How to Fix Twitter

Nov 20, 2022 

John J. Schaub

We are nearing one month since Elon Musk officially completed his purchase of Twitter and it has been an eventful month to say the least. In keeping with his (and previous CEO Jack Dorsey) assessment that Twitter was significantly overstaffed the first month seems to have been focused almost exclusively on reducing headcount and rebuilding the company as a lean development led org. The pace of this change has been surprising to many but what is always missed in these conversations is that when driving massive changes you are far better off to move too quickly and cut too deep than to move too slow. If you move too fast you may be forced to back off and even rehire positions that you cut but if you move too slow there is a real chance that the company fails before you realise your error. Simply put moving too fast is survivable moving too slow probably isn't. With how bloated Twitter was rumoured to be it is entirely possible Elon is taking a reasonable approach as much as the internet hates him for it, time will tell if Elon has been too aggressive and I’ll withhold judgement as I have only anecdotal information to work with.

Where I’m confident offering my two cents is on the projected Product roadmap because I have some real background in that area. In my opinion the number one priority for Twitter needs to be verification of identities and the elimination (or more likely careful containment) of bot accounts. Simply put, if Twitter is to succeed in its role as a global public square users (and advertisers) must have confidence that the opinions being presented are those of real people and not mass fraud networks or nation state backed Psi Ops campaigns. The good news is that verification of identity is not rocket science; it is essentially a solved problem in the banking industry and solutions in place there can be adopted very easily by Twitter. The one wrinkle is that verification tools tend to be jurisdictional and serving a global audience as Twitter does will take a while to roll out solutions on a country by country basis but it can absolutely be done. Once you have a solid lock on the front door the next step is to detect compromised accounts meaning those accounts that were opened by legitimate people but have been taken over likely via malware. It would be a stretch to call this a solved problem in the banking sector but it is very close to solved. There are solutions in production at major Financial Institutions today that are extraordinarily capable of detecting compromised accounts via machine learning and digital fingerprinting (tracking user behaviour). These solutions are even better if you happen to have an app installed on a smartphone (as Twitter often does) as that provides vastly more user identifying data to fingerprint on. This is not to say that you need to block bots or parody accounts. They can absolutely be allowed to exist; they just need to be clearly marked as such to maintain the trust in the system.

With the bot plague under control you can then begin to focus on the deeper problem facing Twitter and frankly Social Media as a whole (there is nothing stopping these projects from running in parallel). That is the lack of any reliable indicator of reliability or reputation for a given poster. IE is the person posting this comment a globally recognized expert on the subject or a drunk teenager or professional Troll. The major weakness on the current incarnation of Twitter is that the only reputation indicator is follower count which given the bot problem is borderline useless but the fact is even if you get rid of the bots followers alone are a very poor indicator of whether someone's opinions are worth giving any credence to. My suggested killer feature for Twitter is simply a prediction market where users are prompted periodically to make predictions about real world events (Who will win the next election, Who will win the Superbowl, Where will the DOW be in six months sort of thing). A poll or two a week over the span of six months is going to provide an incredible data point on the objective reasoning skills of a given user and allow Twitter to decide what replies to rank higher or lower as needed. By allowing for those users with a better grasp of objective reality to rank higher you will vastly diminish the louder is better crowd and hopefully establish some sense of normalcy to the conversations on Twitter. You can continue to refine this process by identifying tweet topics and prioritizing those that have demonstrated fluency in that topic so in effect doctors get boosted on tweets about medicine and engineers get boosted on tweets about rockets. Over time an algorithm of this sort would become a piece of IP akin to the Google search algo and could be monetized directly or made available as a public good. 

If you have no background in how a prediction market of this sort could be leveraged start here. I'll note the actual prediction score can be a part of the overall reliability score and if you go down the path of building a marketplace there are other obvious ways to identify reliable users as well.

Ok so now we’ve solved the major issues it is time to focus on how we make back the $44B we spent. The first thing to recognize is that account verification and a real effort to restore civility and objective reality to the conversation will go a long way to bring in advertiser dollars. The next step is to allow long form and video content and allow creators to charge for access to that content. Yes this could put Twitter in competition with OnlyFans and whether or not you go down that road will be a question that needs careful consideration but there is also the ability to create a microtransaction based news and entertainment ecosystem that would have massive revenue potential. Beyond that since you now have verified identities and a reliability score you can also open up a Marketplace and P2P Payments solution both of which can be monetized in a straightforward way. There are actually a lot of very real monetization paths for Twitter and once you have the ship floating and headed in the right direction you can choose between them. 

In summary the first month hasn’t been great but it is likely nowhere near as bad as the zeitgeist would have you believe, there are absolutely real problems to be solved but they can be solved by borrowing solutions from adjacent fields and does not require any vast amount of creativity or superhuman effort, and finally there exist a myriad of very viable monetization paths to be explored. All in I don’t think things are nearly as bad at Twitter as people seem to think but time will tell.