Change, it had to come
We knew it all along
We were liberated from the fold, that’s all
And the world looks just the same
And history ain’t changed

[…]

We don’t get fooled again

So the classic lines by The Who. Enjoy the full 8:33 of the song here.

And of course the not getting fooled again, in this blog entry, is about our social media landscape in 2023.

After we have seen Twitter go to shit in late 2022 and early 2023, are we not in the danger of being fooled again if we go for Bluesky as an alternative now? Cory Doctorow certainly sees the danger in doing that. Does the presence of Jack Dorsey on the Bluesky board, and the fact there is basically one instance of the network running (bsky.app), not leave that network open to the same single point of failure we saw with Twitter and Musk, despite the reassuring words to the contrary in the Bluesky FAQ?

I set about trying to work that out, but there is a further rock classic that likewise has some lessons for us here – You Can’t Always Get What You Want by The Rolling Stones:

But if you try sometimes
Well, you might find
You get what you need

This is the crux.

In our search for an alternative to Twitter in the post-Musk-broke-it period, we are in danger of letting our dreams of what we want in future obstruct what we need – and what is available now.

Is it worth giving Bluesky the benefit of the doubt right away, I asked myself, because it could be better than Mastodon / ActivityPub in the medium term?

Trying to answer that, and my digging into the AT Protocol that is behind Bluesky led me to a different conclusion than I had initially expected. The protocol is not better or worse than ActivityPub that is behind Mastodon, but the priorities currently are very different.

As Jean Hominal elegantly put it, “there is no way for me to spin up software to host my posts”, and focus of decentralisation in Bluesky so far is about ownership of my own identity on the network, not about a decentralised infrastructure for hosting content. Jean also pointed me to this – that how decentralisation will work in practice is far from clear. My fear then is not as I see it whether Bluesky would go to shit, Musk style, because of a malevolent single actor, but is a more fundamental question of network design and how the AT Protocol will work – and whether the design choices made in the development of Bluesky are going to be an adequate guard against malevolent users, and whether moderation can be handled at scale – or is even designed to be handled at scale.

The decentralised data / instances model of ActivityPub its upsides and downsides, but the central feature – that it gives you, the generator the content, the control over where that content is hosted – is of prime importance. You can even host an instance yourself if you want that. And it gives the administrators of instances freedom and flexibility in terms of what other instances to federate with (or not), and what rules for moderation to apply (or not). If moderation policies are too tight or too loose where you are, you move. The downsides are an absence of network-wide search, and if you need to migrate from one instance from another, you need the old instance to play along for thirty days – not insurmountable, but not perfect either.

And so I am back not to what we want, but what we need. And what shortcomings we are ready to live with.

However good the community on Bluesky is right now is not the issue. The bots will come. The malevolent actors will come. It is a matter of time. In fact however good the functionality of Bluesky is right now is probably not central either (because although the custom feeds are nice, the rest feels rather raw to be honest). And fearing what Jack Dorsey may or may not do is possibly a red herring as well – and at least is likely to not rear its head right away.

The questions instead are more fundamental.

Does Bluesky offer data portability? It might, theoretically, in future – but for now does not. And Mastodon does offer this. Does Bluesky offer the tools for effective content moderation? So far we do not know, although initial experience is not promising. Mastodon, albeit with the major caveat of needing trustworthy instance admins, at least has an array of systems for this. Is the user experience of Bluesky so consistently better than anything else to mean this aspect wins out over all other concerns? Again, no. Although theoretically it could get better.

So to conclude: Bluesky is not yet the network we need. It might improve, but the hurdles to overcome are considerable. I will keep an account on Bluesky, but my posting will be limited and sporadic. For the moment Mastodon – although far from perfect and lacking some functionality I appreciated on Twitter – looks closer to the tool I think we need, and hence will remain my predominant social network for the foreseeable future.