A U.S. judge in New York ruled in May that President Trump may not legally block Twitter users from his account on the social media platform based on their political views. (Reuters)

President Trump likes impressive-sounding numbers. Did you know that he won 306 electoral votes in the 2016 election? It’s true. (Well, sort of.) Did you know he had 110 million people following him on social media as of June 2017? It’s . . . well, it wasn’t true. But it sounded good.

There’s also a tendency toward hyperbole displayed by Trump’s critics at times. Given any particular action or assertion by the president, you can find someone who sees in that action or assertion the fingerprints of rampant corruption or an endless conspiracy. Up to and including Trump’s claims about his social media accounts.

Since the campaign, there has been speculation that Trump’s Twitter follower count in particular has been subject to inflation. The idea is that someone — a Trump staffer? a Trump supporter? a Russian?! — is creating automated Twitter accounts, “bots,” which then follow Trump.

Consider this tweet from commentator Jacki Schechner on Monday afternoon.

They do not look like real people, no. But there’s little reason to think that they are bots.

I created a script that pulls in the last 200 followers of any Twitter user and checks to see if those followers look like the accounts above: no avatar (the picture in the circle), no background image (in place of that blue bar) or no listed biography (which would appear under the username). There’s also speculation that accounts whose usernames end in a string of numbers are more likely to be automated, so I checked for that, too.

At the time that I ran the script — Trump adds followers constantly, so this is probably no longer  true — about two-thirds of Trump’s most recent followers had no background image and no biography. An additional 40 percent had no avatar image. More than a quarter lacked all three.

About 20 percent had usernames that ended in strings of at least four numbers. Twelve percent of the total had a string of numbers and none of the customized elements above. That’s 40 percent of Trump’s most recent 200 followers.

But is that a lot?

I ran the same script on other prominent and popular accounts. For example: About half of Kim Kardashian West’s most recent followers lacked a custom background image. An additional 30 percent had no custom avatar. About a fifth had usernames ending in four or more numbers.

Trump’s recent followers were more heavily “bot-like,” using the definitions above. But other political accounts matched or surpassed those figures.

About half of the most recent followers of Prime Minister Narendra Modi of India, for example, lacked a bio, an avatar and a background image. About a quarter also had usernames ending in at least four digits. Hillary Clinton’s total was close to Trump’s: About 38 percent of her most recent followers lacked all of that identifying information.


So what’s going on? The answer almost certainly has nothing to do with “bots” and everything to do with setting up a new Twitter account.

To test this hypothesis, I set up a few new accounts. The process goes like this:

  1. Give a name and phone number or email address.
  2. Set a password.
  3. Sometimes — but not always — Twitter will ask you to upload an avatar.
  4. Tell Twitter what sorts of things you’re interested in.
  5. Peruse some sample accounts to follow.
  6. Turn on notifications if desired.
  7. Start tweeting!

The key point here is in step 5. New users are asked to start following other accounts before they customize their Twitter accounts. Some will have uploaded an avatar, but not all. (Across the board, it was more common for recent followers of the users we checked to lack a background image than a photo avatar.) In other words, if people start following other accounts as soon as they sign up, those new followers will look the way the accounts in Schechner’s images look.

Why might Trump and Modi see more of these new followers than other accounts? Probably because they’re more likely to be recommended.

When signing up for a new account from near New York City, here’s one of the suggestion screens I was shown during the sign-up process. That this is customized to the region (probably based on IP address) suggests that in other places (like India), new users will see other recommendations.


After arriving at a new user’s blank timeline, there’s another prompt to get suggestions for whom to follow. Here’s what I was shown at that point.


No wonder so many of Trump’s most recent followers haven’t been customized at all.

Oh, about those eight numbers? When you give Twitter a common name, it generates a username for you automatically. It’s some compression of the name you gave — with eight numbers appended. Like so:


It’s worth stepping back from all of this, though. There’s not really a good reason that’s ever been offered for why anyone would bother creating a bunch of automated accounts to follow Trump. Could they then retweet him en masse? Sure, but no real people are going to follow them and see the tweets. If you perhaps think that this an effort simply to boost Trump’s retweet count, it’s worth pointing out that Trump rarely has much trouble getting a lot of retweets for the things he posts. If the idea is that he wants to brag about how many followers he has? Well, as noted above, he already misrepresents those numbers, anyway.

Yes, more of Trump’s recent followers exhibit this behavior than, say, Katy Perry’s. But it’s about the same as Hillary Clinton’s. Are the Russians or the Whoevers making bots to inflate her counts, too?

A few weeks ago, Twitter decided to crack down on automated accounts, running through its system and removing accounts that they thought indicated bot-like activity. In that purge, Trump did indeed lose more than 300,000 followers which were apparently bots. Modi lost 264,000.

Barack Obama lost 2.3 million.