My theory (as someone who's worked on a variety of open source projects, in various scenes) is that "moderately technical user" is the most problematic group to offer open source projects too. The heavily technical, software for software devs, side of things generally has you interfacing with people who have some experience with open source development, and the worst you'll run into are people who are aggressively trying to "help". On the opposite extreme, you've got stuff like VLC player, where a large chunk of the userbase aren't technically inclined enough to hunt down your feedback mechanisms.

The "power user" demographic simultaneously lacks the experience with OSS development to sympathize, and possesses the motivation/ability to hunt you down to reach out to you. Even worse if you end up stumbling into the realm of technically inclined children (Minecraft modding can only be done by completely blocking any form of player feedback).

A friend at a previous company revealed after years of working together that he was the primary developer on a popular console emulator. He uses an alias and keeps it all quite secret because the level of harassment that alias endures is horrible.

The idea of spending hours every day for years straight on something you have to keep secret from almost everyone just... really sucks.

It isn't unique to the emulator scene by any means, though I'm sure some domains are worse than others. (e.g. due to the amount of interest from unsupervised children, among other factors)

I think it's just simply that there is a tiny percentage of the population that is dysregulated (and a large portion that is dysregulated a tiny part of the time)-- once your project is seen by enough people eventually some of these people cast their interest your way.

The communications mediums we use and prevailing cultures (e.g. see "geek social fallacies") are highly vulnerable to abuse, and a few abusers with axes to grind can easily enlist additional abuse from large mobs particularly where communication is public and durable. As contributors patience wears thin they become more exposed, both due to reduced kindness trigging more abusers and their justified intolerance of abuse looking more unreasonable to outsiders.

For most open source development the incentives to contribute are pretty thin. People do it because they enjoy working with co-contributors, helping out the public, and getting some positive recognition. It doesn't take much to turn that net-negative.

There is some level of attack that probably improves friendships and social cohesion. When it's just one wingnut that shows up on your mailing list the community can nuke them from orbit and everyone (except the wingnut, I suppose) can feel good about it ("Can you believe that guy?" "I know, right?!"). But when the attacks are mild enough that it's not obvious if you should ban them, when they come in with a twitter army, when it just won't stop-- it can really sap the energy out of projects.

I think projects could improve by moving more of their regular workflow to invite-only mediums-- private repositories, issue trackers, etc. If someone wants to complain about your project then they can do it on their own forum, they don't get to use the projects tools as a platform to crap on the project and deprive the contributors of their freedom to ignore the noise.

But I'm not sure, the highly open culture today has advantages in gaining new contributors but even if you're willing to eschew those advantages the change in norms makes that less viable. In 1998 you totally expected to email some patches to the developers of a project then get invited onto their email threads or a private mailing list. Today because the norm is some open github repository it's not clear how many people would be willing to work the old way.