Hacker News new | past | comments | ask | show | jobs | submit login
A Group Is Its Own Worst Enemy (2003) [pdf] (gwern.net)
88 points by froasty 40 days ago | hide | past | favorite | 14 comments

I think fukuyama's book on high + low trust societies is about these topics

heard someone claim in an interview that low and mixed-trust societies generally turn to surveillance / censorship

'which attributes of a society enable free speech' is a question people we'll ask seriously + continuously about online communities

What's the most important insight in this essay that HN does not line up well with [1]? When I look at his key points [2], the ones that HN seems to line up well with are:

The place that was founded on open access had too much openness. They had no way of saying, “No, that’s not the kind of free speech we meant.”

Technical and social issues are deeply intertwined. There’s no way to completely separate them. Having good software isn’t enough.

Constitutions are a necessary component of large, long-lived, heterogeneous groups.

There is always an informal piece of the Constitution. The informal part is the sense of “how we do it around here.”

Handles the user can invest in. A way for there to be members in good standing, some way in which good works get recognized. The penalty for switching doesn’t have to be total, but if I change my handle, I have to lose some kind of reputation or some kind of context.

Here are points we line up with, but not as much. Note how the first one overlaps with the last one above—that's because HN straddles this issue somewhat [3]:

I need to associate who’s saying something to me now with previous conversations. Weak pseudonymity doesn’t work well.

You need some barriers to participation, however small. You have to have some cost to either join or participate, if not at the lowest level, then at higher levels. There needs to be some kind of segmentation of capabilities.

I found only one main point where HN differs significantly:

You have to find a way to spare the group from scale. The dense, interconnected pattern that drives group conversation and collaboration isn’t supportable at any large scale. Less is different—small groups of people can engage in kinds of interaction that large groups can’t.

You might think HN was a good match for this too, because we've never tried to juice it for growth, and it's a medium-sized forum by current standards. However, when Shirky says small he means "larger than a dozen but smaller than a few hundred". He recommends finding ways to factor larger groups into smaller ones so that richer interactions can happen. This is something we explicitly do not do, and since HN has millions of readers and tens of thousands of commenters, it's massive by the standard he was writing about.

This is the non-siloed property of HN [4]. It's probably the single most influential aspects of the site's design, and it has many counterintuitive consequences, which I've been writing about lately [5].

[1] I asked this in 2016: https://news.ycombinator.com/item?id=12208054

[2] Several of these quotes are spliced from multiple passages.

[3] https://hn.algolia.com/?query=by:dang%20community%20identity...

[4] https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

[5] https://news.ycombinator.com/item?id=23716395 and https://news.ycombinator.com/item?id=23308098

I feel that HN somewhat does the 'factoring into smaller groups' by having each topic's conversation be pretty clearly separated from each other one's, and quickly expiring off the frontpage. So the group of people who actually have a conversation in any given comment page is much smaller than tens-of-thousands.

During the onset of COVID there were a lot of people jumping to Amazon's defense of their warehouses. With seemingly deep information about changes there.

Could have been just amazon employees, might not have been.

I feel similarly about how the discussion goes when people slam Apple.

Are these infiltrators? Or do we have a lot of Amazon and Apple employees? Do developers usually care that much to protect their company online? (I've never been in that situation so I can't say).

These perceptions are extremely easy to imagine based on seeming patterns that usually boil down to nothing when we look at the data. That's why the site guidelines ask people not to post insinuations of infiltration, astroturfing, shillage, etc., at least if there's nothing concrete to go on.



It's natural for people to post about what they know about from work, and they're naturally biased in favor of their employer, so that's a thing. Even more of a thing is just that each $bigco has a fanbase and, let's call it, a foebase, and the two of them go at it in every $bigco-related thread.

It could be shareholders, who could also be employees.

I couldn’t identify any other insights in the essay that HN does not line up well.

However, I could argue that HN lines up well with one pattern mentioned in the essay: Identification and vilification of external enemies. For example, HNers typically vilify external groups and people who act or actually are ignorant about encryption and privacy issues -- such as, say, the politicians and industry associations who drafted and promoted the EARN IT act. I don't think this particular example is a negative for HN (i.e., I think EARN IT is horrendous), but I do think the similarity is worth mentioning. I hope HN seeks to minimize vilification in general.

Also, I wonder if HN is susceptible to a pattern that (IIRC) is not mentioned or discussed at all in the essay: Influential communities with large audiences attract ill-intentioned members who infiltrate the community only to influence the larger audience. I hope HN has safeguards against this sort of behavior, though.

The value of a group is based on the quality of its individuals.

Unfortunately it's not that simple. High-quality individuals frequently get into low-quality interactions, especially online.

People with high knowledge about math don't get into low-quality interactions about math, no matter the medium. They'll find a way to make it informative. If an interaction is not informative, it's always because one or both participants don't know their stuff.

Moderation is important, but it solves a different problem: letting people who don't know their stuff also participate and benefit without destroying everything.

IANA expert in social interactions but...

To me it seems like people with a high-quality knowledge can definitely still get into low-quality interactions, even when it's about a subject they know inside and out.

Lets say we have two hypothetical mathematicians who both know more math than the average person. They are both _very_ proud about the amount of math they know. They happen to bump into each other in a discussion about $INTERESTING_THEORY (I don't know enough math to know what this would be). They each pick opposing sides in this discussion, and each become more and more nonplussed with the others opinion, eventually derailing the conversation and devolving into a flame war.

Maybe the average person is nicer/more focused on maintaining a useful conversation than I think, but I have observed this pattern to many times to think it doesn't show up. That being said, I haven't seen it very often on HN, so there's evidence for your point still.

What's the basis for this claim? Perhaps the individuals you're thinking of are not as high quality as you think.

There is a good chance that it's exactly the opposite.

Our behavior is determined by the environment, and there is not element in the environment more important that our peers.

If you take a Roman and a barbarian an exchange them, you get a new barbarian and a new Roman (because how could they survive otherwise?). And, I suspect, if you exchange the Romans and the barbarians slowly enough you will have the previous Romans behaving like barbarians and the previous barbarians like Romans.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact