Jump to content

New Search page how to change it back?


SheriBalto
 Share

You are about to reply to a thread that has been inactive for 1026 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

10 hours ago, Sharie Criss said:

Relevance calc is still strange to say the least. (It seems to favor parcel name more than anything.)

As long as SL has had a search engine (the legacy search isn't a search engine), a parcel's name has been one of the two most important and powerful ranking factors. The other one may or may not still be a powerful ranking factor. I haven't checked. I explained the reason why the parcel's name is so powerful many years ago, and it's all still there, but in the forum archives.

To improve a place's rankings, modify the parcel's name.

That's slightly off-topic, but I thought it worth mentioning.

Edited by Phil Deakins
Link to comment
Share on other sites

On 11/23/2021 at 2:04 AM, Lucia Nightfire said:

Relevance calc is still strange to say the least. (It seems to favor parcel name more than anything.)

A search for `Anime` brings back as first result someones private parcel with the name

'Anime F..a 🐔for Non-Anime B...eaches'

Some creative censoring going on there. The parcel name starts with 'Anime' and infact contains 'Anime' twice, however I have watched the parcel and it is largely derelict, and I highly doubt it is what people were hoping to find when they searched 'Anime'.

The description also contains 'Anime' multiple times.

Because I am genuinely curious to see how broken the search is, I have made some edits to my sims name/description.

Before

Name

 

Kokoro Academy - Anime School RP

Description

Anime themed roleplay sim!
アニメ  , 学校

After

Name

Anime Anime Anime Kokoro Academy - Anime School RP

Description

Anime themed roleplay sim!
アニメ  , 学校
Anime anime anime anime
(We're proving the search is broken)

 

 

At time of writing, we are the 9th search result for Anime with the before name/desc. I'm not sure how often the search results update. We'll see where we rank once it does...

  • Like 2
  • Haha 1
Link to comment
Share on other sites

30 minutes ago, Extrude Ragu said:

Because I am genuinely curious to see how broken the search is...

What you described is the search engine working just fine - not broken at all.

It's good to do tests though, because you can learn how the search engine works, and what to do to rank more highly in its results. If the results matter, it is important to work with the search engine and not expect it to work with you.
 

Link to comment
Share on other sites

Just now, Extrude Ragu said:

Wrong, an engineers expected behaviour =/= working behaviour

If your #9 ranking move up after your changes, then you will have proved that the search engine is working just fine. If it moves down, then it could be broken, but not necessarily. Judging by what you wrote in your previous post, you seem to think that, if your ranking moves up, then search is broken. If that's what you're thinking, you couldn't be more wrong.

Also, if that's what you're thinking, you need to learn how search engines work before deciding whether or not it is broken. In the other recent search thread, I posted a link for anyone to learn.

Link to comment
Share on other sites

7 minutes ago, Phil Deakins said:

If your #9 ranking move up after your changes, then you will have proved that the search engine is working just fine. If it moves down, then it could be broken, but not necessarily. Judging by what you wrote in your previous post, you seem to think that, if your ranking moves up, then search is broken. If that's what you're thinking, you couldn't be more wrong.

Also, if that's what you're thinking, you need to learn how search engines work before deciding whether or not it is broken. In the other recent search thread, I posted a link for anyone to learn.

You are thinking in terms of engineering and not in terms of the needs of LL and its user base.

Measuring the quality of software on whether it does what it is programmed to do is a very poor metric. Most software does what it is programmed to do.

But does the program solve the problem it was designed to address? The answer is no.

 

  • Haha 1
Link to comment
Share on other sites

22 minutes ago, Extrude Ragu said:

You are thinking in terms of engineering and not in terms of the needs of LL and its user base.

I am thinking in terms of how search engines actually work. They are unable to produce results in the various ways that individual users might prefer. They are not AI programmes. I'm talking about search engines from Google all the way down to the SL one.

As long as the search engine produces good relevant results, it works just fine, regardless of whether or not it's what you think it should produce. I said in a post further up this page that you have to work WITH the engine because it can't work with you.

I also said that, if you want higher rankings, work on the parcel's name. That will do the most good. If you want to know why the name is so important to rankings, I don't mind posting it. I've done it before, but it's way back in the archives. Understanding how search engines work helps a lot.

  

22 minutes ago, Extrude Ragu said:

But does the program solve the problem it was designed to address? The answer is no.

You didn't design it, so you can't say that it "doesn't solve the problem is was designed to address". I have no doubt that, like all other search engines, the problem it was designed to address is producing relevant results for the searchterm, and from what I've seen, and from what you posted even, it does that just fine. It isn't broken.

Edited by Phil Deakins
Link to comment
Share on other sites

1 minute ago, Phil Deakins said:

As long as the search engine produces good relevant results, it works just fine

The results are neither relevant or good. Do you really think that when people search on 'Anime' they are going to find some edgelords private parcel on the mainland the most relevant result? Or does my sim suddenly become more relevant to the user because I've written 'Anime' into the name another four times? The answer is no

The fact of the matter is, I am able to game the search results, and users are receiving low quality results. This results in users thinking that no high quality anime sims in SecondLife exist, and quitting, this is bad for Business. It does not serve the needs of SecondLife or its residents, therefor it can be adequately described as broken.

And listen, I get it. I too work as a Software Engineer, I spend about half of every waking second of my life writing and maintaining software, I am more than aware of how indexers work, how boolean queries, synonyms, stemming, fuzzy filters, relevance ranking algorithms, term weighting, facets etc I have dealt with plenty of search solutions in my lifetime. It's just all too easy to forget that software is just a means to an end, and if it does not get you to that end, it's not doing the job it's intended to do. I should not need to know how to game the system, nor should the user expect to find gamed results that are useless to them when they search.

It is up to the creators of a product to ensure that when a user uses their product, they are not disappointed by the product. Failure to understand this basic principle will be at your peril, most users do not understand and have no plans to learn how to trick your software into working if it does not work the first time they try. If you have to convince your user after they have used your software to use it again, you have already failed.

  • Haha 2
Link to comment
Share on other sites

11 minutes ago, Extrude Ragu said:

The results are neither relevant or good. Do you really think that when people search on 'Anime' they are going to find some edgelords private parcel on the mainland the most relevant result? Or does my sim suddenly become more relevant to the user because I've written 'Anime' into the name another four times? The answer is no

As far as a search engine - any search engine - is concerned, yes, that parcel is the most relevant. And yes, your sim will become more relevant with the changes you made. I'll say it yet again, understanding how search engines work really does help enormously!

You are wanting the search engine to do something that it cannot do. Even Google cannot do it. That is, rank parcels according the quality that the user sees in each parcel. That can't happen. Maybe when an engine is programmed that's capable of visiting each parcel and evaluating its quality, then it could happen. But no search engine can do that, and there's no use in expecting one to do it.

You really do need to understand search engines and work with them. They can't work with you.

11 minutes ago, Extrude Ragu said:

The fact of the matter is, I am able to game the search results, and users are receiving low quality results.

Of course you can game the results. You can game the results of any search engine on the planet. Why would you think that the SL one is better than the rest?

11 minutes ago, Extrude Ragu said:

It is up to the creators of a product to ensure that when a user uses their product, they are not disappointed by the product.

Generally speaking, users are not disappointed. Users generally know that the search results are not 100% perfect in every way. They expect it. On the whole they are pretty good though.

The only thing that is broken is your expectation. You need to accept what a search engine actually is and work with it. What you expect the SL engine to be, it simply cannot be.

 

Edited by Phil Deakins
Link to comment
Share on other sites

53 minutes ago, Extrude Ragu said:

Wrong, an engineers expected behaviour =/= working behaviour

But if you can't define working behavior in engineering terms, don't know how anyone expects an engineer to fix it.

37 minutes ago, Extrude Ragu said:

Most software does what it is programmed to do.

All software does exactly what it was programmed to do, to the letter, without exception.

A failure of search to match user expectations is not well demonstrated by singular specific examples, especially when it comes to singular search terms or opinions on result ordering.

The engineering view is simply that the user doesn't know how search is actually working, or that they have a simplistic expectation that if applied globally would produce less relevant results outside of the one specific example.

This is especially relevant in cases where the user is expecting a simplistic linear list of results. Tailoring search as a whole to work linearly for simple terms would resolve all the issues raised in this thread, but it would result in a far worse search for all other use cases.

13 minutes ago, Extrude Ragu said:

The results are neither relevant or good. Do you really think that when people search on 'Anime' they are going to find some edgelords private parcel on the mainland the most relevant result? Or does my sim suddenly become more relevant to the user because I've written 'Anime' into the name another four times? The answer is no

Search relevance can only be derived by the data it has access to. A search engine has no concept of what might make one location better over another based on your opinion of data that exists outside the search engines scope of operation.

How is it supposed to determine that one location is (in your opinion) an edgelord ****hole and another is not.

A more context aware search improves results for end users, but it would also be far more frustrating for people trying to get their preferred location to the top of the list, and so far, discussions about search ordering seems to be entirely based on that exact frustration.

Or in short "Search is broken because I, specifically, can't game it".

Citing fictitious groups of people and outcomes is not helpful and actively undermines any discussions.

Edited by Coffee Pancake
  • Like 2
Link to comment
Share on other sites

Just now, Theresa Tennyson said:

If you're relying on the order of search results to determine quality you're setting yourself up for a disappointing life.

When you say `you` understand you are not talking to me personally, but the users of SecondLife. I am intentionally positioning myself from their point of view. I am more than capable of scrolling through a few extra results and if I wasn't I wouldn't still be playing SecondLife.

If SecondLife as an organisation are relying on users knowing that their search is substandard and scrolling through 10's of results to  find some reasonable ones, SecondLife is setting itself up for high user churn.

Considering how much SecondLife struggles to keep new users as it is, don't you think that blaming new users, for not knowing how bad SecondLife's search system is, is ultimately going to lead to SecondLife's demise?

Am I the only person who can see that maybe blaming the user for poorly designed software, will just end badly for everyone?

  • Haha 1
Link to comment
Share on other sites

4 minutes ago, Extrude Ragu said:

Am I the only person who can see that maybe blaming the user for poorly designed software, will just end badly for everyone?

I don't know that answer to that, but you are one of the few, perhaps the only one, who believes that search is "broken" in the way you've described. I'm beginning to suspect that you are simply fault-finding for the sake of it.

ETA:
Maybe you should use the phrase 'doesn't work as well as I'd expect it to' instead of 'broken' which it isn't. Not in the way that you described, anyway.

Edited by Phil Deakins
Link to comment
Share on other sites

3 hours ago, Phil Deakins said:

As long as SL has had a search engine (the legacy search isn't a search engine), a parcel's name has been one of the two most important and powerful ranking factors. The other one may or may not still be a powerful ranking factor. I haven't checked. I explained the reason why the parcel's name is so powerful many years ago, and it's all still there, but in the forum archives.

FYI, you attributed a quote to me that I did not write, I was quoting someone else.

Link to comment
Share on other sites

8 minutes ago, Coffee Pancake said:

How is it supposed to determine that one location is (in your opinion) an edgelord ****hole and another is not.

Actually the answer to that question I think is pretty simple.

Don't rely on a computer to do a humans job - If a user follows through on a result and clicks teleport, chances are that result was relevant to the users query. Use that to help inform the search results.

  • Haha 1
Link to comment
Share on other sites

20 hours ago, Silent Mistwalker said:

 

The main reason for doing it is to target a younger demographic and make it more appealing to them. Another word for it is marketing.

This search design is supposed to appeal to the new generation? They dont even want to watch YouTube videos over 10 minutes in length. Unless its clips of vine, tik tok, or fails. A 30 minute how to on how to get better usage from your Iphone? never gonna watch it. "ugh, just give me the fast way to fix it. or better yet, do it for me."

18 hours ago, ChinRey said:

I don't know. LL obviously has put some thought into this. That alone is an improvement compared to so many other changes they've made to the UI over the last decade or so. I know this may sound like sarcasm but it isn't.

Come on now, we all know they just throw darts at ideas on the wall.

 

 

Personally, i hate dark themed stuff. It literally hurts my eyes. I get after images for hours. Not having the choice to use light theme is a deal breaker for me. 

  • Haha 1
Link to comment
Share on other sites

1 minute ago, Extrude Ragu said:

When you say `you` understand you are not talking to me personally, but the users of SecondLife. I am intentionally positioning myself from their point of view.

No, you very much aren't.

This is argumentum ad populum.

1 minute ago, Extrude Ragu said:

If SecondLife as an organisation are relying on users knowing that their search is substandard and scrolling through 10's of results to  find some reasonable ones, SecondLife is setting itself up for high user churn.

You have no data to make or support this assertion, it is likely to sink any points you might have when made to people (in this case Linden Staff) who actually do.

1 minute ago, Extrude Ragu said:

Considering how much SecondLife struggles to keep new users as it is,

Again, you're not in a position to assert this.

1 minute ago, Extrude Ragu said:

don't you think that blaming new users,

Fictitious new users are not new users.

1 minute ago, Extrude Ragu said:

for not knowing how bad SecondLife's search system is

Opinion.

1 minute ago, Extrude Ragu said:

, is ultimately going to lead to SecondLife's demise?

Chicken little.

 

  • Like 1
Link to comment
Share on other sites

4 minutes ago, Extrude Ragu said:

Actually the answer to that question I think is pretty simple.

Don't rely on a computer to do a humans job - If a user follows through on a result and clicks teleport, chances are that result was relevant to the users query. Use that to help inform the search results.

Google does that, but it's not just click-throughs. They won't help. It's how long it is before the person is back in the search - how long they stayed at the place. LL could possibly incorporate something along those lines into the engine.

Link to comment
Share on other sites

3 minutes ago, Coffee Pancake said:

No, you very much aren't.

This is argumentum ad populum.

You have no data to make or support this assertion, it is likely to sink any points you might have when made to people (in this case Linden Staff) who actually do.

Again, you're not in a position to assert this.

Fictitious new users are not new users.

Opinion.

Chicken little.

 

It really does not take a statistician to realize that users who receive low quality results first are less likely to click through to high quality results than those who receive high quality results first.

To discredit this on the basis of a lack of statistics is disingenuous at best.

It is a bit like saying we have no data to prove that people are more likely to get wet when it rains, therefor it is not true until proven. Do you really need that data?

  • Like 3
  • Haha 1
Link to comment
Share on other sites

6 minutes ago, Drake1 Nightfire said:

This search design is supposed to appeal to the new generation? They dont even want to watch YouTube videos over 10 minutes in length. Unless its clips of vine, tik tok, or fails. A 30 minute how to on how to get better usage from your Iphone? never gonna watch it. "ugh, just give me the fast way to fix it. or better yet, do it for me."

 

Every day a new generation comes into existence. I can't raise them all and I can't force the ones that already do exist to be decent human beings.

Link to comment
Share on other sites

Just now, Extrude Ragu said:

It really does not take a statistician to realize that users who receive low quality results first are less likely to click through to high quality results than those who receive high quality results first.

But framing search dissatisfaction in these terms and expecting people with such data to take you seriously is not going to go the way you want.

Just now, Extrude Ragu said:

To discredit this on the basis of a lack of statistics is disingenuous at best.

 

38 minutes ago, Extrude Ragu said:

This results in users thinking that no high quality anime sims in SecondLife exist, and quitting,

uqlCjJs.png

  • Like 1
Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 1026 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...