Do not display Traffic in 'Search Places', but sort by keyword matching relevance

- Traffic is not relevant when searching, except for 'Popular Places'
- Probably 90% of Second Life is empty because giving priority to traffic in the search results in gathering avatars on a small number of sims. This measure will better balance the charge over the grid.
- This measure will put an end to camping

It is not even relevant then, since "Popular" in most people's mind is not the same as "most campers" or "most alts" or "most Traffic-gaming"

Keyword relevancy is just as easy to game if not more so. Dependant on how its done. Plus it eliminates any way to tell the difference between a six prim house with a Linden Library chair in it and a build like Apollo if both places are using the keyword "Romance".

And lets look at keyword relevancy. Lets assume they use a system that cannot be gamed at all. Its all random, its not affected by what character you use, all that matters is the words in the search.

Take something simple and applies to the whole world like "clothes".

That search term alone would turn up hundreds of results.

How do you tell the difference between clothes by a newbie who's basically just making simple t-shirts with ripped band graphics on them and a designer doing extremely high quality clothing like Rebel Hope or Ambyance2 Anubis without knowing who the good designers are already?

Both are using "Clothes" as their keywork search....so you basically get a list of results somewhere near 200 - 300 results long...what do you do? TP to every place on the list? Thats tiresome.

And what happens if said Newbie Band T-shirt designer comes up at the top of the list? Is it fair that their content is presented first over something designed by Starley Thereian? I sure don't think so.

The traffic system is easily gamed, it is unfair how people are exploiting it in the fashion they are to draw ridiculously impossible numbers simply to attempt to present their content as the best in Second Life.

But without it the artists and content designers of Second Life have no way at all to give the masses a way to visually judge the possible quality of their content at all. Oh...except maybe mass advertising all over Second Life....turning it into an even greater marketing hell hole than it already is.

Basic facts of the situation is the top of the list is the top of the list, and thats where people are going to go first, you take the traffic rating away....and then you just place all the content in the game into a bin marked "generic" and force the artists and designers to take it upon themselves entirely to market their creations to the public in an extremely aggressive manner.

I don't want any more ads in Second Life...theres already more than enough of them. Content developers deserve at least something that gives the world out there a visual indicator of the quality of their work. Don't punish the actual artists and developers just because the non-developer side of the game that simply traffics in uncreative chattel and vice is unbalancing the system.

It doesn't matter how useful traffic is *potentially*, in practice it's completely useless, and even a gamed keyword system would work better... even if it didn't work *at all*. Because it would at least not rip sims to shreds as people load them down with laggy camping holes to try and boost their rating.

People already spam the keywords in the classifieds, so doing pure keywords isn't going to work.

What is needed is some kind of ranking system that has some entropy inserted into it, and isn't 100% transparent. Something like a weighted average of unique visitors for some cycle time linked to the particular keyword(s) searched. Simlar to Google Pagerank but less complex. It would still be capable of being gamed, but more difficult as the workload to game it would be more significant than "sit here for L$4 every 15 minutes."

I posted a while back a topic about 'tagging' keywords to parcels. Well, I will summarise as I've been refining the idea a bit.

Basically a parcel can have some number of keywords, lets say 10 as a rough figure. You enter these keywords, causing your parcel to come up in searches for these keywords. This has the advantage also that we don't have to rely on descriptions being carved up sensibly, and since descriptions no longer need to be checked for keywords they can be ignored and return to being DESCRIPTIONS of plots.
Anyway, if I search for several keywords, then the plot or plots that match the most of these are returned higher up in results. If I go to one of these however and find that it doesn't have anything to do with what I was searching for, then I can re-open my search window and hit the 'Report tag abuse' or some such button. This simply files a point against that keyword for that parcel. If a large enough percentage of users visiting the parcel report the same keyword then it is removed, and the parcel loses a keyword from its allowance for say a month (so it'd be down to nine).
If the parcel goes for a long enough time without losing keywords through this method, then it's keyword allowance increases. So a plot which doesn't 'trap' users into visiting it by using false search-words will get more search words allowance. While plots which are maliciously using search-words to try and force its way into markets it doesn't cater for are penalised as users self-police and remove their tags.
To avoid people simply altering the plot to try and gain keywords back, there can be a delay before a plot is listed, or black-marks for gaming the system can be held against the avatar who added the tag (this is my preferred actually, so if I behave then ALL my plots can have more keywords).

The only main issue is with the potential for a malicious group to report tags on purpose to try and have popular ones removed. But I don't see this as a huge issue so long as an avatar has to be ON the parcel they are reporting and have searched for that keyword within the last hour or so. Since key-word policing would be based on a percentage of visitors who reported the word. So if 10% of all visitors went there thinking it was something it was not and report the word that matched, then that word would be removed.
Another note is that as your keywords go up (from not misusing them), so does the percentage of complaints versus visits per keyword amount. So a starting plot may have a keyword removed if it receives 10% complaints or more complaints from visitors who searched those keywords, with a minimum of 10 complaints before action is taken. However, if it has had a good record for a long period of time, then this may go up to 20% for example.

A similar system can also be used for classifieds

I actually posted a similar complaint on the feedback forum, about keyword search being a necessity in all of the search functions.

/142/0f/167871/1.html#post1418447

In my case, I started a group called Non-Fiction Writers of SL. If you were going to do a search for such a group, what words would you use? "Writers," "Non-fiction," or even the alternate spelling of "nonfiction," right? None of those will bring up my group.

However, if you do a search for "non," yep, JUST "non," you'll see my group. What SENSE does that make?

Oooh, if they got REALLY fancy, they could include the option of either relevant, OR popular results. Now THAT would be technologically advanced...

Balianne Masala

Yes, traffic numbers suck. In my opinion one of the main causes of lag in the system, because they promote free accounts to go and sit on camping pads.

Instead of getting rid of the traffic numbers, they need to change the way they are calculated. Perhaps base it on avatar activity within the region... this way, an inactive camper, who is using an anti-idler and not even at their computer, won't increase traffic numbers, while a person hanging out and chatting or interacting with something/someone will.

Wouldn't this be easy for Linden Labs to implement into their traffic calculation?

Sure, it would create a new batch of anti-idler devices to make the avatars continually "say" random things, but then that would become a nuisance, and land owners would have to kick them out or not have very popular area, hence defeating the purpose of having camping pads in the first place.

No, at that point I'd grab the OS client and poke arount with the incoming chat sections so I can read what's coming in and the outgoing chat so I can talk. Then grab something like the AIM bot for a chat AI and connect it up with chat input and chat output of the client. Of course, I wouldn't use AIM bot, it's not smart enough. But in the end I'd have a client that would be able to communicate with other people on a general enough level that most people would be unaware of it's bot-ness.
Or heck, how about I implement an outgoing SMS port? Use the client's ability to know who is in the area and the land information and when the landowner shows up, I get a text message on my phone and I pop over and sit down for a chat. If it's a landowning group, it may be more difficult, but I could probably come up with something.

Issue with anti-idle traffic is that if you prevent them counting while stationary, then someone will just come up with a camping device that allows avs to run around and get money so long as they remain in the land/simulator. It'd be using slightly more complex scripting and the avatars would then be physical (due to not sitting down), more lag, same problems.

If we're to go with a simple solution to traffic as a stat then my vote is with only counting premium accounts.
If we're to go with a feature that also makes a useful change to searching (which is pretty basic and frankly poor at the moment) then tag-based searching is a must. It's more complex to implement, but not by a huge amount, the main complication is policing it, which is why I like self-policing by residents, either directly effecting a person's ability to create tags, or simply passing it up the chain to a Linden who then issues the final result.

That would probably be better. My suggestion would be only to count premium accounts.
No, because scripts could "say" things on other channels than 0.

Ok, so perhaps my idea of changing the traffic calculation wouldn't work so well as you guys have pointed out.

I really like the idea of only counting premium account avatars for traffic numbers though. Excellent, yet simple and effective!

Is there a voting proposal in the system for the "counting only premium accounts for traffic" idea? If so, I will put my votes behind it.