Welcome Yang supporter! This post kicked off a firestorm on Twitter, which I suppose isn’t a surprise, and you bring up many of the same points. Since I can no longer pretend to keep up with the Twitter thread I’m going to end my response here

“Andrew Yang has a 95% real following”
This is great news, but also irrelevant. What Twitter Audit is looking for is fake followers, which is a different issue. While accounts that are used to promote or hijack hashtags (which is what this article describes) might ALSO be used to pad people’s following counts, they are not likely to come up on Twitter Audit or similar services. These services define fake accounts as low activity, disjointed ratios between following/followers, etc. This is explained directly on Twitter Audit’s page:

Each audit takes a sample of up to 5000 (or more, if you subscribe to Pro) Twitter followers for a user and calculates a score for each follower. This score is based on number of tweets, date of the last tweet, and ratio of followers to friends. We use these scores to determine whether any given user is real or fake.

So Twitter Audit did not give Andrew Yang a 95% real score because they inspected each of his 1 million plus followers but because .5% of them appeared to be active with equal followers to friends. This is not proof that Yang Gang bots do not exist, because bots promoting/hijacking hashtags do not need to be followers to do their job, but it is good news.

“Not sure who you support in the race…”

What’s ironic about this situation is that I actually like some of Andrew Yang’s policies. His use of VAT to raise funds for his programs is a much smarter idea than Warren’s “wealth tax”. Wealth taxes have failed in every country that has tried them largely because if you are rich enough to be subject to them you are rich enough to hire a fleet of lawyers and accountants to get out of them but I digress…

“Myself and the many other yang gang you’ll see in real life have been inspired by this man to actually pay attention to what’s happening in politics.”

This is no doubt true. One of the reasons why I wanted to write this post is because it was not clear whether the Yang Gang was the victim or the perpetrator. I don’t think the fake accounts are part of a nation state attack. They lack some of the hallmarks of that and my instinct is that these accounts are created by Yang Gang supporters unaffiliated with the campaign who are misguided in their attempts to help. But we know that nation state attackers masquerade as supporters of BOTH SIDES of political issues. So while I understood that by focusing on Andrew Yang in this post I was inevitably going to incur the wrath of Yang supporters, it’s weird to me this attitude that Yang Gang accounts cannot be fake because they are Yang Gang accounts. As if there’s some magic force field that prevents attackers, even attackers who are only trying to promote their hip hop boy bands, from typing in a hashtag.

Your movement is vulnerable to being targeted. That’s just a fact of life. Putting sunlight on this issue is a good thing, researching these behaviors and how they work is a good thing, because it helps keep legitimate Yang Gang members from increasing the influence of these attackers by interacting with them.

“I have a new account that posts X times a day and I’m very real!”
(In fairness you didn’t say this, but since this is my last post on this subject I figured I’d throw it in to my response) A bunch of people misunderstood what I’ve been doing and assumed that when I said the software focuses on accounts that post more than 70 times a day, everyday that was the ONLY criteria. What I meant by that is that I am not at this time attempting to find more sophisticated bots that go below that threshold, not that I labelled every account that mosts more than 70 times “a bot”.

That’s also not a criteria I just invented. It’s the standard set by the Atlantic Council’s Digital Forensic Research Lab. 72 posts is suspicious, more than 144 per day is highly suspicious.

One of the reasons why I focused this post on specific stories of accounts and activities that I had found while tracing a network of Yang Gang accounts is because the line between abusive use of Twitter and acceptable use of Twitter gets so muddy around inauthentic behavior that virtually ANY method of bot detection is likely to make mistakes. The bot detector would find accounts that fit my criteria THEN I WOULD REVIEW THEM. Which means if I had an account that posted 100 times a day but when I opened it up those RTs included original commentary, I took that account out of the network. I’m not interested in studying the behavior of Andrew Yang supporters who post a lot on the internet. I’m interested in studying the behavior of accounts that post HUNDREDS OF TIMES a day, in an automatic fashion without any original content.

The weird thing is when people started complaining and rending garments about how unfair I was being to them and their high Twitter use, I pointed the bot hunter at their accounts to see what would happen. Every single one of the people who assumed that their very real accounts would register as a bot came up as completely human. The person who was the most abusive, obnoxious and convinced that I had maligned her registered as 12 posts per day. That’s so far below the threshold it is laughable. People don’t realize how hard it is to post 100 times a day, everyday, without some kind of automation.

Anyway, to any Yang supporters that read this: I’m sorry I hurt your feelings. I realize this election cycle is intense and emotional but that was not my intention AT ALL. That being said you have to understand that the larger and more powerful your movement becomes, the more you will be targeted by attackers. Trying to shut down research into those attacks will not keep them from happening.