The REAL New Search

What are your thoughts on this?

From SL Dev Mailing List:

Hello!

You've been asking us to do better with inappropriate content and content classification and we've been listening. In an effort to make our abuse reporting system easier to access and more effective, we are planning to introduce a flagging system for our search results - and we'd love to get your feedback.

The new flagging system will help Residents to flag inappropriate postings for rapid removal, while preserving everyone's ability to express themselves freely in world.

How it will work

* First, when you post a classified, group, event or parcel listing, you will have to self-declare if your content is mature or not.
* Second, every listing will have a "Flag This" link. You can use this to report inappropriate or mis-classified content or nominate great content for the showcase.

Flag Categories

* Mature - the listing is adult only content
* Prohibited - the listing contains content that is prohibited on SL
* Spam - irrelevant listings (examples: commercial posts in non-commercial venues, event postings for non-events)
* Showcase - nominate something for possible inclusion in the Second Life Showcase (http://secondlife.com/showcase)

[All listings will still have to follow the Community Standards and Terms of Service.]

Types of listings that can be flagged

* Parcel listings
* Classifieds
* Events
* You will not be able to flag avatar or group profiles

What happens when something is flagged

* Every flag counts as one vote for that flag category. While one (or few) votes does nothing, if a listing receives enough votes in that category, it will be auto-classified as reported.
* Prohibited and spam content will be taken down immediately when it gets enough votes
* When content is flagged several times, the content owner will be notified by SL support team and may be penalized according to Community Standards

Anything else?

* Residents will be allowed to flag a particular search listing only once. You cannot change your vote, so flag with care
* There will be a limit to how many times a day a Resident can flag search listings
* Anonymous basic residents or residents whose accounts are less than x days old will not be able to flag content
* Residents will still be able to report abusive content from inworld via the Abuse Reporting system under the Help menu
* Of course, no system can be perfect, so if you think something was flagged in error, you can always file a support ticket for review

Thats it. We hope that this will make content flagging more effective and will help the community better regulate what is appropriate and what is not.

Please reply to this email with your thoughts and feedback on this proposed design!

Cheers,
Jeska, Kalpana and the LL Search Team

I can just imagine now how this can be abused. I hope you're not paying more than 50L for an ad. Top spot ads can be flagged out of position by alts.

I hope "immediately" means "after human review" and not "automatically" .

"Anonymous basic residents" = no payment info on file?

Well, it says anonymous basic accounts won't be counted. So I think you would have to be payment info on file at least. Someone's alts could theoretically be identified by LL in such a case, no?

Well, read my blog for my thoughts on 'automatic'

If they do it automatically for something as serious as one's account, why not for something as trivial as a classified?

It is good to see that they are looking at ways to prevent gaming the system!

Yeah.. that was sarcasm.

hmmm... places that use traffic bots can now, not only increase their traffic with the bots, they can use their bots to mass-flag and kill your search and classified results.

I'll put my few thoughts in. On the whole I think it's good. I do think a few will abuse it. I do think things will be taken down automatically.

Mostly I think LL needs to set some clear guidelines for what they consider to be Mature, as thats likely to be the one most open to interpretation. They never did define that back when Age Verification was first brought up, and it still bears clarification now.

You mean the anti-fraud automatic suspensions for L$ transfers?

That's still quite different though. If a store owner accidently triggers the anti-fraud watchdog then they'll get suspended, but their store is still there and sales continue like nothing happened and they'll be back in-world as soon as a Linden looks at it.

If a classified/places listing gets delisted automatically, sales will drop though. If it takes a few days before a Linden puts it right again that could be quite a loss in income.

Depending on the answer to that question, this proposal is either a great idea or a lousy idea.

My guess is that Linden Labs will keep with standard operating procedure, not spend resources on human review, and leave it at automatic. That's the way they treat Abuse Reports isn't it- if a resident gets enough abuse "votes," ban now and ask questions later (or never).

So there you go- grief your competitors by flagging their ads.

Now if there is human review once an ad gets enough abuse "votes," then this is an okay idea, and long overdue.

That introduces two problems with this idea.

Not only are residents going to over-report, because they have no clear guidance on what the categories (like mature) mean.

People placing ads believing their ads are honestly within the guidelines are going to get burned if their guesses turn out to be wrong.

Lack of guidelines, along with lack of human review by Linden Labs, will turn what is the kernel of a good idea into a tool for griefing and a punishment for honest merchants. I hope that doesn't happen.

I think it's excellent. I can arrange to take out my competitors

Seriously, it's total rubbish. In a utopia, the concept is great, but in reality the working of it, as it is written, is rubbish.

1. It *will* be used against competitors.

2. It *will* be used against people for personal reasons.

3. The removal from search is automatic, and there is no suggestion that anything will even be checked.

4. There is no suggestion that anything will be looked at or done about malicious voting. It isn't even considered to be a possibility.

As it stands, it's rubbish.

This is a joke, right?

I already have a certain "competitor" who has multiple alts that he uses to harrass over on my SIM, that LL does nothing about. I`m sure he`d have a field day with this thing.

The one thing I do like is the showcase votes. Even though that has potential for gaming it takes the cries of Linden Favortism out of the equation.


Man, the griefing potential though. The Lindens need to review these things before they just pull them. There also needs to be a stiff punishment for anyone who abuses the system for griefing to hurt someone's business.

Sometimes I really wonder if this stuff is thought through by the Lindens and if so are they laughing about the chaos it can cause? (The second part is not a serious question, you get the idea).

If Linden Labs were to consider adding a human review element to this procedure, how many hours would they be paying staffers to review these anyway?

Granted there are lots of search listings in Second Life. But they are finite.

Inappropriate "ad-abuse" reports would indeed waste a staffer's time when it comes to review. However, (a) clear guidelines would reduce mistakes by otherwise honest reporters, (b) human review would reduce the effectiveness of grief-reporting, which in turn will reduce the incentive to grief report, (c) the restrictions on "abuse voting" (type of accounts, vote limits per account, vote threshholds for review) will prevent (at least cap) the amount of spam-reporting that can be done.

If a listing needs x number of votes to be considered for removal, then Linden Labs does not have to worry about removing every inappropriate listing. They just have to remove the inappropriate listings that dominate searches. Listings that may be inappropriate, but are otherwise obscure and don't pollute searches, get a pass- and why not, who cares about the listing that no one sees?

Even with the current lack of guidelines, I think an overwhelming majority of the abuses I see every day could be eyballed. Ten-second review, removal, cut-and-paste response would work for most. My guess it that it would only be a small percentage of ads that would require more careful review and more detailed responses.

The initial workload in reviewing listings under a new policy would decrease after a while. If residents can't break the rules with impunity, and actually get caught and corrected, they will have less incentive to try to break the laws. Right now, it's easy to break all the rules, because there is no downside to it. In fact, if you are an honest merchant following the rules when placing listings, you're placing yourself at a considerable competitive disadvantage to this who ignore the rules when placing listings.

There would be great benfit from human review by Linden Labs. So what are the costs of review? I can't imagine that it would be huge. Of course, I don't have all the information about it that Linden Labs does. And then it also appears to me that Linden Labs is so poorly managed, that it's possible no one thought to consider a question like, "What would be the costs and benefits if we had Linden Labs staff review these abuse reports before removing listings?" My guess is that management has a standard reaction to the thought of assigning staff to do something that isn't a sexy job that everyone wants to do. "I don't want to pay someone money to do a new job, and I don't want to have to direct the staff for which I am responsible."

maybe the same Lindens that will be reviewing the Showcase nominations can review the submissions at the same time? Unless that is automated too, but I am sure being the Showcase the Lindens will be being sure it is quality stuff to not send the residents to bogus places. And unlike AR reports I don't believe there will be hundreds of reports a day. Most that submit classifieds do it right. This is in theory only for the out of place ones. Furthermore if people knew that if they file bogus submissions they face penalties (maybe not being allowed to list classifieds to stop competition from false submitting, etc) then the likelihood will lessen.

It may work.

Regarding Spotlight (aka Showcase) - I believe only non-profit venues and events should be included when this gets put on a tab to replace the current Popular Places tab in Search, sometime in the next couple of weeks.

For-profit stores or other endeavors for profit should not receive the enormous and unprecedented advantage of free advertising and promotion on a special Search tab of Linden picks, or the windfall of profits and customer share that will result.

When this Showcase tab comes into SL, we will have a situation where customers and new residents will be funneled to some shops at the expense of others.

Spotlight for non-profit places - yes.

But not for stores - hair and fashion shops, for instance.

There should be no free LL advertising and promotion, or what is essentially sponsorship, offered to particular resident stores and other for-profit resident businesses.

Let people nominate anything they want for Spotlight, but if something is a shop, or has a shop in it, or charges admission, it should not be eligible unless and until the actual venue itself is non-profit.

If you agree, please join the traffic group (click on Jeska Linden's profile to find it), and attend one of the several meetings being offered next week and let them know.

Or write an e-mail to Jeska about it.

coco

I would think it's great if "spam" included "keyword spam" but I'm not really convinced it does from the way it's all phrased.

It seems to be aimed more at the "24/7 'yard sale'" events than anything else or anything that can determined by looking at the listing only rather than "physically" investigate the parcel.

Yeah, I forgot that the Showcase would be "editorially" reviewed.

So of my two theories, I guess that blows the theory that Linden Labs is too pinched on staff resources to have someone actually review listings for compliance with Linden Labs's own rules.

That leaves the second theory: The listings-compliance project isn't sexy enough to attract staff members to want to work on it, and management isn't about to tell any of its staff to do something they really don't want to do.

Geez, now I want to work for Linden Labs. I've always wanted a job in which I only did what I wanted to do, and not have a boss or a client tell me what I had to do. I had just never thought it possible before.

I would volunteer for the listings-compliance project, because that would make a hero to honest merchants in Second Life.

You are probably aware that I agree with all of that Coco. And I do intend attending one of the meetings, but it may not come up. The meetings are about the future of traffic.

I know, Phil, which is part of the tragedy of this. They bundled this bit IN with the traffic changes, sort of like an unpopular rider tacked onto a popular bill in congress.

I hope that you will bring it up in the meeting you attend.

coco

The problem of search listings complying with rules is necessarily linked to the problem of traffic.

The problem of rule-noncompliance and traffic is that they litter Search results with trash that obsure relevant results.

If Linden Labs does not deal comprehensively with Search, then they will fail to develop Search into something useful.

Hopefully, Linden Labs's traffic-project team is communicating with its listing-abuse-vote team. If they aren't, then hopefully someone suggests that such communication and/or collaboration is a good idea, and the suggestion will make sense to them.

If the meetings are anything like the group IMs, it would be swamped in a free for all, but I intend to find out, and voice that opinion if possible.

I saw a chat log of that group on a blog. I don't know how anything will be figured out if that is the normal flow of the thing.

As for the in world meetings... I think if they use the speaking cube (soapbox) (you have to touch it to be in the que to talk) and ignore all other chatter they may do OK.

I think this system's vulnerability to abuse is tied into the number of flags necessary for a listing to be x'd. If I read the specifics of the announcement right, it would cost a competitor US$90 to flag you ten times - assuming his normal account is premium (if not, it'll cost him $100). People's traffic bots, not being premium accounts, couldn't be used to flag anything. Would ten flags be enough to axe a listing?