Technology

Because there are the easy to measure Yandex web scraper bots, and the difficult to measure social manipulation bots. The former dropped noticeably.

It may also be that the latter dipped, but thats more difficult to say, and it would behoove a manipulative political entity to spread such operations to as many countries as possible so that they appear more organic and grassroot.

Imagine being a social media company with a bot problem though, how honest would you be? There are a lot of groups with motive to mislead in this scenario, so to me it makes sense that we see conflicting accounts regardless of what the truth is.

They're trying to use the out that it was search bots, and not social media bots. Which is half true. But the social media bots also measurably dropped. Just look at the likes on Elon's tweets before and after. They dropped by half.

And the social media drop was noticeable elsewhere too. So yes, search bots did in fact drop AND social media bots did too.

1. Because people still misread a single CloudFlare graph that shows a huge drop in bot traffic. The thing is, that graph only shows traffic between one specific subnet and CloudFlare. If you bomb a data center, of course that data center's traffic drops. But for the internet overall, that's less than 0.1% of the total bot traffic and social media bots don't target CloudFlare. They target Twitter, Instagram and Facebook. There is no reason they would ever even show up in CloudFlare's statistics.
2. Barely anyone seems to actually measure the amount of bot activity on social networks. They see less traffic so that must mean fewer bot-written posts.
3. Even the one linked here is misleading. Seems like they did analyze message contents but a very small amount (less than 5000 per day) and their "84% drop" happened for 1.5 days then immediately recovered.
4. For those who profit from the bots, it's a great narrative. Now they can claim that everything positive about them that you see on social media must be true because after all, the evil bots are dead and only true Americans write stuff about Trump, a couple of weeks before the midterms. Or about the German AfD fighting against being banned.
5. Cognitive biases. Many of us (including me) desperately want this to be true so we are less likely to critically examine the details. I don't think Jowi Morales who wrote the Tom's Hardware article that many other articles quote wanted to mislead anyone. He saw something that fit his worldview and went with it. It's the same reason why we've seen "the AI bubble is about to collapse any minute now" articles basically every week for over a year and yet it's still going.

i mean you could read it

Following the attacks that resulted in the complete shutdown of ‘Yandex’ data centers in sasovo and vladimir, two sharp declines in activity among the analyzed accounts were recorded on Facebook and Telegram. On October 8, 2026, the overall volume of messages decreased by 83.6%, and on October 12, 2026, by 42.6%. In both cases, the number of active accounts also declined. The same pattern – decline, recovery, and a second decline – was observed among accounts that disseminated destructive content

If you really look they might even be posted by the same user.

For example, AfD politicians' activity did not drop. This is while they support Russia, they are real people. The likes on some people's accounts went very low, though

There's a nuance. Bot activity dropped, and everybody basically accepts that. The issue is that most of it is crawlers, and the disagreement comes into how many were social "influencers". Those tend to pretend to be human, so it's harder to measure directly.

Because there’s money on both sides.

Bots did get taken out. A search engine's data centre got targeted. But it's the same thing that would've happened if it was Google getting their data centres blown up. What we haven't seen evidence for is that Russian misinformation bots have dropped off.

Fake news funding go brrrrr

Russiagate 2.0

This one comes from a Ukrainian propaganda outlet that used AI to “prove” this so I imagine that’s why it seems to conflict with other sources.