Author here. I needed to split this up into multiple posts because it required a lot deeper dive than I had originally expected.
What I'm really interested in, however, is help with answering the following questions:
What would open sourcing the Twitter algorithm actually look like?
Would it be possible to abstract away all of the engineering complexity that it takes to run a global, production system like Twitter and produce an OSS spec or API that is actually useful?
Would it be possible to produce meaningful results without access to Twitter’s full data set?
What does meaningful even mean here? How would we define success?
What would need to happen in order to make this a reality?
What are some practical proposals that would help improve the status quo?
Agreed! This is one of the biggest changes that Netflix did to its UX to help users understand why they were seeing certain recommendations. I'd need to look up the source, but iirc it dramatically increased user perception of it's recommendation system.
These are exactly the types of questions I'm trying to get more people to think about — and the reason I wrote the article above.
It's not well-defined, but the goal imho should be to allow more transparency and optionality around Twitter's main feed UX.
Open sourcing twitter's algorithmic feed would be a monumental task that's honestly unlikely to ever come to fruition, Elon or no Elon. BUT twitter could certainly do more to move in that direction.
> What are the actual requirements that you're looking to satisfy by doing this?
If it improves transparency around how twitter's algorithm works.. if it introduces more optionality into twitter's main UX (and isn't just a chrome extension, for instance), if it's coming from twitter itself and not a third-party.. if the effects of different factors on relevancy signals are made more transparent (like being able to play around with a simulated tweet under different conditions and seeing the resulting relevancy scores for a set of example users).. there's just so much that twitter could do here to improve transparency around how their feed works. And I'm generally in favor of anything what will move the status quo.
In Elon's words “Civilizational risk is decreased the more we can increase the trust in Twitter as a public platform.”
Actually the more I trust Twitter the more risk I am exposed to. Avoiding Twitter and discounting anything to do with Twitter is a better risk management policy for me.
What I'm really interested in, however, is help with answering the following questions:
What would open sourcing the Twitter algorithm actually look like?
Would it be possible to abstract away all of the engineering complexity that it takes to run a global, production system like Twitter and produce an OSS spec or API that is actually useful?
Would it be possible to produce meaningful results without access to Twitter’s full data set?
What does meaningful even mean here? How would we define success?
What would need to happen in order to make this a reality?
What are some practical proposals that would help improve the status quo?