Mitch Kapor turned me on the other day to a well known essay by R.H. Coase (there is a link to it here), which was written in 1937 about why companies should exist at all. If market forces (free people setting their own prices for goods and services) generally produce almost perfectly efficient results, why do we see these unusual collections of people (Linden Lab for example) all agreeing to work together without any of the usual haggling, agreeing instead to a fixed wage and percentage of the company. It is a fascinating question to re-examine in our age of hyper-communication, and one that is put in a particular spotlight by things like the feature voting page for Second Life that we put up today (if you are a SL resident, you can find it here)
The answer Coase gives is that for many types of group efforts, the ‘coordination costs’ of markets (haggling over price, etc) exceed the value that can be produced by the group – making it more sensible to agree to fixed terms of employment to create a product. This certainly makes sense if you imagine an early 20th century company with a complex product like an automobile manufacturer – a collective market effort would have had a hard time competing with Henry Ford! But the striking advances in communication offered by digital systems like the internet, combined with better education (most people can now do many different things) and globalization (there are very large available labor pools for a given skill) has made things different. Open source projects (in which there is no formal company) can now rival products created by large coordinated firms, and if you look at the way projects work in Second Life you can easily imagine ever more fluid environments in which coordination costs drop to almost zero.
Extending this idea of ‘coordination costs’ to examining the boundary between company and customer is what leads me to the discussion of feature voting. Historically, the information one could get from one’s customers was very weak. You could mail out surveys or canvas or maybe (in a really sexy modern view) even make phone calls! But if you were talking about a clothing line or a new car, you were really unlikely to get very much useful information. You couldn’t really ‘see’ what was going on – the cost of getting a lot of data across that boundary between you and your consumers was too high. You had to rely instead on the ‘experts’ you had hired to guess about what people thought about your product. But in Second Life, you can just GO THERE. It takes only seconds to be discussing the impact of a change or asking people what is going on. The cost of very deeply involving people in SL in designing SL is very small – so small that I think it will change things in a profound way.
By putting up a page where thousands of people can cast a fixed number of votes to prioritize (or modify) a fairly specific work list of features and changes for upcoming versions of Second Life, we are further blurring the boundaries between the ‘company’ of Linden Lab and the residents of Second Life. We are asking for help (and I suspect comitting ourselves substantially to what we hear) in what is generally a very private and hallowed process – the setting of development priorities.
A really common behavior of companies is to believe internally that they can better estimate the priorities of their customers than their customers can. I doubt that in most cases this is true, if information systems are used wisely to help organize the process. There is an interesting body of knowledge suggesting that for many many cases, the aggregated opinion of a diverse audience can outperform any single human decision maker. A great book summarizing these studies is ‘The Wisdom of Crowds‘, by James Surowiecki.
It is certainly the case that innovation is discontinuous and often the brainchild of a single person, but valuing the potential impact of even something fairly hard to imagine will be (I bet) better predicted by a group of diverse users. Let’s wait and see – as of this writing, the two “must-do” features that we had internally identified as top priorities in our next large feature set are also receiving the highest votes. We are still early in the process (a few hours) with only a hundred or so voters, but still it is interesting to see convergence between our gut and the larger community. Where I think the voting will be really helpful to us is assigning relative priority among smaller, less obviously valuable work items.
Very exciting to wait and watch the numbers come in!