The ongoing Elon Twitter saga: "insert demographic" melts down

[Werewolf]

I looked on Github, i couldn't find any recent update to the algo.

Commits · twitter/the-algorithm

Source code for Twitter's Recommendation Algorithm - Commits · twitter/the-algorithm

github.com

Is there anyone left at twitter with access to submit that now?

is it still policy under musk to do it?

It's not exactly proof that it hasn't been changed.

 

[Trusty]

Is there anyone left at twitter with access to submit that now?

is it still policy under musk to do it?

It's not exactly proof that it hasn't been changed.

Those large scale sackings happened before the last commit, so i would assume there is.

 

[Mr Jack]

I dont think it really matters. If he tested it, he tested it i suppose.

If he made any real attempt to test it - and given the broad range of his crankery I don't see any reason to trust that he did. Even if he did, it doesn't make any difference: Media Matters have proof. These ads were served by Xitter on the content they say they were.

When you see stuff like below and see the difference in outrage from people, i do wonder whether there is an ulterior motive.

Frankly, that's shocking, and I'm not particularly surprised to see it on TikTok. TikTok is surely a more troubling platform than Xitter, even in under the Muskian era of deliberately enabling purveyors of hate speech. But TikTok's issues are not really the subject under discussion in this thread.

 

[Trusty]

If he made any real attempt to test it - and given the broad range of his crankery I don't see any reason to trust that he did. Even if he did, it doesn't make any difference: Media Matters have proof. These ads were served by Xitter on the content they say they were.



Frankly, that's shocking, and I'm not particularly surprised to see it on TikTok. TikTok is surely a more troubling platform than Xitter, even in under the Muskian era of deliberately enabling purveyors of hate speech. But TikTok's issues are not really the subject under discussion in this thread.

Sankari posted it above. But it seems Sean Maguire agrees that it's gamed and load of nonsense from Media Matters. Hmm

https://twitter.com/i/web/status/1726666604677849441

 

[Mr Jack]

Sankari posted it above. But it seems Sean Maguire agrees that it's gamed and load of nonsense from Media Matters. Hmm

Yes, this was posted and responded to before.

They're investigative journalists investigating how ads are served. Of course they refreshed to get more ads. That's what they're looking at. Of course they looked at what ads were served on dubious accounts that's what they were looking at.

This is absolutely zero defence. It's literally "investigative journalists did investigative journalism".

And the 50 out of 5.5bn is utterly laughable apples to oranges guff.

 

[dlockers]

Yes, this was posted and responded to before.

They're investigative journalists investigating how ads are served. Of course they refreshed to get more ads. That's what they're looking at. Of course they looked at what ads were served on dubious accounts that's what they were looking at.

This is absolutely zero defence. It's literally "investigative journalists did investigative journalism".

And the 50 out of 5.5bn is utterly laughable apples to oranges guff.

Err excuse me, why aren't you investing all of your spare time defending billionaires?

 

[Trusty]

And the 50 out of 5.5bn is utterly laughable apples to oranges guff.

But it isnt though is it, is it ever going to be zero? I'm not sure you could do that considering the natural fluidity of accounts and what they post.

 

[Mr Jack]

But it isnt though is it, is it ever going to be zero? I'm not sure you could do that considering the natural fluidity of accounts and what they post.

It is, because they're falsely claiming that it's only 50 out of 5.5 billion, which of course they have absolutely no idea about (if they did, they'd not be serving these ads). Twitter could, as other social media companies have done, responded by explaining how they were going to increase their efforts to deal with such occurrences. Instead we got a bonkers legal threat, not carried through on about how Media Matters were being mean to them with some nonsense defence.

And, more importantly, all this comes against a background of Musk actively inviting posters with a bad history of posting back on to the platform, removing verification posts, and Musk himself describing an anti-Semitic post as "the actual truth" among numerous other dubious posts.

I doubt Media Matters report has had half as much impact as the other things.

 

[Trusty]

It is, because they're falsely claiming that it's only 50 out of 5.5 billion, which of course they have absolutely no idea about (if they did, they'd not be serving these ads). Twitter could, as other social media companies have done, responded by explaining how they were going to increase their efforts to deal with such occurrences. Instead we got a bonkers legal threat, not carried through on about how Media Matters were being mean to them with some nonsense defence.

And, more importantly, all this comes against a background of Musk actively inviting posters with a bad history of posting back on to the platform, removing verification posts, and Musk himself describing an anti-Semitic post as "the actual truth" among numerous other dubious posts.

I doubt Media Matters report has had half as much impact as the other things.

I imagine it's a bit like whack-a-mole, the fluidity of accounts and what they post means that some bad stuff will fall through the net. But the algo is open source and transparent, if Media Matters are gaming the investigation then i don't see how that a legit way of doing proper investigative journalism, it won't be anywhere close the average experience.

 

[StriderX]

Sankari posted it above. But it seems Sean Maguire agrees that it's gamed and load of nonsense from Media Matters. Hmm

https://twitter.com/i/web/status/1726666604677849441

Yet twitter admits it's the truth that ads have been featured next to such content... how is it gamed if it's the truth?

The fact is that advertisers may accept that there's going to be mistakes, what they don't like however is having their brands blasted on multiple headlines as being in any way connected to extremists and with Musk's behaviour recently that's going to keep being highlighted.

 

[Trusty]

Yet twitter admits it's the truth that ads have been featured next to such content... how is it gamed if it's the truth?

It's never going to be zero, is it?

 

[unwashed potato!]

It's never going to be zero, is it?

What is the point you're making?

 

[Sankari]

I looked on Github, i couldn't find any recent update to the algo.

Commits · twitter/the-algorithm

Source code for Twitter's Recommendation Algorithm - Commits · twitter/the-algorithm

github.com

Looks like there's a rollout lag. The July 2023 update was introduced in August, so perhaps the November update is still being finalised.

 

[unwashed potato!]

Are you saying the algo has been changed? Where have you seen that?

To be clear, you're suggesting at a time of heightened worry from advertisers, that xitter has made not a single change to how ads are served? Nothing at all? Do you think that little of xitter that they'd be so inept to not even announce that they are improving how ads are served, improving their algorithm etc?

Also, how can I refresh the page to reproduce what media matters have done, given that it was IBM advert that appeared, and IBM have left. And Apple, Google, Microsoft... Is anyone left to have their adverts appear next to these posts?

I don't get this obsession over how media matters investigated and found the system to serve these ads. It's a weak defense.

Also this idea of "but look at tiktok", surely it should be "also Tiktok sucks too".

I'm sure most sensible people in this thread see that these ads were and can be served against Nazi posts, and think that xitter and all social media sites need to do more to combat that. They need to not just do more, but appear to do more.

Xitter has an issue that not only do they appear to be doing less, but the owner down right refuses to accept they could do better and changes will be made, down right refuses to reassure advertisers that things are moving in the right direction in terms of how their algo works, and down right refuses to appear like an adult.

As much as we can discuss this, these massive companies have decided that the evidence is enough and pulled back. And I imagine it wasn't just because of media matters. But also Elon's tweet. And the last year of Elon. Everything is moving in the wrong direction. Xitter can't keep itself out of the news for all the wrong reasons. At any other company, if it had the ceo constantly dragging it down and getting negative news, and unable to respond to that bad news in a professional way, they'd be gone. Anyone remember BP back when it had that oil spill and how long the CEO lasted during that event, and the final nail was him comment "i just want my life back". But Elon can't be removed easily. So advertisers appear to say it's not worth dealing with this company. Too risky. Better options else where. And it's true. And it'll be more true tomorrow than it is today.

 

[tang0]

https://twitter.com/i/web/status/1726747531843915965

 

[unwashed potato!]

Let's dissect that.

Not a single authentic user on X saw IBM’s, Comcast’s, or Oracle’s ads next to the content in Media Matters’ article.

what is she considering authentic user? how many users saw the ads? why can she list how many users saw apple's ads next to nazi tweets, but not how many users, what she considers authentic or not, saw the ads. purposefully excluding this? if they consider the type of account that media matters made to be inauthentic, why do they still show these companies ads to them? can the xitter staff not figure out an authentic/non authentic users from each other?

Only 2 users saw Apple’s ad next to the content, at least one of which was Media Matters.

on top of IBM, there was also apple's ads shown. but what about the 1000 companies not listed that also advertise on xitter. isn't the issue that ads besides nazi tweets do and can happen. How many "mypillow" equivalent ads were shown next to those tweets?

I'm all for explaining the situation, but it seems like she's worded it to be vague and sound like she's saying "no one saw these ads against tweets" when in reality she's saying "no one we consider authentic saw the ads, but they did appear", and "we wont tell you how many times 3 big companies ads appeared, but we will tell you how many times apple's ads appeared", and also "we wont tell you how many other companies ads do get shown".

Perhaps she could expand a little.

 

[Trusty]

What is the point you're making?

The point is it'll never be zero. Just because it's a tiny percentage doesn't mean that it's bad. All social media companies have issues, it'll never be zero.

 

[dowie]

Yes, this was posted and responded to before.

They're investigative journalists investigating how ads are served. Of course they refreshed to get more ads. That's what they're looking at. Of course they looked at what ads were served on dubious accounts that's what they were looking at.

This is absolutely zero defence. It's literally "investigative journalists did investigative journalism".

And the 50 out of 5.5bn is utterly laughable apples to oranges guff.

Did they report it that way though? Like how was it put across?

I've not looked at this media matters story yet though he did manage to get the ADL to back off very quickly with a lawsuit threat after they'd been up to some shenanigans with advertisers.

 

[unwashed potato!]

The point is it'll never be zero. Just because it's a tiny percentage doesn't mean that it's bad. All social media companies have issues, it'll never be zero.

we can all agree on that. but isn't the issue that more than just that it does happen, however infrequent it may currently be?

 

[Trusty]

we can all agree on that. but isn't the issue that more than just that it does happen, however infrequent it may currently be?

Not when you're gaming the experiment. The chances of being struck by lightening are tiny, but if you go round trying to get struck, can you whinge that you eventually did? I think it's poor form. But if there is a ulterior motive, then it's done it's job