|
|
|
|
|
|
by poronski
5941 days ago
|
|
|
> ...which would address the same problem. I don't think it would. Points determine the rank, so people would still be inclined to upvote their favourite users' comments if they see them engaged in a debate with someone else. Anonymizing can be implemented by simply randomizing usernames, but in a way that keeps users uniquely identified in a scope of a single discussion. For example (a) randomly generate discussion key (b) hash(key + username) (c) use the hash to pick a word from the dictionary and (d) use this word as a username substitute. For extra fun use the list of registered HN usernames as a dictionary :) |
|
|
I think you underestimate the ability of people to identify others based on what English teachers call their "voice". I mean, here's a test you can do in forty-five seconds in Ruby or your language of choice: grab the RSS feeds from searchyc.com for pg and any other user you want to play the game with, have your script randomly spit out a recent post from one of the two lists, see if you can guess who said it. This turns out to be really, really easy, particularly for good writers who one spends a lot of time reading. (You can do this with Bayesian classification, too. In the "Pick who wrote this out of two possibilities for which there are many samples" case it's like taking candy from a baby in terms of classification tasks.)
Also, do you think you or anyone else is going to have much trouble telling who baby_rutabega_spoon is when he talks about his bingo business or ruthless_carrot_fork when he gets exasperated about password hashing?