Author Topic: More screwed up Best Match (Read 15588 times)

zorki · « **on:** November 12, 2007, 11:41 »

Anyone else following the discussion over at IS about the Best Match search engine. http://www.istockphoto.com/forum_messages.php?threadid=60039&page=1

Try typing in "christmas tree" into the search and use Best Match. Looks like a couple of uploaders hit the jackpot!

Also, Sean Locke mentions the "ratings gangs" that you all dismissed as not having an effect on the best match searches.

zorki · « **Reply #1 on:** November 12, 2007, 11:59 »

Even worse, do a search on "blueprints" and tell me the search engine hasn't been hijacked... I wish I knew how he did it!

massman · « **Reply #2 on:** November 12, 2007, 12:38 »

I doubt if IS will do anything about it, they very rarely do. Files I came across and reported for keyword spam over a year ago still contain spam, most of the files I spotted were exclusive IS big hitters. Maybe thats another perk you get when you go exclusive.

zorki · « **Reply #3 on:** November 12, 2007, 12:45 »

I don't know this time, some of the biggest uploaders to IS are pissed. Just interesting to see where it goes... I'm trying to read the whole thread and it looks like a couple of guys sharing equipment are uploading together and then to game the system, they are downloading each other's images as soon as they get uploaded. That seems to trick the best match search and put them on top...

Kngkyle · « **Reply #4 on:** November 12, 2007, 15:16 »

I sure as hell hope they fix it.

madelaide · « **Reply #5 on:** November 12, 2007, 15:57 »

I wonder is they will ever clarify this matter.

Is what we do here also considered "gang rating"? It seems the problem there, given what I read so far, is a fixed number of people rating each other images (each person rating all new uploads by the other member). Here we do a more modest thing, link A rates one of B, B rates one of C, etc. I often rate a couple of images in my network's latest uploads. I just don't think the rating should have a huge weight, especially in something called "best match". Keywords and ratings can't "match", unless the problem is that main keywords are used in comments and repetition makes them more relevant.

Regards,
Adelaide

Kngkyle · « **Reply #6 on:** November 12, 2007, 16:01 »

Quote from: madelaide on November 12, 2007, 15:57

I wonder is they will ever clarify this matter.

Is what we do here also considered "gang rating"? It seems the problem there, given what I read so far, is a fixed number of people rating each other images (each person rating all new uploads by the other member). Here we do a more modest thing, link A rates one of B, B rates one of C, etc. I often rate a couple of images in my network's latest uploads. I just don't think the rating should have a huge weight, especially in something called "best match". Keywords and ratings can't "match", unless the problem is that main keywords are used in comments and repetition makes them more relevant.

Regards,
Adelaide

No, the main problem is people downloading their own images/having someone download their images as soon as they are approved to get them to the top of the best search result.

They are basically investing the $1 or whatever it costs for the smallest download and getting in return higher commission through the canister levels, and more downloads on their images because they are at the top of the results. It makes sense, but it is really abusing the system and I truly hope Istock fix it so it is fair for everyone.

madelaide · « **Reply #7 on:** November 12, 2007, 16:09 »

Oic, the problem is not just the rating, but the sales raising levels and commissions.

Do dlds count in best match too?

Regards,
Adelaide

zorki · « **Reply #8 on:** November 12, 2007, 16:51 »

After reading through this whole thread at the IS forum it appears that the problem is someone (or two) with two different account uploading images and as soon as they get approved, someone downloads a copy which makes it rise to the top of best match searches. There also seems to be some type of correlation between the size of the image (XXL) and how it ranks in best match searches. They are also rating the images as soon as they get approved. All of this adds up to unfair rankings in best match.

Susan S. · « **Reply #9 on:** November 12, 2007, 16:57 »

Downloads per month seem to be very important in best match - so if you get a download in the first hour or so of the image being up, the downloads per month goes through the roof and your image sits at the top of the best match. (I've had it happen to a couple of images of mine that were quickly bought - not by me I should add! so I've known about the impact already). It's much more important than ratings (which currently seem to have very little effect) and other factors, although other things clearly come into it. Obviously if the file is no good then no one will buy the image and it will gradually fall back. But for the generic good stock that is being uploaded in this case (with lots of dreadful keyword spam too) the continuing downloads that result will keep them up in the first page.

A couple of people (or one person using more than one account) are gaming this system to great effect.

hatman12 · « **Reply #10 on:** November 12, 2007, 18:47 »

Well the offending photographers have had their images removed and accounts suspended. Swift and decisive action by iStock.

madelaide · « **Reply #11 on:** November 12, 2007, 18:50 »

Quote from: zorki on November 12, 2007, 16:51

There also seems to be some type of correlation between the size of the image (XXL) and how it ranks in best match searches.

It's another funny thing to be in a "best match". If size is not required by buyer in the search, why should it be important? Most of my sales in IS are XS, S and M, so it doesn't seem L (I don't have anything larger) is in huge demand.

Regards,
Adelaide

leaf · « **Reply #12 on:** November 12, 2007, 18:56 »

Quote from: hatman12 on November 12, 2007, 18:47

Well the offending photographers have had their images removed and accounts suspended. Swift and decisive action by iStock.

good to see they took decisive action

madelaide · « **Reply #13 on:** November 12, 2007, 19:00 »

Quote from: Susan S. on November 12, 2007, 16:57

Downloads per month seem to be very important in best match - so if you get a download in the first hour or so of the image being up, the downloads per month goes through the roof and your image sits at the top of the best match.

Interesting, and that may explain why some images of mine perform so well when they are downloaded quickly. Right now for instance this one, which is nothing extraordinary, appears well when searching for "two orchids" or "orchids satin". And it hasn't received any rating.

Regards,
Adelaide

leaf · « **Reply #14 on:** November 12, 2007, 19:07 »

Quote from: madelaide on November 12, 2007, 19:00

Quote from: Susan S. on November 12, 2007, 16:57
Downloads per month seem to be very important in best match - so if you get a download in the first hour or so of the image being up, the downloads per month goes through the roof and your image sits at the top of the best match.

Interesting, and that may explain why some images of mine perform so well when they are downloaded quickly. Right now for instance this one, which is nothing extraordinary, appears well when searching for "two orchids" or "orchids satin". And it hasn't received any rating.

Regards,
Adelaide

it has a rating now mwwwhhaaa ha ha ha.... your plan is foiled

madelaide · « **Reply #15 on:** November 12, 2007, 19:50 »

Quote from: leaf on November 12, 2007, 19:07

it has a rating now mwwwhhaaa ha ha ha.... your plan is foiled

Be careful. You may be banned.

Regards,
Adelaide

fotografer · « **Reply #16 on:** November 13, 2007, 01:45 »

Quote from: zorki on November 12, 2007, 16:51

. There also seems to be some type of correlation between the size of the image (XXL) and how it ranks in best match searches

I think that was just because these people or person was uploading xxl images so as they had so many images on the first page it looked like xxl was being given priority. So don't all rush out and get a camera expecting to get on the first page

GeoPappas · « **Reply #17 on:** November 13, 2007, 07:40 »

Quote from: hatman12 on November 12, 2007, 18:47

Well the offending photographers have had their images removed and accounts suspended. Swift and decisive action by iStock.

Wow. That is pretty unbelievable. The contributor that was at the top of the Best Match for "blueprint" was an exclusive contributor and had over 6000 downloads! I can't believe that they banned one of their own.

The ironic thing is that the same clumping of images by contributor happens at Dreamstime in their Best Match sort order (which is called Relevancy), but nobody seems to care. For example, if you search for "blueprint" on Dreamstime the same sort of clumping by contributor appears. Why is it that if it happens on one site people go crazy, but if it happens on another site people don't care???

sharply_done · « **Reply #18 on:** November 13, 2007, 12:55 »

If they gamed the system so that they had only a dozen or so images on the first page (instead of the entire page) they would probably have never been caught.

Just like a typical 'heist' flick, it was greed that killed them.

digiology · « **Reply #19 on:** November 13, 2007, 13:23 »

I am sure it's only the tip of the iceberg.

leaf · « **Reply #20 on:** November 13, 2007, 14:02 »

Quote from: GeoPappas on November 13, 2007, 07:40

Quote from: hatman12 on November 12, 2007, 18:47
Well the offending photographers have had their images removed and accounts suspended. Swift and decisive action by iStock.
Why is it that if it happens on one site people go crazy, but if it happens on another site people don't care???

well i think the istock crowd is a bit more reactive. Perhaps since people are exclusive with istock as well they take things like that a little more personally

yingyang0 · « **Reply #21 on:** November 13, 2007, 14:22 »

Someone finally capitalized on a flaw that has been obvious since they lowered the ratings value in best match searches and they banned them? Even though they hijacked the best match results I don't think they should have been banned. People are always trying to find ways to "beat the system", in every endeavor. As devious as it may have been, this seemed like it was within the rules.

I know you can't download your own images but is there a rule that you can't purchase a friends?

StockManiac · « **Reply #22 on:** November 13, 2007, 14:41 »

Quote from: yingyang0 on November 13, 2007, 14:22

Someone finally capitalized on a flaw that has been obvious since they lowered the ratings value in best match searches and they banned them? Even though they hijacked the best match results I don't think they should have been banned. People are always trying to find ways to "beat the system", in every endeavor. As devious as it may have been, this seemed like it was within the rules.

I know you can't download your own images but is there a rule that you can't purchase a friends?

I have to agree with you. They should have just fixed the best match algorithm so that this doesn't happen anymore. They change the algorithm every month, what's one more time?

fotografer · « **Reply #23 on:** November 13, 2007, 14:46 »

Quote from: GeoPappas on November 13, 2007, 07:40

The ironic thing is that the same clumping of images by contributor happens at Dreamstime in their Best Match sort order (which is called Relevancy), but nobody seems to care. For example, if you search for "blueprint" on Dreamstime the same sort of clumping by contributor appears. Why is it that if it happens on one site people go crazy, but if it happens on another site people don't care???

I think that this is because on DT it is because of the way the best match works rather than manipulation by contributors. The DT relevancy search definitly seems to favour images by photographers with a high Dls/image ratio and high acceptance rate so you find a lot by one particular contributor on the first page of the search.

GeoPappas · « **Reply #24 on:** November 13, 2007, 15:13 »

Quote from: fotografer on November 13, 2007, 14:46

I think that this is because on DT it is because of the way the best match works rather than manipulation by contributors.

In both cases, the best match was working as designed. They didn't manipulate the best match algorithm on IS. They just figured out how the best match worked and then used that to their advantage.

But in either case, the bottom line is that clumping of images occurs on both sites. I highly doubt that is what a buyer is expecting when they do a search.

Quote from: fotografer on November 13, 2007, 14:46

The DT relevancy search definitly seems to favour images by photographers with a high Dls/image ratio...

I don't believe that is true. I have a very high DL/image ratio (~ 7:1) and most of my images don't show up anywhere near the 1st page of an image search. My DL/image ratio used to be higher, but it has taken a plunge since they changed the best match algorithm the last time.

DT is now making very little for me from month to month. I just hope that they will come to their senses and change the best match to something that makes more sense.

fotografer · « **Reply #25 on:** November 13, 2007, 15:29 »

My dl/ image ratio is over 9 and in one search which produces 2500 results 12 of the first 13 are mine. I can't think of any other reason they are there other than a high dl/ image ratio. My acceptance rate is about 81% which is fairly high but not over high so if it's not that then I don't know what it is as some of the images haven't had many dls yet.

madelaide · « **Reply #26 on:** November 13, 2007, 15:53 »

Quote from: GeoPappas on November 13, 2007, 07:40

Why is it that if it happens on one site people go crazy, but if it happens on another site people don't care???

If the same is happening in DT, or elsewhere, why not report it to the staff?

Regards,
Adelaide

Susan S. · « **Reply #27 on:** November 13, 2007, 16:00 »

Quote from: yingyang0 on November 13, 2007, 14:22

Someone finally capitalized on a flaw that has been obvious since they lowered the ratings value in best match searches and they banned them? Even though they hijacked the best match results I don't think they should have been banned. People are always trying to find ways to "beat the system", in every endeavor. As devious as it may have been, this seemed like it was within the rules.

I know you can't download your own images but is there a rule that you can't purchase a friends?

The suspicion was strong, given that the photographers concerned used the same subject matter props and models, that they were in fact the same individual - which is a direct violation of istock policy. (and infact you can purchase your own file - not that I have done so - one of the irritations to some people about istock is that if you want a copy of one of your own files for some reason, unlike some other sites, you have to purchase it.)
Given their sales numbers, I was suprised that they were outright banned so quickly - one has a little suspicion (but of course it is only a suspicion) that when their accounts were looked at more closely there were other things wrong. To be honest they were rather silly downloading every single image they uploaded straight way. It makes it very obvious.

zorki · « **Reply #28 on:** November 13, 2007, 16:55 »

IMHO, Best match searches should be based on keywords, otherwise make it completely random. They should offer a true Best Match, Random, Newest, Download for the search engine. Far too many people use Best Match to make their decisions on what to use and to include criteria like age of the file, upload acceptance ratio, how quick someone downloads, Exclusivity, etc, etc into the best match search is just going to cause too many problems and open it up to manipulation.

GeoPappas · « **Reply #29 on:** November 13, 2007, 17:57 »

Quote from: fotografer on November 13, 2007, 15:29

My dl/ image ratio is over 9 and in one search which produces 2500 results 12 of the first 13 are mine. I can't think of any other reason they are there other than a high dl/ image ratio. My acceptance rate is about 81% which is fairly high but not over high so if it's not that then I don't know what it is as some of the images haven't had many dls yet.

I believe that it is your high acceptance rate that is getting your images to the front page. 81% is quite high. There was a thread where contributors stated their acceptance ratio and the average was around 62%. Here is the thread for reference: http://www.microstockgroup.com/index.php?topic=1177.0

hatman12 · « **Reply #30 on:** November 13, 2007, 19:12 »

Acceptance rate has already been confirmed by Achilles as being part of the search preferences.

DL/image has not been confirmed, but it would be logical to include it because it hints at quality/popularity.

yingyang0 · « **Reply #31 on:** November 13, 2007, 22:13 »

Quote from: Susan S. on November 13, 2007, 16:00

The suspicion was strong, given that the photographers concerned used the same subject matter props and models, that they were in fact the same individual - which is a direct violation of istock policy.

Ah...didn't know that.

Quote from: Susan S. on November 13, 2007, 16:00

....To be honest they were rather silly downloading every single image they uploaded straight way. It makes it very obvious.

Common trait of people who game (or cheat) systems, they are greedy and their greed leads to their demise.

fotografer · « **Reply #32 on:** November 14, 2007, 01:01 »

Your probably right then. I have the same acceptance rate on IS and when there was a thread about it I didn't seem to do very well against lots of the people posting their rates so I assumed that DT was the same but then maybe only those with very high rates bothered posting.

Quote from: GeoPappas on November 13, 2007, 17:57

Quote from: fotografer on November 13, 2007, 15:29
My dl/ image ratio is over 9 and in one search which produces 2500 results 12 of the first 13 are mine. I can't think of any other reason they are there other than a high dl/ image ratio. My acceptance rate is about 81% which is fairly high but not over high so if it's not that then I don't know what it is as some of the images haven't had many dls yet.

I believe that it is your high acceptance rate that is getting your images to the front page. 81% is quite high. There was a thread where contributors stated their acceptance ratio and the average was around 62%. Here is the thread for reference: http://www.microstockgroup.com/index.php?topic=1177.0

zorki · « **Reply #33 on:** November 14, 2007, 09:27 »

Quote from: hatman12 on November 13, 2007, 19:12

Acceptance rate has already been confirmed by Achilles as being part of the search preferences.

DL/image has not been confirmed, but it would be logical to include it because it hints at quality/popularity.

If that's the case, I sure wish I could start over again! My early rejections are dragging down my percentage!

yingyang0 · « **Reply #34 on:** November 14, 2007, 13:49 »

Quote from: hatman12 on November 13, 2007, 19:12

Acceptance rate has already been confirmed by Achilles as being part of the search preferences.

DL/image has not been confirmed, but it would be logical to include it because it hints at quality/popularity.

I was not aware that acceptance rate was part of the best match search, however DL/image was confirmed and has been part of the best match for as long as I've been at iStock. When did Achilles say that acceptance rate was part of the best match search?

There's no economic justification for including acceptance rate (not that I care because my 80.46% is good) because it doesn't impact relevance of the image to the search, nor does it reflect popularity of the image.

fotografer · « **Reply #35 on:** November 14, 2007, 14:03 »

Achillles is DT not istock
[/quote]
I was not aware that acceptance rate was part of the best match search, however DL/image was confirmed and has been part of the best match for as long as I've been at iStock. When did Achilles say that acceptance rate was part of the best match search?

There's no economic justification for including acceptance rate (not that I care because my 80.46% is good) because it doesn't impact relevance of the image to the search, nor does it reflect popularity of the image.
[/quote]

digiology · « **Reply #36 on:** November 14, 2007, 15:03 »

Just discovered more best match mysteries. Search for the word "trashed" and under best match the same artist appears several times on the first page but sorted under downloads and the results are far more varied. If D/L per image is taken into account for best match then shouldn't the results be simular to the download sort?

zorki · « **Reply #37 on:** November 14, 2007, 15:24 »

I'm sure IS has their hands full with this problem. They will be trying to undo all the people trying to take advantage of the situation until they figure out what they are going to do with best match. The only two big stock houses that I hear people consistently complain about search and that is IS and Alamy...

yingyang0 · « **Reply #38 on:** November 14, 2007, 17:23 »

Quote from: digiology on November 14, 2007, 15:03

Just discovered more best match mysteries. Search for the word "trashed" and under best match the same artist appears several times on the first page but sorted under downloads and the results are far more varied. If D/L per image is taken into account for best match then shouldn't the results be simular to the download sort?

Sorry, I thought we were talking about dl/mo.

digiology · « **Reply #39 on:** November 14, 2007, 18:01 »

Quote from: yingyang0 on November 14, 2007, 17:23

Quote from: digiology on November 14, 2007, 15:03
Just discovered more best match mysteries. Search for the word "trashed" and under best match the same artist appears several times on the first page but sorted under downloads and the results are far more varied. If D/L per image is taken into account for best match then shouldn't the results be simular to the download sort?
Sorry, I thought we were talking about dl/mo.

Oh were we...sorry I missed that.

Then I guess that makes more sense why the sort results differ.

sharpshot · « **Reply #40 on:** November 15, 2007, 10:48 »

So have they changed the best match again now? My sales were near new highs but have crashed again the last few days.

pixelbrat · « **Reply #41 on:** November 15, 2007, 10:54 »

Yep, they changed it. My sales have gone down slightly too but I'm going to give it a few more days to see if it could just be ebb and flo. I sure will hate losing my record high sales if this is the cause...

All part of the game I guess.

zorki · « **Reply #42 on:** November 15, 2007, 12:28 »

That's funny... I've had a couple of really nice days after a few weeks of crap... sometimes I feel like the Wizard of OZ is behind the curtain pulling random strings trying to get things right, but never quite gets it right...

MicrostockGroup Sponsors

Author Topic: More screwed up Best Match (Read 15588 times)

digiology

digiology

digiology

Related Topics

Sponsors

Microstock Poll Results

Sponsors