wu :: forums
        (http://www.ocf.berkeley.edu/~wwu/cgi-bin/yabb/YaBB.cgi)
      

        riddles >> general problem-solving / chatting / whatever >> why cheating is good for human computation games
        
(Message started by: amichail on Jul 2nd, 2007, 7:52pm)
Title: why cheating is good for human computation games
Post by amichail on Jul 2nd, 2007, 7:52pm
After reading about Luis' new game for identifying beautiful images, it occurred to me that cheating would enhance human computation games.

The reason being is that cheating adds a new dimension to these games, namely strategy.

So in the game for identifying beautiful images, you can try to get into your partner's head based on the results of previous image pairs.  This adds a new dimension to the game that is missing in the ESP Game.

So instead of combating cheating, why not encourage it explicitly by providing features to help players cheat?  For example, in this beautiful image game, one can show the results of past image pairs while the game is in progress.

Even in the presence of such cheating, one may still get lots of useful data for subjective problems.  In fact, one may get more as the games would have a new dimension to them.
Title: Re: why cheating is good for human computation gam
Post by towr on Jul 3rd, 2007, 3:41am
Isn't it the point to eliminate dimensions, rather than add them?

Who's Luis anyway?
Title: Re: why cheating is good for human computation gam
Post by amichail on Jul 3rd, 2007, 6:58am

on 07/03/07 at 03:41:42, towr wrote:
Isn't it the point to eliminate dimensions, rather than add them?

Who's Luis anyway?

Could you elaborate?

See this article:

http://www.wired.com/techbiz/it/magazine/15-07/ff_humancomp
Title: Re: why cheating is good for human computation gam
Post by Three Hands on Jul 3rd, 2007, 7:34am
I imagine a lot of players already start looking for common themes in the answers of others, in order to synch up and so obtain a higher score, or just show more agreement with the other person, and so validity for their assertions. Whether you consider it "cheating" or not, I think it is more a kind of meta-gaming - playing a game beyond just the game of "matching" the other person's answers.

In any case, it struck me as being very similar to Wittgenstein's "Language Games" - terms gain meanings through agreement within certain circles, rather than through any private understanding - but in different kinds of language. While it follows some sound thinking, I do still have some reservations of just how perfect the agreements are, and how much can be concluded form the results.
Title: Re: why cheating is good for human computation gam
Post by towr on Jul 3rd, 2007, 10:33am

on 07/03/07 at 06:58:27, amichail wrote:
Could you elaborate?
Well, it seems to me that in a classification problem you want to find the relevant features (implictly or explicitly) of what you're classifying, not introduce new components that are irrelevant.
The question is somewhat: do you want a classification of your images, or do you want people communicate through images ignoring the task you set them.
Title: Re: why cheating is good for human computation gam
Post by amichail on Jul 3rd, 2007, 10:35am

on 07/03/07 at 10:33:37, towr wrote:
Well, it seems to me that in a classification problem you want to find the relevant features (implictly or explicitly) of what you're classifying, not introduce new components that are irrelevant.
The question is somewhat: do you want a classification of your images, or do you want people communicate through images ignoring the task you set them.

It might be that the loss of precision is worth it because a lot more people find the game fun with cheating added/encouraged.  So you get quite a lot more data.
Title: Re: why cheating is good for human computation gam
Post by towr on Jul 3rd, 2007, 11:36am

on 07/03/07 at 10:35:53, amichail wrote:
It might be that the loss of precision is worth it because a lot more people find the game fun with cheating added/encouraged.  So you get quite a lot more data.
It is not my experience in playing online games that they get more popular when people cheat; quite the contrary in fact.

If it needs added fun, it should be fun. While preferably not obfuscating the information you're tryign to extract.
Title: Re: why cheating is good for human computation gam
Post by amichail on Jul 3rd, 2007, 12:03pm

on 07/03/07 at 11:36:16, towr wrote:
It is not my experience in playing online games that they get more popular when people cheat; quite the contrary in fact.

If it needs added fun, it should be fun. While preferably not obfuscating the information you're tryign to extract.

It's not cheating if it is part of the game.  

From a player's perspective, the information being extracted is irrelevant.  They are not playing the game to help you.

The question here is whether this variation I proposed is worth exploring.  It seems like it would be fun to get into other people's heads.
Title: Re: why cheating is good for human computation gam
Post by towr on Jul 4th, 2007, 1:35am

on 07/03/07 at 12:03:15, amichail wrote:
It's not cheating if it is part of the game.  
Then you named this thread badly.
However, I would say it is cheating if the players consider it cheating. If the game nevertheless allows it, it is unbalanced or flawed.


Quote:
From a player's perspective, the information being extracted is irrelevant.  They are not playing the game to help you.
They are playing the game to help you, they just might not know it. The game should be designed such that they want to act in the game in such a way that overall they help you the most (as a group).


Quote:
The question here is whether this variation I proposed is worth exploring.  It seems like it would be fun to get into other people's heads.
Well, there's only really one way to find out; and that's by exploring it.


Powered by YaBB 1 Gold - SP 1.4!
Forum software copyright © 2000-2004 Yet another Bulletin Board