Comment by benatkin
20 days ago
They chose apps that had personality. For instance, one of their flagship apps, Rocket.Chat, had a slash command for lenny face. https://docs.rocket.chat/docs/slash-command
This was good, and goes against the Zombocom problem.
However, as you said they didn't address self-hosting as well as they could have. I don't think it was because of other domains, but because they were envisioning people sharing them or paying for a cloud host, and didn't try to emphasize only apps that don't guzzle CPU, memory, or storage. Rocket.Chat is a MongoDB app. GitLab is another one: https://apps.sandstorm.io/app/zx9d3pt0fjh4uqrprjftgpqfwgzp6y...
Similarly on Dokku you wouldn't have apps sharing a database instance. If you had two apps that needed a database, you'd start two postgres instances.
I don't think they ever failed to see the Zombocom problem or attempt to avoid it. Curating apps for security made since. However, it didn't see enough use for them to add lots more apps to their library.
Adding an easter egg command to your chat app is trivial. Actually making it good, stable, usable is not, especially if you prioritize easter eggs.
Slack has /shrug which isn't that far off.
Idk though, I stand by it. Any company that wanted a chat app could get slack or skype or teams instead, and they are backed by a big business who assures they're safe, come with desktop apps that have pop up notifications, and were quicker to boot up, and frankly are better.
And, personality or not, any app that was ported to Sandstorm you can still run off Sandstorm too.
The killer features are one single account for everything (ala Microsoft's suite which also has that), the potential for some very cool interoperability that never quite reached its full potential, and better security, which not enough people/companies were willing to sacrifice other desirable traits for, apparently.