Re-rank

[Cryptoprocta]

41 minutes ago, Reimar said:

I was going to re-do my search with creative and RF added as you had done.  Before that, I refreshed my search for "relevant" and RM at 120/page and now find the image on page 8.

I'm sure it wasn't there before.  I think something at Alamy is trying to mess with my head.

Ah, I just left the default settings, as although I have only RM photos here, I very seldom see [RM] specified in a search. Also, I'm searching on 100/page, which is why I see it on p10.

[Martin P Wilson]

I gave up on BHZ a year or more ago as my results bore no relation to real world searches. My BHZ were on the last few pages but on my more realistic test searches I have pictures on the first page of much larger search results than BHZ. In fact I am no 8 on the first page for one search (close for others) as opposed to being page 20+ for BHZ  (100/page). Doesn't seem to make much differences to my sales results, or zooms, sadly .

[BobD]

12 minutes ago, Martin P Wilson said:

I gave up on BHZ a year or more ago as my results bore no relation to real world searches. My BHZ were on the last few pages but on my more realistic test searches I have pictures on the first page of much larger search results than BHZ. In fact I am no 8 on the first page for one search (close for others) as opposed to being page 20+ for BHZ  (100/page). Doesn't seem to make much differences to my sales results, or zooms, sadly .

 

Me too for the same reason.

The BHZ results bear no relation to normal searches for a few reasons.

1. the contributors that play BHZ if a very small proportion of Alamy's total contributors you will normally compete with.

2. Some who play the game use normal tags some use super tags.

3. some contributors BHZ tag more than 1 image in the same pseudo.

So all in all the results are a complete distortion of normal searches.

 

There are more important things to worry about. 

[Niels Quist]

2 minutes ago, BobD said:

 

Me too for the same reason.

The BHZ results bear no relation to normal searches for a few reasons.

1. the contributors that play BHZ if a very small proportion of Alamy's total contributors you will normally compete with.

2. Some who play the game use normal tags some use super tags.

3. some contributors BHZ tag more than 1 image in the same pseudo.

So all in all the results are a complete distortion of normal searches.

 

There are more important things to worry about. 

 

Couldn't agree more.     Life is way too short 

[Martin P Wilson]

Alamy has always insisted BHZ is pointless so it wouldn't surprise me if if Alamy tweaks the results for BHZ so as not to give anything away about the ranking algorithm, I probably would

[Matt Ashmore]

I do wonder if Alamy have dropped the 'big bang everything changes overnight' re-ranks and implemented more of a 'steady drift over time' approach with images changing position based on new algorithms which also aim to rotate images a bit.

[NYCat]

A lot of people on page 1 have put bhz in their caption and that will cause you to shoot to the top if it is in addition to a super tag. I tried it. I have taken it out of the caption to see something I think is probably more realistic for me. I think Alamy now has ways to consider subjects. In areas where I have sold images I get better placement.

 

Paulette

[Cryptoprocta]

5 hours ago, BobD said:

 

Me too for the same reason.

The BHZ results bear no relation to normal searches for a few reasons.

1. the contributors that play BHZ if a very small proportion of Alamy's total contributors you will normally compete with.

2. Some who play the game use normal tags some use super tags.

3. some contributors BHZ tag more than 1 image in the same pseudo.

So all in all the results are a complete distortion of normal searches.

 Additionally, two other factors affect real searches:

1. Caption outweighs keywords in search (I've seen a file with 0 keywords at #1  of a several hundred pic result )

2. diversity algorithm outweighs accurate keywording.

This has been the case for a long time, but I've just been analysing 'leopard on road' and that clearly shows up both of the above. Alamy says (to parpahrase what they have said) that the keyword 'leopard' will show above the keyword phrase 'leopard tortoise', but in fact, as leopard tortoise is correctly in the caption, and the search algorithm isn't sophisticated enough to distinguish, then leopard tortoises are dotted through the search, along with bharals (on  a snowy crag, not a road) which are the prey of snow leopards and were photographed on a 'road trip', ('snow leopard' an 'road trip' in keywords though not in images) Florida Leopard Frogs, a woman wearing a spotty blouse, cyclists from a team with Leopard in their name, a town which featured in a book about a leopard, the Leopard Inn (I suspect the new system split a lot of keywords on that file) a bus with leopard in its model name ... and two genuine pics of leopard on road are at the very bottom of the search.

Must be frustrating for buyers. And also for suppliers. We know from Measures that a proportion of buyers don't look past their first page of images (often 100, which is the default). If these irrelevant files were removed, probably that particular search would be <100, so all relevant files would be seen.

[Allan Bell]

 

Just so that everyone including contributors new to Alamy are using the same search criteria.

 

Search BHZ (bhz)

• RM only
• small images
• 120 images per page
• Relevant (Actually there does not seem to be any difference between relevant and creative but there is a big difference if you click and search on NEW.

There should be 2,921 images in brackets at left side. Unless any others decides to join in the fun.

 

Allan

 

 

[Avpics]

2 hours ago, Allan Bell said:

 

Just so that everyone including contributors new to Alamy are using the same search criteria.

 

Search BHZ (bhz)

• RM only
• small images
• 120 images per page
• Relevant (Actually there does not seem to be any difference between relevant and creative but there is a big difference if you click and search on NEW.

There should be 2,921 images in brackets at left side. Unless any others decides to join in the fun.

 

Allan

 

 

And they also need to add BHZ as a supertag and nowhere else if they want to get an idea of their rank. As Martin and others have pointed out, there are so many factors that will skew the results but it should give people an idea of where they are in comparison with any other contributors who have followed the same 'rules'. Of course it shouldn't be taken too seriously, but it's about the only method of gauging where you are against some form of level playing field. If you search using keywords that apply to some of your images it'll put you high on the list if you're particularly skilled in captioning and keywording - ie, exactly what you want in the real world - but it won't give you an idea of your rank.

[Allan Bell]

2 minutes ago, Avpics said:

And they also need to add BHZ as a supertag and nowhere else if they want to get an idea of their rank. As Martin and others have pointed out, there are so many factors that will skew the results but it should give people an idea of where they are in comparison with any other contributors who have followed the same 'rules'. Of course it shouldn't be taken too seriously, but it's about the only method of gauging where you are against some form of level playing field. If you search using keywords that apply to some of your images it'll put you high on the list if you're particularly skilled in captioning and keywording - ie, exactly what you want in the real world - but it won't give you an idea of your rank.

 

Whoops! Yes, I had forgotton that point. Well spotted. (Just like your images.)

 

Allan

 

 

[Cryptoprocta]

11 minutes ago, JeffGreenberg said:

WAIT !!!

Alamy tells us repeatedly:  BHZ use-less...

What is the convincing counter-evidence???

That's why it's the BHZ-game, not the BHZ-proven-science.

It only shows our relevant rank for BHZ, which might be an indication of our overall rank.

If most of the people above us shoot totally different subjects, we may be higher for any given search.

I have searches where my highest file in the search is on the top line, and others (sadly, including in my favourite genre) where my highest-ranked file is pages down.

[Martin P Wilson]

1 hour ago, Avpics said:

And they also need to add BHZ as a supertag and nowhere else if they want to get an idea of their rank. As Martin and others have pointed out, there are so many factors that will skew the results but it should give people an idea of where they are in comparison with any other contributors who have followed the same 'rules'. Of course it shouldn't be taken too seriously, but it's about the only method of gauging where you are against some form of level playing field. If you search using keywords that apply to some of your images it'll put you high on the list if you're particularly skilled in captioning and keywording - ie, exactly what you want in the real world - but it won't give you an idea of your rank.

 

If my images are well captioned and appear high for real world searches why should I be concerned about some artificial measure of rank (whatever it is) that does not match what I see with real world searches?

 

It is of marginal, passing interest only. So I junked BHZ, I can find plenty of other things to waste my time on

[Reimar]

1 hour ago, JeffGreenberg said:

WAIT !!!

Alamy tells us repeatedly:  BHZ use-less...

What is the convincing counter-evidence???

Well, it's been quite useful to me over the years, and remains a tool I may use in the future for testing, although lately the relative rank from BHZ has been very stable (thank goodness).

I'm convinced page position is PROPORTIONAL to rank based on my three pseudo results.  It's not rank per se, which we will never get, but it's close enough to gauge relative position movements.  For example, it's been useful to check the impact of caption, supertags and tags and combinations of these.  This was the case during a transition period after Alamy "adjusted" the search algorithm.  Less so now.