If you believe only 5% of all Twitter accounts are spam or fake, boy HOWDY do we have a bridge to sell you. And that’s, of course, what Twitter has been trying to sell Elon Musk … and why he is trying to walk away now. He asked them to provide information showing he wasn’t buying a bunch of fake users.
Hard to prove that’s not the case when it looks like there is NO WAY IN HELL that only 5% of Twitter’s users are fake. Hell, this editor probably deals with at least 5% bot accounts every day, and that’s just one user’s mentions.
Gurbaksh Singh Chahal put together a thread breaking down the data and OUCH, so much ouch, Twitter.
From.
Day.
One.
Since the Twitter and @elonmusk agreement was signed, new information has emerged that calls Twitter's oddly static claim in SEC filings that less than 5% of its accounts are fake or spam into question.
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
Oddly static because it’s BS, as he so colorfully put it in the first tweet.
I'd like to walk you through how @Twitter currently defines its (mDAU) process:
"Human reviewers randomly sampled 100 accounts per day (less than 0.00005% of daily users) and applied unidentified standards to somehow conclude every quarter for nearly three years that fewer…
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
Humans.
Not an algorithm.
Not AI.
Some blue-haired, tatted-up, mouth breather who likely marks a lot of accounts THEY don’t agree with as bots while ignoring obvious bots promoting their agenda. They love to make it look like the progressive agenda is HUGE when in reality? Not so much.
Keep going.
…than 5% of Twitter users were false or spam."
Here's a screenshot that walks you through how they determine their analysis. You really don't have to be a data scientist right there to realize that this right here is a total downright scam. pic.twitter.com/08gZqte5Yv
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
A total downright scam.
Here's what we did:
There are probably far more intelligent people who follow me who are experts in data analysis. So there's a simple technique called "data segmentation" that you can use to determine whether some "low quality" users are real, fake, or bots.
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
This is not rocket science, but rather very "basic" knowledge.
So, how would you go about doing that? Easy. You basically create types of categories that you suspect could be "low engagers," so you can check them out to see if they are problem accounts.
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
The default variable I've started with is "Years since Tweeted" as an important indicator, since I believe that after three years, someone on Twitter will simply "RT, reply at some point" to get some engagement.
After 3-10 years, it's probably not real or active, right? pic.twitter.com/lkhXNOwev5
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
Fair. Very fair.
To drill down even further, we created a list of people with "private accounts" and sorted them by "Years Since Tweeted" because they are even more likely to be low engagers because, well, their account is private, and if they haven't tweeted in 3-10+ years… pic.twitter.com/ou0Hutb3ur
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
Wow.
…that means they are likely not real/fake/active.
To make things even easier, we've narrowed a segment into:
"Years Last Tweeted, Private Profiles with Less Than 5 Followers" – I wouldn't be surprised if this included a large number of "troll farms," which open accounts… pic.twitter.com/XpA2zOIL5w
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
See why their claim of only 5% fake users is an obvious lie? It doesn’t take a rocket scientist.
Thank goodness … heh.
…quickly in order to terrorize individuals and groups, as well as large populations, in order to sway broader opinions their way.
I know, I know, there are still a lot of people who want to believe there are a lot of lurkers so they can keep their followers.
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
We’re thinking this really got going with Obama and his Organizing for America nonsense back in the day. Politicians figured out how to work the system, especially Leftist-leaning, and Twitter just looks the other way.
But, the sad news is, you're just not famous, bro.
If you have any questions about the analysis, we have a lot more in our series of #datadrops coming up. And if you have any questions about this, ask nicely and just don't be a Karen.
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
Just don’t be a Karen.
Excellent advice.
You can RT the original thread because @Twitter suppresses all of these posts. Furthermore, they have not processed our Twitter Ads payments (even though it's an AMEX CC, auto-charge), so I am unable to run a promoted post for wider distribution.
— Gurbaksh Singh Chahal (@gchahal) July 18, 2022
Shocker. Twitter doesn’t want people to see this thread … so you KNOW we’ve gotta write about it.
***
Related:
Join the conversation as a VIP Member