?

1:30:28

And why do they

1:30:30

do that? There's a rationale

1:30:34

But

1:30:34

really, it's kinda like, oh, these are

1:30:36

the exceptions and kind of here's why.

1:30:38

And it's it's not satisfying to hear

1:30:40

the explanation, but eventually, you just learn

1:30:43

the ones that do that. I mean, it makes some sense for, like, calculator because it's just

1:30:45

got one window. And when you close it, it used

1:30:47

to be a desk accessory. What's a desk accessory? We

1:30:49

gotta go into super old man mode to learn

1:30:51

about that. But, like, there are

1:30:53