The REAL New Search

As one example, the event information is supplied to Eventful.com. Events can be searched there. The result is not nearly as useful as it could be because Eventful.com is designed for real world events, not SL events, so the data is shoe-horned into fields that don't quite fit it.

If this data (events, classified, etc.) was made availble to all takers, then folks that run sites like sluniverse.com would be able to create properly customized search systems using the data, tailored for display in the SL embedded browser, on pages made to work without scrolling when they are the size of a reasonable sized dialog box in the SL interface.

There already exist third party SL search sites.

It should even be possible to allow one to pick the source of the data that is displayed in the built-in search result display. You could pick the SLUniverse version of search results or the Caledon version or the NCI version or what have you if the default Linden version didn't suit you.

Yes, something like you are suggesting Kitty would be awesome. That would keep the griefing at bay after it plays out a little bit.

The only problem with it is that if they should decide to do something similar to that, they can't tell us they did it that way .

Otherwise you could proactively report a bunch of things that are clearly spam and then afterwards report a bunch of things you have a grudge against.

Bah! You foiled my nefarious plans!

So now what is "PG" or "Mature" comes down to whether or not x number of residents think, for any reason or no reason, is PG or Mature.

More accurately, unless there is completely unanimity in among Second Life users that content is PG, then it will be deemed to be Mature.

Do you know someone who thinks that everything everywhere is for some reason or another not suitable for children? I bet you know more than one. And I bet there are more than x people like that with non-anonymous accounts on Second Life.

So Linden Labs is getting rid of this "PG" and "Mature" nonsense, which would be good. Except instead of doing it easily and honestly- by admitting that this PG and Mature self-regulation bit hasn't worked- they are taken the hassle-laden road of letting the residents destroy the PG and Mature system.

The clarifications made about this new little system also make clear that removals will be entirely automated- destroy first, ask questions later. Since the reasons for the abuse-votes are immaterial, it will be great empowerment for those who abuse-report with bad nor no reason, and it will create helplessness for those with good reasosn for creating the listings the way they did.

You know what, there is one good thing about this. I was seriously close to opening up an animation shop, and doing serious advertising for it and all the things that a businessperson would do to promote one's business. Forget that, though. I can just see where this Search-listing mess is heading, and I don't want to be a part of it. Maybe in tne end, it will all work out and be fine and peachy keen. But I'll pass and watch how others fare under this little experiment.

But what if I consistently flag as prohibited child pornography parcels and free to play casinos, and flag twice as many child pornography sites than free to play casinos? My reliability rate keeps going up, but I'm consistently mis-flagging some search results!

Matthew

Good idea. In fact, when I was new to Second Life and less immersed in the ways of Linden Labs, I assumed that any sane system they had to deal with abuse reports would include some component of flagging the little boys who cried wolf.

I was operating under the assumption then that there was someone at Linden Labs who read the Abuse Reports that came in. Now I know it doesn't quite work that way.

Since rating the flagger requires human intervention and decision-making on the part of an LL staffer, they'll never do it.

No, I understood that, but unless LL did an automatic penalty, a Linden would have to in addition to deciding whether the offending search entry was prohibited (spam or whatever), would also have to trawl through all X votes, and for each one make a judgement call as to whether it was malicious or a genuine mistake.

Apart from the extra work, there is also scope for the Linden concerned to make mistake and penalise a genuine mistake and fail to penalise a malicious flag.

Similarly



Adds an extra burden on the Lindens.

Presumedly, the rationale behind this is to automated some of the manual labour involved in the existing AR system (since all the things this system can catch are all AR-able issues) rather than increase the manual work involved!

Matthew

You'd only do that because you'd know you were "rewarded" for flagging appropriately though . If you didn't know, you'd likely not bother and just start mis-flagging, hence the disclaimer that we shouldn't be told how it works . And you can figure out a more complex algorithm to establish reliability as well.

The only reason I suggested it is because it doesn't require any Linden overseeing time beyond what they already said they'll do. The less Linden time needed, the more likely they'll go for it .

True Matthew.

Some of us still think of ways to smooth out the system and forget that we can't count on Linden help. They are too taxed as it is. I guess we all like to trouble shoot and talk here of ways to make things work and forget that it is just never gonna happen lol.


At the end of the day it will be automated like ARs. Maybe like ARs there can be an appeal system for some instances? Again more work.. I know.

There might be some clever ways to punish those who misuse the automated system.

However, I think you'll find that any fair method of punishing misuse of the automated system will use for more LL staff resources than would a staff review of a flagged listing before removal.

Reviewing one listing is surely far less work than reviewing the multiple votes for flagging that listing.

By the way, there a huge difference between these two standards of review:

1) Considering the wisdom of the action before taking action.
2) Reviewing whether an action was the appropriate one to take, after the action had been taken.

With the former, one is inclined to ask oneself, "What is the right thing for me to do?"

With the latter, one is inclined to say, "Is there enough justification for the action that I can let it stand without undoing it?"

That's why the "destroy first, ask questions later" policy requires less work than making a fair consideration before acting. Justifying an act already taken is a much quicker and easier than thoughtfully considering an action before taking it.

OK, that was a canned and simplistic example. But in practice a typical user might flag things as prohibited for a large variety of reasons. Let's say that someone has about a dozen criteria they use when thinking a search entry should be flagged prohibited (including child porn, gamlbing, racism, etc.) and most of those are definitely prohibited by LL, a few are in the grey area, and one is fully permitted in SL but the person concerned is mistaken about it being banned (let use take furries as a clearly silly example)

Overall the reliability rating will go up, and because the system you propose is a blackbox automated system, the person will never get corrected that furries are allowed.

Let use suppose that this person is part of a community who are all relatively sane about what is and isn't allowed in SL, apart from this belief that furries are banned. We all go around reporting child porn, gambling, racism, and furries - our reliability ratings still go up as in general we are reporting the right things, but the furry community is no in danger of keep having their entries removed temporarily and reinstated, whilst this community is non the wiser about their mistaken belief.

Again, a somewhat canned example.

The fundamental problem is that you are trying to represent a complex belief system of what is right or wrong, as a single scalar variable - you really need to track whether someone is reliable in reporting a particular type of search entry, rather than whether someone is just reliable.

Matthew

Maybe it's a personality trait, but I don't really see if that way .

If I had to judge whether a listing is related to gambling, I'd make a more "split second" judgement if no action was taken yet, but if it was already taken I'd be likely to stop and actually think about it. I'd feel less pressure to decide one way or the other afterwards while it would feel more urgent to make up my mind beforehand.

There's also the question of which is more harmful: letting prohibited listings remain until a human can investigate it, or wrongly removing innocent listings.

For gambling I'd think it's rather harmless whether the listing is there for an extra day and it would be better to wait. For child pornography the sooner the listing (and content and account for that matter) vanishes, the better and you don't want to wait.

While I'm not naive enough to think the latter is non-existant on SL I would think (and certainly hope) that it's extremely rare to the point where the shoot-first policy would do more harm than good.

*nods*

I realize all the flaws, but with most things I suggest I keep the following in mind:
* if it requires a lot of development work, it won't be considered
* if it requires interactive Linden time, it won't be considered
* is it better than what's there now or what is being proposed

I can't really think of a situation where you'd be guaranteed to be worse off with a singular scalar rating. It's possible you might exclude people who later learn the error of their ways, but you could account for that somewhat. You almost certainly would eliminate some people who are malicious and you'd have to deal with those who report a mix of good and bad if you didn't use the rating anyway so no change for better or worse.

To me that adds up to a net win, if a far from perfect intermediary solution, hence the suggestion .

---

Edited to add that I'm also assuming that flagging an entry is very different from filing an AR in that you'd only be able to pick a category and click a "Report" button and no description field so there would really not be any way anyone but the reporter would know why they flagged it, just that they did so.

I should have looked at the SLdev thread before posting, blah .

You missed a rather important quote from Jeska, Darien .

I've seen it in practice. Let's say there's a law office, which does a steady amount of business drafting documents. Sometimes there is a young associate who reviews the documents before they go out to the clients. Sometimes there's a hurry to get the documents out, and the young associate reviews the documents after they've gone out.

When the documents are reviewed before they go out, the young associate may give it a careful lead, flag problems, suggest solutions, and even correct spelling. When the documents are reviewed after going out, they get a glance to determing, "Okay, there's nothing that can be considered malpractice in this document." Even if one does find a mistake after the fact, managing partner may not want to mention it to the client- no use in embarassing oneself for a mistake that will probably never make a difference anyway.

Answering the question after the ban has taken place, rather than before, does pyschologically have the effect of reframing the question. I'll suggest to you that when it comes to making decisions, in the absense of other factors, people are more likely to take the road of least resistence. When the decision is made before removing the ad, the road of least resistance will favor the one who placed the ad. When the decision is made after the fact, the road of least resistance will favor the abuse-reporter.

If you think it would be different for you, that the timing of the decision would not affect the way you considered it, then you would be a very conscientious person. That's absolutely a good thing, and there are some people like that. It would also run contrary to the human tendancy to follow the road of least resistance.

But what I think I know about human nature isn't really important. Regardless of human nature, Linden Labs, as an organization, strikes me as ultimatenly committed to the path of least resistance.

I think there are very few listings that by there existance are so devastating that the mere risk of leaving them remain until the next business day will create irreparable harm to SL users or Linden Labs. And if such a listing did exist, then why wouldn't a Linden Labs staff member want to consider it immediately anyway? Even if you wanted to carve out the special case for child pornography, that's not what Linden Labs is contemplating. Child porn gets even treatment with the claim that an event advertisement for live music doesn't really have live music.

Based on what I think I've learned about Linden Labs since I've been using Second Life, I think it's far more likely that the intent of this new procedure is to leave it reliant on the automation, and the after-the-fact "review" is that meaningless rubber-stamp to approve the automated act.

It'd be a great thing if I were wrong about this. But everything I've seen or read makes me confident in my prediction.

I haven't read the whole thread yet but just wanted to say that this is my concern. I am sadly already kind of a target of a 'hate campaign'. Everything I do is mimicked to a point where I keep notes on the dates and details of every single little thing I create. But not only that the friends of this person have joined the campaign as well.. so I guess I know where their votes are going!

I refuse to be drawn into 'profile wars' as I know myself how stupid that looks when you see that in a businesses profile..so instead I have to silently put up with it.



Thanks LL.

OK, if it is a choice between the LL proposal and your proposal, I might favour yours as being slightly better (but I'm afraid I still feel in practice that it isn't that much of an improvement on the LL proposal - but that is more indicative of problems in their proposal than yours)

However, if it is a choice between the status quo and the LL proposal, I'd take the status quo.

The things that really shouldn't never ever ever be in search (e.g. child porn) tend not to advertise publically anyway; as for things like gambling etc. appearing in search, the normal AR process would work well enough (perhaps add an AR button to search entries?)

Google has a SafeSearch filter which could be used in addition to mature tagging for those who don't want to see mature content (although we are all meant to be adults here so the occasional slip through shouldn't harm anyone).

Spam should be dealt with by improving the relevancy score of non-spam. This is where I think the real effort should be.

Matthew

Where is this SLdev thread?

The shoot first, ask questions later idea is rubbish - total rubbish. The question is, when is later? It could be minutes later or, more liekly, it could be days or weeks later which, in some cases, would unjustly cost a person a significant amount of real money.

The system is all about LL abdicating responsibility for producing good results. The idea is to put the onus on the residents, and let them do the work, and if a few people lose money through malicious flagging, that's the way it is. Kitty's suggestion won't fix that. A group of people with malicious intent will easily be able to get a place removed for a while, costing the owner real money. They'll be happy enough with that - they won't need to get the place taken down permanently. They won't need to do it to the extent of being black-flagged - just a one-off will satisfy them, and another effort can even be organised later, using different people.

It's a *very* bad idea to put such power in the hands of residents.

This is a terrible idea.

People pay a lot of money for their classified ads, The only one with the power change them is the hopefully impartial Lindens.

Sounds like this is just an attempt to save their AR enforcement team some work.

Why not be more specific about what content is Mature, what counts as spam, and what sorts of things are allowed where.

A lot of the more casual misuse exists because people do not know the actual distinctions.

I agree that this will create vigilante and harassment reporting rather than geniune. Can't LL set up a filter...if you do an event notice that has the forbidden words in it, the notice cant be posted. So if the rule is no yard sales, no land rental or sale ads, no ongoing events listed over and over...just build that in on the front end. So, those things can't even be IN the events listings to begin with.

I do think the idea of flagging an offensive post of any kind is a good tool. But to use it to basically pit one av against another for any number of good or bad reasons is NOT my idea of building community.

Or as i have said before...why not have a tab FOR yard sales, land rentals, etc...that way people can post those things and get them out of the general events area.

It's a mailing list you subscribe to, but you could just follow this particular thread in the archive if you don't care about all the technical discussion that goes on there .

https://lists.secondlife.com/pipermail/sldev/2008-May/thread.html#9456 - for the thread

https://lists.secondlife.com/cgi-bin/mailman/listinfo/sldev - if you want to actually sign up (do read the guidelines)

I only posted the first post, in it's entirety. That must have been from some subsequent post, which i haven't posted. It wasn't me.

All I can say about that quote is, LL doesn't have a good track record for enforcing anything, why should I believe this would be any different?

But as I stated before, I think the system overall is good, it just needs clearer guidelines about what constitutes Mature.