• 03May

    Background: Linden Lab has announced that they are planning to implement a system where search results can be flagged as mature, prohibited, spam, or worthy of being showcased.

    I’m quite glad to see that some of the Lindens have started to give some heads up about their plans. The “surprise announcements” on the blog come off as arrogant and aloof (“Feedback? We don’t need your stinking feedback!”), and the suddenness of the announcements also triggers an instinctive opposition to change, the gut feeling people get when suddenly presented with something they don’t have the time or information to understand.

    So, these advance notices are a step in the right direction. However — and this makes me quite sad — very few of the pre-announced systems undergo any significant changes before they are rolled out, even in the face of legitimate criticism (setting aside the usual wall of bitching and moaning).

    You can read my jocular comment describing of a way to game the system, abusing it to harrass specific targets. Of course, the potential for abuse isn’t limited to harrassing others. You could also game the system by paying newbies to flag your store listing as showcase-worthy, analogous of camping under the current traffic/Popular Places system.

    The obvious problem here is that there is no disincentive for mis-flagging search listings, nor an incentive for accurately flagging search listings. (I’m not talking about monetary incentives, e.g. L$10 for each accurately-flagged listing, which would likely lead to more gaming of the system.)

    As a way to encourage accurate flagging and discourage inaccurate flagging, one might propose that each account had a “relevancy” score, which would change in small amounts (e.g. 1% of the current score) for each accurately or inaccurately filed listing.

    The accuracy of the listing would be determined by the Lindens who eventually review the flagged listings. That is, when the listing comes up for review due to a large number of flaggings, the Linden or Lindens reviewing it would decide whether it was actually mature/prohibited/etc. If they decided that it was, everyone who flagged it as such would gain some relevancy; if they decided it wasn’t, everyone who flagged it would lose some relevancy.

    The relevancy score would affect how much weight is given to your flagging; if you perpetually abuse the system, eventually your flagging would count for very little. On the other hand, if you consistently make accurate flaggings, your opinion would become more relevant, and weighted highly.

    (Having a system like this certainly isn’t a new idea. If I recall correctly, Slashdot employs a similar system for its comments, for better or worse. I’m sure other, less damning, examples exist. Can you tell I’m not a Slashdot fan?)

    I’m not sure whether the relevancy score would be best as completely hidden, shown only to that avatar, or made public (e.g. in the profile). On the one hand, if it were made public, it could be a “badge of honor” (or a mark of shame, for low scores). Honest Residents could pride themselves on their high score, and feel like they were being recognized for their dutiful service. Relevancy score might also be a way to judge the character of a person you don’t know, although that’s not necessarily a point in favor of the system.

    I’m hesitant here, because a public relevancy score starts to stray into the territory of the old Rating system (pay a small fee to rate an avatar on their appearance, building skill, helpfulness, etc.), which was eventually removed. One important difference here is that the relevancy score would not be determined by other Residents, only as the result of the accuracy of your search flaggings. You could not pay another Resident to increase your relevancy score, and your relevancy score would not be a “popularity contest”.

    Another factor to consider in whether relevancy is made public, is the potential of gaming the system, although I’m not sure off-hand what someone would gain by gaming it. I suppose you could boost your relevancy rating by accurately flagging listings, and then capitalize on your high relevancy score (“Will flag showcase 4 lindenz”). I could see farms of bots being orchestrated to flag listings at their owner’s command.

    But, to keep your relevancy high, you’d have to do at least as much good as harm; and if inaccurate flaggings decreased your score by more than an accurate flagging increased it, you’d have to do more good than harm, so even gaming the system would result in a net increase in accurate flaggings.

    So, as you can see, this is a complex and nuanced issue. I doubt there will be one solution that is both effective and immune to abuse, but some solutions are better than others. I hope Linden Lab will take the opportunity to review their plans in light of this feedback, and make some changes to eliminate at least the most obvious problems. (But I’m not holding my breath.)

    Posted by Jacek Antonelli @ 4:21 pm

One Response