58 comments

[ 2.1 ms ] story [ 116 ms ] thread
(comment deleted)
It's possible a bunch of people quit the platform and simultaneously a bunch of people join.

I'm sure some of that happened to some degree.

However, when you look at the numbers the sudden dramatic changes are staggering.

This could be excluded by looking at the activity on the accounts that increased-- if the amount of responses increased, are those responses by newly active accounts?
What’s the argument against this becoming an “Eron using the shredders” situation??

I’m with you, this definitely sounds like normal population fluctuations - but I am curious, what would make the above not true?

Edit: a comment on the controls or logs in place are a thorough and appropriate counter for what I’m looking for here.

Elon doesn't yet own or control Twitter, and won't for months to come.
If it isn't mysterious data manipulation, conservatives are arriving in greater numbers than liberals are departing.
What's strange though is many people are reporting there accounts are suddenly un-banned and they are allowed to post and use Twitter again.
Is this not fake news? Seems a little too fast and without the new twitter god claiming responsibility.

A taste of the misinformation to come.

This will probably get downvoted and flagged (who wouldn't like to shoot the messenger) but here's an example [0] of a person posting a long chain of twits that would almost certainly warrant a ban under the old management, and not getting shot down so far.

[0] https://twitter.com/SaraGonzalesTX

It doesn't look like they've been banned, unbanned, or changed their tone in the past month.
You can still be liberal and not a fan of censorship. For example, I don’t support Russia’s interference in Ukraine. However, I enjoy periodically reading RT.com which I know is biased propaganda. I read it to see the slant/arguments an entity I disagree with makes. At the same time, many are calling for the outlet to be censored. Maybe I’ll gather more perspectives, at the very least it will help me better support my own viewpoint. The heart of liberalism is the open exchange and debate of ideas.
I suppose the question this raises, is are these numbers ordinary daily fluctuation numbers or do they reflect a trend or is it "an adjustment period" after which we'll revert to ordinary daily fluctuations?
One of the biggest media events for Twitter in the past few years, I'd be surprised not to see people joining.
(comment deleted)
I can see the case for conservatives gaining followers (if they trust the service to treat them fairly)

I don't see the case for liberals experiencing a mass exodus of followers. Why would a follower of an account tilting liberal suddenly not want to follow the liberal account they have liked? Especially considering many who "lost" followers are somewhat miffed at the losses --so it seems that although Twitter is changing ownership these voices still see it as a valuable and useful medium whereas for the numbers to line up, the followers would have to have the opposite view from the accounts they follow. Time will tell.

I’m guessing if someone deletes their account it’ll remove them as a follower from any counts
I dunno, I would not expect to join or leave in droves so quickly --even before signatures on the deal have dried. But possible, I guess --just seems unexpected, if true.
They may have deleted their account. Certainly many people said they would if Musk bought Twitter. Perhaps they did.
That would be somewhat a kneejerk reaction before seeing how the new ownership manifests itself --but you could also be right. Hard to say unless TWTTR provides some data to give people the background to these numbers.
(comment deleted)
I would put my money on big liberal PACs and PR firms deleting botnets and cutting ties with botfarms that exist solely to boost engagement on their candidates posts.
Hopefully some light is shed on what's behind the numbers. This is as good a guess as any at this point. I mean, sure could be actual people rage-quitting Twitter, but it may also be interested parties cutting their losses.
I think they are covering their tracks and removing the bans, shadow bans, and censorship before Musk takes over and revealed it.
That doesn't explain the losses from left leaning accounts.
Is this sarcasm? Because, if so, that's funny!
If those numbers were padded by fake accounts, it does.
Twitter ToS says they ban accounts for quite a few reasons. These include promoting violence, harassement, wishing or hoping that someone experiences physical harm, promoting terrorism or violent extremism, etc https://help.twitter.com/en/rules-and-policies/twitter-rules

Maybe Twitter has started enforcing these rules universally, banned some accounts, which affected these numbers?

That wouldn't explain the increase in conservative numbers.

What seems more likely to me is left leaning people leaving the platform and right leaving people returning to it due to a perceived impending shift in the platforms policies and culture owing to the political views of the new owner.

> left leaning people leaving the platform

It’s believable they dislike the change and leaving Twitter. I don’t believe thousands of them have actually deleted their accounts.

People are generally lazy. In my experience, this quality is universal, and applies to 100% of political spectrum. Closing an account takes quite a few clicks.

It’s much easier to stop using Twitter, and uninstall the mobile apps. Doing so doesn’t affect these numbers, at least not until the 6 months specified in the Twitter's “inactive account policy” document.

Covering their tracks by making a huge obvious splash. Air tight theory you got there...

More likely that the bot farms decided to go into overdrive while no one is at the wheel. Cyber warfare is an arms race, and one side just skipped a turn.

> Covering their tracks by making a huge obvious splash. Air tight theory you got there...

If they purge the code that says "underreport Ted Cruz's follower count by 10,000", then Ted Cruz's follower count will go up by 10,000 overnight, but they'll have plausible deniability that it naturally happened due to a bunch of conservatives deciding to join Twitter because Elon bought it.

But if they don't purge the code that says "underreport Ted Cruz's follower count by 10,000", then there will be nothing noticeable at first, but then they won't have plausible deniability once that code does eventually get found.

I also heard Carl Sagan kept a dragon in his garage.

Seriously though, it's incredibly disheartening to see so many people fall victim to rampant misinformation. It's like everyone completely forgot the basic scientific method from middle school, and instead flock around unfalsifiable, sensationalist conspiracies. They're much more interesting, I'll admit, but that's what fiction is for. We live in a demon haunted world.

Are any of these numbers statistically significant? 10,000 followers change.. is that meaningful if you have 10M+?
That's not how statistical significance works. It's perfectly possible for a small variation on a large number to be statistically significant.

The question would be more whether the variance on a given day is statistically significant compared to the variance on any other day, or the average variance.

That is exactly what I’m asking! Is it unusual to have these types of swings?
The screenshots show typical daily swing is a few hundred
It's small, but imagine an account with 10,000 bots interacting with it every day. It would appear extremely active, and you wouldn't see the same username twice unless you were really paying attention.
Given the publicly announced feature freeze, it wouldn’t surprise me if the pro-free speech faction within Twitter has been able to achieve a level of control sufficient to start disabling banlists and other suppression systems.

I greatly doubt this is an Enron style paper-shredding exercise.

Why does this have to be some conspiracy theory or anything to do with the algorithm?

Elon buys Twitter, people who left the platform rejoin when they learn that news.

The story doesn't have as much impact when you look write it this way:

Popular conservatives see follower count go up by ~1% after announcement that Musk has bought Twitter.

I have no idea how this sort of thing works at Twitter's scale, but is there a remote chance that some kind of huge cache got rebuilt?
It appears that some accounts are losing lots of followers. Are these people all quitting Twitter?

  https://socialblade.com/twitter/user/hillaryclinton
  https://socialblade.com/twitter/user/berniesanders
  https://socialblade.com/twitter/user/senwarren
If these numbers have been artificially manipulated, I hope Musk nails them to the wall.
If you convert this to percent changes it's not particularly interesting.

- @TuckerCarlson +62,083/5.1M = 1.2%

- @JoeRogan +63,313/8.6M = 0.7%

- @DonaldJTrumpJr +87,296/7.7M = 1.1%

- @tedcruz +51,405/4.9M = 1%

The liberal examples all have lower numbers of people unfollowing and higher numbers of people already following. So all of their changes are less than 1%. Hilary Clinton's change is 0.05%... not particularly wild.

So, for starters it's not the mass migration to/from the platform the raw numbers look like.

In addition I'm sure there is some relationship to the number of existing followers you have and the rate that you acquire/lose them. That is, it's reasonably to say that the rate you add followers grows until about 10M and then slowly starts to shift to you slowly losing followers as you head back towards an equilibrium point.

I doubt that latter explanation really explains these numbers away, but there is a fair bit that I would want to see controlled for before jumping to the "conservatives are rushing towards the platform while liberals flee!" explanation.

edit: I do want to clarify, that I don't at all doubt that something is happening. Obviously Twitter going private is going to change the userbase. Just that the raw numbers, at least for me, make the effect seem more much dramatic than it is.

There are reports and tweets of many people being unbanned today.
This can be equally well explained by both software changes and the usual patterns of the public attention.

When any kind of high-impact news breaks out, it immediately sparks peak interest lasting a few days, and then exponentially dropping back to near zero, as the attention becomes split between other topics. It doesn't matter if the original event is still ongoing, it's just how the public attention works. You can see it quite well on Google Trends for any kind of large-impact event (e.g. [0]).

The attention to Twitter buyout will follow exactly the same curve. Most people who would ever react to it (by joining or leaving) will do so in the first few days, and then the change numbers will start dropping exponentially.

[0] https://trends.google.com/trends/explore?q=Ukraine%20war

Maybe I'm naive, but the implication that this is Musk favoring conservative accounts just seems stupid to me. I'm not saying Musk wouldn't do that, but I just think it's all too fast.

I have no insight into the inner workings of Twitter, but I'm almost certain they don't have a big bias dial that they can turn from "liberal" to "conservative" as they see fit. Based on the decade I've spent working in tech, I'd be shocked if just the development part of a change like that that could be done in a day or two. When you factor in existing priorities, coordinating with stakeholder, design document reviews, code review, QA testing, deployment time, and all the other non coding things it takes to get code from idea to production, it feels like this type of thing would take weeks or even months.

Beyond that though, let's say you could just ignore all that and we could get the code from idea to production in a single day. It's not like Elon is just going to whip out his laptop and start making commits on his own. There would probably be multiple levels of management as well as one or more product teams in the loop on this. Word would travel fast inside the company and I'm sure at least some people at the company would disagree with the decision. We would almost certainly have a "anonymous source from inside the company" leak details.

How does this thread imply this is because of Musk?
It doesn't imply that directly but there's definitely the implication that it could be linked somehow to what's happening now regarding the Twitter acquisition.
So you're suggesting it's purely coincidence?
On the other hand, there may be a toggle that in essence reads: apply manipulation (algorithmic filters)/remove manipulation (algorithmic filters).
Cleaning out the bot followers of the people who Twitter likes and promotes, and removing hidden followers from the people Twitter has tried to suppress. It's very simple to understand. It'll be very interesting when Musk reveals Jack lied in his testimony to Congress.
> removing hidden followers

Wouldn't that be revealing/unhiding, not removing?

I’m sure Twitter will be perfectly profitable with My Pillow ads.

Where are the liberal billionaires starting an alternative and offering options for twitter employees to bolt? Seems like a no brainer if those unicorns actually exist.

What's funny is that if the trend were reversed, with right-wing numbers going down and left-wing numbers going up, all else equal, every thread on HN would be full of people screaming bloody murder about leftist sabotage or somesuch.