With Elon Musk set to roll out new Twitter verification options, the need to ensure that accounts cannot impersonate companies, people or trademarks has been heightened. While some of the impersonations have been parody accounts, an activist was able to impersonate Eli Lilly, the pharmaceutical giant, and cause a six percent fall in its stock price. Organizations, both private and public, use Twitter to broadcast news and other information. For that reason, there is a need to ensure that impersonation does not occur. In that respect, Twitter handles for organizations and trademarks at least, need to be governed by a similar set of rules that govern domain names. ICANN requires that trademarks not be appropriated by others for domain names. That means no one can own a, say, Nintendo-labelled domain name other than Nintendo. The same should be true for Twitter handles.
The old Twitter verification option provided a partial fix. It used a manual process to verify accounts held by organizations and individuals. This gave rise to the famous blue check. Critically, this meant that, while others could hold accounts that used trademarks, a blue check would indicate that due diligence had identified an account as official.
The problem is that you want a process that can do this at scale. It is cumbersome to verify individual accounts. Instead, you want an automated process.
The good news is that game theory — in particular, the field of mechanism design can provide a way to automate this process.
The Wisdom of Solomon
Richard Holden and I have recently written a paper showing how a Solomonic-inspired mechanism can surface the truth. The advantage of the mechanism is that it is simple to understand and cheap to operate. We have already used it to propose a solution to blockchain front-running.
Here is the mechanism we propose for Twitter. It involves two stages.
Stage 1: Dispute Identification
- Someone, the initial claimant, creates a Twitter handle and pays a PRICE to claim it is verified.
- A challenger, claiming the initial claimant is illegitimate, pays a nominal amount (FEE) to mount the challenge.
- The initial claimant then has an opportunity to relinquish their claim. If this occurs, the handle is allocated to the challenger, who is refunded FEE. If the initial claimant chooses to reassert their claim, we move on to the challenge stage.
Stage 2: Challenge
- The challenger is given the opportunity to withdraw their challenge.
- If the challenger withdraws, the initial claimant retains the handle.
- If the challenger reasserts, then the handle is withdrawn from usage by any person.
We demonstrate that this mechanism results in legitimate claimants obtaining the handle and illegitimate ones not only being denied the handle but being deterred from trying to obtain it in the first place.
There are two cases to consider: (I) that the initial claimant is legitimate and (II) that the initial claimant is illegitimate.
Case I: Suppose the claimant is legitimate. They will pay a PRICE to obtain the verified handle. Any challenger, who they know is illegitimate (as does the challenger themselves), has to consider whether it is worthwhile paying a FEE to try and obtain the domain name. It is easy to see that any challenge will lead to a challenge stage with the legitimate claimant continuing to assert their claim. So the “best” outcome for the challenger is to have the handle withdrawn. Doing so is unlikely to gain them anything, and so all they have to show for their attempt is the loss of a FEE. Thus, the legitimate claimant retains the handle.
Case II: Suppose the claimant is illegitimate and pays a PRICE to obtain the verified handle. The legitimate claimant then can mount a challenge by paying a FEE. Why is this worthwhile for them? Well, either the initial claimant withdraws their claim, or they assert it, and the handle is parked. These are outcomes that are preferable to the legitimate claimant than having the handle fall into someone else’s hands. Moreover, given these likely outcomes, it is not worthwhile for an illegitimate claimant to pay a PRICE only to have no control over the handle. Thus, the legitimate claimant ends up with the handle.
This mechanism can easily be implemented in code. The PRICE and FEE do not need to be very large for this to work.
There are some vulnerabilities. For instance, a malicious person who wants to deny an official organization the use of a handle can challenge it and have the handle parked. That could be disruptive. Twitter, of course, could then have a likely more expensive process to sort such issues out manually.
Verification is a critical feature of social media sites. We need innovations that will allow this to be done at scale. As can be seen, being imaginative with game theory and economics provides a way of coming up with such innovations.