A recent Jakob Nielsen Alertbox post on Participation Inequality: Encouraging More Users to Contribute (October 9, 2006) led me to an odd and unexpected smile of recognition. It seems that User Generated Content (UGC) sites like YouTube or Wikipedia or my favourite web design forums have something in common with the Parent teacher Association at my local school, the local football team, or other such real-life volunteer efforts.
Nielsen's article on participation inequality says that in the sites he has studied, a few contributors (maybe 1% or less of the membership) do nearly all the work, with a somewhat larger group (maybe 10%) contributing sometimes and the remaining (say 90% or more) not contributing very much. Mathematically, this distribution of effort is what is called a Zipf curve (of which more later). My smile of recognition is that when I have been involved in real-life volunteer groups, there is a tendency for me to wind up on the committee, and then typically we spend some time musing about exactly this pattern - how great things would be if we could only find a way of energizing the least energetic members of the community to do more! Talking to other groups, it always seemed that there was the same problem - whether the group was sporting, religious, charitable, political, trades union or whatever. Shortly after huffing about this, I got my comeuppance and an insight when I moved onto more demanding assignments at work and had two small children, and found that there was much less left in the tank for volunteer work than before life had these complications.
Therein, I think lie two reasons for the uneven contribution to user generated content sites - first I think that some people are contributors by personality; and second, most folks have other priorities. Not much can be done about this. Nielsen's article has some ideas about whether this skewed participation is a bad thing and what to try to level things up, and I have some more.
Clearly the behaviour of the heavyweight participants is a key thing - you want to be blessed with a few key contributors who are knowledgeable, write well enough, and are polite and tolerant of occasional or new contributors (including those who ask for quite basic information). This is, I think what makes my favourite formus my favourite ones. If you have had the bad luck to be kidnapped by a special interest group, or a clique that excludes others, then you are in trouble. Moderation of the group is the traditional way of preventing bad behaviour.
Other barriers to entry are also worth examining. Many user generated content operations require you to go through a sequence of steps e.g. to become a member in some way. Some potential contributors will fall at each step in the process - either they can't work out what to do next (a usability problem), can't be bothered to do it (a problem with the product or its marketing) or a mixture of both. You can draw a funnel-shaped chart of number of customers you see at each stage. So E-marketers talk of a "marketing funnel" (or "customer life cycle funnel" or "sales funnel", and many of the software packages that analyse web statistics can draw these funnels for you). The diagram shows a possible funnel for joining a forum - at each stage , some potential customers fall away, until most potential contributors have given up before they become heavyweight presences on the site. Clearly it makes sense to do what you can to make it as easy as possible for people to proceed down the funnel. However, this will not be a cure-all: Wikipedia has one of the easiest funnels - you can just start making changes to articles without any login etc. and still only a tiny proportion of visitors actually do any editing or writing (according to Nielsen's article). This ought to leave you with the potential that more people can contribute as soon as they want, and an inequality based on what people can be bothered to do - probably one of the less harmful inequalities in a world that has many.
The Zipf curve turns up elsewhere on the web. For example, it can turn up in:
- The distribution of traffic between different sites on the web as a whole (something that allows you to estimate your competitors' web traffic - or your competitors)
- The distribution of referred traffic to your site (so that a few sites send nearly all of the traffic to your site).
Should you wish to make a closer acquaintance of the Zipf curve, Nielsen has a friendly introduction to the Zipf curve and Wikipedia has a more mathematical article about Zipf's law.The Zipf curve is also one of the mathematical distributions that gives a "long tail" - with the business consequences originally discussed in "The Long Tail" ,an article in Wired by Chris Andersen (2004); discussed further in this Wikipedia article on the Long Tail, and of course the subject of a book (which I must get around to reading....) and Chris Andersen's blog . [The basic argument - the Internet has reduced the costs of being very specialist, as it is now easier and cheaper to sell to unusual needs and tastes. For example, Amazon can afford to supply very obscure books, but your local bookstore cannot for lack of shelf space].