Working on a reputation system, looking for feedback

Well, I've been trialling Dale's system and it seems to work just fine. I don't quite understand the full under-the-bonnet workings even though Dale spent quite a while explaining it all to me. I can see the potential in his system and the benefits it could bring to keeping areas of the grid 'griefer-free' while remaing as open as possible. Yes, it could be abused or gamed, but with a large enough userbase such activity ought to reamain within acceptable parameters of 'noise'.

"Hot or not"

I would definitely be for payment/non-payment people to be indicated differently. Perhaps also, include a method in the future code that only let's the person do X if non-payment if their reputation is pretty high. Or something... (early in the morning, sorry)

I have thought about a very similar kind of system recently as well. I guess I was beaten to the punch.



The largest advantage to this type of system is that unlike Sims Online a large group of griefers would have basically no effect on your score to anyone but themselves.

If you don't trust the griefers, then you just won't care about how they rate people and the score you see for any one person will be completely independant of what the griefers do.

If Alice trusts bob and sue and doesn't trust (or maybe doesn't even know) joe:
- If alice or bob trust sam, then alice would see a 'positive' rating for sam
- If alice or bob don't trust sam, then alice may see a 'negative' rating for sam
- If sam has a million alt accounts that all rate sam positively it will have no effect - because alice doesn't trust those alts.
- If joe and his million alt accounts all "don't trust" sam it will have no effect - because alice doesn't know or trust joe or his alts.

When you can extend this out a couple of levels of depth you can get in interesting 'trust' score.

Some comments:
* DON'T implement 'enemy of my enemy is my friend' or 'friend of my enemy is my enemy'. That does give control to griefers. You want the path to completely break at people who are not trusted.
* I'm personally against being able to see any scores but your own. I should be able to see (and change) how I rate people, and be able to see the 'rating' from my viewpoint only. Publicly visible lists of aggregate votes or the links between people can give griefers power or cause witch hunts respectively.
* My personal opinion is that 3 levels is all that is needed. My trust, The trust of those I trust, and the trust of those my trusted trust. More than that gets too far from the source I think.

Just my 2c.

As hilarious and insightful as some of the posts are in this thread, I'd like to test this as well.

This reads as though it could be a useful tool eventually.

I know, from experience, that even though you have someone in your F-List, it doesn't necessarily mean they are a "friend" and that, after time, you may even forget why the hell they were there in the first place.


People get busy in SL all the time and some times new "friendships" fall to the wayside due to time contraints and the ever-present Second Life Time Distortion Phenomenon or STD for short. XD


From the way this reads, invariably, I may be able to look at my "list" on this item and at-a-glance know who a person is, if I've had contact with them before, was it good/bad/nuetral?


Perhaps an interface with this allowing an individual to create notations assigned to a persons name would be nice.

I'm not sure if you're familiar with WoW (World of Warcraft) but, as an example, I point at the Guild list interface in that game.

You can, as an officer in a guild for example, click a members name and create a note that all members of that guild can see. You can also create a personal note viewable by the creator only.


From my experience in SL, having the ability to do that in correspondence to the friends list would be a very handy tool indeed. Especially if you consider it from the perspective of say...managers of a store or club owners, ect.


For example, a griefer (or some other undesirable character) comes into your establishment and creates a problem. They may not do anything to warrant a ban but, after a couple of days or even weeks, you might completely forget that incident or at least recall something involving that person and that it was a negative experience. You could simply create a note in this good/evil/neutral list thing with that person's name so there’s no blurred lines when it comes to the details of why you like or dislike that person.


There is obviously a lot of potential to abuse something like this, yes, but that is like, with just about anything in SL that categorizes an individuals person, place or property.

I would advise, to those of you who view this "idea" as a negative to read between the lines, open your mind and take the service at face value without bias created by your own bad personal experience/s.


I know for me, I have more issues than National Geographic when it comes to the number of instances I can think of where this "kind" of service could have been a huge irritation. But that BS aside, I see a lot of potential for this if developed and used responsibly.


*kicks soap box out nearest window*

The scanner shows the payment status along with name, distance, age and score.

That's exactly how it works


In principle, I'm making it so that alternative scoring systems are possible. But the current one is going to stay the default. If additional systems are made they'll have to be specifically chosen. 'Enemy of my enemy is my friend' is definitely not going in.

Search depth will definitely be configurable.


That's how it currently works, you wear a scanner that calculates scores from your POV, and there's no way to make it do it from another. I don't plan allowing it either.


Yup, I agree with that as well. Besides, extra levels would add tons of extra load on the server. 3 levels seemed to be a nice balance between a high likeness of providing something useful and not being too brutal on the server.

Sort of. People you rated directly always have a score of either +10 or -10. People you haven't rated get their scores calculated from the scores of people you rated positively. If nothing is found, you get a 0.0 result.


Yes, I will allow attaching comments to a rating.

BTW, you can currently add notes like that already, only viewable by yourself. Look at the last tab of anybody's profile.


No, I don't play (or even have ever played) anything of the MMO kind besides SL.


Yes, that's definitely going in, along with an option to set an expiration time for negative ratings. I'll probably make it tell you when your rating is about to expire and ask if you still think the same way.


Actually, I don't see that much potential for abuse, but maybe I'm missing something. If you have some scenario in mind then I'd like to hear about it.

As has alazarin, i tested this system out too. I thought it was awsome. Took a bit of time to figure it out but it was simple after that (like all new things).

Only people who have registered and paid can either rate or be rated. That is essential to restrict rating spam attacks and rating grief attacks. It won't prevent them of course, but it will impose a rl financial limiter on such abuse.

Ratings are *relative*, therefor it's up to the end user to decide who to trust.

I'm sure they are relative. But given the ease with which a rating system could be abused with alts (which only cost time for the abuser, and not even that if they are sufficiently skilled scripters), any rating which allows teh views of a free account to be considered will become meaningless once those guys cotton on to abusing it.

If that is an option 8alongside only listening to the ratings from paid for accounts), cool. But go in with your eyes open, and be aware that a free for all system will be gamed to death so that it vwill be meaningless.

There is no external view. All your ratings are calculated "on the fly"-ish based on who you like and dislike. YOu could have a million alts positivley rate you, and it wouldn't matter because noone trusts those alts. The system can't be "Googlebombed"

Suppose those million alts also rate each other, and an organised group of a dozen or so people in on the hack also rate each other and those alts. Still sure of yourself?

Poisoning of that scale would be quickly spoted by the community. If I as a member of the graph maintain my connection to the scammer not through someone in on the "scam", I can be assured of clean rating.
The moral is, don't be loose with your ratings.

Well, let's see. When you first start using the system, you're a lone point. As you start rating people, you create a graph. A group rating each other would create a bigger and bigger graph. A second group doing the same would create another, independent graph. And when they start rating each other, they will create connections between them.

Still, all of this in no way affects your ratings, unless you rate one of the alts or the "people on the hack" positively. And if you do and notice you're now getting inaccurate results, just remove that positive rating.