Podcast
Root Causes 259: What Went Wrong with the Twitter Blue Check Marks


Hosted by
Tim Callan
Chief Compliance Officer
Jason Soroko
Fellow
Original broadcast date
November 30, 2022
The Twitter authenticated identity blue check marks made a big splash and then quickly went away. In this episode we explore the intent of these check marks and why they failed. In particular, we detail the challenges involved in authenticating and vouching for the identity of an individual or organization.
Podcast Transcript
Lightly edited for flow and brevity.
It’s strange. Originally, the blue checkmark was something that mysteriously appeared after I guess Twitter would reach out to certain kinds of people, celebrities, companies, etc. I don’t even know if they had a set of rules in place. Nobody really knew what they were. But it seemed to be a bit of a place of pride to have one. It kind of meant you were somebody as decided by somebody at Twitter, and I guess rules changed once Elon took over. So, I think the point of it was supposed to be that it was supposed to vouch for the authentic identity of somebody famous. And yes, what’s your definition of famous. But, you can see where this makes perfect sense. If I’m some kind of a celebrity or a famous politician or a very powerful person in industry that somebody going on Twitter pretending to be me, putting up a picture of me – which are trivially easy to get if you are that famous – and then saying things that people are gonna think that I’m saying would be bad. And you can see a lot of reasons why it would be bad and so this blue checkmark was supposed to be, authenticate, that this was authenticated and that this was really this person or this company or something along those lines.
And then so, of course, in the early days of the new era of Twitter, the company announced that everybody could just get a blue checkmark for whatever identity they want and certainly not surprisingly to me and I think we’ll return to this, it turned out that this thing was very widely abused and so we had an account that purported to be Coca-Cola saying that Pepsi was better. We had an account purportedly to be I believe it was Pfizer saying that they were gonna make certain products like insulin for free and various other things. And the consequences of this were really bad. Some of them were reputational damages. Some of them actually cost companies money because things happened like share prices dropped. And so, the stakes were really high and then after a couple days, I think recognizing that the whole thing was ill-formed and unconsidered and not put together in a reasonable way, Twitter dropped it. Dropped the whole thing.
And the reason I say not surprising and what I’d like to focus on, Jason, if you don’t mind, is because authenticating the identity of an individual or an organization or an individual in alignment with an organization, is not at all a trivial task. And in fact, there is a whole industry – I happen to be part of it. It’s public CAs. – that have been committed to this for decades. And if you go back to the very beginning – why did we create SSL certs at all back in 1995? It was to vouch for the identity of an organization on the worldwide web. So it was in order to say, this is really this bank. Or this online retailer. Or this online brokerage. And to do that, there was an exercise that CAs would through that was codified where they had specific rules scheduled and they would go through their rules and they would ensure, they believed to their best of their ability or to a sufficiently high degree of accuracy they believed at the end, that they could really say this really is – name your favorite bank. And all of that, from what I can tell, was just missing. Completely missing from the Twitter blue checkmark.
And so now if you start to say well this is a problem because you’ve got somebody out there with this name and it’s also the name of the famous person, ok, that’s a problem that’s bigger than Twitter or a social media platform. That’s a problem that’s bigger than a certain authentication scheme. That’s a problem that has to do with the fact that the way we name ourselves is fundamentally not unique and that’s a fact that you have to live with and until you live with that fact, any scheme like sticking blue checkmarks on accounts is gonna be fundamentally subject to flaw. But, of course, this didn’t even get close to that. Like there is a lot of ways that this particular strategy could have been much stronger and much better, and it would have been using the benefit of nearly 30 years of public CAs as a starting point.
But in this case, it was just hurried to market, extremely hurried to market and nobody did that and you could imagine a Twitter or another major social media platform, a Facebook, etc., sitting and doing the homework and doing it right, possibly attaching themselves to something like certificates. They could require an EV certificate or they could do the equivalent of an EV authentication. They could require an eIDAS certificate. There’s a number of ways they could do it which at least would vouch to the degree that a human in society can be shown to have a certain name that that human has that name. Now is it the human with that name who is the same individual with the same DNA as the person you think of with that name who is famous? Not necessarily. But that’s even another level of difficulty that needs to be dealt with. But again, the blue checkmarks didn’t even get there. They didn’t even get close.
So, right. But what if someone takes that seriously. Probably a good place to leave this one for now. I for one wanted to point out the difference between being able to computer science-wise set up a blue badge that appears on a certain account and being able to make claims about who really controls that account. And I think we saw that fall apart in a big way, but this isn’t the last time this is gonna come up because, you know, how do you map an online entity or an online brand to something in the physical world. It’s not a problem that’s going away.

