Further thoughts on GreenScanner and info collaboration sites

A couple of days ago I posted a blog entry about GreenScanner and our very own Ken Fasimpur of the IT Borderlands blog added the comment that: “the quality and quantity of the data would seem a major issue. On a whim, I just added an environmental review of a CD I’m listening to. That’s now public information. If I have a grudge against Omni Consumer Products, can I report that their merchandise is produced by child labor and is responsible for the elimination of 1000 acres of rainforest per day? If I lie, will I be corrected?” Ken raises an interesting point which is ignored or at best overlooked in many public information collaboration sites: How do you ensure truth and accuracy?

The first part of a solution is distinguishing between the comments from registered (with double opt-in) users and those from non-registered users. A major discrepency between the ratings of registered and non-registered users would be flagged as suspicious. Secondly, vendor involvement. The system should warn vendors when they get a very negative response to a product and allow them to defend themselves with their comments being very obviously displayed. Thirdly, track the responses of each registered user looking for patterns of anomolous negative feedback. This is a complex task but my guess is that it would share some of the same logic as credit card fraud detection. I’m sure I’m missing some tricks and techniques but a system engineered in this way would definitely make “gaming” the system much harder. The biggest problem is whether people using the site would actually weigh the various comments instead of just accepting the majority opinion which makes me wonder if this kind of data is suitable for use ‘on the go.” Perhaps an editorial team is required to interpret the comments or a more structure comment entry system is required to “normalize” the context of comments. BTW, I didn’t, in my previous posting on GreenScanner bother to note that the site’s design is pretty rough — sort of an alpha design. What intrigued me was the idea. And it just occurred to me: What about making the GreenScanner database portable? Wireless connections from inside many warehouse-type structures tend to be pretty poor or non-existent. Should be easy enough to fit it into a couple of gigabytes …