The tragedy in the four bombings in London yesterday makes me worry for mankind. My heart-felt sympathies to all affected by this.
Community and social computing
rawn 100000R0P5 684 Views
rawn 100000R0P5 627 Views
As our developerWorks Community Editor, I'm faced with issues which are part talent management and part information management. Aside from all the regular issues of focusing effort and developing content to build a stronger community, I'm facing a happy problem of having so many enthusiatic contributors: what happens when there's too much to read?
It isn't hard to see this problem is evident; just try googling on a topic and see how many hits you get. Would you really want to read anything more than 100 entries?
In talking with James Snell and others a few weeks ago regarding corporate blogs, wikis and other such tools, we agreed that there are typically two ways of looking at who should contribute. We both agree everyone should be able to do so; but when you start running into large numbers, how do you organize this information?
First, there's the top-down approach where you pick the topics and find people to contribute to each category. This is quite commonly used; dW does this in our many zones.
Then there's also a bottom-up approach where you want topic experts to self-emerge from a population.
This second approach is harder to figure out, but the answer may not be as complicated, especially when you have a large population and numbers on your side.
I think the idea lies in building a Reputation network, modeled on trust. Each individual essentially would rate how a particular interaction with a potential expert turned out (in two types of trust, which I'll discuss later). You can then average all the trust ratings per candidate to determine their reputation. The number of interactions combined with this average trust rating describes their reputation.
Those with higher reputations automatically emerge as "experts" in whatever categories they want to focus on. The process is democratic and self-emergent within the community. For that matter it also automatically helps to define topic areas that people are interested in.
That's the basic principle. I'm quite sure there are many ways of looking at this, so I'd be curious to hear counterpoints.
rawn 100000R0P5 601 Views
The basis of community in my mind is in the interaction between people. The tools, mechanisms, processes, and communications we use are the outward expression of these interactions. A community really lies on a subliminal level in the form of the trust we develop directly with other members of the community.
For the most part, this trust is an implicit factor that we associate mentally with another entity (a person or a group). However, to externalize this factor into a meaningful measurement that many others can relate to can only be a rough approximation. Beyond that, to normalize that measurement across anyone, anywhere reduces it to a lowest common denominator of trust.
This is still important, however. By indicating how you would rate trust between yourself and another person, and by sharing this rating with others, you describe the relationship network around you. Metcalfe's law states that the usefulness, or utility, of a network equals approximately the square of the number of users of the system. This is useful in understanding the context in interactions between people. Context after all makes the difference between princes and purloiners.
You can build a complex directed graph by combining the various relationship networks of all the individuals in a community. However, that just becomes increasingly complex with each person you add. From Metcalfe's law, the possible number of connections is squared, making huge numbers of connections.
A simpler way is to create an aggregate trust rating per person across all the ratings they receive from others. Each entity thus has a single rating factor.
Actually, two rating factors really. You need to ask after the result of any interaction:
1. Helpfulness - Was this knowledge source helpful? Do you trust that they were providing the information in good faith and for your benefit?
2. Usefulness - Was the information provided by the knowledge source useful to you? Did the information help? Can you put it to use?
A person can be a source of useful information but be too busy to help people get that info. On the other hand, a person can truly want to help but not really know much about a subject.
How do you apply these ratings? Keep it simple: rate 1 through 5, lowest to highest, on each of these factors per interaction you have with someone.
What you really end up getting over time, is a measure of the reputation of an entity as well as the number of people they interact with. By measuring recurring interactions with the same knowledge source, you can also measure how loyal people are to that source.