I think if I had to do it all over again with Spectra, I would’ve done the following things differently:

1. Data Caps By Default: everyone can have, say, 25gb or reserved storage per account, with the option to purchase more reserved storage in the app as necessary. There are streamers on the instance who upload hundreds of videos that are easily 1 to 3 hours long each.
2. Premium Paid Features: I wouldn’t restrict any of the main PeerTube functionality, but certain features like syncing up your entire YouTube account with every video it’s ever made would make a lot of sense for a monthly plan. At one point, I tried to offer this to everyone, and loads of people tried to import over a decade’s worth of old videos.
3. Better Outreach: The whole “if you build it, they will come” strategy of cultivating online communities can be inherently flawed. Historically, I’ve had pretty much no strategy with reaching out to creators who might be interested in posting their videos to the Fediverse, because
4. Better Infra: This is a vague thought, not really defined with specifics. Sometimes, you can’t predict what your server is going to look like years and years down the road. At the risk of sounding like some know-nothing, I probably would’ve looked more into a Docker / Kubernetes setup? The instance itself is not hard to run or do maintenance on, but a more decoupled setup optimized to scale up would probably be a good idea.
5. More Curation: I take a very…relaxed perspective on how to manage an instance. Block bad actors, uphold community standards, help people with issues when they crop up. But for the most part, I don’t do any curation or community-building. The instance often just feels like a bunch of random stuff, which is kind of a bummer. The least I could do is dig in and add fun stuff to a playlist or something.

I’m saying all this not because I’m in some bad place or lamenting my decisions, just more reflecting on the fact that running a community instance for media is really demanding sometimes.