I set up offworld.fedisonic.cloud when it was apparent that Twitter was going to start circling the drain pretty quickly. I had an idea of running a Mastodon instance for anyone in my circle who wanted it, but with my usual overambition I started to think that it could be a business of some kind.
I first looked at Mastodon when it was announced back in 2016 or so, when it was becoming apparent when Twitter was getting gamed, and put up an instance on a VM, but the discovery process wasn't there and it was hard to find people and I took it down again. I made a mental note to watch ActivityPub and developments around it. I joined mastodon.social (and chaos.social, which is the opposite) when the toxicity in Twitter was starting to become the norm but didn't find it compelling, but still kept an eye on what was going on.
When That Guy squandered enough money to end world hunger on being That Guy on Twitter every day, enough was enough. I had a couple of underused servers so threw up a instance, invited people to see what the fuss was about and imported my followers as everyone was doing the same.
Now it's just short of a year later and I've closed that server. I'm running one for myself. set up in a better way, and of the people I invited, one has stayed on Mastodon.
Over the last year I've realised a few things:

  • Mastodon is not the Fediverse. It's shorthand for the network because it is the most popular platform, and we like to have an easy handle for these platforms.
  • Mastodon the application is technically simple to set up. I am a sysadmin of several million lifetimes' experience to be sure but it is made of common components and the documentation is well written, and if you can set up WordPress or Drupal it's not too much of a challenge.
  • Mastodon the service is challenging to maintain. The abillty to block instances and indeed users at the server level is powerful, but a year ago it relied on a hashtag which of which it quickly became apparent was open to abuse and indeed straightforward politics.
    Now, common blocklists have emerged and operators can choose what to use, but with any kind of success comes abuse, and it's obvious that public instances have bigger operating overheads than scaling their hardware.
  • Mastodon the software is resource hungry. At scale it is probably very efficient. With two (two!) active users it's overkill and things like scheduled deletes of busy follows could grind my server to a halt if that server was in the same timezone (if that makes sense). As a system that's designed with privacy in mind it's also difficult to troubleshoot as there are no logs by default (they can be switched on of course) so the main source of information is the job queues.
  • Technically, use cloud storage rather than on server storage, or at least don't use the same disk device. Cache as well, also not on the same disk device. It creates a lot of I/O.
  • Legal implications are getting more onerous across the world, and while they're a grey area for small instances, in the UK at least, we're not far from operators having to take a lot more responsibility for their users.

Running a small server with and for other people is a committment. Doing it for money would be something else again. I'm glad that there are businesses setting that up
On the actual social side of social media (remember that?), my experience of the Fediverse has been generally good. My feed seems to be largely cat pictures (especially on , but finding good people has been relatively easy. I'm not a big socialiser at all but I've long liked how the Internet works in that respect, and as ActivityPub is basically marshalling RSS feeds, we're back where were before we were rudely interrupted, just doing it in a way that makes more sense now.
I am going to try a couple of other platforms as I don't think Mastodon is the best option for a single user fediverse instance, or even two or three users. and work out how to aggregate different applications, like WordPress for example, so that's the next year sorted.

A touch of : I worked at Easynet in the early days of the UK commercial Internet, and we packaged browser in our welcome pack as it was initially free for non-commercial purposes and we stated we were providing it as an option for browsing the web rather than supported software.
The pricing model changed in February 1995 and the company had a visit from a sales person at Netscape's UK office, who presented them with a bill for the retail price of the number of copies of the browser they had distributed.
A deal was negotiated based on that (it might have been zero, I never found out the details) and the pricing policy changed within a year, quite possibly because they'd had the same response in many ISPs.
Netscape's position as most popular web browser lasted as long as it took for Microsoft to license the same code base (Mosaic, licensed from Spyglass, who had in turn licensed the code from the University of Illinois) and improve on it and include it with Windows. Their other corporate products, Web Server, Identity Server, etc which were pretty good at the end of the 90s, also got passed around various companies and survive as open source versions that are arguably more successful than their commercial antecedents.
That could have been Mark Andreesen's legacy, but it was all born out of a want to get rich, and as he's got older, he's gone the same way as so many of the early tech bros, and like many of them, not actually doing anything useful as a means.
This piece puts it in a much better way.