Crowdsourcing is awesome. Take reCAPTCHA -- not only does it protect you from robots, but it nearly effortlessly accomplishes an otherwise extremely tedious task. For more complicated content, like encyclopedias, I really want to see an implementation which combines wiki-like submission with ratings (a la eBay seller rankings, digg, amazon products, PageRank). Throw in some kind of accreditation system, whereby legitimate "experts" assign more points, and I believe it would go a long way towards making things like wikipedia more trustworthy.
It would need several pieces: content system, rating system, and metric/analysis system.
@ Content would handle submission / display. This should be simple for most things -- text articles (wikipedia), photos (geotagging, reference guides), numbers (data collection for environmental monitoring).
@ Rating -- other people / users "vote" the item up or down, and points are assigned to the submission's score based on users' "credibility" rankings. The rankings themselves would function like Google's PageRank, and be determined (in the Metric system) by how useful the user's other submitted content has been scored, or they get a bonus factor if they are a registered professional in an appropriately related field (like a physicist would count more on physics content, but not necessarily on interior design content). This may also be subject to certain predefined limitations (on requested content), so that data collection for air quality near Crabtree Valley Mall is only accepted from people who are registered for that area. Or it could also involve an invitational aspect, so that only certain areas (like profession, location) are "invited" to comment. I do think that for some applications of professional information, layman opinion should be considered just to shake things up; possibly a larger gap in 'credibility' rating would be factored in, so that a professional opinion is worth 100 random opinions.
@ Metric or Analysis system: used to determine how the ratings are calculated. This would probably need to function differently for different kinds of content. For example, with wikipedia articles, it would function mainly on user voting (subject to professional moderation? oy), but for "tedium-reducing" work like geotagging or reCAPTCHA, it would have to be based on some kind of quality measurements -- "usefulness" or "appropriateness". I'm not sure which would be more difficult -- the second requires programming and data analysis, while the first is affected by social factors (who's paying attention?).
I know many elements of this have already been implemented, but I don't think they've been combined into a standardized whole. Has it? Can this even be done? I'm open to your comments...
2008-11-22
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment