The rise of social governance

I just read an interesting article by Kevin Kelly. It’s called “The Bottom is Not Enough” and its premise is basically that a system (he uses Wikipedia as an example) needs more than just user-generated content, more than un-shepherded crowdsourcing. It also needs a small amount of editorial control from experts. This not only improves the quality of the content, but it also helps the system grow much faster due to enhanced overall intelligence and reputation. In practical terms, a mix of the top-down and bottom-up approaches gives content systems like Wikipedia a way to not be overrun by spammers and misinformation artists; it helps Wikipedia stay relevant and valuable as a source of reliable information. I’ve been thinking a lot about this recently and I think these concepts are part of a field I like to call “social governance”.

A specific example is keeping a comment system clean. Comments themselves are a bottom-up approach to building content. The problem is, there are spammers who want to clog your system and trolls who want to ruin the mood with their rudeness and foul mouths. It’s part of the fun of the Internet. It’s also a major pain. In most cases, it’s impractical to have editors review every comment. For sites with scale, this is just too slow and expensive. So, many sites allow users to flag comments as spam or inappropriate, etc. This is basically crowd-sourced editing, and it often doesn’t work because not enough users care about rating comments (an example is youtube.com - users don’t care because there’s no reward for helping as an editor). With the right incentive system however, such as the one used by Digg.com, it works well.

How it works is fairly straight-forward: usually, member votes are assigned a point value based on their reputation. Those vote point values are tallied and when they reach a certain threshold, the action that was voted for happens.

The trick is, you have to figure out the reputation of the members doing the voting. You can only judge that if you know something about them. You can start with their history with your site. Each site will have different criteria, different elements it can track. Sometimes you can infer a user’s good standing based on articles or comments they’ve written that others have found useful. You can determine usefulness by whether or not that chunk of content is voted up or visited often. But you can also factor in a whole host of other elements: do they visit often, are their site usage patterns consistent with a real human being, do they have customized profiles and a certain number of friends? These are the types of elements that go into making an algorithm that can help you measure a user’s reputation.

It’s interesting to me that this these types of algorithms are not a new thing. Google’s PageRank algorithm is a perfect example of a program that collects and weighs many different variables to determine the authority of an entity (in this case, a piece of content). They have the most relevant results because they have the smartest algorithm. They have the smartest algorithm because they collect the right data and then weight and relate the individual pieces of data in the right way. As with any algorithm used to gain a competitive advantage, it must evolve to stay ahead of the competition. We’re all familiar with the constant struggle betwen Google and spammers and scammers who are trying to game Google by finding a shortcut to the top of their search results.

There are many great products that rely on algorithms to rank information and determine reputation. The blog comment spam filter Akismet has saved me from most of the spam that would otherwise engulf this blog. A very small percentage of the spam makes it through, but it’s a small enough number that I have the time to moderate the comments and approve them for display or not. Digg is another great example of an algorithm-based system. They rely on votes by users to determine whether a story should make it to the home page. But as most Digg users have noticed, often stories with a relatively small number of Diggs can make it to the home page. This is because the users who did the voting are perceived by the system to have a high level of authority/reputation. Digg tracks and weights many different variables for each user in order to determine that user’s strength or authority. If that sounds like a classic role playing game, it should. The idea of using a point system to rank user strength is very much based in game theory.

What I’m interested in is taking crowd-sourced editing to the next level. This is what I call “social governance”. I am trying to understand the elements needed to design a system where users care about their reputation, the environment is tolerant, expletives are where they’re are wanted (for instance, not in public YouTube comments), and where new users can’t be bullied by more powerful users. In games, that last element is often accomplished by forbidding combat initiated by the far more powerful player. In this way, bullying can be avoided. In a comment system bullying may come in the form of an unwarranted flag.

This is where social governance needs to step in. Just as in our justice system, I believe commenters who feel they were wrongly flagged should be able to appeal the flag. This doesn’t mean that the makers of the site have to perform the judgment on this appeal. I think it’s possible to identify and recruit tolerant long-time users, and then reward them for serving as judges in these situations. You could set it up like the supreme court where a group of more than five all vote on the case, majority opinion wins. You could even have a further layer of appeal that goes to the site creators. If an appeal is won by the defendant, the system could grant that user more power, and immunity from the flaggings of the users who were in the wrong.

This social governance layer is needed to make sure users are treated fairly, and technically it wouldn’t be that hard to create. The truly hard part is designing a site where users have a reason to care about their reputation, a reason to care about helping evaluate cases appealed by other users, a reason to participate at all, etc. Digg has figured out a great reason for users to care — the more you participate on Digg, the more powerful and influential you become. The direct result is that it is easier for you to get the stories you like on Digg’s homepage. I don’t think Digg is the only site where influence could be used effectively as a reward; it could be used on lots of sites, albeit each use will likely require customization to properly support a site’s specific goals. I have some ideas on how exactly to use influence as a motivator, and on other ways to motivate users, but I’m using them in my own startup, so I have to keep them on the down-low for now.

What about the bigger picture? What if motivation and reputation were somehow standardized? We don’t have an overarching reputation system for the Internet, not yet anyway. But I think this is where we are going. What I find interesting is the idea that if you have enough data from independent sources, you can come up with a great overall reputation score for any given user. For instance, imagine if the OpenID platform had a rep system that allowed member sites to say how they determine the rep of their users. A non-biased board would review their rep system and approve their rep scores for inclusion in the open rep system. Scores from different sites could be weighted differently, but an overall picture would begin to emerge that could tell a lot about a user.

This rep system doesn’t have to be big brotherish either. If we vet the scoring systems of member sites, the only public info will be the points totals. And if a user doesn’t like where their rep is headed, they can start over and build a new one from scratch. If there are real benefits to the user for having a good Internet rep, there will be an incentive to pay attention and build a good rep. A vetted internet rep could have real world benefits like lower insurance costs and better interest rates. All of a sudden, the time you put in to your carefully cultivated blog and its niche audience could pay off in many new ways.

For site builders, the immediate benefit of a rep system would be that it would be easy to know that a new user is a real human with a certain level of authority. With the right data, you could even determine things like follow-through, trustworthiness, civility, and influence. Also, imagine if sites like YouTube could automatically filter out the trolls and spammers who far too often clog their comments. They could do this if they only allowed comments from users with an Open Rep of a certain score. Or if they weren’t willing to set a standard, at least you could tell your browser to only show you comments from users of a certain standing.

I think in the end it will have to be the user’s choice to reveal their point totals to any Open rep system. If they perceive it as beneficial for them to have a good rep score, then they will build one, just like people build out and post their resumes. If open reps become valuable in any way, there will have to be fair systems of social governance to appeal the inevitable mistaken demerits to a person’s score. I also think there will always be enough totally private people to make sure there’s no stigma for not having an open rep score, at least I hope so.

Whether or not we ever see an Internet-wide rep system, I think more and more sites will implement internal rep systems. If you then offer the right incentives to users with good rep, you can tap their brains to help make your site the kind of place you and they want it to be. A system of tolerant social governance could help keep comments clean, new content flowing in, and that new content vetted for accuracy. The right social governance system can help make an online community a vibrant place indeed.

Of course, it’s easy to identify the need for a social governance system. It’s quite another to design the system to be easy and desirable for users to understand and use.