iStockPhoto Best Match

iStock's best match was a really good idea. The idea that you can search on any subject and you will get the "best" results is fantastic for designers. But in practice, does it work.

No, not really.

First of all, iStock doesn't define the "best". That's because if they did everyone would try and cheat the system to try and get it to work in their favour. When it was discovered previously that ratings played a good part in getting files on the first page there was the discovery of "ratings gangs" that were small groups of people all rating each others files with 5's in order to get them to the top of the searches. Needless to say, iStock took action and dealt with the people that were playing the system.

Then we had another reprisal, but this time it was discovered that high downloads per month would push your file to the front. So then it was discovered that there were a few people who were working together in the same studio (or even the same person?) using the same props, same camera etc, both exclusive, who were buying each others files in order to push them up to the front page. Of course, this was very clever when you think about it. If an exclusive file comes through the queue in two days after it's upload, and you buy it asap, then it's going to have an instant download per month figure of around 28, which is higher than most files ever reach. This instantly pushed them near the front of the search.

That's a really good way of doing the best match, since images that are being bought on a regular basis are surely going to be better than average. But without meaning to, it puts non-exclusives at a disadvantage. It takes non-exclusives a week on average to make it through the inspection queue. That means that even if it was bought instantly after it was approved (this does happen legitimately!) then it would only ever get a downloads per month figure of around 4, which is very low compared to the 28 figure that is possible for exclusives! In many peoples opinion iStock should set the downloads per month recording from the moment it is inspected, not from upload. Of course exclusives should get perks, and they do get many. But files in a "best match" search should compete against other similar files on equal footing. Most designers will not care if a file is exclusive or not, they just want a file that suits their needs. iStock is shooting themselves in the foot by doing this. There are many exclusive photographers on iStock churning out mediocre work (the same as many non-exclusives) and photos should be judged on their own merits alone. If designers want exclusive images they can use the exclusive only search.

Of course another problem with the best match is that the key-wording on iStock is pretty damn awful! If I search for "doctor" then the first result I get is a stethoscope! A stethoscope is not a doctor, and already I bet we've already put a few users off. A doctor is also not a nurse or an abstract background or a computer keyboard. So in order to clean the searches up iStock introduced the keyword wiki. Fantastic idea! You simply click the key-words of a file, deselect the irrelevant ones and then it gets sent off to Calgary HQ for an admin to look at and clean up. Or so it should. The problem is, there are not enough staff to deal with the sheer amount of badly key-worded files! If I remember rightly, this was introduced around 18 months ago. When it first started I had some images that were placed reasonably well in a search for "rock climbing". What was holding them back though was pictures of kids climbing trees and climbing plants that were in front of them. So I wiki'd everything in the "rock climbing" search that was irrelevant. They've still not been done. I've even had some of my own files under keyword review (which I changed the keywords on as soon as I knew they were incorrect) for 14 months now. 14 months is totally unacceptable to wait for the keywords to be corrected. All this time the keyword spammers are benefiting from the exposure at the front of searches that they should be in.

The other huge problem with keyword spam, is that half of it isn't spam! When disambiguation was first brought in, those with huge portfolios were promised bulk disambiguation tools. We were told well over a year ago that they were almost ready for release. Guess what? No one has seen them! Also, if you search for an ambiguous term, for instance "orange" you get the choice of colour or fruit. The searches quite often default to one or the other, which actually penalizes users that correctly disambiguate to the correct tag! Spammers will show up in both, therefore searchers won't bother to check off a term. The search needs to show everything or nothing, not decide for itself what the user is searching for. Another good one, was when you used to search for "London" the default was "London, Ontario". I bet more people around the world, if searching for "London" would want the UK city!

Sour grapes? Maybe. But there seems to be alot of people saying that the best match doesn't work as it should.