icon Sqweegee's Blog
It’s Rank! Alamy's Punters Get Their Algorithms In A Twist

Be careful what you wish for. Professional contributors to Alamy have long complained about the size of the agency’s seven million plus picture collection, and in particular the perception that their work is buried among other, lesser, images in search returns. For over a year now rumours have circulated about the agency’s plan to introduce a new ranking system, which many felt would go some way to solving the problem of search returns. Put simply, the good stuff – that’s all of mine – would rise to the top in the new system, while the bad stuff – that’s all of yours – would sink to the bottom.

The launch of AlamyRank in early October – like much else about Rank the exact date has become obscured by the fuss surrounding it – did indeed produce an immediate and dramatic change in search results: but not what many had anticipated. Imagine the reaction of those who discovered that Alamy’s new algorithms hold them in considerably lesser regard than they do themselves, as their pictures were overnight moved so far down the search results as to become invisible at the bottom of a 10 million strong pile.

It’s been extremely hard to find any professional Alamy contributor with a good word for Rank: reactions have ranged from initial nervousness through increasing concern to outright fury and rather entertaining paranoia. But professionals bitch about everything, and anyway they’re not the only – perhaps not even the majority – contributors to Alamy: a substantial number of the agency’s 15 million images are produced by amateurs.

So in the interests of balance EPUK sent a reconnaissance team into the world of the Alamy Lists, a universe with an atmosphere so rarefied, inhabitants so alien and opinions so fantastical that an Alamy executive once commented that he could only bear to watch while hiding behind the settee. Like the agency itself, the lists are populated not only by professionals – hence the rather avant-garde opinions – and may therefore give a more accurate picture of how all Alamy contributors view the changes.

Sadly the Alamy lists seem to have had an attack of normality since we last visited. There was a disappointing lack of discussion on how to fiddle taxes, fake model releases and cheat on keywording: Rank however had certainly made an impact. Almost inevitably the ill-named AlamyPro list launched a poll on members’ opinions of the new ranking system. Given that it’s far too early for Rank to actually do anything this was a pretty pointless exercise: rather like reviewing a car you haven’t driven or a movie you haven’t seen yet.

Nonetheless the poll is revealing insofar as contributors at AlamyPro are so far largely in favour of the new system: 64% for and 26% against. That’s certainly the opposite of elsewhere, but not especially surprising given AlamyPro’s membership profile. What was surprising was that the vote certainly didn’t reflect the comments being made at the forum, which were mostly unfavourable.

It was largely the same elsewhere: contributors weren’t just upset, they took the changes distinctly personally, which is pretty odd when what we’re talking about here are computer algorithms. Some even took to poring over the Rank patent application seeking clues on how to outwit the cunning geekery. “So much for my keywording” was a common refrain, ignoring the possibility that perhaps the keywording in question was the source of the problem. And of course in true elephant in the living room fashion there was one thing no photographer dared say out loud: perhaps, after a nine-month trial by people who actually buy photographs, Rank had concluded that their pictures sucked.

Not that Alamy themselves are blameless for the some of the confusion: contradictory statements from the company only served to heighten the bewilderment and frustration evident amongst contributors. Alamy CEO James West has on more than one occasion publicly stated that tools will be made available for contributors to tweak their keywording and improve their ranking to enable their work to fight its way up past 100 million others. However when one photographer inquired about these recently Member Services told him that there was no such tool, but that, rather helpfully, his suggestion for one would be passed on to the relevant department.

And the company even seemed unable to decide whether Rank is now complete, or still a work in progress. On the same day last month the same member of staff managed to tell one contributor that Rank was the final version, then told another that it wasn’t. This being the internet, within hours both contributors – and a few thousand of their closest colleagues – had read both versions and were left scratching their heads.

But in the long term the most interesting recent development at Alamy may turn out to be something entirely different from the new search algorithms. In the midst of all the ranking around one Alamy contributor apparently confused Alamy’s homepage with Google when looking for an alternative agent and searched ‘Corbis’. Guess what popped out of Rank? Tens of thousands of images apparently from Corbis, but hosted and sold by Alamy.

The sudden appearance of huge numbers of images from a supposedly rival agency would certainly help explain Alamy’s rapid expansion to a collection of 500 million images, but is also likely to give individual Alamy contributors further cause for concern.

Of course whether or not anyone will be able to find any of those Corbis images amongst Alamy’s one billion strong collection is another matter.


Comments on this article:

I’d like to point out that at alamy there are also many Getty collections (Medioimages and Stockbyte). There are also many Jupiter collections too (Creatas, Hemera and Bananastock).

Comment #1 posted by Steve Laverga at 17 November, 03:41 PM

The discussion of the AlamyRank patent application was relevant in that the current version of AlamyRank has little in common with what’s described in the application. The algorithms described in the patent application probably would tend to produce more relevant results but all it takes is a few trial searches to see how bad AlamyRank, as released, is.

I can only think that they decided to rush a not completely baked AlamyRank out the door since it has already been long-delayed. That might also explain Alamy employees own confusion about what exactly it is that we have.

Comment #2 posted by Mark at 17 November, 08:24 PM

I don’t understand the epuk outrage at Alamy Rank. If you truly have a specialist collection with accurate keywording and your pictures have visual values that appeal to buyers then you will go up in the rank. If you have a vast collection of many near duplicates of images that alamy’s client don’t want to see or buy then you will go down in the rank.

The truth is that before Alamy Rank each contributor was assigned a rank by a personal judgement of how competant the contributor was. Now the market is defining itself. It was Margaret Thatcher who used to like saying, You can’t buck the market.” Incidently Milton Friedman died today; maybe he coined the saying I don’t know. I didn’t like Thatcher and I’m not sure what I think about Friedman but stock photography is about the market or it’s about nothing. And Alamy is a brilliant democratic concept so much better than the Getty concept of cornering the market to attempt to buck it. And I don’t see it is reasonable for so called specialist suppliers to seek to buck it either. And when it really comes down to it if these ‘specialists’ were so wonderful their pictures would speak for themselves and they’ld be at the top of the rank. It ill behoves them to poke fun at the contributors to AlamyPro because while many are pretty gauche and silly many are not. Moreover what really drives them is a love of the medium. And like the market I have a whole lot more time for those that love photography than for those that love being photographers.

Alamy Rank is democratic. You self select your place in the rank by the quality/saleability of your pictures. What exactly is there to complain about.

Comment #3 posted by Stephen Oliver at 17 November, 11:43 PM

If indeed it worked as claimed, Stephen, I would agree with you but, frankly, some of the results are just plain strange. When obviously wrongly-keyworded images are coming up at the tops of searches (and, yes, amongst some good stuff, as well), it’s not remotely working as claimed. Ranking per image per keyword, which is what the patent describes, makes sense but what we’ve been given is not that.

If I were, say, a professional tennis photographer, do you think I’d have reason to wonder what’s going on if a picture of a bowl of strawberries came up thousands of images ahead of mine in a search? I don’t think “oh well, my images are crap” would be my response.

Comment #4 posted by Mark at 18 November, 03:32 PM

A search on the word “tennis”, I should have said.

Comment #5 posted by Mark at 18 November, 03:34 PM

“So in the interests of balance EPUK sent a reconnaissance team into the world of the Alamy Lists”

EPUK goes incognito into limited membership forums with confidentiality rules & then reports finding for all to read…? Limey mate, I’ll go back to mashing me tea.

Comment #6 posted by Jeff Greenberg at 19 November, 03:41 AM

Mark, Alamy rank could justifiably rank a picture of a bowl of strawberries higher than images of tennis, if a search for ‘tennis’ was made.

Because it is the buyers who determine what is and isn’t relevant, its entirely possible that during Wimbledon, searches for tennis end up with a picture researcher buying a picture of a bowl of strawberries to illustrate ‘Its Wimbledon time’

ie: Your common sense idea about the relevance of a keyword is not what Alamy rank (or Google’s pageRank for that matter) is about. The buyer of the images effectively define the meaning of ‘relevance’. As a result there will always be strange and intriguing rankings in a system driven by algorithms of this type. Photographers might not like it, but the buyers probably will.

Comment #7 posted by Paul Freeman at 1 December, 06:37 PM

I may be a rarity amongst alamy members, but I’ve found that the alamy ranking system has moved all of my pictures up to the front or near front pages. How this will affect sales I’ve yet to see, though November returns were very good. Unless picture buyers have a habit of drilling down way past the early pages of a search, I am most content with the changes. Just thought I’d throw this in to provide some balance.

Comment #8 posted by Rob Walls at 2 December, 02:16 AM

It’s not just ranking that has changed. It’s that single images appear with a ‘more’ option. Sods law states that the likelihood of the system choosing a relative weak picture is near 100%. This is the diversity algorithm. I can see the sense of it but it is likely to hit sales of those with large collections compared to the previous system.

Comment #9 posted by Ian Murray at 2 December, 12:50 PM

The Alamy archive looks like it has been hit by a tsunami of amateurish offal: you can spot a real photograph now and again, but make sure you grab it before you are swept along.It is hard to believe that a busy art director has time to go wading through all that.
If the best images have shot up the rankings then the criteria must be invoked by some visual illiterate.It takes just as long to find a decent image on page 1 as page 101 in many cases.Well over 90% of Alamy’s images are poorly composed, poorly exposed and poorly conceived images that look like amateur rubbish.Professional photographers need somewhere to sell their images without being exploited by getty or whoever (in some deep recess of my mind I seem to remember this was the cornerstone upon which Alamy was founded – must be going senile). Why dilute the resource of creative professionals (which Alamy is lucky to have) with a load of rubbish that could only be conceivably chosen by an art director as visually incompetent as the “photographer” who bothered to send it in or the “editor” who improved its ranking? That isn’t to say amateurs can’t produce great images or even be great photographers, just as professional plumbers don’t have a monopoly on talent in that area! However, for every Alamy amateur eager to have his/her moment of fame there is a true professional with quality images just giving up on the whole amateur way this industry is being run.What is the point of allowing your agents to release your images to Alamy? This completely undermines any strength of negotiation a photographer may still hold.Let’s not forget that this is precisely how the getty monopoly was created. Are you sure you are not helping Alamy achieve a similarly exploitative goal? Before we can all move on we need to eliminate the intermediary parasites, creaming off up to 30% of the sale price for doing absolutely nothing that you cannot do for yourself directly with Alamy.Perhaps if the staff at Alamy had to deal with photographers, they may be forced to learn something about what they are surfing over!

Comment #10 posted by James at 30 December, 07:13 AM

Its a cheat solution to their reluctence to address both their strengths and their weaknesses. CEO says that searches go 20 pages in. Like bollocks they do. Therefore rank is unimportant (ergo therefore no need to rank at all… although actually they always did). So they set up Alamyrank where initially everybody must be ranked AT THE START (God knows how) and that defines the silliness of it from the off. Either rank figures or it does not. And their initial rank will inevitably define their subsequent rank forever there after… Since the very conception of Alamyrank is an admission that search rankings make a difference to search results and sales. Alamy try to have it both ways and they suck. Either rank is important or it is not. If it is important then Alamyrank is concieved on false premises (since initial rank will define subsequent rank) because it takes no account of the image iteself (on a website selling images).... er… only keywords matter…. er …. keywords provided by the contributor who is totally unregulated. If rank is not important then why bother?.... The people at Corbis and Getty must be wetting themselves with laughter as Alamy go in to full on foot shooting mode…. Nice idea…. Pity they f***ed it up by not actually considering what they sell…. DUH!.... IMAGES!..... What we used to call photographs.

Comment #11 posted by Jerome at 31 December, 08:32 PM

Things have gotten bad at Alamy. They’d be wise to consider to convert their entire catalogue to microstock (something along the lines of IStock)

Comment #12 posted by Susan at 4 January, 10:10 PM

Almay has a blog now, and it looks like their contributors are not happy with the Rank system.

Comment #13 posted by Ted at 24 January, 02:27 AM

“It takes just as long to find a decent image on page 1 as page 101 in many cases.”

Dear James:
Please provide some examples.

Comment #14 posted by EkS at 8 February, 10:35 AM

I heard Corbis might buy Alamy and use it to get into the microstock business. Any truth to this?

Comment #15 posted by Leslie at 10 April, 11:57 PM

This whole thing is strange!
50% of the pix on alamy may not be very good but who should decide which 50% is which.
The market thats who!
OOPS is that the time I must dash and catch the bank as another alamy cheque has just arrived.

Comment #16 posted by dave barrett at 21 April, 08:59 PM

Looks like the whole industry is going discount. First Getty, and now Alamy,

Comment #17 posted by Neil at 22 September, 08:41 PM

Echoing Dave Barrett’s “Kerching!”…The market decides what is good. A single picture last month netted me over$4,500 on Alamy.

Comment #18 posted by Rob Walls at 25 May, 01:37 PM

Add your comments here:
(Your email will not appear on your comment, but we cannot accept your comment without one. We won't give your email address to anyone else)
(You don't need to enter your website address, but if you do, your post will link to it)

EPUK reserves the right to edit or delete posts which the moderators feel are irrelevant, offensive, libelous, untrue or just plain nutty; and in extreme cases, to ban those who make them.


sqweegee byline