DPChallenge: A Digital Photography Contest You are not logged in. (log in or register) 
 

DPChallenge Forums >> Web Site Suggestions >> Stop voting from thumbnails and "cherry picking"
Pages:  
Showing posts 101 - 125 of 125, (reverse)
AuthorThread
 05/16/2007 12:28:27 AM · #101
Originally posted by chimericvisions:


I'll admit I do it on occasion. Many a shot I thought looked interesting from the thumbnail turns out not to be, and vice versa. Scores aren't effected. Quit whining.


Scores are affected. To what extent is unknown. Discussion does not equal whining. Your response *IS* whining. So please can it! ;)
 05/16/2007 12:29:04 AM · #102
Originally posted by DJWoodward:
Here's an evaluation from a recent challenge...

Based on this challenge with 305 submissions there is a positive correlation between the averages and the number of votes as depicted in this scatter plot:

This would confirm in this sample that higher average submission receive more votes.

However before getting too excited we should look at the degree of correlation. The value (r) is called PearsonĂ¢€™s Correlation. A perfect positive correlation equals 1. A scatter plot of a nearly perfect correlation looks something like this

Conversely if there was no correlation r=0 it would look like this


You can see that the actual sample data is somewhere between the perfect and no-correlation extremes. The value (r2) represents the proportion of variance in the average score that can be predicted from the # of votes. The r2 value in the first chart indicates that only 24% of the variance in average score can be predicted from the # of votes. I believe the converse would indicate that 76% of the prediction of average score would be left to other factors.

Based on this it can't be denied that a correlation exists but I suspect that the amount of impact on scores is not as significant as some would fear.

By the way...
In this challenge the highest number of votes was 286 (#5 of 305), the average was 261 and the lowest was 234 (#272 of 305). A spread of 52 votes (20% of the average)

One last factor to consider. There's no way to say that this correlation is due wholly to cherry-picking thumbnails. Even when viewing a full size image taken in random order a voter can skip vote an image for many reasons. Some percentage of the lower vote totals are just be abstentions.

Please remember I don't claim to be a statistician and the conclusions I've drawn are based on a sample from only a single challenge


I think you need a 12-steps Statistics Anonymous program.
 05/16/2007 12:30:01 AM · #103
Originally posted by MikeOwens:


Agreed - whilst this is a site we all learn from it should not be taken too seriously. People whinge and moan too much on this site. Its supposed to be for fun! Relax, enjoy and chill out!
You cannot dictate to people how they vote. We have the site rules to follow and they are sufficient enough as far as I am concerned.

Mike


I wish you guys would quit whining about what you think is whining. At least LeRoy has some cheese to go with his whine.
 05/16/2007 12:31:28 AM · #104
Originally posted by digitalknight:
DPC as a major part of self-validation is a hard life to live.


My life is not usually too hard after I self-validate.......
 05/16/2007 12:34:27 AM · #105
Originally posted by stdavidson:
Please, don't change anything.



This I can agree with. Don't change nuttin. Voting is overrated anyway. Most of the big ribboners don't vote that much anyway. They're too busy shooting!
 05/16/2007 02:10:31 AM · #106
Originally posted by GeneralE:
Originally posted by yanko:
I'd imagine the reason DPC puts them in random order is so every photo gets a lot of votes.

Correct -- it is this goal which cherry-picking the thumbs defeats.


I disagree with that statement completely. "cherry picking" has nothing to do with the fact that if photos were displayed in the order uploaded, the first 20% would have significantly more votes.

The randomization accomplishes a NUMBER of tasks - the aforementioned 20% prevention, largely. It also prevents emotional influencing of voters by the ordering of the photos. (A "bad" photo that is displayed immediately following a "good" photo could receive lower scores than if placed next to an average photo or a worse photo. If it is consistently placed next to the good photo for all viewers, it could yield a lower score than it might otherwise have achieved.)
 05/16/2007 02:17:58 AM · #107
Originally posted by DJWoodward:
One last factor to consider. There's no way to say that this correlation is due wholly to cherry-picking thumbnails. Even when viewing a full size image taken in random order a voter can skip vote an image for many reasons. Some percentage of the lower vote totals are just be abstentions.

Please remember I don't claim to be a statistician and the conclusions I've drawn are based on a sample from only a single challenge


Votes is one data point - what about Views During Voting? I think this would be a much more telling point, because even if a photo is viewed full screen in linear voting, there's always the next button.
 05/16/2007 04:33:15 AM · #108
I posted this as an idea to improve the site NOT to bitch and moan that voting is inherently unfair. It may perhaps cause some shots to get 0.005 or so more score and change the order of the ribbon winners. The site has addressed "Friend Voting" and the WPL for that reason and therefore in the same vein, they should address this issue too. It is good that this thread has brought out so much debate.
 05/16/2007 05:11:25 AM · #109
i don't know if anyone proposed this or not, but what if, when the user clicks on a thumbnail of an image to vote, instead of getting that image, the system served up random 'unvoted' image ;-)
 05/16/2007 05:22:05 AM · #110
Originally posted by Skip:
i don't know if anyone proposed this or not, but what if, when the user clicks on a thumbnail of an image to vote, instead of getting that image, the system served up random 'unvoted' image ;-)


you are kraa-zeee! ;)
 05/16/2007 05:23:10 AM · #111


'scuse me while I talk with my mouthful - I am of the "stop trying to fix what aint broke" camp. Some agree cherry picking affects the score, but also agree that nobody knows how much. If the issue is really worth analyzing (which I don't think it is), I would not focus solely on the scores but the more important issue: number of comments relative to number of votes - because it is the limited exposure to critique that is the only legitimate gripe about cherry picking, IMO. My money is on the fact that the correlation between votes and scores is insignificant and same with number of comments - bad photos get less comments because of fear, not lack of exposure. I think this is all much ado about nothing, but it is mildly entertaining. Carry on.
 05/16/2007 05:24:01 AM · #112
Originally posted by Skip:
i don't know if anyone proposed this or not, but what if, when the user clicks on a thumbnail of an image to vote, instead of getting that image, the system served up random 'unvoted' image ;-)

It was suggested I think in a bulleted suggestion somewhere down there.
 05/16/2007 05:25:30 AM · #113
Making good thumbnails and good photos are different arts. I'm not good at either, but still there are times when I get comments like "The thumbnail totally doesn't do it justice." (Thanks, Sandy)
Cherry-picking does not have major effect final standings but it surely can influence my feeling about my entry. I'm with the fatalists who are here to learn so I need comments (and I mean positive criticism, not "lovely"-type comments).
On the other hand, I also need to look at photos, vote and occassionally comment on them, and read the others' comments later. For this, I vote when I have the time to go through all or most of the images. In full size and in the given order. I don't believe that voting quickly (including but not limited to cherry-picking) can be serious.
 05/16/2007 06:16:26 AM · #114
Originally posted by Art Roflmao:
I think this is all much ado about nothing, but it is mildly entertaining. Carry on.

Yes, yes, yes! Thank you.
 05/16/2007 10:29:39 AM · #115
Originally posted by Spazmo99:
Originally posted by DJWoodward:
Here's an evaluation from a recent challenge...

Based on this challenge with 305 submissions there is a positive correlation between the averages and the number of votes as depicted in this scatter plot:

This would confirm in this sample that higher average submission receive more votes.

However before getting too excited we should look at the degree of correlation. The value (r) is called PearsonĂ¢€™s Correlation. A perfect positive correlation equals 1. A scatter plot of a nearly perfect correlation looks something like this

Conversely if there was no correlation r=0 it would look like this


You can see that the actual sample data is somewhere between the perfect and no-correlation extremes. The value (r2) represents the proportion of variance in the average score that can be predicted from the # of votes. The r2 value in the first chart indicates that only 24% of the variance in average score can be predicted from the # of votes. I believe the converse would indicate that 76% of the prediction of average score would be left to other factors.

Based on this it can't be denied that a correlation exists but I suspect that the amount of impact on scores is not as significant as some would fear.

By the way...
In this challenge the highest number of votes was 286 (#5 of 305), the average was 261 and the lowest was 234 (#272 of 305). A spread of 52 votes (20% of the average)

One last factor to consider. There's no way to say that this correlation is due wholly to cherry-picking thumbnails. Even when viewing a full size image taken in random order a voter can skip vote an image for many reasons. Some percentage of the lower vote totals are just be abstentions.

Please remember I don't claim to be a statistician and the conclusions I've drawn are based on a sample from only a single challenge


I think you need a 12-steps Statistics Anonymous program.


I bet you like baseball.
 05/16/2007 10:48:28 AM · #116
Please explain if I am incorrect in these observations.

1) As I've stated in other forums here, below the DPC logo, it does say "a digital photography contest".

2) As far as I know the goal and requirement of any Contest Judges is to pick the Top-n number of Winners.

Winners meaning, the "Best" out of all the submissions. We have thumbnails to preview first, as do most website galleries of photographs. I have been able to pick the top 3 out of my top 15-20 picks.

This is all very typical, thumbnails and contest results.
So if you want to look good for judges make sure your photo has "Thumbnail Appeal".

I am sure in all contests/auditions some things attract attention, even from an eye-squinting thumbnail distance. We are not judging the 30 x 40 Prints in a Gallery room here. This is the Internet and a very common aspect of that medium of displaying art.

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Explaining the main points of my above previous post...
Maybe there needs to be some ways of increasing commenting on the bottom half of the photos submitted. That's a different issue.

1) A Contest is a Contest.
2) Judges have an obligation to pick the Top-Winners.
3) Voters have a right to vote for the best candidates.

This is not a contest where Top-Winners are picked at random.
This is as fair as most "Real" world situations/assignments

05/16/2007 10:57:20 AM · #117
Judges have an obligation to select the top pictures from among all the entries -- how is that possible if the photos are not viewed on an equal basis?

If Photo A is clearly "better" than Photo B at full-size, but looks positively horrible in thumbnail view, there is a good chance that the "better" photo will not win if judges pick which photos to vote on from the thumbnail page.
 05/16/2007 11:00:24 AM · #118
Originally posted by chimericvisions:
Originally posted by GeneralE:
Originally posted by yanko:
I'd imagine the reason DPC puts them in random order is so every photo gets a lot of votes.

Correct -- it is this goal which cherry-picking the thumbs defeats.


I disagree with that statement completely. "cherry picking" has nothing to do with the fact that if photos were displayed in the order uploaded, the first 20% would have significantly more votes.

The "first 20%" is different for each voter, so that if every voter votes exactly the first 20% (or whatever percentage) of their personal voting queue, all the photos should end up with just about the same number of votes.
 05/16/2007 11:51:57 AM · #119
I swear, I'm desperately trying to think of just how I can take a picture of my thumb for one of the challenges.... And I'll make sure it's very vibrantly colored when I do!
 05/16/2007 11:59:14 AM · #120
300 photos of thumb nails.
 05/16/2007 12:08:44 PM · #121
Originally posted by DJWoodward:
Here's an evaluation from a recent challenge...

Based on this challenge with 305 submissions there is a positive correlation between the averages and the number of votes as depicted in this scatter plot:

This would confirm in this sample that higher average submission receive more votes.

However before getting too excited we should look at the degree of correlation. The value (r) is called PearsonĂ¢€™s Correlation. A perfect positive correlation equals 1. A scatter plot of a nearly perfect correlation looks something like this

Conversely if there was no correlation r=0 it would look like this


You can see that the actual sample data is somewhere between the perfect and no-correlation extremes. The value (r2) represents the proportion of variance in the average score that can be predicted from the # of votes. The r2 value in the first chart indicates that only 24% of the variance in average score can be predicted from the # of votes. I believe the converse would indicate that 76% of the prediction of average score would be left to other factors.

Based on this it can't be denied that a correlation exists but I suspect that the amount of impact on scores is not as significant as some would fear.

By the way...
In this challenge the highest number of votes was 286 (#5 of 305), the average was 261 and the lowest was 234 (#272 of 305). A spread of 52 votes (20% of the average)

One last factor to consider. There's no way to say that this correlation is due wholly to cherry-picking thumbnails. Even when viewing a full size image taken in random order a voter can skip vote an image for many reasons. Some percentage of the lower vote totals are just be abstentions.

Please remember I don't claim to be a statistician and the conclusions I've drawn are based on a sample from only a single challenge


Nice work. As one might suspect, there is a lack of really significant correlation, but not a total lack of correlation. You're also right on the money that the stats don't inform us as to how much of the correlation is due to cherry-picking. Nonetheless, I cannot come up with a hypothesis that supports a positive correlation and does *not* include cherry-picking. The data is suggestive of exactly what the anecdotal evidence suggests... cherry-picking happens, but at a moderate level.

Edit:
I'm coming from an industrial stats background... anything less than about 0.9 for r^2 is a failed hypothesis. In the social sciences, statisticians are often ecstatic to see 0.24; this data definitely belongs in the latter field.

Message edited by author 2007-05-16 12:11:03.
 05/16/2007 12:37:54 PM · #122
Originally posted by stdavidson:
Originally posted by cpanaioti:
Originally posted by stdavidson:
Originally posted by ClubJuggle:
If there was compelling evidence that cherry-picking WAS being used as a cheating technique (as an integral part of one, at least) would that change anything?

~Terry

There is evidence? If so, what is it?


He said IF there was .... ;o)

If there is no evidence then it isn't a discussable point.

When someone proposed that the minimum voting percentage be increased from 20% to 50% to prevent cheating Terry himself argued against it because his studies indicated that more than half the "friend voters" caught had voted more than 50% of the images.


Yes but that wasn't the point ...

He said IF there was .... would it change anything? ;o)

So, would it change anything? Probably not.
 05/17/2007 06:25:04 AM · #123
Originally posted by kirbic:
Nice work. As one might suspect, there is a lack of really significant correlation, but not a total lack of correlation. You're also right on the money that the stats don't inform us as to how much of the correlation is due to cherry-picking. Nonetheless, I cannot come up with a hypothesis that supports a positive correlation and does *not* include cherry-picking. The data is suggestive of exactly what the anecdotal evidence suggests... cherry-picking happens, but at a moderate level.


That is quite correct but that moderate level can still lead to different challenge results as the ribbon winners are usually so close to each other in score. The "Friend Voting" with about 5% of voters affected, was addressed, so why not address this (measurable) score-affecting occurrence? The WPL now incorporated in DPC has forced the change that you cannot vote for your team members to advantage them. Similar issue.
 05/17/2007 06:51:28 AM · #124
I cherry pick. There, I said it. :)

I look at the thumbnails and find one I am intrigued by. I open it up, stare at it for a while, then vote. The next pops up and I vote as a matter of course, same with the next few until I get to one that I really don't like/don't understand where I go back to the front page and pick another one to start at.

Sometimes, I just start at the first and vote for a while, after I have looked over the thumbnails.

Sometimes, its difficult to get through even 20%. The subject just completely bores me or I run into a string of images where my only reaction is why?

Lately I find myself not voting more often than voting. So many people complain about how horrible the voters are - I don't want to be one of those! Same for commenting.

I think, if you look over the forums for the past few months, you will see more complaint threads than thank you threads. Rather depressing. How many times will you put out your hand only to have it slapped before you stop offering your hand?
 05/17/2007 08:54:33 AM · #125
I agree with dahkota. Cherry pick. You can see which images you like from the thumbs and vote those. Besides, everybody hates getting 1-3 votes anyway. Cherry picking stops low voting. Unless you are picking out all the low ones. lol

Message edited by author 2007-05-17 08:55:53.
Pages:  
Current Server Time: 08/06/2025 06:50:17 AM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 08/06/2025 06:50:17 AM EDT.