Author Topic: iStock Location Aware search test (Read 16000 times)

travelstock · « **on:** August 12, 2011, 05:04 »

There's a theory that keeps popping up in discussions of iStock's Best Match, that it now includes location data, and that the results you get will vary depending on where you're searching from.

I suggested in another thread that we try to do a small test to see whether this is in fact the case, and if so to get an idea of how much difference there is. I had some response to this, so thought I'd post another thread to see if we can organise a time to try it out, and hopefully get some responses from different locations.

What I suggest is that we set a common time (early next week?) and common search strings, then post a screenshot of the top results for each and see what comes out.

Any takers for participating & suggestions on search strings?

robynmac · « **Reply #1 on:** August 12, 2011, 05:06 »

Count me in. I'm in Australia, and would be very interested to see the results of such a test.

gclk · « **Reply #2 on:** August 12, 2011, 05:18 »

Hi Holgs, good idea - happy to participate too, I'm in the UK. Maybe we should do the test both logged in and logged out to iS.

iStock may be interested to see the results too

franckreporter · « **Reply #3 on:** August 12, 2011, 07:11 »

i'm in, tell me what and when !

I'm in italy

travelstock · « **Reply #4 on:** August 12, 2011, 07:37 »

I'll still be in Spain BTW...

Maybe we should aim for Monday, say: 12 midnight Sydney time, 9pm Bangkok, 4pm Berlin, 3pm London, 10am New York, 8am Calgary.... (+/- 1 hour?) seems to be a time when its not unreasonable anywhere. I think emptying cache and doing a search both logged in and out coule be good. Anyone have any searches they want to suggest? Could go for some generic ones like "Classroom", "Business" etc. (but not too many!)

Opla · « **Reply #5 on:** August 12, 2011, 07:45 »

Will be interesting, I'm in from Holland.
I suggest "pregnant" and "harbor"

gclk · « **Reply #6 on:** August 12, 2011, 09:35 »

Maybe we should include generic search terms which could give local results.

Eg in Spain (and much of the world) results for 'car road' might show more cars driving on the right, whereas in Australia/UK/etc, buyer search behaviour may have made the results skewed towards cars driving on the left.

leaf · « **Reply #7 on:** August 12, 2011, 09:45 »

sounds like a fun idea. I can do the search from Canada. It may be hard to coordinate so many time zones at a 'good' time.

SNP · « **Reply #8 on:** August 12, 2011, 09:52 »

Tyler - where are you in Canada? I'm in Toronto, I would participate in this too...just out of curiosity

Eyedesign · « **Reply #9 on:** August 12, 2011, 10:32 »

Count me in, Berlin!

traveler1116 · « **Reply #10 on:** August 12, 2011, 11:43 »

Sure I'll try, near Washington DC here. I think we should search for "cityscape" or something that should get very different results if the location aware features are working well. Right now a lot of my top results are for US cities NY, Chicago, Atlanta I would be curious if a search in England came up the same.

luissantos84 · « **Reply #11 on:** August 12, 2011, 12:02 »

yep! Portugal will do it too

travelstock · « **Reply #12 on:** August 12, 2011, 13:38 »

Quote from: traveler1116 on August 12, 2011, 11:43

Sure I'll try, near Washington DC here. I think we should search for "cityscape" or something that should get very different results if the location aware features are working well. Right now a lot of my top results are for US cities NY, Chicago, Atlanta I would be curious if a search in England came up the same.

Actually that's really revealing - logged in I get Australian results, logged out I suddenly have Madrid and Barcelona in the mix...

reckless · « **Reply #13 on:** August 12, 2011, 13:45 »

Northeast central Colorado, USA

RapidEye · « **Reply #14 on:** August 12, 2011, 14:30 »

South Africa here. I'm in, as long as the time isn't too crazy.

ETA: Whoops, missed holgs's time proposal. That'll suit me fine.

cathyslife · « **Reply #15 on:** August 12, 2011, 14:37 »

What will also be interesting is the results you get today, compared to results you may get tomorrow or next week, for exactly the same searches.

Anyka · « **Reply #16 on:** August 12, 2011, 15:25 »

Count me in for Belgium. How shall we be reporting?

gostwyck · « **Reply #17 on:** August 12, 2011, 16:03 »

Whatever 'information' you derive from this exercise is only going to be a snapshot of how the system may work (or not) on the day that you do it. Istock may be continually refining the system in the same way that they do with the best match. And, just like the best match, even if you were aware of the factors, the weighting and the effect ... there's still very little if anything that you can do about it to improve your own sales.

If you do find out that Aussie buyers get different results to Spanish buyers how can you amend your workflow to take advantage of it?

madelaide · « **Reply #18 on:** August 12, 2011, 16:44 »

Brazil.

Should we search in our language? It could be interesting to test words that have different meaning in different places (like football).

qwerty · « **Reply #19 on:** August 12, 2011, 20:37 »

cityscape search I get a mix but Australian shots up higher in best match than that I thought.

How come when you put the price slider down to minimum you still get Vetta photos ? (edit driver error) it jumped back to maximum without my intevention

RacePhoto · « **Reply #20 on:** August 13, 2011, 10:49 »

I'm in Central States USA.

Please find something that should bring topical results by area, the car idea was good, fashion, food? I don't care.

And please keep it simple, not all kinds of sliders and agencies specials, vetta or sorting, just what the average person dropping into the site would see when they search for a three or four word phrase.

I wonder if this will bugger up their count when suddenly people from all over the world start searching for "sliced tomato isolated" LOL

At IS HQ: "Hey look we need more sliced tomatoes shots there's a flood of searches coming in from around the world.

bunhill · « **Reply #21 on:** August 13, 2011, 11:10 »

There would (likely I should guess) be other factors apart from geo location which would potentially reasonably result in subtle differences even at the same moment.

Don't mess with the magic , I say

bunhill · « **Reply #22 on:** August 13, 2011, 12:32 »

^ what I mean is that there is no way of excluding other potentially unknown factors and therefore experimenting only with geo location - because other potentially unknown factors are unknown. So the experiment cannot demonstrate anything for sure.

Anyhow the best recipes are a secret. Lobster à la Riseholme etc.

RacePhoto · « **Reply #23 on:** August 13, 2011, 15:33 »

Quote from: bunhill on August 13, 2011, 12:32

^ what I mean is that there is no way of excluding other potentially unknown factors and therefore experimenting only with geo location - because other potentially unknown factors are unknown. So the experiment cannot demonstrate anything for sure.

Anyhow the best recipes are a secret. Lobster à la Riseholme etc.

True, you can't control the other invisible variables, but there is one that can be tested and that is location. That's the theory, that location affects the best match results. It's isn't difinitive and if the best match changes according to the second of the minute, then nothing will ever be known,. We'll be like Heisenberg or who was that other guy? Schrödinger's theory represented the observational data by means of continuously evolving causal processes in space and time.

So lets just look at what we can see at that moment, we may not know how fast it's changing or where it will be in a few minutes, but we have results from assorted locations around the world, at a small window of time.

Hey, what if all of us get the same results? Then it's just as good of a proof that location doesn't change anything!

So the experiment is better done, than saying, it's impossible, it's secret and lets go play outside in the rain and forget about testing anything?

Yeah, on the other one, Nina's (an Indian Restaurant Manager) "Tandoori Chicken", there's one I'd like to know how to make. She broght it in partially prepared, with the seasoning in a baggie, so we never knew how she made it.

cathyslife · « **Reply #24 on:** August 14, 2011, 07:24 »

Quote from: RacePhoto on August 13, 2011, 15:33

Quote from: bunhill on August 13, 2011, 12:32
^ what I mean is that there is no way of excluding other potentially unknown factors and therefore experimenting only with geo location - because other potentially unknown factors are unknown. So the experiment cannot demonstrate anything for sure.

Anyhow the best recipes are a secret. Lobster à la Riseholme etc.

True, you can't control the other invisible variables, but there is one that can be tested and that is location. That's the theory, that location affects the best match results. It's isn't difinitive and if the best match changes according to the second of the minute, then nothing will ever be known,. We'll be like Heisenberg or who was that other guy? Schrödinger's theory represented the observational data by means of continuously evolving causal processes in space and time.

So lets just look at what we can see at that moment, we may not know how fast it's changing or where it will be in a few minutes, but we have results from assorted locations around the world, at a small window of time.

Hey, what if all of us get the same results? Then it's just as good of a proof that location doesn't change anything!

So the experiment is better done, than saying, it's impossible, it's secret and lets go play outside in the rain and forget about testing anything?

Yeah, on the other one, Nina's (an Indian Restaurant Manager) "Tandoori Chicken", there's one I'd like to know how to make. She broght it in partially prepared, with the seasoning in a baggie, so we never knew how she made it.

Unlike istock, I'll bet Nina makes her secret recipe exactly the same way every time. After all the screw-ups with the istock site and their IT department, I find it hard to believe that anything that's being done with their search algorithm is actually planned. Or did everyone forget about that? It's not that it's still not happening...it's just that the current trend is to speak highly of istock so that it appears all is getting better and improving. You know, like the ostrich syndrome...stick your head in the sand and pretend all is well.

MicrostockGroup Sponsors

Author Topic: iStock Location Aware search test (Read 16000 times)

travelstock

robynmac

gclk

franckreporter

travelstock

Opla

gclk

leaf

SNP

Eyedesign

traveler1116

luissantos84

travelstock

reckless

RapidEye

cathyslife

Anyka

gostwyck

madelaide

qwerty

RacePhoto

bunhill

bunhill

RacePhoto

cathyslife

Related Topics

Sponsors

Microstock Poll Results

Sponsors