Honestly the more I hear about AT/Bluesky and some of these other new protocols, the more I keep coming back to thinking of ActivityPub as the one with the most potential for the future.
BlueSky is a better social media service than any ActivityPub system out there right now because of the content recommendation algorithms BS was designed to support.
You need to gather the messages from various huge AP instances for anything close to a good recommendation algorithm.
I think the whole algorithm situation is exactly what's wrong with social media today, but it's also what makes people come back for more. Sadly, I think BS will beat AP in this regard because of that.
I certainly understand where you're coming from but isn't that only demonstrable right now because BS is the only "huge" instance running at the moment? They haven't federated yet so we assume this is just going to work across the entire federation of servers?
Further, development on ActivityPub is not "done," correct? As in, someone could technically still write/develop their own client and attach its own algorithm over the top of ActivityPub, right?
To be clear, I hope I'm not sounding like my opinion on this is definitive or anything and I'm just having a conversation. There are definitely problems with ActivityPub, it just sounds like BlueSky is trying to solve/fix every problem that every individual person maybe/possibly/who knows will have before they even launch which, in my experience, is generally a concerning approach.
AP is set up in such a way that you get the content you want but nothing more. Scraping messages from other servers is slow and cumbersome, the entire system is set up for pushing content to people who are interested rather than pulling in content that may be interesting to you.
That's not a failure in any way. Actually, I think it's a much more respectful approach than what many social media companies are doing. However, it does severely impact the effectiveness in which this network of small servers can push "stuff you might like" to other users. Some major instances may be able to implement a Twitter-like algorithm, but most of them won't have the messages to recommend to people.
BlueSky has chosen a complex model that basically exchanges huge amounts of data. That should allow for the algorithmic crap that AP servers will struggle with.
Don't get me wrong, I consider AP to be better than BS because of the designs and business concerns involved. However, from what I can tell from the people around me, people want the addictive social media algorithms, outrage posts, drama videos, you name it. Every platform with this crap has people addicted and coming back over and over again. Mastodon and friends mostly seem to contain people interested in each other and their work (and not just in good ways) rather than people pushing for high algorithmic scores to boost follower counts.
This is why I think the fediverse will fail in terms of social media. The predatory alternatives will be able to pull in the general public with ways that AP-based servers never would.
