Memetracking is a HARD Problem

Videomeme has a great post on the difficult problem of (specifically) video memetracking:

YouTube’s ranking system is pretty broken right now. None of the ranking metrics – views, ratings, comments – are particularly good at helping you discover new content that you’ll like. And, as YouTube’s star has risen, it’s become increasing susceptible to gaming, ranging from cheap and cheerful spambots inflating view figures to sophisticated, paid-for marketing tactics like those used to elevate the Rocky Balboa trailer to attention. I’ve spoken to plenty of people whose trust has been lost as a result and who think, rightly or wrongly, that these days, you just buy your way into YouTube’s feature lists.

Yeah… memetracking is hard which is why I think it’s important for dedicated sites to provide this functionality.

This is one of the point’s that Arrington makes about the new video memetrackers, which base their results on blog buzz; that they should be less susceptible to voting fraud. To some extent, this is right. It’s harder to set up a fake blog than a fake user account, though only just. And the meme trackers mitigate the risk of fake blogs to the extent that they upweight more authoritative blogs, though at the cost of increased methodological opacity.

The spam blog thing is a major issue which is why it’s important to not index them. The trick is to have decent algorithms which prevent you from indexing spam in the first place. Unfortunately spam is becoming a big problem so I don’t see any major services shipping in the future which don’t take spam into consideration.

The odd thing is that Digg is a spam and gaming magnet. Expect people to use systems like Twitter in the future to continue to game Digg.



%d bloggers like this: