0:01

this is on the media's midweek podcast

0:04

on broke lad stone this week

0:06

bloomberg reported that social media

0:08

posts about israel and hamas have

0:11

led to a sticky cesspool of confusion

0:13

and conflict on elon musks

0:15

x on saturday just

0:17

hours after hamas fighters from gaza

0:20

surged into israel unverified

0:22

photos and videos of missile

0:24

airstrikes buildings and homes

0:27

being destroyed and other posts depicting

0:29

military violence in israel

0:31

and gaza crowded the platform

0:34

but some of the har was actually

0:37

old images passed off his new

0:39

media we've also spotted and many of

0:41

fake video circulating on a line

0:43

in particular claiming to show at

0:46

children captured by hamas

0:48

specially of this appalling a video

0:51

that's been circulating and seen a two

0:53

million times on this post only on

0:55

x some of this content was posted

0:58

by anonymous accounts that carried

1:00

blue checkmark signaling

1:01

that they had purchased verification

1:04

under exes premium subscription

1:07

service some military footage

1:09

circulating on x was drawn from

1:11

video games on sunday

1:14

elon musk recommended to his

1:16

hundred and fifty nine million followers

1:18

to accounts for quote following

1:21

the war in real time the

1:23

same accounts of made false claims

1:25

are anti semitic comments in

1:27

the past ah the asher shapiro

1:30

covers tech for the thomson reuters

1:32

foundation welcome back to the show

1:34

avi

1:35

they throw have a member so

1:37

on tuesday the european

1:39

union industry chief here a break

1:42

tall told elon musk that

1:44

he needed to tackle the spread of disinformation

1:47

on x to comply with

1:49

new edu online content

1:51

rules didn't mosque take

1:54

some stuff down afterwards

1:56

i'm not sure about that i've been looking

1:59

x for the last week or so to see

2:01

how easy it is to find something that's just

2:03

completely made up about what's going on in

2:05

Israel and Gaza and see how easy it is

2:07

to find a verified or big account pushing

2:10

something that's just obviously made up.

2:12

That typically takes me a couple seconds. I

2:14

found this terrible video that was tweeted

2:17

over and over again of a woman being burned

2:19

alive that had been passed off as happening

2:21

in the conflict. And it was a seven-year-old video

2:23

from Guatemala. And then, you know, a day

2:25

later, it disappeared.

2:27

That one really got around. In

2:29

my doctor's office, the doctor's assistant

2:32

said that it had kept her up all night crying.

2:34

That image itself? Yeah.

2:37

Wow.

2:37

Wow. Wasn't that

2:39

repurposed everywhere from India to the Middle East?

2:41

Yes. It tends to crop up during crises.

2:45

Opportunistic people put it online to

2:47

try to generate interest. You asked

2:49

the question, didn't Musk take stuff down? I

2:51

mean, I have no idea why that post

2:54

doesn't exist anymore on X,

2:56

right? There's a real lack of transparency

2:58

about what's going on and how they're tackling these

3:00

issues. There's all sorts of rules.

3:03

The company has stopped issuing transparency

3:06

reports, which are the tools that we as reporters

3:08

would use to sort of parse how policies

3:11

were enforced. And they've threatened to sue

3:13

independent researchers who have tried to measure

3:16

the spread of hateful content on

3:18

the website. It is a bit of a black box

3:20

as to what's going on. And we are reduced to sort

3:22

of looking at the feed ourselves and saying,

3:24

wow, there's a lot of crazy stuff on here that doesn't

3:27

look true.

3:28

It's weird, isn't it? Because

3:30

there's so much true stuff that's

3:33

horrible enough.

3:35

Yeah. I mean, that's what I've been saying over and over

3:37

again. I mean, isn't

3:37

there enough bone-chilling images

3:40

and video to go around? One

3:43

of the things you have to understand is that Musk

3:45

has significantly changed the incentives

3:48

for how people can use his platform.

3:50

Before he took over, there wasn't

3:52

a route to make money as a creator

3:55

on Twitter by creating this verification

3:57

scheme where you pay for verification.

4:00

which allows you to get more reach and get injected

4:02

in people's algorithmic feed. And then you

4:04

can get paid out on the other side.

4:07

If you have a viral tweet, you can get a share

4:09

of the revenue. He has created

4:11

the conditions to incentivize some potentially

4:14

very unsavory behavior in a

4:16

moment like this. And the question is, has

4:18

he created the parallel institutions,

4:21

rules, hired the staff to

4:23

guard against the worst externalities

4:26

of that kind of economic system on the platform?

4:29

I don't know. We know that he fired

4:31

half of the staff when he took over. We

4:33

know he's steadily ground

4:36

down the trust and safety and other teams that

4:38

are supposedly tasked with doing this kind

4:40

of work. He just completely

4:43

acts huge content moderation teams,

4:45

people who had language expertise. Before

4:48

Musk took over Twitter was sort of a hybrid platform. They

4:50

had hundreds of staffers who were doing

4:52

editorial style functions, right? There'd be something

4:55

happening. They would create a moment. They would create

4:57

these sort of carousels where they put

4:59

authoritative sources around a certain

5:01

issue. All of that's gone. It's

5:03

been outsourced to this thing called Community

5:05

Notes. Musk is trying to do what

5:08

they used to pay people to do with volunteers. So

5:11

now they have people who volunteer to

5:13

append labels to tweets. And

5:15

there's stuff that's really good about that. It's

5:17

more democratic. But you had

5:19

a great piece earlier in this week by NBC's

5:22

tech team that got inside of the Notes

5:24

program and saw that they were overwhelmed.

5:27

They didn't have enough volunteers. There wasn't professional staff

5:29

doing the work. And meanwhile, they were racking

5:31

up hundreds of thousands of views, making claims

5:34

about churches being destroyed that weren't, about

5:36

military aid being provisioned that wasn't. And

5:39

this was all while this kind of beleaguered

5:41

team of volunteers who are now on the front

5:43

lines of this are made to label this at

5:45

the pace that they can manage it. To recap,

5:48

Twitter was not a platform where

5:50

you could

5:50

monetize your engagement by

5:52

clicks. Now it is.

5:55

Other platforms have done that, like YouTube. And

5:58

apparently they had a stiff learning. curve

6:00

and figuring out which accounts could be monetized

6:03

under what circumstances and so forth.

6:05

Do we have any clear guidelines

6:07

from X about when you

6:10

can and how you can monetize

6:13

or when they will demonetize

6:15

your account? They

6:16

have a page on their

6:18

website which is called Creator Monetization

6:21

Standards

6:22

which does lay out all of the different

6:25

things that you can do to lose your monetization

6:28

privileges, right? They

6:30

have all sorts of things. If you're promoting a

6:32

pyramid scheme, literally they have a section that says

6:35

if you're promoting miracle cures you

6:37

can lose your monetization. For me

6:39

the key question here is what

6:41

kind of architecture has Twitter built around

6:44

its monetization program to

6:46

actually create good incentives for people

6:49

who want to make money on the platform to not

6:51

go viral posting

6:53

demonstrably false information or titillating

6:56

information that's misleading. I just

6:58

don't know. I think they have an obligation

7:01

to tell journalists and tell

7:03

the public to demonstrate that they're taking

7:05

those rules seriously. Over

7:07

the weekend Musk tweeted

7:09

to his 150 million followers

7:12

that they should follow two accounts for

7:14

updated information on the conflict. He

7:17

was giving guidance. One of them was

7:20

to a place called at War Monitor

7:22

and the other was at Scent

7:25

Defender. These are two accounts

7:27

known for spewing lies

7:30

and although he eventually took down the tweet, 11

7:32

million people ended up seeing it. So tell

7:34

me what kind

7:35

of hard truths are you deriving

7:38

from the propping up of so-called

7:40

citizen journalists?

7:42

There's been a couple instances where people have found

7:44

like strange anti-Semitic things that

7:47

one of the accounts has said and then I think they

7:49

had made mistakes in the past where they've sort

7:51

of claimed certain things had happened and they

7:53

were wrong. What Musk has

7:56

done is said that Twitter put

7:58

its thumbs on the scale. in the past, that the people,

8:01

the executives and the people who worked at the company

8:03

had certain political biases that they were

8:05

not being honest about and that

8:07

he has ushered in a new era of sort of openness

8:10

and democratic horizontalism on the platform.

8:13

But really what he's done is he has just replaced it

8:15

with a new set of preferences that sort of revolve

8:17

around him. Right. And so you'll see that in moments

8:19

like this, where he's like, huh, like, what am I finding interesting

8:22

on the internet? Like, let me recommend it. Like it's a

8:24

very like personalist approach to ruling

8:27

the platform. Right. You'll see it in

8:29

his unilateral decisions about rules,

8:31

but then you also see it in more subtle

8:34

ways around, you know, introducing algorithmic

8:36

feeds. Like in the past, Twitter

8:38

had human beings curating moments

8:41

and authoritative sources. Now they're

8:43

using algorithms. Like, is that really fairer

8:46

or better? Like, no, it's like he's come up with a different

8:48

way of displaying you information that has his own

8:50

set of pitfalls. Right. As we were talking about

8:52

earlier, like, I don't really know why

8:54

a certain thing is in my feed. It's a verified

8:57

account that I don't follow. I've never looked

8:59

at before Twitter has decided to inject

9:01

it in my feed. That's an editorial decision they've made.

9:04

And there's no accountability. I mean,

9:06

that's the difference between his

9:08

brand of citizen journalism,

9:11

I guess, and what responsible

9:13

news outlets do.

9:15

I want to be clear about one thing that I really think

9:17

is important to underline, which is that, you

9:19

know, Twitter before Musk had a lot

9:22

of problems. And like, as someone who reported

9:24

on the platform and tried to bring accountability

9:26

to the platform, there are certain things they hid

9:29

that were particularly frustrating to me. For

9:31

example, you know, although they released transparency

9:34

reports, they didn't release information about

9:37

when a government would contact them and ask them

9:39

to take out information that was violative

9:41

of their terms of service. And that was a real problem.

9:43

Right. So you have to understand that, like, you

9:46

know, these platforms, it's not like there was this

9:48

perfect thing that like Musk came in and like

9:50

throw a wrecking ball into. That

9:52

being said, you know, he has taken a

9:54

lot of steps that have made it even harder to

9:57

sort of assess the social impact

9:59

of his platform. And I think threatening to

10:01

sue researchers who try

10:03

to collect information about the spread

10:05

of hateful speeches, really

10:08

chilling. How are you meant to sort of keep

10:10

tabs on this place that is

10:12

run by the richest man in the world who seems to have

10:14

a trigger finger on defamation lawsuits?

10:17

So how did X respond

10:20

to inquiries about all

10:22

the disinformation proliferating on

10:24

the platform around Israel and

10:26

Hamas? How did this response

10:28

compare to Twitter's initial

10:30

response to the flood of information in the aftermath

10:33

of Russia's invasion of Ukraine

10:35

back in February of 22? I've

10:38

been speaking to people at Twitter who were

10:40

on the front lines when the company

10:43

was trying to respond to the initial

10:45

days of the Ukraine crisis. You

10:47

see a totally different posture.

10:50

They had human rights lawyers on staff.

10:52

They had Trust and Safety

10:54

Council of NGOs

10:56

and groups around the world with experts. They

10:59

were consulting about how to make these decisions.

11:01

They released some groundbreaking new

11:03

rules around images of prisoners of war where

11:06

they were trying to apply international humanitarian

11:08

law to how the company was dealing with

11:11

images coming out of the conflict. So I'm sure

11:13

they made a lot of mistakes. And I'm sure that

11:15

there is plenty of things you could point out. But

11:18

it was a very different posture. At the

11:20

moment now, there's very little information

11:22

coming out of the company. They've tweeted some long,

11:25

mega-length tweets saying that they're taking

11:27

this seriously and that they have staff

11:30

looking at stuff. But they don't

11:32

have the same level of

11:34

granular detail, like long blog

11:37

posts that the company's kind of policy teams were

11:39

releasing in the early days of the Ukraine war

11:41

where they were trying to communicate

11:43

to the public how they were going to handle

11:46

information coming out of the conflicts.

11:48

Okay, so then how

11:51

do more responsible actual

11:53

news outlets try to staunch

11:55

the flow of false wartime

11:58

videos and images?

12:00

There's ways that you can basically

12:02

use satellite imagery, you can use

12:04

mapping technology, you can use

12:07

metadata, that you can look at an image

12:09

that's being posted online and say with a reasonable

12:11

degree of certainty, where was this taken? Using

12:14

context clues, looking at images in the background

12:16

and say, ah, this is actually from Gaza,

12:19

this is not from Guatemala. You can

12:21

look at the metadata inside

12:23

of images or inside of videos to sort

12:25

of get a sense of who might have originally posted it.

12:27

Is that how you figured out that the Burning

12:30

Girl was not from Israel?

12:33

No, Brooke, I figured that out by Googling

12:36

the words Burning Girl

12:38

video. And the first thing that came

12:41

up was a CNN story

12:43

from 2015 that said, here is a viral video

12:47

of this terrible thing that happened in Guatemala.

12:49

It was a two second thing. Whoever

12:52

had posted that, which was a verified account

12:54

that had thousands of followers, had either

12:57

not bothered to do, or they themselves

12:59

had actually re-skinned this video

13:02

and passed it off. I'm not sure the origins

13:04

of it, but you don't need to

13:06

be a whiz to debunk some of

13:08

this stuff, but it's really about the investigative

13:10

power of it. You've seen these investigations,

13:14

they've done incredible, I think

13:17

it was the post that did an incredible

13:19

recreation of the killing of Shri Naboo-Aqal,

13:22

the Palestinian reporter in

13:24

the West Bank recently, which showed

13:27

definitively that it was an Israeli sniper

13:29

bullet.

13:30

Even though the Israeli

13:32

army denied it. Right, and

13:34

how did they do it? There's these firms like Forensic

13:36

Architecture, who could go and recreate

13:39

in digital form these streets and

13:41

the ballistics, and then you pair it

13:43

with images taken from the time. And so they're amazing

13:45

investigative tools. The fact that newsrooms are putting

13:47

these kinds of people into the field does

13:50

all of us a service, because I think we ultimately

13:53

will have a clear-eyed

13:55

sense of what's going on in Israel

13:57

and Palestine.

13:59

that we've seen

14:02

the utter failure of X during

14:04

this crisis. What

14:07

are we missing out on? Because there

14:09

are other sites, Mastodon, Blue Sky

14:11

they've tried, none obviously have

14:13

risen yet, certainly, to

14:16

the influence that Twitter has had in

14:18

terms of being a springboard for awareness

14:21

and protest and pure

14:24

information.

14:25

Well, one of the trends that people are talking a

14:27

lot about is the TikTok vacation,

14:30

or the sort of like discordification

14:32

of social media. That this era of

14:35

sort of like an open platform where

14:37

you could sort of search around and you sort of

14:39

like create your own experience and people

14:41

would sort of post to the world and you

14:43

would go figure out what you wanted to follow,

14:46

that it's ending, right? And then what's replacing

14:48

it is much more either algorithmically

14:50

driven places like TikTok, where you just

14:52

turn it on and strap in, you're like, show me what

14:55

you got. Or these places like

14:57

Discord and Telegram, which are closed communities,

14:59

which are not searchable in the same

15:01

way, where people kind of cousin themselves off

15:04

into different little groups and share their.

15:07

So I think it's possible we are missing this era

15:10

that I think had a lot of positive externalities.

15:13

People kind of got to choose a little bit about what they saw,

15:16

and they could search widely around the

15:18

world and learn about things. There

15:20

was an openness to the design. I

15:23

don't think people are gonna design like that anymore.

15:25

So what do you think are the consequences

15:28

of an even greater amount of misinformation

15:30

than usual on this platform, in

15:33

the context of this conflict?

15:35

What's happening right now is the creation

15:38

in real time of a historical record of

15:41

this terrible, terrible, bloody

15:44

conflict. And flooding

15:46

the zone with BS doesn't help anybody,

15:48

right? There's always been a fog of war, but

15:52

an algorithmically driven fog of war

15:54

that actually injects potentially

15:57

false information in front of our

15:59

eyes. we scroll through the internet is a different

16:01

level of dystopian thinking. Yeah.

16:04

An algorithmically driven fog of war

16:07

does a level of disservice to the public

16:10

discourse that we haven't seen before.

16:13

Avi, thank you very

16:14

much. Thank you, Brooke.

16:15

Avi Asher Shapiro covers

16:17

tech for the Thomson

16:19

Reuters Foundation. Thanks

16:22

for listening to OTM's Midweek podcast.

16:25

And please check out The Big Show, which

16:27

posts on Friday. It's the final

16:30

part of our three-part

16:30

collaboration with ProPublica

16:33

called We Don't Talk About

16:35

Leonard. And it's about the conservative

16:38

movement's well-funded effort to

16:40

take over America's courts and

16:42

the man at the center of it all. You

16:45

can find all three parts, of course, wherever

16:47

you get your podcasts.