Jump to content
Sign in to follow this  
Blondin Linden

Test Out New Search, Delivering More Relevant, Faster Results

Recommended Posts


Pussycat Catnap wrote:

Shoes was not an ideal example - the shoe stores have self policied themselves and people selling sexbeds or jazz DJing don't generally put 'shoe' in their name...

Try 'neko' or 'furry' for mass confusion.

Elemination of traffic from the results will mean that places that are about or for furries and nekos will rise over places that are say, malls that use those terms in search and have only 1 or 2 if any relevancy to neko or furry.

- Neko and furry being just two potential items that tend to get gamed like that to the point of now being valueless in search. 

Now THAT I can agree with. There are words, crucial genre-defining keywords that folks abuse. Part of the problem is that fashion and accessories that folks in the Furry and Neko communities identify as "theirs" also happen to be very popular in general on Second Life. Designers and Builders cannot really be faulted for saying their merchandise fits into those categories .. because in most cases it truly does.

What differentiates "Gaming" from "Optimizing"? It's a very subjective line and attempts to automate it get into some really hairy algorithms .. most of which fall flat as soon as they are applied on inputs that fall outside their "test bench" data. However one tool that can be applied easily and with nearly 100% accuracy is the one tool that Linden Lab seems unable or unwilling to use ... People.

I recently filed an AR against someone for "Spamming Search". To me, the key determinant is fairness. The target of my AR had used techniques they derived by purchasing some of my products and then using me as a free resource to learn everything they could about Search and how it worked. They then turned around and built their own tool and used that to dominate the top 7 slots in a very competitive niche.

Not only had they previously engaged in fraud and strong-arm tactics to extort money out of "Clients" (if the Client didn't pay the demanded weekly "Monitoring Fee" their parcel would suddenly and mysteriously drop from Search) but they then engaged in tactics that clearly showed total disregard for others, for the community and for everyone else that derives benefit from Search.

It is my STRONG belief that that sort of "bad apple" in the barrel is what gives Linden Lab (and many many others) the impression that everyone in the top slots is nothing but a Spammer or is somehow "Gaming Search". In fact that is a perception that is patently false.

The people that are on top of most categories are the Merchants that have spent long hours developing and building the best products they can, garnering a large following of loyal customers, and creating a reputation for value, quality and customer service. AND they have thus had the resources available to apply extra leverage to gain a stronger showing in Search Results. In short, they're good quality Merchants that have spent the money and time to gain that position.

If LL had been able to apply a person to respond to my AR, they would have quickly seen that the person in question was violating the basics of the Community Standards ... Fair Play. They could have rapidly resolved the issue by force-downranking their parcels. Shoot, just force-downrank all but ONE of their listings if you truly want to be fair about it. But it takes a human to honestly look at, evaluate and understand these things on a case-by-case basis .. and yeah, that's expensive.

So instead we have yet another variation of Search Engine designed to provide today's definition of "Relevant". I truly believe the folks behind this have the best intentions at heart. They WANT to make Search work for people and they WANT to serve the needs of the paying community. But they're applying technology to fix what is at its heart a social engineering issue. They may have all the latest and greatest technological tools at their disposal, but the one tool they NEED is the one they aren't allowed ... a human-staffed resource that can judge the edge cases and take appropriate and decisive action to remove and "adjust" those that will not ever obey the unwritten rules of "Sharing".

Share this post


Link to post
Share on other sites

Logged on - no chat bubbles? No names over av's heads? And still, no HUDS (they're worn - and don't show even in show transparent).   Didn't even get to search once I realized I couldn't communicate.  Relogged with Snowglobe (the only viewer on this iMac, OS10.6.7). Now I can't interact with anything and I'm a cloud. Yes - I did all the normal routines and got every trace of V2 off the computer.

I'll do a clean install of Snowglobe and that should solve it. 

 

Share this post


Link to post
Share on other sites

@Ciaran:

Parcel size IS relevant, because it denotes content amount.

example: 3% keyword match for a search term included for a full region that includes multiple search terms vs 100% match for the only search term on a 512 (or heavens forbid, smaller). which is likely to have more content for the search term? you can have 128 x 512 parcels in a single region.... keyword space is limited... if the region lists it it's very likely to have content related to it.

I'm not saying it's and end all/be all stat, but it should be in there.

@more general

another one is traffic, which be so much harder to game with a slight change to how it's calculated... constant avatars should be counted at a logarithmically reducing rate with a minimum, with a slight  precalculation bonus for repeat visits. avatars spending between x and y time in the area would be unaffected and get a bonus for unique within a time frame visits, and avatars spending <x time would not be counted at all. the diminishing returns from constant avatars would be quickly outweighed by unique visits of a certain timeframe... it could still be gamed, but it would be infinetly harder... especially if the behavior were tracked on the avatar/account side, rather than the by the location.

@Thread

the non-obvious wildcard set up is annoying, but nice to be aware of, but exact match searches are an absolute must, bar none. title first, description second.

Share this post


Link to post
Share on other sites


Void Singer wrote:

@Ciaran:

Parcel size IS relevant, because it denotes content amount.

example: 3% keyword match for a search term included for a full region that includes multiple search terms vs 100% match for the only search term on a 512 (or heavens forbid, smaller). which is likely to have more content for the search term? you can have 128 x 512 parcels in a single region.... keyword space is limited... if the region lists it it's very likely to have content related to it.

I'm not saying it's and end all/be all stat, but it should be in there.


No, sorry, the parcel size should be irrelevant, if a larger parel has more relevant content because it has more space that is a completely different issue. Why should a store on a 512M parcel that specialises in, say steampunk, be less relevant than a full sim that has less Steampunk items than the 512M store?

If the full sim parcel has more relevant items, then of course they should score higher than the smaller parcel.

I've been through this Void, I had parcels cut on a full sim and a quarter of sim so I could focus on the right products, in the right parcels and pay L$30 for each of those parcels to show in search and then LL changed something and they were all irrelevant in search terms, so I merged them into larger parcels again because it was the only way to be seen, but now I can't identify and focus so easily, it's broad based search terms so now I can't advertise to people looking for x,y and z that I have that, because there's not enough room in the parcel description to do so.

That change messed search up badly.

Share this post


Link to post
Share on other sites

why would a ful region, with what you admit is limited description space list a search term for a category that it holds less than a relevant amount of content for? if it has less than a 512 parcel worth of items, then that means that less than 1% of it's business is devoted to that... there would be no reason to list it, when other things are more common to it that could generate income and need the limited space available... sorry no I don't buy that argument at all

and I was very careful to say that it is NOT the only stat that should be looked at... parcel items should also count, although frankly they are hugely limited in that it just encourages keyword stuffing of every single prim you can, and actually hurts most smaller plots since they tend to need to rely on vendors, which reduces their for sale item counts compared to what a bigger plot can afford to waste on box vendors

it is not the stat that is at fault, but how it is weighted and applied. when they first applied it, it was given too much weight, and that is something I think we can both agree screws up rankings.

Share this post


Link to post
Share on other sites


Void Singer wrote:

why would a ful region, with what you admit is limited description space list a search term for a category that it holds less than a relevant amount of content for? if it has less than a 512 parcel worth of items, then that means that less than 1% of it's business is devoted to that... there would be no reason to list it, when other things are more common to it that could generate income and need the limited space available... sorry no I don't buy that argument at all


So they can be at the top for several different search terms, this is a well known tactic. When you're specialising, you need less search terms. LL cut the number of characters in classifieds precisely because of this issue.


Void Singer wrote:

and I was very careful to say that it is NOT the only stat that should be looked at... parcel items should also count, although frankly they are hugely limited in that it just encourages keyword stuffing of every single prim you can, and actually hurts most smaller plots since they tend to need to rely on vendors, which reduces their for sale item counts compared to what a bigger plot can afford to waste on box vendors

it is not the stat that is at fault, but how it is weighted and applied. when they first applied it, it was given too much weight, and that is something I think we can both agree screws up rankings.

Vendors have long been borked in terms of being useful in search, keyword stuffing in prims is more of an issue on larger parcels where they can hide hundreds of the damn things, and we have seen this happen too.

The parcel size on its own shouldn't be a weighting factor, a larger parcel of content a should beat a smaller parcel of content a by virtue of having more content, but that's not what always happens.

Share this post


Link to post
Share on other sites

I have a 768m2 that ranks high in search. I also have another parcel in the same region that in group adds lots of prims so the 768m2 has lots of items set to search. More than a 768m2 allows. And that parcel ranks very high in the associated searches.

 

Parcel size is not up ranking. The amount and diversity of content that is not spamming is what ranks.

Share this post


Link to post
Share on other sites


Pussycat Catnap wrote:



Shoes was not an ideal example - the shoe stores have self policied themselves and people selling sexbeds or jazz DJing don't generally put 'shoe' in their name...

Try 'neko' or 'furry' for mass confusion.

Elemination of traffic from the results will mean that places that are about or for furries and nekos will rise over places that are say, malls that use those terms in search and have only 1 or 2 if any relevancy to neko or furry.

- Neko and furry being just two potential items that tend to get gamed like that to the point of now being valueless in search.

It occurred to me on my drive to work this morning that I -still- wasn't using the best example.

Lets go to 'product X' (because we're in generalizations at this point and specifics just cloud the discussion with annectdotes).

Store Y is all about product X and exists on its own lot somewhere. It is one of the more popular makers of X in SL.

Store Z sells product X as well, but through a small outlet vendor wall at club XXX. Club XXX has insane traffic because well... its full of 10-15 bots at all hours scripted to dance on poles, greet people, spam gestures while running a DJing a animation, and a few on sexbed poseballs off to the side... You know this place, its the bane of the grid... You've AR'd it 2321 times but it just never seems to go away.

Store Z has a miserable selection of product X items. Half of them are copybotted. The other half are from freebie malls.

 

Guess which store shows up better when traffic comes into play?

 

 

Share this post


Link to post
Share on other sites


Darrius Gothly wrote:


Pussycat Catnap wrote:

[snipped mad-kitty babbling]

- Neko and furry being just two potential items that tend to get gamed like that to the point of now being valueless in search.  >

Now THAT I can agree with. There are words, crucial genre-defining keywords that folks abuse. Part of the problem is that fashion and accessories that folks in the Furry and Neko communities identify as "theirs" also happen to be very popular in general on Second Life. Designers and Builders cannot really be faulted for saying their merchandise fits into those categories .. because in most cases it truly does.

Proper relevancy would therefore ignore terms like 'Furry' and 'Neko' and 'Elf' unless it was for things like avatar kits / skins, roleplay locations (a fantasy sim for elf for example), or subculture hangouts (like luskwood).

- And that takes humans to refine the AI until it learns how to analyze this.

(This is why google and yahoo and bing all employ very large teams of human contractors from around the world to go through search engine results in different languages and culture groups and tell the AI how well it performed - feeding the data back into the AI in a never ending cycle so that it keeps getting better and better.)

 

Share this post


Link to post
Share on other sites

Where to begin with the problems? Here we go...

Search for: AIRPORT

(Any measure of the popularity of a location ought to be based on traffic. Searching in the old viewer in PLACES with the search term AIRPORT will review the following traffic numbers and rankings (note: obvious non-airports are ignored such as Sex sites that have the word aiport in their teaser terms at the end and thus become top rated airports since the traffic number is for the sim).

 

Traffic        Rank         New Viewer Rank        Name

18899        1                      35                             Olds Air Force Base

10548         2                     30                             Hollywood Airport

101              100                   7                             Hollywod Airport Terminal

1668            19                      1                             SLPG Prague Airport

So what does the new viewer show? After you delete all the high traffic listings that put dozens of teaser terms at the end of their descriptions to get fake top ratings in any of a number of fields, including airports we are told......

The two most popular airports in SL today (1- Olds Airport, and 2- Hollywood Airport) are ranked 35th and 30th respectively in the new search.

A parcel that is virtual barren (Hollywood Airport Terminal) is ranked #6 in the new search.

The top rated airport in both the old ALL web search AND the new search (SLPG Prague is 19th in actual traffic.

Pardon me if i call the SL search system FUBAR (Futzed Up Beyond All Recognition).

 

 

 

 

Share this post


Link to post
Share on other sites

it would be very useful if we could have Marketplace search results in it, for the following reasons:

Marketplace is unknown for many residents, because usually when people download a software, they rarely come back to the website they download it from.

this reduce significantly the number of potential customers for Marketplace merchants, to only the users who come back to the website and have find out about it.

having Marketplace products listed in search results would have the following benefits:

  • it will make the Marketplace products available to everyone who uses search, expanding dramatically the potential customers to the Marketplace merchants.
  • this will make the Marketplace a very useful tool for Second Life business, becoming a mass media for their products.
  • this will expand the range of products offered, giving the users of Marketplace much more choices.
  • it will give a boost to the Second Life economy, since merchants would have more revenue to invest in making higher quality products, and it will be easier for them to sustain paying for land in a long term.
  • the user will benefit in that it will have the choice of buying the product right there, or teleport to the store if s/he choose so.
  • search will benefit from the filters that are already functional in Marketplace, such as, number of prims, price range, and permissions
  • it will refine search results, reducing the gaming of it, making it even more relevant, since merchants wont have the need to include the products they wish to sell in the description of the place.
  • computer resources will drop down, increasing the speed of the viewer, because it will not be necessary anymore to have a web browser and the viewer open at the same time.
  • it will improve the immersion experience, by not needing to get out of the viewer for another Second Life related function.


hope you consider this feedback weighting in how much would this improve the user experience.

congratulations for your hard work Linden Lab!

Share this post


Link to post
Share on other sites


ZenWarrior Fuosing wrote:

Although SL is a world comprised solely of information and finding only the information needed is absolutely essential, LL has neglected search for so long or done it so poorly that except for the denizens of these forums, those 7 of the 10 people (Lindens?) surveyed, and so-called vendors, LL's misteps with search mean no small number of "average" residents don't even care anymore. That's what happens when something as vital as search meets the following management and customer service attitude:

It has been a while since we last updated you on search and that’s because we’ve chosen to take a big leap forward versus smaller, incremental improvements.
  ~ LL blog posting, 25 May 11

WT_?! Talk to us! 

Personally, I essentially gave up on SL search a while ago after realizing I could find whatever I needed within the first 5 returned results using any decent seach engine and a properly constructed guery, with the exception of whatever LL used. Sorry, LL, but you lost me and 90% of my business long ago when search should have already been fixed. So guess what? Fixed or not, I don't care about search anymore! Too many years without decent search capabilities taught me not to even bother logging in at times. (Yes, Virginia, there are many reasons concurrency is down or flat. Did anyone in LL ever think being unable to find items and events was causing people to either leave SL or log in only once to never return?)

As I've been saying for over a year now, SL's actual decline started when they implemented the GSA6. Stats do not lie. Personally, I'm amazed that my store has lasted this long and still trucking along, despite all the crazyness and complete neglect of the system.

Things are looking up tho. Search will be key to any recovery. It looks nifty to me. The results are wack tho. I mean any basic search engine given the same info would return better results. I can't really understand LL's reasoning on the algorythms. It makes no fricken sense to me. For me tho, it does not really matter. My strategy for search from the beginning never really centered on 1 word, so even with all the problems, I would show up in lots of other results. Stick around ZenWarrior, things are still kind of in flux but I'm hoping for a good rest of the year, or maybe checkout another grid.

Share this post


Link to post
Share on other sites

What a coincidence, I just left today, then again, viewer 2.6. It was impossible for me to work with it. I had it for a week, an improvement over the 4 days I lasted before. 

Just couldn’t find people in the search, all returned can’t find. Since I use a lot the notes for customer’s requests, couldn’t accept not having it. Plus, the program stopped working 3 times, while I was trying to retrieve something from my inventory, plus its huge 96.1mb, enough as to have one of my CPUs at 100% all the time.

I rarely use the search in the website, I do my 99.9% search inworld, and is not working. Unless today was a special day, because they were working on it?

So, my opinion, search in the page seems to be working fine. Nice design, even though is not the same when you try something just to try, than when you really need to use something. In the other hand, for me is useless, I search inworld.

Share this post


Link to post
Share on other sites

 


Void Singer wrote:

@more general

another one is traffic, which be so much harder to game with a slight change to how it's calculated... constant avatars should be counted at a logarithmically reducing rate with a minimum, with a slight  precalculation bonus for repeat visits. avatars spending between x and y time in the area would be unaffected and get a bonus for unique within a time frame visits, and avatars spending <x time would not be counted at all. the diminishing returns from constant avatars would be quickly outweighed by unique visits of a certain timeframe... it could still be gamed, but it would be infinetly harder... especially if the behavior were tracked on the avatar/account side, rather than the by the location.


This.

Share this post


Link to post
Share on other sites


MarkTwain White wrote:

 

The two most popular airports in SL today (1- Olds Airport, and 2- Hollywood Airport) are ranked 35th and 30th respectively in the new search.

A parcel that is virtual barren (Hollywood Airport Terminal) is ranked #6 in the new search.

The top rated airport in both the old ALL web search AND the new search (SLPG Prague is 19th in actual traffic.

Pardon me if i call the SL search system FUBAR (Futzed Up Beyond All Recognition).

 

With all due respect, where the hell have you been for the last year and a bit? The issues you're complaining about have been going on with viewer 2 search and viewer 1.23 all search for over a year. Traffic has been diminished as a ranking factor for over a year in the new search. (The latest one is the new new search).

You highlight some pros and cons of the old old search, genuine traffic is a sign of popularity, people using terms that have sod all to do with their parcel was a problem.

Share this post


Link to post
Share on other sites

It's overall pretty nice but that is not useful feedback so I will point out the criticisms.

Two feedbacks that sort of bothered me about it...

1. Search itself only appears in "Advanced" mode (unless is it hidden somewhere? I couldn't find it). Well hidden or nonexistent... its same effect. This is very important as it means all "Basic" users will have much trouble going anywhere but a handful of Destination Guide locations. Even most of the destination guide is inaccessable from the basic mode! I thought basic was directed at casual users and shoppers... kind of hard to shop if 99.99% of second life content is inaccessable :matte-motes-impatient: Search should be easily discoverable in all viewer modes.

2. The other negative is the new search (like web profiles) is super-duper slow. Some engineer work could make it more speedy. Regular search we have today is snappy in comparison. I waited for it to load but impulse buyers want things instant... this is gonna lose people all over the place.

 

Look and layout is great. The search is noticably smarter too which is good. Sorry to point out the specific negatives but if nobody does point out the bad it won't improve much :matte-motes-asleep:

Share this post


Link to post
Share on other sites


Canoro Philipp wrote:

it would be very useful if we could have Marketplace search results in it, for the following reasons:

 

Marketplace is unknown for many residents, because usually when people download a software, they rarely come back to the website they download it from.

 

this reduce significantly the number of potential customers for Marketplace merchants, to only the users who come back to the website
and
have find out about it.

The last thing we need is more reasons to kill the inworld experience. So I hope they never do this. We should be pushing the other direction, towards shifting marketplace people back into inworld shopping.

 

Share this post


Link to post
Share on other sites


Ciaran Laval wrote:

So they can be at the top for several different search terms, this is a well known tactic. When you're specialising, you need less search terms. LL cut the number of characters in classifieds precisely because of this issue.

and it worked... so why then abandon parcel size as a stat? this actually gives smaller parcels a edge since they can cover their content more thoroughly while the larger parcels cannot, which balances nicely against a parcel size modifier


Vendors have long been borked in terms of being useful in search, keyword stuffing in prims is more of an issue on larger parcels where they can hide hundreds of the damn things, and we have seen this happen too.

The parcel size on its own shouldn't be a weighting factor, a larger parcel of content a should beat a smaller parcel of content a by virtue of having more content, but that's not what always happens.

keywords/m^2 is an easy calculation to cut stuffing, and remember that stores are not the only things in SL, parcel size matters a great deal to events, hangouts, RP areas, sandboxes, educational venues, etc etc.

Share this post


Link to post
Share on other sites


Void Singer wrote:


Ciaran Laval wrote:

So they can be at the top for several different search terms, this is a well known tactic. When you're specialising, you need less search terms. LL cut the number of characters in classifieds precisely because of this issue.

and it worked... so why then abandon parcel size as a stat? this actually gives smaller parcels a edge since they can cover their content more thoroughly while the larger parcels cannot, which balances nicely against a parcel size modifier

The reason there was a problem in the first place was keyword stemming. which was giving the top paid classifieds upto three times as much word coverage. Keyword stemming hasn't gone awat. We're back to me asking you why a 512m parcel that is dedicated to steampunk should be less relevant than a full sim that has only a few steampunk items?


Void Singer wrote:


Ciaran Laval wrote:

Vendors have long been borked in terms of being useful in search, keyword stuffing in prims is more of an issue on larger parcels where they can hide hundreds of the damn things, and we have seen this happen too.

The parcel size on its own shouldn't be a weighting factor, a larger parcel of content a should beat a smaller parcel of content a by virtue of having more content, but that's not what always happens.

keywords/m^2 is an easy calculation to cut stuffing, and remember that stores are not the only things in SL, parcel size matters a great deal to events, hangouts, RP areas, sandboxes, educational venues, etc etc.

On my RP Sim we have an events stage in the sky, we have no room on the ground for it, the sky area isn't as large as the parcel it's on.

The only parcelling I have really done is for the tavern and market area and that's because I insist we have them on different parcels, I could merge them. The market would suddenly be on a full sim parcel, why should it get an advantage for that?

When I'm searching for something I want to find the content I'm looking for, I don't want to spend time trawling around a hair store that has a small line in mens clothing when there's a much better mens clothing store with more content on a smaller parcel. Parcel content is what people are after, not parcel size.

Share this post


Link to post
Share on other sites

You know, I was kind of torn between whether parcel size should really matter. Now tho, I think it makes for perfect reasoning. Not soo much the size, but the prim limits that come with the size. Essentially, you can look at it like this. We pay for the land based on size and prim limits. Prims can be seen as advertising tools. So for every prim you use, you are advertising. How is this any different than a classified ad where you just pay to have your ad higher? It only makes sense that some1 who has a larger parcel should have more chances to advertise. When look at in this way, there is no need for any policing. If some1 wants to spam a certain word, and has nothing in their store related to it, the customers will learn and the merchants will be considered a spammer by the shoppers. It is just as easy to remember your favorite store, as it is to remember the 1's that annoy you with spamming.

Whatever algorythms that LL creates to hurt spammers will eventually hurt legit merchants, besides the other unforeseen consequences.

Share this post


Link to post
Share on other sites

why Marketplace has to be against sim location shopping? why cant they coexist?

the merchant gets the lindens, the client gets the product, everybody happy.

what im saying is to bring the Marketplace inworld, because right now is outside, that will work in favor of increasing the inworld experience because people would not have to open a web browser and stop using the viewer to buy the product from the Marketplace.

buying in the Marketplace or in a sim store its about choice, i prefer very much to buy in the Marketplace, i have filters in number of prims, price, i dont have that searching in sim stores, i have to face lag, i have to face rezzing all those posters of products also with all the textures of the store and people around with their increase of lag, its about choice, we should have that choice, you could go to sim stores and buy your things, the rest of the people who prefer it my way, should be given the option. many dont know the Marketplace exists, for the reason i stated in my other post and that reduce their choice as to only shop in a sim store, why would that burden has to be imposed on them when they can buy things more easely?

is not fair that because we like something one way we should stop everybody else to do it another way.
in fact if possible we should have as much choices as possible to buy a product.

Share this post


Link to post
Share on other sites


Ciaran Laval wrote:

The reason there was a problem in the first place was keyword stemming. which was giving the top paid classifieds upto three times as much word coverage. Keyword stemming hasn't gone awat. We're back to me asking you why a 512m parcel that is dedicated to steampunk should be less relevant than a full sim that has only a few steampunk items?

On my RP Sim we have an events stage in the sky, we have no room on the ground for it, the sky area isn't as large as the parcel it's on.

The only parcelling I have really done is for the tavern and market area and that's because I insist we have them on different parcels, I could merge them. The market would suddenly be on a full sim parcel, why should it get an advantage for that?

When I'm searching for something I want to find the content I'm looking for, I don't want to spend time trawling around a hair store that has a small line in mens clothing when there's a much better mens clothing store with more content on a smaller parcel. Parcel content is what people are after, not parcel size.

 for the exact reasons you stated... both have limited space, but a region has more literal content and more room which equate to more potential people interaction.. that's hugely important for all sorts of non-store and social locations.

the space available to a region to describe it's content has to be focused to it's main content, but can cover a wider range, so the owner has to cherry pick it's majority content... that's always going to be over the focused content of a 512 plot... and if it isn't the region owner is throwing away business, because people will come in, not see what they were after and not come back. when they could have been generating leads for content that they do have a majority of.

content quality is a completely different case, and I don't think there is anything realistic that search can do about that right now. Also, not all content is measured in prims, social and RP venues find themselves in the position of providing space and content, but largely, space. more space = more interactions, more people = more popularity and driving traffic to them.

I'm not suggesting that the size is more important than other stats, merely that it IS important for a wide range of venues, and should account for something. Throwing out metrics is never a good idea... if they become broken, reduce their weight or refactor how they are calculated, but you never throw out a useful metric... you just rebalance it.

Share this post


Link to post
Share on other sites

Void Singer wrote:

 for the exact reasons you stated... both have limited space, but a region has more literal content and more room which equate to more potential people interaction.. that's hugely important for all sorts of non-store and social locations.

the space available to a region to describe it's content has to be focused to it's main content, but can cover a wider range, so the owner has to cherry pick it's majority content... that's always going to be over the focused content of a 512 plot... and if it isn't the region owner is throwing away business, because people will come in, not see what they were after and not come back. when they could have been generating leads for content that they do have a majority of.

Wait wait, there was a time when you could have small parcels with relevant content generating leads and getting people to those areas, then something happened, this is when the parcel size bias rumours started which LL would neither confirm or deny but it seemed that larger parcels were getting a boost, this change hit me to such an extent that I gave up trying to have smaller parcels, I still had the same land area parcelled or not but the small parcels were completely and utterly buried in search, which made the whole exercise in trying to parcel, pointless.


Void Singer wrote:

 content quality is a completely different case, and I don't think there is anything realistic that search can do about that right now. Also, not all content is measured in prims, social and RP venues find themselves in the position of providing space and content, but largely, space. more space = more interactions, more people = more popularity and driving traffic to them.

I'm not suggesting that the size is more important than other stats, merely that it IS important for a wide range of venues, and should account for
something
. Throwing out metrics is never a good idea... if they become broken, reduce their weight or refactor how they are calculated, but you never throw out a useful metric... you just rebalance it.

Agreed about quality of content, I've actually had that dispute with people before, search cannot measure quality. I know about RP sims and spaces, trust me, I've had people questioning why I have a market and wouldn't that space be better used for RP space, the answer remains no because markets help with funding and I like to get merchants involved at some point anyway, it's good networking.

There are a limited amount of useful stats, maybe parcel size could be useful as a tiebreaker, when two parcels score equal everywhere else. Maybe there should be different relevancy states depending upon which search category you're in, hangouts have different scoring methods to shopping places, although I think this would be a nightmare for the folk running search to configure.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...