“Given time, [a] system will reach a state of thermo dynamic equilibrium, at which point the entropy will have a maximum value and the deviation of free energy will be greater than or equal to zero.”
-Wikipedia’s entry for second law of thermodynamics
This week’s musings were triggered by two things: A discussion on a tech-interest mailing list I belong to, and research I did for my Network World Web Applications newsletter about a company that aims to make reputation a common and easily determined quality.
The list discussion started with a comment about how many of us use Google to cure CRS (can’t remember, er, stuff) Syndrome, then turned to Wikipedia (described as “a giant anonymous Internet bun-fight”) and how people use it.
Some observed that the open editing of Wikipedia, with all of the ongoing partisan arguments, disputes over objectivity, and unintentional errors (the term “benign demolition” was used), is increasing the entropy of Wikipedia’s content. Put another way, the information content is tending towards zero.
Of course, you have to be careful of Wikipedia content, just as you have to question any data source you might reference. An interesting study by the science journal Nature last March compared the number of factual errors in Wikipedia to the number of such errors in the Encyclopedia Britannica and found that in 50 selected articles Britannica had 123 errors while Wikipedia had 162.
Assuming it survives over the long term, as Wikipedia grows and becomes more diverse, it eventually will eclipse every other reference source for sheer scope and depth. That said, Google always will have the edge on new stuff; the difference between Google and Wikipedia will be between looking for data and looking for information. (Remember Peter Drucker’s definition: “Information is data endowed with relevance and purpose.”)
So, as Wikipedia grows and gets an improved management infrastructure, the tension among its accuracy, range of topics and timeliness of articles will grow. It’s a variation of the Gibbs Rule of Multiple Constraints: Given three codependent constraints, you can only control two.
In response to a glaring inaccuracy that slipped between the cracks last year, Wikipedia stopped allowing anonymous postings. But the accuracy problem remains. Just because no one has complained doesn’t mean it is accurate, and that means the entropy can increase.
What is needed is a public reputation system. This would let readers see what kind of public credibility authors have – then they’ll know how confident they can be of the accuracy of the author’s postings.
It is not only in the Wikipedia world that we need reputation checking. For all sorts of online transactions, from purchasing to receiving e-mail, reputation checking would streamline and enable a whole new way of interacting.
For example, if I don’t know you and you want to send me e-mail, a public reputation system would let me apply very useful criteria about whether I should accept or reject your message.
If a message claims to be a personal, noncommercial message from someone who has the blessing of 96% of his 1,024 references (the other 4% are ex-girlfriends), and 32% rate him as trustworthy for personal transactions, and 16% trust him for business transactions, that might be enough for me to accept the message.
A real and useful reputation system adds a social dimension to other identity systems, such as digital certificates. So, we will know you are who you claim to be, but if you are a dog, we and everyone else on the Internet will know that, too.
Decrease my entropy. Write to backspin@gibbs.com or on Gibbsblog.
