A Consumer Watchdog in Second Life?

Yeah, and this wasn't a good thing. That's why Linden got rid of it, and diminished the importance of it.

'Considerds this a good way to say..No longer wanted..Terrar'..I'll have to remember it!

Another interesting occurance that could happen, David:

"Customer: If you don't give me this item for free (or at a discount) I'll report you negatively at the SLBBB (or whatever the site will be called.)"


People on SL are very good at, if anything, finding ways to manipulate systems to their gain. I don't think the above is all that unthinkable.

I'm just trying to raise these questions (as were raised in another thread, actually) to bring them up -before- a system becomes entrenched.

Surely however, a monitored voting service that is closely watch would be better then something that is basically free-for-all.

They are good questions, I suppose we could impliment a complaint or an inspection system to find out if the report it just or unjust.

Yes, but monitored by who? Whoever monitors it would have to remain ever vigilant, it'd be a full time job. You couldn't just get tired of the investigations and say "fine, whatever was entered, that's what stands".

You'll essentially have 'reputation griefers' that you'll have to deal with.

EDIT: I'm not saying this is a horrible idea in its 'ideal', but just that to create a fair and objective agency, that has creedence, credibility, and a good reputation *on its own* it takes a lot of work.

As soon as the system was gamed, or it became basically a flame board or a medium for griefers, the site's credibility would disappear: nobody would take any stock in it anymore.

For something like this to work, fairness *has* to be your #1 "product", and that takes a LOT of work to maintain.

This surely depends on how big it gets, obviousaly it'd be ideal to be big and beautiful..I've seen certain business that are mostly automated..With moderation..I don't think it'd be a full time job, but I do think it would be a bit of a big task.

Thing is, the more automated it is, the easier it'll be to manipulate. A small business could live or die on an unfounded rumor. (As could a large one, really.)

Perhaps an intricate system like 'resellerratings.com' uses, where the business owner can comment after each rating.

This, of course, won't stop team or mass negatives. While a couple of pages of "-5 OMG U GIZE ARE FAGgS LOL!@#" probably wouldn't be taken seriously, a "summary" of (-2500) because of those ratings, and a position at the very bottom of a rankings list, likely would have a negative effect even if it wasn't deserved.

To avoid this would require extra human intervention, as well as a judgement call as to what entries are frivolous.

There, we open up the pandora's box of bias and personal opinion of the moderator -- and the appearance of such -- even if the moderator is not actually biased.

Perception of bias leads to reduced credibility, which leads to diminished efficacy of a service like this. Just like how before 1.7, you couldn't *really* judge how "bad" a person was by their received negatives. The system essentially obsoleted itself because of lack of oversight and its completely arbitrary (and automated) nature.

Well if there is a register system where people register to vote, it'll be alot easier to tell if they are group voting..And alot easier to ban the individuals who are doing it, rather then shutting down the voting system.

LOL! I never really thought people would read it so I may have skipped some of the more boring details. The complaint was valid, the complaint was against him. He was terrible at getting back to anyone, he was also bad at dealing with anyone in person, I have a few stories about that too, but those are for a different day and thread.

Not sure about that. I (and some other SL users) have recently been "working" at a site, Amazon.com's "Mechanical Turk" - a site where you are given tasks that only humans can do (computers cannot; i.e., 'identify the pizza store in this picture') - and you get paid per work unit.

The system uses voting logic, to determine what the 'correct' answer (or 'correct' picture in a group of several, i.e., the one that actualy -does- contain the pizza store) actually is.

Over the weekend, a few "enterprising individuals" found a way to script bots to hammer the site with random answers.

Now, the entirety of the data is dirty, and no 'humans' can actually do the work units because they are all 'contaminated' and taken by the bots.

The bot script was posted, and now hundreds of users are simply running bots on the off chance that even a 'low' percentage of their hits are correct - with the current system (until it is shut down, which it undoubtedly will be) they will at least get some pay for the bot's "random correct answers".

The point here is that "mass" votes or negs can't always be tracked or filtered out, and sometimes (often) a system is 'gamed' in such a way that it isn't expected.

If you are the type of person who doesn't like a certain category of person or product, you may be compelled to vote negatively at random times, or simply when browsing categories, even if you have never had a transaction with this person or used their product.

This would be very hard to track. Not all "attacks" on a system happen in bursts, come from the same location, or even show a solidly identifiable trait.

Granted, most businesses won't have to worry about this. There aren't entire categories or teams of people who hate, for example, shoe vendors.

But some businesses certainly will suffer due to large numbers of people who simply want to make trouble. (However, similar problems could be caused by 'competitors' and their 'shills' as well.)

I agree completely with you. any type of system used to rate products/people would need to have some sort of checking system to verify the complaints/comments/ratings or it would turn into the same crap we had to deal with while we had the +/- rate system for avatars. Apparently, spending money to give bad ratings won't deter people from doing it.

Well, OK. But I've heard plenty of stories where local BBB chapters actually do act in rather draconic manners, having little understanding for startups or even being accused of political bias.

Bias happens whenever an organization is run by humans. Manipulation happens whenever an organization is run by machines which are operated by humans.

Exactly. This is what I originally said in my complaints of the rating system: that they have to be transactional versus arbitrary. (Philip later supported this exact notion - requirement of a transactional nature for reputation - in a speech at SLCC.)

Right now we have no API or method to verify that a transaction between two people has taken place on SL. Therefore any SLBBB would have to rely on an arbitrary data collection or rating system, which by its nature is easy to game or manipulate without constant and intensive human moderation or oversight, which inherently brings in the possible factor of moderator bias.

I am not a programmer, but couldn't a system based off of the XML implementation in SL be used for these types of functions? Or did I totally botch this train of thought up too?


I seem to be batting -5,000 today....I hate Mondays!

They certainly could be, but it'd have to be a standard, and univerasally acepted, and used.

I used to toy around with this overall idea myself some but i decided the overall work load for such a system would probably be too high.

Generally what has to happen is the complaints have to be 'factually' verifired before they are ever entered into the system, aka there has to be specific human oversight, or at leas a semi-automated way of contacting people when complaints are raised and allowing them to factually verify the very basics... a) did so and so actually come by on day X, b) did so and so purchase any product or service, c) when was this transaction made, d) when was complaint Z raised, e) when and how was the complaint 'handled' by the seller... can person X who raised the complaint, if the seller denies, actually produce the sold object for inspection?

etc etc...

Now im not saying this is a bad idea, overall, i think its interesting and could be used for alot of good things, but it really does need to have alot of direct hands on investigation and resolution of complaints otherwise its just going to be a grief/extortion tool.

one potential option would be to mandate that any complaints tendered must have an agreed upon log of the support session before being made 'public' to the site (agreed on as the factual log of the conversation by both parties) etc...

again i'm not saying that any of this is bad, just that its going to require alot of careful thought and planning, if its really going to be a good tool to help the people of secondlife

Sounds interesting, David. FFRC meetings are on Sundays, 1PM SL. There are obvious synergies here.