Hacker News Reader

Comment by fredley | Hacker News Reader

Although is not anything like at the scale of YouTube/Twitter, the model they have for moderation (volunteers, supported directly by staff, BYO bots, as well as company-owned bots for grunt work like spam removal) works amazingly well. Much much better than any other site with the same volume of new content.

It does make me wonder why companies like Google, Facebook and Twitter, which have much much greater resource to throw at the problem, still struggle with content moderation. Just hire good people, give them the right tools, and engage the community.

Disclosure: I am a volunteer moderator on a Stack Exchange site.

View on HN

I too was a moderator years ago for a Stack Exchange site. I was always amazed by the sheer amount of tooling that I had access to. I could instantly tell when there were sock-puppet accounts, voting rings, auto-flagged low-value postings, etc.

One thing that Stack Exchange does very differently from other sites is that they emphasize professionalism. Since Facebook and YouTube allow memes and off-topic material, there is an expectation that low-value postings will be tolerated by the community. And that's exactly what those places get.

SE sites prohibit garbage. On the Quant Finance site, for example, we required people who post there to actually work in the industry. That metric made it very easy to distinguish ourselves from other QF forums; we didn't have "How do I become a quant?" or "What programming language should I use?" for every other question.

Another community that has some bar of professionalism is Hacker News. There's no requirement that people be entrepreneurs or programmers here, but there is an expectation that anything posted must valuable and high quality.

The thing I see about StackOverflow's volunteer moderation is that unlike Facebook, Tweeter, etc.:

  + StackOverflow has clear criteria for expertise.
  + Reputation is based on being helpful to other people.
  + "How do I make baby" is off topic.
  + Content has long term value.

Unlike Facebook, Tweeter, etc. self promotion is bounded because of the purpose of the site. Imagine if you will, a Facebook where reputation is based on the quality of a person's Likes and Friendships as determined by community voting. What gets measured gets optimized. Facebook's user facing metrics don't encourage skills transferable to moderation.

To me, a big difference with StackOverflow is that it is a bounded community from which people do not graduate. Facebook was a bounded community until it expanded beyond people with .edu emails. When Zuckerberg's Harvard cohort graduated and started producing grand-children, the platform had to expand to grand-parents to keep people on the platform. Now that those children need messaging, the platform is expanding to keep people there.

StackOverflow has expanded laterally by adding siloed communities around expertise as the basis for reputation. While its systems have morphed and evolved, there has been a set of underlying ideas that prevent ad hoc change.

Mark Zuckerberg is 33; his classmates are not having grandchildren yet. Facebook did not have to do anything to 'expand to grandparents'; grandparents could create accounts on Facebook as soon as anyone with a general e-mail address could.

His classmates' children are somebody's grandchildren. That's why the platform expanded...emailing photos to grandparents was a pain point that the expansion of the user base solved. It was the killer app that got old people onto the platform.

The number of FB comments must be multiple orders of magnitude larger than SO comments; the average user's education and comment quality much lower, so easier said than done.

meeh number of comments to process do not really make any difference for bots.

Is it reasonable and scalable to expect people to volunteer their time and expertise largely for the benefit of for-profit businesses? (Disregarding the benefits to the communities themselves.)

Due to network effects and inertia I imagine building communities in walled gardens only makes it more difficult to leave if the platform changes their terms. Patreon's recent fee change being one example.

Structure and Purpose matters.

For Stack Overflow in particular, and some Stack Exchange sites (usually the more successful ones) there is a clear purpose and we can all judge whether a Question, an Answer, a Comment is useful towards that purpose. Without a clear Purpose everything becomes a popularity contest. There is also a useful Structure, the Q&A itself is structure, and there are tags putting related material near by.

Examples:

1. WowHead is (was? I don't play any more) a site for the MMO World of Warcraft, built by scraping all the data that powers the game and then with the ability for its members to comment on anything from the resulting data set.

The scraped data provides Structure, linking related things together and putting some useful bare-bones information in for everything. So you would visit Wowhead even without expecting to read any human contributions that were relevant, because the data on its own might be helpful. The game could track whether an item had significantly changed since you wrote a comment, and automatically differentiated things with similar names based on their underlying ID.

The game provides Purpose. I write a two paragraph description of how to perform a tricky but rewarding quest correctly, someone else makes a pun about the name of the quest, I get up-voted more because it's more helpful.

2. community.letsencrypt.org uses Discourse in an effective way

c.l.o is weaker on Structure, because it's very often being used by inexpert users seeking help, the same sort of cognitive shortcomings that can cause someone to not read and understand an error message can cause them to decide "Issuance Tech" is the right category for their question about how to configure Apache.

But the Purpose of c.l.o is pretty strong, most new topics on c.l.o start with a user who has a clear goal ("I want my web site to work in Firefox") and the community can help reach that goal without getting too distracted ("Why does your site use Comic Sans?") because the central topic, Let's Encrypt, is very clear and not a huge political minefield.

3. Youtube comments for _very specific channels_

Sometimes a Youtube channel ends up bringing together people with a very clear purpose. The only "Structure" here is provided by the videos themselves, comments about Episode 3 belong with the Episode 3 video, duh. But Purpose can be brought by a community.

Minecraft players focused on challenge gameplay (e.g. people building Agrarian Skies, or Project Ozone 2) attract a community of other people playing the same or similar challenges and the comments are naturally attracted to suggestions. "Use a Loot recycler, then merge all the lesser bags, nothing less than Epic is useful to you".

Reply on news.ycombinator.com