Bluesky early thoughts

11/12/19

Subscribe to my newsletter here or check out the rest of my writing here. My email is sriram@sriramk.com or you can always find me on Twitter at @sriramk


Twitter and Jack Dorsey shared a intriguing proposal today to design an open standard for social media. As someone who has spent many years working on these systems, I wanted to quickly write up some thoughts. 

I highly recommend first reading Jack’s tweets and the linked Wolfram and Masnick pieces which are incredibly thoughtful and cover the below in detail (and not written over 20 minutes on a choppy flight like this one is)



Disclaimer: this is my own interpretation and I haven’t spoken to anyone at Twitter about this - caveat emptor.

The easiest mental model is this: imagine if Gmail was the only email client ever developed and SMTP+IMAP had never existed. Gmail could have defined all things email: what is spam and isn’t. Or how various clients can work if they can work at all. But that is not where we find ourselves since SMTP does exist. These are open, documented standards which have evolved over years of open development. They allow any client to talk to Gmail’s servers and both interop and more importantly innovate without permission. Superhuman can build an innovative experience on top of email without ever having to ask for permission from Google.

Now, compare this with the current state of any social media product. You get a full “stack” - some of the most complex, intricate systems ever built in tech - which combines everything from relevance algorithms, content policies, verification policies, identity, incentive structures among many other things. When you open up any modern social app be it Twitter or TikTok or Instagram, you get this entire “stack” as designed and built by that particular company.

 BlueSky (bluesky? Bluesky? @bluesky?) imagines a world where things work a bit differently.

First, this is developed out in the open. Instead of opaque, closed system implementations, you get a community-built effort and incentive systems to match. You reap the benefits of 20 years of the tech community understanding how open source communities work.

Second - more importantly - you get to see various implementations of the core pieces of a platform. 

Here are a few theoretical examples.

  1. Alternate ranking models: every social platform uses a complex feed ranking algorithm optimizing for a combination of factors - community interaction, engagement, likelihood of spending time, etc. However as a customer you don’t get a choice of using different models. Imagine opening up any social platform and switching to a ranking that optimizes to show you only insightful content - or joyful content - or the most recent content. Or one that ranks content from people who typically don’t get attention. Or one that just shows you Keanu Reeves content. 

You can imagine a marketplace of “ranking models” and you get to plug and play any one.

  2. Alternate verification policies: Verification is a hard problem for any social platform (and no, I can’t help you get verified on any platform). You have to thread a needle between verifying people who are impersonated, notable in the real world, active on the community and be mindful of how you are giving social currency to them. 

Imagine a world where you can plug in alternate verification policies - perhaps you only trust a reputable independent third party to verify accounts. Perhaps you only want to see accounts verified if they’re backed by real world paper ID. Whatever you choose, you can again, imagine a wide variety of implementations of the “blue check”.


  3. Alternate content moderation policies: Heading into 2020, it’s clear that what content gets to stay and what gets amplified will be the key problem for social platforms to tackle. From customers to media to academia to ad partners to governments, there are multiple equity stakeholders who will all be deeply involved in shaping this.

BlueSky allows innovation and multiple variants of this. Imagine a world where a marketplace of policies and implementations can decide what gets amplified (working in concert with #1) and what doesn’t.

Like any open ecosystem, the above are just examples of what innovative tech is possible. Ideally this opens up multiple competing models with customers and the market letting the best ones rise to the top - and they’re ones we can’t even imagine today.

Now, all of the above isn’t without some hard challenges.

For starters, open systems are notoriously bad at building user friendly and centric experiences. To put it bluntly, no average user is asking for this right now. It is on these implementations to demonstrate real value for this to matter for regular users.

These systems are some of the largest and most complex systems ever built in tech. Interoperability, adapting existing systems will be non-trivial or perhaps impossible.

Additional questions will have to be answered around privacy (how do you prevent the next Cambridge Analytica)? Business models and incentives: what’s in it for any platform other than Twitter to help or adopt these? Do relevance systems provide a strong enough moat for a core ad supported business? For new developers: what’s going to motivate startups to build on top of this?

This is a non-trivial effort which may very likely fail. But I believe it is too important to not try to innovate here and I for one am excited to see how this evolves.

Would love to hear thoughts - reply to @sriramk or shoot me a note at sriram@sriramk.com.