0:00

So Mike, I've been browsing the internet

0:02

for some shelves, right? You know, as

0:04

one does. As one does, you know,

0:07

this is the thing I do in

0:09

my spare time now. My house renovation

0:11

is almost coming to an end.

0:13

I've talked about a bunch of this

0:16

podcast, I hope listeners on board. But

0:18

I'm basically at the shelf stage, so

0:20

I was looking around for some shelves,

0:23

and naturally I ended up on eBay,

0:25

and I thought we haven't used

0:27

eBay on control or speech. Oh. And

0:29

you can take that as a kind of material

0:32

search, or you can think of it

0:34

as a kind of philosophical prompt. Well,

0:36

I was going to say there's

0:38

a few different ways, but I

0:40

can search for a different government

0:43

for the US, but I think I would

0:45

like to search for more time. I

0:47

feel like I don't have nearly enough

0:49

time for anything. Yes, we're just discussing

0:51

in terms of like how much stuff

0:53

is going on. I could use more

0:56

time. So if I could buy more

0:58

time on eBay. I'd go with that.

1:00

Yeah, okay. What about you? Maybe a

1:02

person assistant. Search for anything then. If

1:04

I was to search for something,

1:06

I would search for the kind of

1:08

dying vestiges of a free

1:11

speech platform and that links

1:13

to something that we'll talk

1:15

about in today's podcast. So

1:17

I'll explain more. Hello

1:27

and welcome to Control All Speech,

1:29

your weekly roundup of the major

1:31

stories about online speech, content moderation

1:33

and internet regulation. It's March 20th

1:36

2025 and this week's episode is brought

1:38

to you with Financial Support from the

1:40

Future of Online Trust and Safety Fund.

1:42

This week we're talking about AI Slob,

1:44

the Online Safety Act coming into force

1:46

and Metis' attempt to build a better

1:48

community notes than X. My name is Ben

1:50

White Law. I'm the founder and editor of

1:53

Everything in moderation. And one with a very

1:55

time-poor... Mike Masnick. How much would you

1:57

pay Mike to have an extra hour?

1:59

every day. Gosh, I don't know. That

2:01

is a good question. Just one hour.

2:03

Yeah, one hour is not enough. I

2:06

think I mean an extra day every

2:08

week. Yeah, okay. Where nothing happens, right?

2:10

So... I need everybody else, like everybody

2:12

else can stick with the seven day

2:14

week and I need that extra day

2:17

when nothing is happening and it's just

2:19

perfectly quiet and I can just catch

2:21

up on stuff. I think that's what

2:23

I would, yeah, someone get me that,

2:25

someone get me that. So you're paying

2:27

for the whole world to stay still,

2:30

essentially. Yeah. No, I mean, you know,

2:32

in their lives, like, they can just

2:34

think that the world still has seven

2:36

days. But I want, I want, like,

2:38

this pause, that I can still do

2:40

stuff. Give me that extra day. I'm

2:43

not greedy. You know, I don't, I

2:45

don't need two days. Just, just one.

2:47

It reminds me of a children's TV

2:49

show in the UK called Bernard's Watch.

2:51

Okay. And if you ever came across

2:54

this. No. It's kind of unhinginged. click

2:56

and stop and he went about his

2:58

day-to-day life and everyone else is frozen.

3:00

Yeah, that's it. You want to burn

3:02

its watch, don't you? I think, yeah,

3:04

yeah, I think there was a movie,

3:07

like a Hollywood movie on that premise

3:09

too, but I didn't see it. I

3:11

remember the preview with that preview with

3:13

that premise, but no, I just need

3:15

that, but like I wouldn't do anything

3:18

nefarious with it. I would literally just

3:20

sit and work, right, right? I hope

3:22

you don't think this is work. This

3:24

is our time. This is our time

3:26

to decompress my. I think that's a

3:28

sensible thing to buy. My purchase would

3:31

be something very not sensible. And I

3:33

was referring to it in the opening

3:35

today. I want to buy the Twitter

3:37

logo that's up for sale again. I

3:39

don't know if you saw this. The

3:41

big bird. used to sit outside of

3:44

Twitter's office back when it was called

3:46

Twitter in San Francisco has gone up

3:48

for sale again and it is the

3:50

perfect analogy for what has happened to

3:52

Twitter slash X since it was bought

3:55

by Elon Musk. So it was, I

3:57

don't if you remember in 2023, Musk

3:59

sold it as part of a kind

4:01

of fire sale, like a kind of

4:03

garage sale. Is that what you call

4:05

it in the US? Yeah. Sure. Yeah.

4:08

And it sold for $100,000. It's currently

4:10

up for sale now. and you just

4:12

looked right and it was 2700 dollars?

4:14

27,000. 27,500. Okay, so yeah, just as

4:16

Twitter's X's Valley plummeted over the last

4:18

several years, although there's been some sort

4:21

of correction this week, so has Larry

4:23

Bird, the giant blue logo that used

4:25

to say outside the office in San

4:27

Francisco. Maybe I could use it as

4:29

a shelf. Yeah, there you go. You

4:32

take two birds with one stone. But

4:34

you look at that. But the, yeah,

4:36

I mean, I guess it's just been

4:38

sitting in storage in a storage facility

4:40

in San Francisco. Yeah. And so you'll

4:42

have to pay for shipping to the

4:45

UK. Yeah, yeah, if you get it.

4:47

But I will note that it says

4:49

there's about six hours left as we

4:51

record this. So by the time this

4:53

comes out. You may only have a

4:55

few minutes left to bid on it

4:58

and purchase it for Ben. In case

5:00

any of our listeners want to make

5:02

a donation to control all speech and

5:04

to Ben shelving fund in particular, you

5:06

can purchase the bird and send it

5:09

off to London. Yeah, you would get

5:11

regular photos if you contribute. Yeah, I

5:13

wonder what your wife would feel about

5:15

a giant bird showing up for your

5:17

newly renovated home. Yeah, I probably wouldn't

5:19

love it, but it's funny to see

5:22

such a kind of old vestige of...

5:24

what Twitter used to be come up

5:26

for sale again in such circumstances. Yeah,

5:28

cool. So purchasing power aside, you know,

5:30

our wish lists, we've kind of dealt

5:32

with today now, but we have a

5:35

lot of stories that we want to

5:37

get through. We have some really interesting

5:39

kind of notable stories that we're going

5:41

to talk through and explain, just as

5:43

we always do in the podcast. Before

5:46

we do a request for some help

5:48

from our listeners, we haven't had a

5:50

review of the since, I think, September

5:52

2024, Mike. Yeah. Isn't that tragic? Yeah.

5:54

So if you don't hate listening to

5:56

this podcast, that's, that's, that's, that's, that's

5:59

mixed it up a bit. If you

6:01

don't hate listening to this, very low

6:03

bar, you're sitting there. I know, but

6:05

it might prompt people to think, actually,

6:07

this is me, you know, I'm somebody

6:09

that doesn't hate this podcast. Maybe I'll

6:12

spend some time helping Mike and Ben.

6:14

in their quest to have their podcast

6:16

discovered across all of the platforms on

6:18

which it is disseminated every week. If

6:20

you don't hate it, Spotify, Apple podcast,

6:23

wherever you get the podcast, leave a

6:25

little review. Doesn't need to be very

6:27

long. Yeah, exactly. And you know, we

6:29

look at the numbers and we've seen

6:31

like our download numbers increase. So there's

6:33

more of you. There's definitely some of

6:36

you who weren't listening back in September

6:38

of 2024. So someone out there must

6:40

be listening to this and thinking. I

6:42

have not written a review. for control

6:44

speech and I could, right? Like think

6:47

positively. You can do it if you're

6:49

listening to this and you haven't written

6:51

a review. And we'll give you a

6:53

shout out in next week's podcast if

6:55

you do. We won't shame you. There's

6:57

no shame in submitting a review, but

7:00

we'll give you a shout out. There

7:02

is a risk, Mike, that some of

7:04

our listeners perhaps are connected to our

7:06

first story of this week. I'm hoping

7:08

that every single... download and listen is

7:10

a real human person who works in

7:13

the industry. There's a small chance. There's

7:15

a few... AI generated bots. That's how

7:17

the internet works. I'm sorry to say

7:19

it to you. And the first story

7:21

from 404 media really goes deep on

7:24

this idea of AI slop and content

7:26

being kind of created as a result

7:28

of AI generated systems and platforms. I

7:30

got so deep into this, explain what

7:32

it is while you picked it. and

7:34

why it's important to our listeners. Yeah,

7:37

I mean, obviously 404 media does such

7:39

wonderful work, but this is such a

7:41

wonderful story, kind of laying out. Everybody's

7:43

heard about, you know, the rise of

7:45

AI and AI Slop and 404, and

7:47

in fact, just done a lot of

7:50

reporting in general on. AI Slop and

7:52

the fact that you know social media

7:54

especially meta properties in particular are sort

7:56

of filling up with this stuff and

7:58

there's been this like confusion over why

8:01

and and how but part of what's

8:03

really interesting to me here is just

8:05

like not just how much is getting

8:07

filled up with just nonsense AI stuff

8:09

but the fact that there are these

8:11

people these content creators who are basically

8:14

saying like were fighting the algorithm and

8:16

we're trying to figure out the best

8:18

way to get the most click throughs.

8:20

And the best way to do that

8:22

is take as many shots on goal

8:24

as we possibly can. And AI will

8:27

just generate a ton of stuff. We

8:29

don't have to put work into it.

8:31

You look at the really successful content

8:33

creators like Mr. Beast or whatever, who

8:35

spends millions of dollars on every video

8:38

and then is able to pay that

8:40

off because he gets tons and tons

8:42

of views. But these guys are saying

8:44

like that space is too expensive. There's

8:46

too much capital costs up front to

8:48

get there. and it's too competitive. And

8:51

so the better thing to do is

8:53

just use this bullshit generator engine to

8:55

generate as much bullshit as possible, put

8:57

it all out there. If stuff doesn't

8:59

work, it's no loss to us because

9:01

it doesn't cost us anything, and therefore,

9:04

let's just get it out there until

9:06

we find something that hits. And so

9:08

now you have a whole bunch of

9:10

people doing that. And so that's sort

9:12

of like the motivating... behind a lot

9:15

of this is this idea that, well,

9:17

we need to get attention and the

9:19

best way to get that attention is

9:21

just to try as many things as

9:23

possible and see what actually clicks. And

9:25

then some of these guys are like

9:28

teaching other people how to use AI

9:30

Slop and that's becoming like a, you

9:32

just have this whole ecosystem of nonsense,

9:34

which is all about trying to game

9:36

the algorithm. Yeah, the bit that I

9:38

really like about this piece is the

9:41

comparisons to a kind of brute force

9:43

attack. Yeah, which is obviously a kind

9:45

of cybersecurity term. You know more about

9:47

it than me, but like, talk us

9:49

through like why AI Slop is kind

9:52

of like a brute force attack. Right,

9:54

I mean it's just a question of

9:56

like trying to get through, right? And

9:58

so like a normal brute force attack

10:00

is like, the simplest version of a

10:02

brute force attack is like trying to

10:05

guess a password or something, right? And

10:07

so you just will generate as many

10:09

random, not random, but you know, passwords

10:11

that might get into an account. Or

10:13

if you're like looking for some other

10:15

way into a website, you'll go through

10:18

like. scrape the entire thing and try

10:20

and find like every possible way in.

10:22

It's just like trying as many things

10:24

as possible. It's not very sophisticated. It's,

10:26

you know, it's brute, like, as it

10:29

says as I did. But this is

10:31

that for the algorithm. And the thing

10:33

that it kind of reminds me of

10:35

in a lot of ways is... When

10:37

Google came along, there were other search

10:39

engines in the 1990s and the rise

10:42

of the web, and none of the

10:44

search engines were that good. And then

10:46

Google sort of solved search by using

10:48

this page-rank system. And they said, you

10:50

know, we have the system to actually

10:53

make sure that the best stuff gets

10:55

to the top. And the way it

10:57

was originally done was based on how

10:59

many links with the idea that things

11:01

would only get links if they were

11:03

good. But pretty quickly, people figured it

11:06

out like, like, oh, here's an. to

11:08

our advantage. As more people go to

11:10

Google to figure out where to go,

11:12

if we can just get more links

11:14

and then became this whole crazy race

11:16

and one-uping each other of search engine

11:19

optimization and often what sometimes referred which

11:21

is black hat methods, sort of sneaky

11:23

methods, try and get around it. I

11:25

mean, not a day goes by when

11:27

I don't get emails from spammers asking

11:30

to pay me to put links on

11:32

Tector, because Tectorate has, you know, strong.

11:34

reputation within Google and they will offer

11:36

me money to put links because they

11:38

want those links to gain the system.

11:40

But you know you could use that

11:43

money to buy more time. I thought

11:45

you were going to say to buy

11:47

the Twitter. Yeah, exactly. Either one would

11:49

do. I will say I every once

11:51

in a while when I'm really like

11:53

in a cynical mood. I will respond

11:56

to one of those guys and say

11:58

sure it's a hundred million dollars. And

12:00

I've usually they disappear after that. Occasionally

12:02

I've had one. back to be like

12:04

no what's the real number and I'm

12:07

like a hundred million dollars I'm not

12:09

joking about this like yeah I'm gonna

12:11

destroy the the reputation and value of

12:13

my site and might as well get

12:15

paid for real you know not whatever

12:17

they're offering they're always offering like a

12:20

hundred bucks or something right that you

12:22

know but every system gets gained and

12:24

it's just that in the past like

12:26

gaming social media algorithms was tricky right

12:28

I mean you had to put in

12:30

the work which is what they're like

12:33

you know there's some inexpensive YouTube channels

12:35

or Instagram things that have been successful,

12:37

but the really successful ones are a

12:39

little bit more produced. And so what

12:41

the AI is enabling them to do

12:44

is to like make it cheaper to

12:46

game the system. So now everybody's doing

12:48

that. And so to me it sort

12:50

of then becomes this arms race, you

12:52

know. We saw this with Google and

12:54

like at some point in the sort

12:57

of mid to early 2000s, Google completely

12:59

revamped their ranking algorithm because they knew

13:01

that they had so much search spam

13:03

and they referred to it as search

13:05

spam that was clogging up the results

13:07

and making the experience worse. And so

13:10

they completely redid the algorithm. Some companies

13:12

that has sort of been built up

13:14

around search spam got destroyed by it

13:16

and there was like the big shakeout.

13:18

I'm getting. we're going to have to

13:21

see something similar happen in the social

13:23

media space? Yeah. And that there's got

13:25

to be some way that meta in

13:27

particular YouTube to some extent have to

13:29

figure out how to deal with this

13:31

stuff and whether or not people actually

13:34

want this content because I think yeah

13:36

like people are clicking on some of

13:38

it but then I think people are

13:40

getting annoyed or like people are clicking

13:42

on it because of the novelty and

13:44

like why am I seeing this nonsense

13:47

you know? Yeah. to kind of emphasize

13:49

the point and what the piece does

13:51

really well is just emphasize the insanity

13:53

of some of this content. Oh yeah,

13:55

it's bonkers isn't it like there's crazy

13:58

stuff like there's like giraffes eating people

14:00

and spawning on subway platforms there's like

14:02

it's almost it's bonkers you know there's

14:04

like grannies kissing llamas and there's like

14:06

heads appearing out of sharks and models

14:08

sitting on the street of markets being

14:11

fed food by like market traders there's

14:13

a man with I can't even explain

14:15

how crazy it is yeah but the

14:17

weird thing is it's kind of watchable

14:19

yes you kind of like can't look

14:21

away which obviously is why that content

14:24

gets recommended to other people because people

14:26

are engaging with it to some degree

14:28

the thing that I you talked about

14:30

the kind of speed with which this

14:32

has happened Mike it took years for

14:35

people to figure out how to game

14:37

Google and to an extent game other

14:39

social platforms. The speed with which this

14:41

has happened is I think what's shocking

14:43

to me. And I wondered like if

14:45

this is about the shift from social

14:48

platforms that are based on following other

14:50

people and then the shift to people

14:52

being grouped into cohorts or segments and

14:54

just being served content because that shift

14:56

means that AI slot can kind of

14:59

proliferate, right? If this stuff had happened

15:01

back in the day and you were

15:03

following somebody on Twitter and they started

15:05

sharing insane videos of a man, you

15:07

know, a pizza-doh man rolling out his

15:09

own stomach, which is one of the

15:12

ones in the... in the in the

15:14

article if somebody did that you would

15:16

unfollor them right or you know you

15:18

would you have the option to opt

15:20

out of that you don't have that

15:22

option when you're on tick-tock or whether

15:25

you're on reals on Instagram when you're

15:27

just served this ongoing fire hose of

15:29

posts is that what's changed I think

15:31

yeah, it gets to this issue of

15:33

like how social media for the most

15:36

part, not entirely obviously, but for the

15:38

most part switched from this thing that

15:40

it was the people that you followed,

15:42

which then led to the rise of

15:44

the algorithmic feed. And the algorithmic feed

15:46

originally on social media was designed to

15:49

sort of highlight the content from the

15:51

people that you followed that would be

15:53

most relevant to you because otherwise it

15:55

was just undifferentiated. from people you follow,

15:57

but you would miss stuff. And so

15:59

it was an attempt to do that.

16:02

The big shift then was to go

16:04

beyond that to the sort of tick-tock

16:06

style for you page, which was beyond

16:08

the stuff that you were directly following,

16:10

having that algorithm, try to pull in

16:13

other stuff. And you can understand where

16:15

there's like some element of value there,

16:17

because like maybe you don't know about

16:19

some other people. But then that has

16:21

introduced this thing, where now it's entirely

16:23

the algorithm. And so where the shift

16:26

is that it went from. You saying

16:28

who you want to follow being the

16:30

signal for the kind of content you

16:32

wanted To now the algorithm beginning to

16:34

take on not just that as the

16:36

signal But also what things are you

16:39

watching and then what might similar people

16:41

want to watch right? And so you

16:43

have this the concept of cohorts and

16:45

everything like that Whereas then you could

16:47

opt out of following certain people, you

16:50

can't opt out of your cohort that

16:52

they sort of built for you. And

16:54

that makes for a very different experience.

16:56

And so I mentioned this to you

16:58

before the podcast. I just yesterday as

17:00

we're recording this, I just interviewed Chris

17:03

Hayes, who's the cable news journalist, really

17:05

interesting guy who just came out with

17:07

this book called the sirens Call, which

17:09

is all about how our world is

17:11

now based. on attention and writing for

17:13

attention and business models built on attention.

17:16

And I think all that plays into

17:18

this. That interview will be on the

17:20

Tector podcast soon, maybe next week. We

17:22

got to figure out which thing is

17:24

rolling out when, but probably next week.

17:27

But we had a really interesting discussion

17:29

about this very thing and how algorithms

17:31

have sort of taken over our lives.

17:33

But, you know, one of the things

17:35

that we really tried to explore in

17:37

that conversation, which went, I thought went

17:40

pretty deep. was this idea that algorithms,

17:42

like a lot of people right now

17:44

want to blame the algorithms, and so

17:46

I think it's really easy to look

17:48

at this 404 story and say like

17:50

this is the algorithm's fault, right? Like,

17:53

oh, you know, algorithms, and you sort

17:55

of, I'm not saying you did that,

17:57

but you were saying we've switched from

17:59

this thing, where was the people you

18:01

followed to the algorithms, and that leads

18:04

some people say, well, you know, the

18:06

answer that then has to be like

18:08

outlaw the algorithms, like the algorithms, users

18:10

don't actually really like that. Like the

18:12

pure following feed doesn't work as well.

18:14

It's like a good algorithmic feed. The

18:17

issue to me, and this is a

18:19

lot of what I exploit with Chris,

18:21

is who controls the algorithm and for

18:23

what purpose, not so much the existence

18:25

of the algorithm themselves. And as I

18:27

said, like the early point of the

18:30

algorithms was good. It was to sort

18:32

of highlight the content that would be

18:34

more valuable to you and would be

18:36

more useful to you in the same

18:38

way that like the success of Google

18:41

early on was that it found the

18:43

stuff that you were searching for that

18:45

was better and more useful. It's just

18:47

the problem is that then people start

18:49

to game it. you're dealing with a

18:51

bunch of different incentives. You have the

18:54

company's incentives, they're trying to make money,

18:56

you have the people who are trying

18:58

to game the system, they're trying to

19:00

make money, and then you just have

19:02

like little old you who just is

19:05

trying to find interesting information. And you're

19:07

just kind of like at the whims

19:09

of all this other stuff. But if

19:11

you had more control over that algorithm,

19:13

right? So I mean you were saying

19:15

before, like it was... people you were

19:18

following, that is still an algorithm, but

19:20

it was one that you had control

19:22

over so you could remove someone. And

19:24

so the thing to me that I

19:26

keep coming back to is like the

19:28

real issue is who controls the algorithm

19:31

and what their intent is. So when

19:33

it's a company that's just trying to

19:35

keep you engaged or trying to sell

19:37

you stuff or trying to really sell

19:39

your attention to advertisers, which is the

19:42

way that most of these companies work

19:44

right now, then their incentive is misaligned

19:46

with you that leads to the sort

19:48

of insitification. And so that fits in

19:50

neatly with then a bunch of crazy

19:52

people with access to these AI tools

19:55

who can just pump out so much

19:57

slop and just throw it into the

19:59

system doesn't matter to them if it

20:01

doesn't work. Because all the incentives are

20:03

messed up for the users. Yeah. So

20:05

what you're saying is kind of the

20:08

algorithms have algorithms as a term have

20:10

been kind of associated with very centralized

20:12

platforms. Yes. The ones that we've used

20:14

for over a decade now, but what

20:16

you're saying is actually if those algorithms

20:19

are. control by users and there are

20:21

dolls and levers and levers and you

20:23

can tweak them, then actually the algorithms

20:25

themselves aren't the issue. That makes sense.

20:27

The kind of issue that I was

20:29

wondering about was like something you touched

20:32

on is when is the course correction

20:34

going to happen? You know, when our

20:36

platform is going to limit the visibility

20:38

of this kind of AI content, do

20:40

you think they have incentives to do

20:42

so? And where does labelling fit into

20:45

this? Because I remember last year, around

20:47

this time, meta. which is where 404

20:49

media focus much of its reporting in

20:51

this piece, came out and said we're

20:53

going to label AI content. We're going

20:56

to be able to detect. They did,

20:58

didn't they? Yeah, they said, we've got

21:00

this covered, don't worry about the rise

21:02

of AI content. We're going to detect

21:04

the signals and we're going to be

21:06

able to label to users that this

21:09

is AI generated content and we're going

21:11

to work with other platforms and other

21:13

partners, going to kind of embed ourselves

21:15

into some of those existing initiatives as

21:17

one called... C2PA which has got a

21:19

lot of the platforms signed up and

21:22

the suggestion was that they were going

21:24

to make it very easy for us

21:26

to kind of opt out of that

21:28

content or for that content. to be

21:30

kind of downgraded if it was less

21:33

quality. It doesn't feel like that's happening.

21:35

No. Do you think it will? Based

21:37

back to incentives. Yeah, it's a good

21:39

question, right? I mean, Mark Zuckerberg seems

21:41

to have this view that like AI

21:43

content is actually a good thing for

21:46

the platform. He's making sort of a

21:48

bet on that and he sort of

21:50

talked about that, right? And they've talked

21:52

about building their own, like AI generated

21:54

users on social media that could interact

21:56

when we've talked about that in the

21:59

past. His view, and therefore the company's

22:01

view, is whatever keeps people coming back

22:03

and using the platform, is good. I

22:05

do wonder how successful this could be

22:07

long-term, right? You know, if you're just

22:10

getting all this garbage content, like sure

22:12

it is entertaining and amusing the first

22:14

few times, like, you know, check the

22:16

shit out, basically. But after a certain

22:18

point, you have to be like... you

22:20

know, come on. Why am I spending

22:23

all my time looking at this when

22:25

I could have more sort of realistic

22:27

engagement somewhere else? And so I don't

22:29

think this works long term. I sort

22:31

of feel, if anything, the most likely

22:34

scenario is that it follows the path

22:36

that Google followed at the time where

22:38

they suddenly decided, okay, we need to

22:40

crack down on this. And then it

22:42

becomes a big project and they come

22:44

up with some system to crack down

22:47

on it. And that only lasts for

22:49

so long. I mean, I think a

22:51

lot of people would argue... probably correctly

22:53

today, like Google is a bunch of

22:55

garbage again and the search engine optimizers

22:57

have really won and Google's made a

23:00

lot of really bad decisions that have

23:02

really sort of been shidified the results

23:04

in some cases. You know, it's a

23:06

constant struggle and a constant battle and

23:08

you hope that they'll reach the point

23:11

that they'll fix it. But I sort

23:13

of feel that meta in particular is

23:15

going to have to realize that this

23:17

this kind of content is not a

23:19

long-term sustainable success, especially as there are

23:21

more and more alternatives that are sort

23:24

of trying to take a different approach.

23:26

And you know, one of the things

23:28

that, and this is another thing that

23:30

came up in the conversation that Chris

23:32

and I had, which is that more

23:34

and more people seem to be moving

23:37

away from social media platforms like this

23:39

to like group chats, where there is

23:41

no. algorithm and it's much more authentic

23:43

and you're able to engage and the

23:45

feeling of fun that people had in

23:48

the early MySpace Facebook Twitter years people

23:50

seem to be getting that feeling more

23:52

from group chats than they are from

23:54

global social media yeah and it's because

23:56

you have all this other stuff sort

23:58

of crowding their way into social media

24:01

and I think there's gonna have to

24:03

be a reckoning of some sort by

24:05

meta with that yeah So you're right,

24:07

it could be the users vote with

24:09

their feet and say actually we want

24:11

to spend more time as well. But

24:14

it could also be that regulation dictates

24:16

and governments dictate actually what content should

24:18

be labelled as AI slot or otherwise.

24:20

And you found a couple of stories

24:22

that relate to that this year, right?

24:25

Yeah. So this was interesting where Spain

24:27

announced that they're going to change the

24:29

law so that there would be finding

24:31

companies for platforms if they don't label.

24:33

AI generated content. And then there's another

24:35

one saying China is doing the same

24:38

thing. And so there is this part

24:40

of me that thinks, as I often

24:42

do, like if Western countries are doing

24:44

the same thing that China is doing.

24:46

Maybe think about that for a second.

24:48

It's funny, the article, the one on

24:51

China is from Bloomberg, and it has

24:53

the lead of China join the EU

24:55

and US in rolling out new regulations

24:57

to control disinformation by requiring labeling of

24:59

synthetic content online. And I was like,

25:02

what are they talking about regarding the

25:04

US? Because we don't have a regulation

25:06

requiring that at all. So I think

25:08

the reporting on this is a little

25:10

bit confused. The Spain one. It was

25:12

just from writers is interesting where they're

25:15

saying they have this new bill that

25:17

will impose massive fines on companies that

25:19

use content generated by AI without properly

25:21

labeling it as such. And you understand

25:23

the thinking behind it. The Chinese approach,

25:25

like it's pretty clear that that law

25:28

is designed to like stifle certain kinds

25:30

of speech and use AI labeling as

25:32

an excuse. The Spanish law I think

25:34

is meant in good faith because they're

25:36

scared about deep fakes, but I find

25:39

these approaches also to be kind of

25:41

silly and probably ineffective in part because

25:43

like where do you draw the line,

25:45

right? I mean everybody who's doing content

25:47

creation these days uses AI tools in

25:49

some form or another at some point.

25:52

Okay, maybe not everyone, but a large

25:54

percentage, right? So if you're writing and

25:56

you're using like grammarly to check your

25:58

grammar, that is an AI tool. Do

26:00

you need to label it? That it

26:02

told you you had a typo? Or

26:05

maybe it suggested, you know, not ending

26:07

a sentence in a preposition, right? Like,

26:09

is that now AI generated? You know,

26:11

I think most people would say no,

26:13

but then that raises the question of

26:16

where is the line? But same thing

26:18

with like images and videos, like if

26:20

you're touching up an image, if you're

26:22

changing a video, if you're editing something,

26:24

a lot of those tools are really

26:26

AI. So where do you draw the

26:29

line? between what needs to be labeled

26:31

and what doesn't. It feels like this

26:33

is the kind of thing, this is

26:35

my complaint with lots of regulations where

26:37

it's like, okay, I see this thing,

26:40

it is a problem, therefore let's just

26:42

outlaw it without thinking through, like, does

26:44

that really make sense? Is that really

26:46

targeting the thing or are you wiping

26:48

out a whole bunch of other useful

26:50

stuff with it? Yeah, I think the

26:53

line is pizza men rolling out their

26:55

own stomach. Yeah, how do you write

26:57

that into law though? Easy, easy. No,

26:59

I agree. Just ask Ben. Ben is

27:01

the decider. Once you see it, you

27:03

know, let me tell you that. I

27:06

will say that, I've been thinking a

27:08

lot about like what the terms we

27:10

use to refer to content on social

27:12

media platforms in the age of AI

27:14

slop. And I wonder if we can,

27:17

I'm going to use this opportunity to

27:19

pitch you, Mike, a change in terminology.

27:21

You know, we've been using the word

27:23

feed feed. for as long as social

27:25

media has been around. Yeah, I think

27:27

it's time to end the use of

27:30

feed. Now that we're consuming all this

27:32

AI slop, I'm thinking we use the

27:34

word trough. I think we, you know,

27:36

I'm going to, we can say, I

27:38

mean, I'm going to the trough to

27:40

consume my news. I'm going to open

27:43

my, you know, my app and go

27:45

to my favorite trough. I think that's

27:47

kind of befitting of where we're going

27:49

with content on the internet, right? And

27:51

we can all, you know, like, little

27:54

pigs around a trough, just consume all

27:56

the crazy images of pizza dough men

27:58

and insane human seals and whatever else

28:00

the AI generators the AI generators have

28:02

got for us. Yeah, so Ben, just

28:04

for you, I'm going to use an

28:07

AI generator to generate a picture of

28:09

pigs around a trough with social media

28:11

content flowing through the trough, all right?

28:13

So I think that's going to work.

28:15

Yeah, coming to a platform near you.

28:17

Okay, well, we went a bit mad

28:20

Denmark, but I said, fascinating story. Do

28:22

go and read that 404 media piece

28:24

would include it in the show notes.

28:26

Our next story is one that we've

28:28

touched on a lot since the start

28:31

of the year. It's... that big old

28:33

question of what Mete is doing now

28:35

that it's not fact-checking, right? And many

28:37

of you listeners will have seen this

28:39

week that the company has announced that

28:41

it's rolling out its community notes function.

28:44

It's decided that this is the way

28:46

forward. It's starting to test it with

28:48

around 200,000 people that have signed up.

28:50

And there's some interesting details in there,

28:52

Mike, in the press release this week,

28:54

though I wanted to kind of talk

28:57

a bit about and get your thoughts

28:59

on. Basically the beta is in six

29:01

languages. It's only focused in the US,

29:03

but it's covering six different languages. And

29:05

what's going to happen is that people

29:08

are going to start to see the

29:10

community notes underneath particular posts, and they're

29:12

going to be able to contribute to

29:14

them. So if you're one of the

29:16

200,000 that's signed up, you're going to

29:18

be able to give you a thought.

29:21

Every community note is going to be

29:23

500 characters. It has to include a

29:25

link. And the community notes will be

29:27

kind of, I think, if validated and

29:29

rolled over time. the very end of

29:31

the for these, it says that the

29:34

community notes are just in the US

29:36

right now, but there's a plan for

29:38

it to be all over the world.

29:40

There's a global domination in aspects to

29:42

the community notes, so it's just the

29:45

US now, but there's clearly plans to

29:47

do it beyond that. And I think

29:49

there's a couple of really interesting points

29:51

to this month, but for me there's

29:53

the missed opportunity, and then there is

29:55

the fact that they've somehow bungled this,

29:58

that it's even worse than X's community

30:00

notes. Okay. So I didn't realize this.

30:02

Did you know that when Community Notes

30:04

are applied on X on Twitter that

30:06

actually it doesn't affect the distribution of

30:08

that post? Did you know about this

30:11

probably? I think I had heard that.

30:13

I think I kind of do that.

30:15

But I hadn't thought deeply about it.

30:17

Yeah, I mean, so Facebook is meta

30:19

is copying. X, Twitter in a number

30:22

of different ways. It's using the open

30:24

source code that Twitter X has been

30:26

putting out into the ether. It's taking

30:28

a similar broad approach, but it's also

30:30

deciding not to downrank posts that have

30:32

community notes that contrast to the post

30:35

itself. And I'm just thinking that's a

30:37

clear missed opportunity. What is the point

30:39

of getting people to input on a

30:41

post and have the process that goes

30:43

on behind the scenes on a community

30:46

notes post where you have people from

30:48

different parts of the political spectrum. inputting

30:50

and potentially disagreeing and then showing that

30:52

community notes if you're not going to

30:54

then take a kind of view on

30:56

the distribution that seems like a mistake

30:59

to me or you know a decision

31:01

at least the matter is taking in

31:03

the way it's doing community notes so

31:05

I take umbrage with that then there's

31:07

also the fact that at least on

31:09

X there are community notes on ads

31:12

you know some of the best community

31:14

notes have been on random ads on

31:16

the platform where it's calling out gambling

31:18

companies for crazy claims or really low

31:20

quality e-commerce sites for the bonkers stuff

31:23

that they sell. And you get some

31:25

kind of funny interesting community notes. ones

31:27

I've seen the most at least. Meta

31:29

is deciding not to do that on

31:31

its ads and clearly there's a kind

31:33

of commercial imperative or incentive to do

31:36

that but I'm like that's another big

31:38

opportunity like at least if you're going

31:40

to take a lot of the broad

31:42

approach off Twitter and X at least

31:44

do the kind of coverage of community

31:46

notes that Twitter has. So Casey Newton

31:49

on platform has written a bit about

31:51

this as well this week. He says

31:53

that it's flawed approach. It's unclear how

31:55

it's going to affect the experience of

31:57

people on platforms, like Instagram and WhatsApp

32:00

and Facebook, but from the press release

32:02

alone, you can see that there's not

32:04

been as much thought-bin into this as

32:06

there could have been. Yeah, I mean,

32:08

there's a question of how much thought

32:10

went into all of this, right? I

32:13

mean, the sort of the entire new

32:15

approach, right? I mean, a lot of

32:17

it really felt like... Zuckerberg just sort

32:19

of giving up on contact moderation generally

32:21

and just being sick of it and

32:23

just not wanting to deal with it

32:26

and sort of saying, well, you know,

32:28

people seem to like community notes, so

32:30

let's just go with that. And in

32:32

fact, you know, just the fact that

32:34

they're using the same code, which itself

32:37

was based off of other open source

32:39

code, is kind of interesting. Yeah, this

32:41

is what happens when I engage in

32:43

ideas and good faith, Mike. You're right.

32:45

This whole you turn was never... was

32:47

never done for good reasons for better

32:50

outcomes. So yeah, I don't know that

32:52

necessarily I like trying to parse out

32:54

like why exactly are they doing it

32:56

this way? I'm not sure there's a

32:58

good answer for that. You know, I

33:00

think also like just the fact that

33:03

on X, the fact that community notes

33:05

does apply to ads may be a

33:07

legacy issue as well, and that ads

33:09

on X were based on just taking

33:11

a tweet and promoting it. And therefore,

33:14

you know, probably separating out that code.

33:16

the community notes originally birdwatch code from

33:18

those tweets probably was more difficult than

33:20

it was worth and so people were

33:22

like let's just leave it and then

33:24

Elon Musk fired everybody anyway so there's

33:27

like nobody to actually do that I

33:29

think if it became a big enough

33:31

issue, they would probably remove it too.

33:33

Whereas meta is much more of an

33:35

ads-focused business, you know, that is their

33:37

business, and so I'm sure they were

33:40

much more conscious of how that would

33:42

play, and also their ad system is

33:44

a little bit different, and a little

33:46

bit more involved. And so I sort

33:48

of understand probably the way that came

33:51

down, but yeah, it does sort of

33:53

give you a suggestion of like, this

33:55

is not really about... You know, the

33:57

whole sort of embrace of community notes

33:59

was Zuckerberg to say, like, here's a

34:01

solution, it's like power to the people

34:04

nonsense, allows him to say something that

34:06

is not entirely like we're turning off

34:08

all of our moderation. Yeah. And then,

34:10

you know, sort of just hoping for

34:12

the best. So, you know, the fact

34:15

that it's not floating back to the

34:17

recommendation algorithm. interesting. I mean, it sort

34:19

of goes into our first story and

34:21

the AI Slop, are there going to

34:23

be community notes on the AI Slop?

34:25

That'll be interesting to see what happens

34:28

there. And so I could see there

34:30

are arguments for having that play into

34:32

the algorithm or not. You know, one

34:34

of the arguments against it is like,

34:36

If it does impact the distribution algorithm,

34:38

then you're giving more incentives for people

34:41

to game the community notes. And community

34:43

notes, again, is designed to be resistant

34:45

to gaming. It's actually very clever in

34:47

terms of the way it is set

34:49

up and the sort of consensus mechanisms

34:52

that are built into it. But that

34:54

doesn't mean it's impossible to game. And

34:56

so if you were to put it

34:58

into the distribution part of the cue,

35:00

that would create a larger... attack surface

35:02

that people would go after and then

35:05

you would see brute force attempts on

35:07

the community notes feature. Yeah, that's a

35:09

theme already of the podcast. I actually

35:11

went to look at what the eligibility

35:13

criteria is for users on X who

35:15

want to sign up as contributors to

35:18

Community Notes. Meta have said you need

35:20

to have an account for six months

35:22

before you can join the program. Really,

35:24

actually... If you go to the page

35:26

on x.com that says become a part

35:29

of the community of reviewers you get

35:31

this 404 cannot access Basically it's broken,

35:33

right? Yeah. So it does tell you

35:35

the eligibility requirements, which are also the

35:37

six months and a verified phone number,

35:39

but then, wait a second. Oh, not

35:42

anymore. So we checked this right before

35:44

you can check now, Ben, but you

35:46

had me check right before we started

35:48

recording and I got the 404 page

35:50

and an error. And I just went

35:52

again to click to open it again

35:55

because I wanted to read the exact

35:57

error message. And now it is showing

35:59

me. a button to join community notes.

36:01

Oh, okay. So somebody's been listening into

36:03

a pre-pod cost preparation. There was definitely

36:06

an error before. Okay. Because you sent

36:08

it to me and I clicked on

36:10

it and I got the error message

36:12

and now I'm like, okay, so here's

36:14

what it is. If I go directly

36:16

to the link that you sent me,

36:19

I get the error message and it

36:21

says, oops, something went wrong, please try

36:23

again later. If I follow the low...

36:25

through reading the rules and then click

36:27

to go it does give me the

36:29

chance to join Community Notes and I'm

36:32

actually going to click this button I

36:34

have no idea what is going to

36:36

happen. Do you know what it is

36:38

Mike? I think it's the twitter.com versus

36:40

the X.com. It could be. So I

36:43

think there's a page on the Community

36:45

Notes help center article that directs to

36:47

a old URL that's Twitter./flow/join birdwatch and

36:49

that clearly is an old euro well

36:51

I do not I have now found

36:53

out I do not qualify to join

36:56

community notes yeah I thought that it

36:58

does it says so there there are

37:00

five categories and I have checks on

37:02

four of them but I am rejected

37:04

because my phone carrier is not trusted

37:06

oh So I have no recent rules

37:09

violations. I joined at least six months

37:11

ago. I have a verified phone number,

37:13

which I probably should remove. And I'm

37:15

not. associate with any other community notes

37:17

account, but they do not trust my

37:20

phone carrier. Interesting. So I am blocked

37:22

from joining community notes. Well, we are

37:24

all worse off for that, because I'm

37:26

sure you would have left some very

37:28

good community notes. Sorry, sorry to say,

37:30

this is embarrassing for you to be

37:33

recorded, isn't it? You know, a man

37:35

of your standing being denied. I would

37:37

expect nothing less. I would expect nothing

37:39

less. Well, fingers crossed. I will say,

37:41

meta did pop up a thing inviting

37:43

me to join their community notes. Oh,

37:46

you in? And well, I clicked and

37:48

I said, okay, and then it said,

37:50

well, we'll let you know more and

37:52

I haven't heard anything more. So you

37:54

probably went here back if... Yeah, I

37:57

might not. If Twits is anything to

37:59

go by. I may get a notice

38:01

that my phone... It's not trustworthy. Yeah.

38:03

Indeed. Great. So they were our two

38:05

kind of big stories that we selected.

38:07

Obviously we go a bit deeper on

38:10

the chunky interesting stories that we feel

38:12

merit most discussion this week. But there's

38:14

also a third story this week that

38:16

we both touched on, you know, happy

38:18

online safety act week. Yes. I don't

38:21

know how you'll celebrate. See what we

38:23

celebrate, you know, other, you know, regulatory

38:25

acts are available. So on Monday. the

38:27

USA in the UK came into force?

38:29

I'm almost surprised we're allowed to talk

38:31

then. You're in the UK and I

38:34

don't know, our podcast recording doesn't violate

38:36

the Online Safety Act. Well, I mean,

38:38

if a lot of the coverage is

38:40

anything to go by this week, then

38:42

you would think so. You would think

38:44

so. There's been a whole tranche of

38:47

op-eds and reports about what the duties

38:49

are of the USA. You know, we've

38:51

touched on it a bunch of times,

38:53

but this is the UK's... Long in

38:55

the making regulation about platforms and intermediaries

38:58

having to remove illegal content and reduce

39:00

the risk of UK users on the

39:02

platform, if they don't, you're liable for

39:04

an £18 million fine or 10% of

39:06

the global revenue, whatever's greater, that old

39:08

chestnut that we've seen from other regulations.

39:11

And we've touched on the OSA Mike

39:13

in recent weeks because a number of

39:15

smaller sites and forums and communities have

39:17

started to talk about the fact that

39:19

they're nervous about whether they are going

39:21

to be caught up in the OSA.

39:24

And we've seen a couple more this

39:26

week. My favorite was a website called

39:28

lemi.zip. I don't know if you came

39:30

across this. Finney. It's a finished news

39:32

aggregator, kind of decentralized platform, quite kind

39:35

of quirky, text only, and it's got

39:37

a very long and quite scathing review

39:39

of the OSA in which it says

39:41

the act does not meaningfully protect users,

39:43

the offcom, the regulator of the act

39:45

has demonstrated a lack of technical understanding

39:48

when drafting and enforcing these rules, very

39:50

very kind of pointed in its criticism

39:52

of the act. continues a lot of

39:54

what we've seen in red, but you

39:56

found a really interesting counterpoint to that

39:58

might, which is a forum that you

40:01

found that's taking a different tact that's

40:03

going a different way. Yeah, and so,

40:05

you know, there are a number of

40:07

different forms that are sort of reacting

40:09

different ways, so one that I had

40:12

seen was from lobsters. dot r s

40:14

the lobsters part is dot r s

40:16

at the end and it's sort of

40:18

a techie focused forum it's been around

40:20

for a while they had announced way

40:22

back when that they were planning to

40:25

block u k users the founder of

40:27

it just you know explain that it

40:29

would be effectively impossible and then this

40:31

week in the lead up to this

40:33

they announced that they had disabled the

40:35

geo block. They were going to geo

40:38

block UK users, but they had disabled

40:40

and they wrote this very long and

40:42

kind of interesting analysis of it where

40:44

I mean there are a few different

40:46

interesting elements to it to me where

40:49

they said basically we're going to do

40:51

this in part because we don't think

40:53

they're they're based somewhere in the Pacific

40:55

Northwest. I think they're based in Oregon

40:57

and you know have no connection to

40:59

the UK and even though the Online

41:02

Safety Act claims to cover every internet

41:04

site that is accessible in the UK.

41:06

Part of what this guy is claiming

41:08

is that. they have no jurisdiction over

41:10

me and I doubt they're going to

41:12

go after me because I'm a small

41:15

forum but if they do I'm sort

41:17

of willing to kind of fight the

41:19

extraterritoriality aspect of this. Okay. But what's

41:21

interesting is like he notes that he

41:23

like reached out to off-com people multiple

41:26

times trying to like have a discussion

41:28

with them and he goes through like

41:30

all the reasons why it's effectively impossible

41:32

to comply with how the fines are

41:34

ridiculous and disconnected from reality for this

41:36

which is a hobby forum that he...

41:39

He just runs on the side, he's

41:41

not making any money from it. But

41:43

it's basically like, it's just not worth

41:45

it. And there are too many questions

41:47

and too many things to deal with.

41:49

And so he just sort of gave

41:52

up on the idea of geo-blocking it

41:54

and just figures he's effectively going to

41:56

take his chance. So there is this

41:58

thing in here, which is that at

42:00

one point, offcom replied and said, when

42:03

we write to you to say that

42:05

you're in breach, that is, you are

42:07

in breach. Which, you know, like, even

42:09

though, you know, and I, this struck

42:11

me as interesting because for years now,

42:13

Offcom has been running around and I

42:16

saw them say this directly at TrustCon

42:18

that they were not going to be

42:20

this like. iron-fisted enforcer of the law,

42:22

but that they wanted it to be

42:24

this sort of back and forth, and

42:27

if they thought you were not in

42:29

compliance, they would reach out to you

42:31

and have a conversation, and they would

42:33

be the friendly regulator. And yet this

42:35

message... that even though it's kind of

42:37

interesting that it's in this post where

42:40

he's saying he's just gonna ignore the

42:42

law they say when we write to

42:44

you to say that you're in breach

42:46

you are in breach and basically saying

42:48

don't wait until you get that breach

42:50

letter and says reach out to us

42:53

work with us so they still have

42:55

that a little bit like work with

42:57

us but it's like they're not shining

42:59

away from the fact that they could

43:01

try and find you for millions of

43:04

dollars and so I thought it was

43:06

a really interesting analysis saying look we're

43:08

not going to comply and we're just

43:10

going to hope that they don't go

43:12

after us. Yeah I mean that's with

43:14

a lot of the regulations come in

43:17

that's really the role that it plays

43:19

right it's the kind of damakly sword

43:21

of yes that hangs over your head

43:23

and that they can kind of dispatch

43:25

at any moment. and the question will

43:27

be is I is often willing to

43:30

let that sword go against a relatively

43:32

small forum based in the northwest of

43:34

the state's like probably not so we're

43:36

seeing more and more reporting of implications

43:38

of the act which is I think

43:41

interesting because right now it feels like

43:43

people still don't understand implications of it

43:45

and how it affects their access to

43:47

the internet and communities that they've probably

43:49

long been part of, such as the

43:51

cycling forum and such as the kind

43:54

of developer website we talked about in

43:56

previous episodes. And the hamster forum. And

43:58

the hamster forum has shut down. Oh

44:00

man. And it says we are sorry

44:02

the forum is closed down. You can

44:04

now find us at Instagram. And so

44:07

if the idea was, you know, again

44:09

this is the point that I've raised

44:11

a bunch of times when you have

44:13

these online regulations. they often give more

44:15

power to the big companies that supposedly

44:18

these regulations are designed to deal with

44:20

and so yeah the hamster form a

44:22

forum about hamsters I'm sorry to down

44:24

and move to Instagram instead to to

44:26

listen to control speech that are users

44:28

of the forum and who are hearing

44:31

about this now I'm sorry yes it's

44:33

not great but yeah this is a

44:35

theme that we are seeing we won't

44:37

figure out the online safety at this

44:39

week Mike We all going to talk

44:41

about it again and again and again,

44:44

I'm sure. I wanted to bring a

44:46

story to you that, it's actually a

44:48

kind of positive story, a story that

44:50

I really liked. I almost sent it

44:52

to you as soon as I read

44:55

it. It's a Guardian article about how

44:57

tech experts keep their children safe online.

44:59

And I love it because it says

45:01

so much, it's kind of sensible stuff

45:03

about kids safety and children's use of

45:05

smartphones and internet devices. And in particular,

45:08

it says a couple of things that

45:10

I think are just so worth reading.

45:12

If you're a parent, if you're not

45:14

a parent... There's so much good stuff

45:16

in this article. The corruptive it is,

45:18

you know, and there's a few experts

45:21

in there that they both say the

45:23

same thing, the corruptive it is talking

45:25

about the internet to your child early.

45:27

And it's exactly what you've been saying

45:29

in previous episodes of control or speech.

45:32

There is no simple way to protect

45:34

kids on the internet. They're going to

45:36

see inappropriate content from time to time,

45:38

just like they fall over and graze

45:40

their knee. actually the only way you

45:42

can deal with it is by broaching

45:45

with them early working with them almost

45:47

as as adults and understanding that they

45:49

know as much as you or sometimes

45:51

more than you it's just such a

45:53

great article I've very rarely seen anything

45:55

like it so what did you think

45:58

about it was yeah it is a

46:00

fantastic article I'm doing you think about

46:02

it was yeah it is a fantastic

46:04

article I'm definitely going to use this

46:06

and send it to people as well

46:09

it makes the point that I've been

46:11

trying them know like teaching them how

46:13

to recognize that they may come across

46:15

things that are dangerous and learning how

46:17

to do it and you can handhold

46:19

with them but the idea of like

46:22

trying to lock down the internet and

46:24

make sure that they never see anything

46:26

bad is not only impossible it is

46:28

ineffective because it also doesn't teach kids

46:30

how to use things appropriately and so

46:33

there's a lot of like be open

46:35

to conversation, let kids know that they

46:37

might come across stuff if they have

46:39

problems, to come talk to the parents,

46:41

to have that open line of communication,

46:43

which I think is the most important

46:46

thing, and to just let them know

46:48

that there are problematic things on the

46:50

internet and teach them how to use

46:52

it appropriately, maybe work with them to

46:54

see how they're playing stuff, or using

46:56

different sites or different games or whatever

46:59

it is, and make sure that you

47:01

have that line of communication open is

47:03

the most important thing, because anything that

47:05

you do. that is like trying to

47:07

cut them off or try and block

47:10

them or it even says like you

47:12

know kids are going to break the

47:14

rules that you set for them they're

47:16

gonna get around whatever blocks but like

47:18

not freaking out about that and not

47:20

coming down hard on them either but

47:23

like figuring out like what is it

47:25

you're trying to do like do you

47:27

just want to watch something a little

47:29

bit more like let's talk about that

47:31

and this is something that we've done

47:33

in our family as well. Like we

47:36

do have time limits set on things

47:38

but we have the ability and this

47:40

comes up relatively frequently where my kids

47:42

will be using computers and they'll run

47:44

out of the time and we'll say

47:47

like hey can I get another half

47:49

an hour I was in the middle

47:51

of this or whatever and we talk

47:53

about it and figure out yeah you

47:55

know it's usually okay yeah and like

47:57

that's what this is giving this sort

48:00

of common sense approach where the big

48:02

thing is don't lock down everything that's

48:04

not effective but have those open lines

48:06

of communication and it's It feels unsatisfactory

48:08

for some people, but it's super important.

48:10

Yeah, and I'm not saying at all

48:13

that it's not hard for parents to

48:15

figure out how to set up particular

48:17

platforms or set up their home Wi-Fi

48:19

in a way that screens out most

48:21

of the risk. You talked there about

48:24

how you've done it in your home,

48:26

and I've got friends who have parents

48:28

and they spend a lot of time

48:30

thinking about this, but this article is

48:32

great for acknowledging that, but also saying

48:34

that there are alternatives to it too.

48:37

So, great peace. onto our last story

48:39

Mike which is is a little bit

48:41

linked it's it's about how teams use

48:43

social media and this one scared the

48:45

hell out of us and I'm gonna

48:47

I'm gonna pose it to you directly

48:50

okay so I want your kind of

48:52

honest answer are you a half-swipe guy

48:54

well I will say I did not

48:56

discover the concept of half-swiping until last

48:58

night when I read this article so

49:01

I did not know this was a

49:03

thing So, Snapchat had implemented this feature

49:05

at an affordance within the app where

49:07

if you can see whether or not

49:09

you've looked at the thing that was

49:11

sent to you on Snapchat, the person

49:14

who sent it to you can see

49:16

whether or not you've looked at the

49:18

content that they sent you. So, Snapchat

49:20

had implemented this feature at an affordance

49:22

within the app where if you don't

49:24

fully swipe in, but you have swipe,

49:27

you could see the message that was

49:29

being sent to you without indicating back

49:31

to them. that you had seen it.

49:33

So no read receipts. No read receipts.

49:35

There's a sort of a way of

49:38

getting around the read receipts. is you

49:40

could do that. So people couldn't tell,

49:42

but then they added this other like

49:44

premium feature for some users, like if

49:46

you paid, you could see if someone

49:48

was effectively doing the half-site. And then

49:51

you would have the information, they'd say,

49:53

oh, they're not really committing to seeing

49:55

my content. And so. There's a lot

49:57

of talk in here about how for

49:59

teenagers who are in different social relationships

50:02

or you know interested in people romantically

50:04

how they're like freaking out about this

50:06

like are they half-swiping or are they

50:08

not replying to me do they not

50:10

really like me if they only half-swipe

50:12

all of this stuff that's just like

50:15

all this social dynamic stuff that is

50:17

heightened by the affordances specifically of the

50:19

app that the developers of these apps,

50:21

I'm sure I haven't really thought through.

50:23

I think about this in the context

50:25

of, there's all these talks now, even

50:28

with Snapchat, there have been a bunch

50:30

of lawsuits about specific affordances within the

50:32

app and how they're sort of negligent

50:34

in some form or another. But you

50:36

can see how these kinds of things

50:39

develop. They're not developed to like. make

50:41

a teenage girl have her heart broken

50:43

because someone half-swipes instead of full swipes.

50:45

Or like they have this example where

50:47

this one girl wanted to hang out

50:49

with a boy and sent him a

50:52

message and he apparently half-swipe but she

50:54

had the premium account so she had

50:56

the premium account so she could see

50:58

that a half-swipe and he could see

51:00

that a half-swipe and he didn't reply

51:02

for a whole hour and then he

51:05

did say he wanted to hang out

51:07

but she still upset because he took

51:09

an hour and he took an hour

51:11

and he half-wipe and he took it.

51:13

this is this is teenagers right like

51:16

this is always the way teenagers have

51:18

been and I talked about I grew

51:20

up in a time pre-internet effectively and

51:22

calling we had to use the phone

51:24

and call and there was all this

51:26

nervousness and if their parents pick up

51:29

what do you say and how do

51:31

you address and all this kind of

51:33

stuff I think some of this is

51:35

just the nature of like being a

51:37

teenager yeah but is sort of turned

51:39

into this like big deal because of

51:42

the app itself yeah right And so

51:44

to some extent I think this is

51:46

an exaggeration, but it's also kind of

51:48

funny to see like how these things

51:50

play out in the nature of a

51:53

teenager who's sort of like going through

51:55

the things that teenagers go through. Yeah.

51:57

So true, it's the importance of product

51:59

design and how it contributes to interactions.

52:01

It's not necessarily a safety thing, but

52:03

you can see how that half-swipe and

52:06

that delay in somebody replying could lead

52:08

to the kind of anxiety and stress

52:10

that might lead to safety issues, to

52:12

bullying, to people to take decisions that

52:14

they don't want to take. So it's

52:16

a light-hearted story that has a serious

52:19

element to it. And it goes back

52:21

to the story we touched on that

52:23

was publishing the Guardian about importance of

52:25

conversations. and of course of kind of

52:27

talking to children and anyone frankly about

52:30

the kind of different affordances of different

52:32

platforms and technologies. There's no way out

52:34

of it otherwise. Yeah. And this has

52:36

been a conversation that has been helpful

52:38

for me Mike and hopefully will be

52:40

one for listeners as well. That's a

52:43

point I think we should wrap up

52:45

this week. We've really enjoyed talking about

52:47

this week's news. Thanks again Mike for

52:49

joining. It's been great to chat to

52:51

everyone. We'll see you next week. Don't

52:53

forget that review. We'll see you in

52:56

seven days. Thanks

52:59

for listening to Control Alt

53:01

Speech. Subscribe now to get

53:03

our weekly episodes as soon

53:05

as they're released. If your

53:07

company or organization is interested

53:09

in sponsoring the podcast, contact

53:12

us by visiting Control Alt

53:14

Speech.com. That's CTRL Alt Speech.com.

53:16

This podcast is produced with

53:18

financial support from the Future

53:20

of Online Trust and Safety

53:22

Fund, a fiscally sponsored multi-doner

53:24

fund at Global Impact that

53:26

supports charitable activities to build

53:28

a more robust, capable, and

53:30

inclusive trust and safety ecosystem.