I came across the following blog post where the author is actually talks about social media networks today. While at it, I wondered the problem might be very much relevant to communities as well.
In summary, users are given 1000 points (label the currency as you wish) when they join and they get to see feed:
Every ten minutes, the system slightly reallocates points. Think of is a tax on users with high points that get reallocated to users with fewer points. This algorithm’s goal is to eventually return users to an equilibrium state of 1,000 points each (their starting conditions).
I highly recommend for the folks at discorse to give this article a read. It has do with teh posts curation mechanism. I do not know how content is curated for users today on forums, thought this might ignite a valuable discussions.
That’s really interesting. I’d be curious to see how it would work in the real world.
I think the distributed nature of Discourse (many small communities focused on particular interests) makes it less prone than large scale social networks to the issues that are described in the article. Discourse also doesn’t use either a follower model or an algorithmic model for ordering topics in the topic lists. By default, topics are displayed in the order they are created, with the most recently updated topics at the top of the list.
Probably the closest thing to Discourse topics being ordered by a “points” system would be to view a Discourse topic list ordered by likes, for example: https://meta.discourse.org/?order=likes. Using the likes filter, scores are calculated on a per topic basis, not on the reputation of the user who created them, so it’s not a great analogy of the points system that’s discussed in the article.
I think it would be possible to use Discourse to prototype a points system similar to what’s outlined in the article. I wonder how large of a community would be required to really test it out though.
I don’t think forum communities suffer from the problems the article addresses. For instance, here are two problems the article suggest the system will solve:
Of course, none of this would be interesting if it didn’t actually address the issues with today’s social platforms that I mentioned above. So let’s take a look at each of them and why DK addresses them at their core:
Follower-Based Distribution - Unlike the traditional follower model, here distribution is based on the short term perception of the quality of a user’s content.
Accumulated Advantage - No one is able to sustain their distribution benefit for long without continuing to create high quality content.
(“DK” is the name of the fictional network.)
These two problems aren’t particularly relevant to most (all?) Discourse communities because the way ideas are distributed has nothing to do with followers or artificial systematic advantages. Instead there are two main paths:
Search engine traffic based on queries people are actively using.
Views from organically browsing the forum—latest, pinned posts, scrolling through topic lists, bookmarks, notifications, suggested/related posts, etc.
The way to “win” in the forum system is to write lots of high-quality content. It’s still possible for bad ideas to circulate, of course. But the solutions tend to be activities such as flagging or responding with better ideas. More generally, solutions can come from the culture of the forum. A community has defenses against the problems this article references that aren’t technical.
Not that I think the solutions would work particularly well for social media either. It’s not an algorithm problem, but a scaling problem. When you have a global chat system where people can (and are even encouraged to) wade into a discussion that’s happening in some other community, ordinary cultural defenses don’t apply. Systematic solutions fail too because there’s always an incentive to find the loophole. A version of Goodhart’s law applies:
All metrics of scientific evaluation are bound to be abused.
So with the decay system, there’s an incentive to continually post. One method might be to post new content. But another solution could be to recycle content. Sure there might be algorithmic protections built in, but some users will find ways to skirt the safeguards if there’s some incentive to do so. Humans are much better at cheating (and detecting cheating) than systems are at keeping everyone in line.
We also have the scoring calculation used for the Top list, though that’s also not really analogous with the OP version either - How are "Top" Topics Calculated?
local queries? these can be semantically searched locally now too.
In a Q&A setup, you definitely can game it. This is outside of any “follower” system. I remember when we had a local Stack Exchange enterprise installation at my old firm, some people got very addicted to the points system. This wasn’t “social media” in the broader sense - you could argue it was basically a forum with a Q&A driven dynamic. I’m sure you could explore this further with Discourse and the current Q&A plugin, possibly with extension.
It would be interesting to know if the Discourse Follow plugin had any influence on a large forum …
Ultimately no system, even a well-designed system, can keep people on mission. But they can be tools for scaling up a community, which is why we still need these systems.
Haha, love it. A blog post, the antithesis of a quick win on Stack Exchange, about Stack Exchange. Great post.
I used to answer questions until I was safely in the Front Page of top users, then I would back off and go back to my proper job.
Stack Exchange is a classic, though and I will remember those days of having a firm one fondly.
Funnily enough … the guy who ran it got another kick out of it - he installed it on our intranet and administered it. I remember thinking about this when I first built my own install of Discourse - this must’ve been the pleasure he felt.
It really is a great feeling to get a server up and running. It’s even better when it’s something like Discourse that other people can contribute to. One of my regrets at my last job was not insisting on having an internal Discourse instance where our team could store things we’d learned and discuss next steps.
