The hiding of absolutes, so you only see relative “if you like this person you’ll like that person” or “if you don’t like this person you won’t like that person” as someone suggested is a good idea and solves the griefers giving each other good points.

For people worried about RPing evil, there is no problem. People can RP evil and still be non-griefers and not jerks.

Then I didn’t explain it very well. It’s not like that at all.

>Say two pirates are fighting each other or competing over the same star system. They give each other negative rep. I\’e2\’80\’99ve been killed in the past by one of them and so I\’e2\’80\’99ve given him negative rep. Now when I encounter the second pirate they will show up as friendly/nice/good people

No, they won’t. People don’t get to rate their own characters.

If you encounter someone (let’s say X) with a rep unknown to you, then this is what happens:
1) Every player’s opinion list is scanned, and those that have an opinion concerning X are long-listed.
2) Each long-listed opinion list is compared against your own opinion list. If there’s no intersection, they’re removed; otherwise, they’re put in the short-list.
3) The short-listed opinion lists are ordered based on how much they agree with your own opinions. Ideally, there would be a perfect match (positive or negative) concerning those individuals you’ve grouped with, but if that’s not the case then you’d go with the closest match instead.
4) The opinion list matching your own opinion list the closest is taken as the reference. You are told how this opinion list rates X. For extra fidelity, you could be told how close a match the list was; you could also ask what the second-closest and third-closest opinions were.

Now from this, you can see that the opinion list used for rating any individual could vary every time. It’s also entirely anonymous – I may hate the person whose list is being used, but neither of us know the other is involved.

In your pirate example, I want to know what people think about X. Y has rated X as bad, but the fact I’ve rated Y as bad doesn’t mean I’ll take the inverse of Y’s opinions. If it happens that out of all the players in the game, Y’s opinions match mine the closest then I’ll be told that X is bad. I’ll never be told that X is good based on Y’s opinions; I may be told that X is good if the best-matching opinion set to mine is by someone who thinks X is good (and Y is bad), but if they do think that then the chances are that I’d actually agree with them anyway.

Richard

http://www.darniaq.com/wordpress/2006/05/general-gaming/character-accountability-idea/

http://whatwouldmattdo.com/2006/06/20/bartle-stole-my-idea-ok-not-really/

I’m just kidding, but it’s along the same lines. And as I said on my blog, the real reason I bring it up is because I want someone to implement it already! Hurry! WoW alone would be ten times as good if I could easily find good people to group with…

But that could be combined with any LFG system to corss reference automatically. I.e. You get a group invite and the system automatically cross referrences the people in that group with your self updated contacts lists (friends, ignores, general info) based on account UID, not character name.

As already mentioned, I’m good at deciding if I like someone or not. The trick is remembering the next time I meet them that I don’t like them. What I want is a client-side database that tracks my notes/descriptions of each player based upon the UID of the player. (NOT based on character name, as that is prone to error & typos)

Rather than investing enormous effort on building automated schemes to propogate this meta information (meta information which, incidentally, is a lot more nuanced than Like/Dislike!), leave it to the players to share rep as they wish. If Bob has an opinion of Jane, I’d rather judge myself how that affects my opinion of Jane. This, again, is what humans are good at.

The resulting system is still griefable – nothing stops would be griefers from spreading lies while pretending to be good honourable people. The difference is that there is no computer algorithm deciding to spread the lies – it is instead only a result of each individual having chosen to accept that griefer as authoratative. Again, this is something which humans have had thousands of years of experience with, for good or for bad, so can mostly deal with.

In conclusion, I’d like to dismiss all the fancy “ranking” schemes in exchange for a simple way to:
1) Make notes about other players from within the game. Notes are plain text, “Stole my kill”, “Owe 50 gold for favour”, etc.
2) Flag other players and have the flags visible whenever the character is. Flags are *all* user defineable, not a mere “good”/”bad” flag.
3) Ability to talk to another player about a third party with the UUIDs in a relatively error free way.

#3 could be as easy as being able to click on someone in your database and have their name + UUID pasted to your text as a uneditable coloured string, allowing the receiver to bring up their corresponding entry and add their own notes. This allows people to alert others about those that they like or don’t like. The people receiving the name can act as they wish on the info.

Please note that character names should not be unique ids! It is just too gameable to rely on unique character names. It should be impossible/difficult for users to type in the unique ids – they should have to acquire them in game somehow to avoid accidental references.

Admittedly, that’s also not always true, but it’s got better odds of being so than the old platitude does.

The medals would contribute to a “character” value in the recipient\’e2\’80\’99s viewable stats.