My take is, if a community is constrained by quality (eg moderation, self-selecting invite-only etc) then the only way it grows is by lowering the threshold. Inevitably that means lower quality content.
To some extent, more people can make up for it. Eg if I go from 10 excellent artists to 1000 good ones, chances are that the top 10% artwork created actually gets better.
But eventually if you grow by lowering quality, then, well, quality drops.
I suppose for very small societies, they may be limited by discoverability/cliquiness and not quality, so their growth doesn’t mesh with quality and so they could also get better with size.
Note, “quality” doesn’t have to mean good/bad but also just “property”. When Facebook started, it was for kids from elite schools. It then gradually diluted that by lowering that particular bar. Then it was for kids from all schools. Then young people. Then their parents too. Clearly, it’s far from dying in absolute terms, but it’s certainly no longer what it initially was. To many initial users, it’s as good as dead though.
There's another side effect: Whenever a site or community grows, it becomes a more and more attractive target for bad actors. The site finds a way to separate signal from noise, this attracts an audience, and that attracts bad actors who learn to mimic the signal in order to access the audience for their own gain. Unless the site finds a way to expand whatever it did in the first place to isolate signal from noise, to also isolate signal from mimicry, then the site will go into quality decline.
We've seen it on Amazon with fake reviews and a flood of cheap crap, we've seen it on Facebook with clickbait, bots and manufactured outrage being counted as "engagement", we've seen it on Etsy with cheap crap passed off as homemade crafts, we saw it with StackOverflow-copying SEO-spam on Google, and just blogspam and SEO-spam in general. Whenever there's a commonly-relied-upon signal, there's someone trying to mimic it, thus lowering the value of the signal.
> Whenever there's a commonly-relied-upon signal, there's someone trying to mimic it, thus lowering the value of the signal.
Goodhart's law: When a measure becomes a target it ceases to be a reliable measure.
The other day someone was lamenting the disappearance of the downvote indicator from YouTube videos, which makes it impossible to use an up/down ratio as heuristic for quality. Sometimes the measure is so good that it's the custodians of the community themselves that become bad actors.
I believe the parent commenter did not mean that StackOverflow itself became bad, but rather that its content gets copied and cloned by bots and posted as much lower quality content on Google.
So I guess the critic goes to google content filtering algorithms then.
Quality overall dropped somewhat over they years (my subjective impression) but they grew by expanding to many different niches. They also skillfully connected them, so that each one feels like a tight community, yet easy to join.
This actually harbors a good point. The quality stays high if the interactions users receive are strongly weighed towards the high-quality content/people.
I think HN fits this to an extent by weighting voting power according to account age and other factors, which would dramatically slow the speed with which an explosion of new users would change what is seen on the homepage. User comment quality remains a hard problem to solve though.
I find myself continually amazed at how HN continues to be one of the more overall generally "sane" discussion forums in existence. Still by far just about my favorite source of interesting links and equally interesting discussions about those links. "Kudos", "props", and "mad respect" at all those folks who help it stay that way.
But with traditional social media structures, as the user group grows, the quality of voting/interaction suffers - you get more people willing to like/upvote memes, clickbait, and controversy.
Maybe what we need is more media/discussion curated by people who care about quality instead of curation by the masses
The OP seemed to be speaking more broadly about any community being able to grow without ruin, and there are plenty of examples of communities that grew and improved in quality for arbitrary amounts of time. When we talk about online, fast growth, low-indoctrination, easy-to-join communities (i.e. they grow from without vs. within,) those are prone to the eternal September effect and to losing/changing identity. There are all sorts of circumstances that can effect a groups growth dynamic, like how hierarchical it is, how communications are prioritized and presented, etc.
I think fast growth and low barrier to entry are a recipe for becoming a microcosm of general human behavior/cooperation issues, but that there are mitigating factors that may allow it to work better or worse.
To some extent, more people can make up for it. Eg if I go from 10 excellent artists to 1000 good ones, chances are that the top 10% artwork created actually gets better.
But eventually if you grow by lowering quality, then, well, quality drops.
I suppose for very small societies, they may be limited by discoverability/cliquiness and not quality, so their growth doesn’t mesh with quality and so they could also get better with size.
Note, “quality” doesn’t have to mean good/bad but also just “property”. When Facebook started, it was for kids from elite schools. It then gradually diluted that by lowering that particular bar. Then it was for kids from all schools. Then young people. Then their parents too. Clearly, it’s far from dying in absolute terms, but it’s certainly no longer what it initially was. To many initial users, it’s as good as dead though.