| Author | Thread |
|
|
02/23/2005 06:54:35 PM · #101 |
Originally posted by kirbic:
Originally posted by scottwilson:
If anyone thinks this is a just a coincidence then you are simply not looking at the data. This photo got 3.33 times more 1 votes then it did 3 votes. It is hard to find any photo that gets more 1 votes then three votes and certainly not a photo that scores higher then a 5. A ratio of 3.33 to 1 of off the scale, it never happens.
This is not a simple case of artistic differences. |
The logic here is seriously flawed. There are plenty of photos (and high scoring ones) that get more one votes than 2s or 3s. Why? If the photo generates a wide variety of opinion, the normal distribution of it's votes gets "chopped off" at on end, or the other, or both. The votes simply pile up in the last bin. You can find this effect on the high end as well, but noone ever complains about that, LOL.
I have done graduate level work in statistics, so I know enough to be dangerous. I have never found evidence of more than a few "spurious" votes on any image I've analyzed, including the one currently in question. |
The only photos that I have seen even come close to this are ones with a very controversial subject and even then I don̢۪t remember any that were this heavy with ones and light on 3s and 4s.
I make my living in large part in mathematics including statistics, but I don̢۪t think it takes a mathematician to see that something is very odd here.
|
|
|
|
02/23/2005 07:00:06 PM · #102 |
That's it's "odd" is in no way evidence of cheating, and that's what you're saying is going on.
Over in the other thread someone was nice enough to analyse the votes, and it was still statistically "normal" which means it's evidence of nothing.
The photo won a ribbon! |
|
|
|
02/23/2005 07:06:25 PM · #103 |
| Can't we all just get along? |
|
|
|
02/23/2005 07:10:08 PM · #104 |
Judging by the paranoia in here there's some seriously good dope going round. Who's going down the late shop for munchies?
|
|
|
|
02/23/2005 07:14:47 PM · #105 |
Originally posted by scottwilson:
I don̢۪t think it takes a mathematician to see that something is very odd here. |
To a mathematician it would look odd. Math can not account for human emotion/opinion/actions... |
|
|
|
02/23/2005 07:19:28 PM · #106 |
Originally posted by GeneralE:
That's it's "odd" is in no way evidence of cheating, and that's what you're saying is going on.
Over in the other thread someone was nice enough to analyse the votes, and it was still statistically "normal" which means it's evidence of nothing.
The photo won a ribbon! |
That is very odd is evidence that something is going on, but then we don't have to stop there, there are more tests that can be done to see if something very odd is going on.
As for looking that the Standard deviation, that is a weak test, for example if a photo got only even votes, 2s, 4, 6.. etc the SD could look very normal but still this distribution would not be expected. The same holds true when looking at the 1 vote, 10 1 votes will not have that much impact on the SD, but still it can be way outside the norm for a voting distribution.
|
|
|
|
02/23/2005 07:26:20 PM · #107 |
Originally posted by scottwilson:
Originally posted by GeneralE:
That's it's "odd" is in no way evidence of cheating, and that's what you're saying is going on.
Over in the other thread someone was nice enough to analyse the votes, and it was still statistically "normal" which means it's evidence of nothing.
The photo won a ribbon! |
That is very odd is evidence that something is going on, but then we don't have to stop there, there are more tests that can be done to see if something very odd is going on.
As for looking that the Standard deviation, that is a weak test, for example if a photo got only even votes, 2s, 4, 6.. etc the SD could look very normal but still this distribution would not be expected. The same holds true when looking at the 1 vote, 10 1 votes will not have that much impact on the SD, but still it can be way outside the norm for a voting distribution. |
Something going on... like there are 10 people on this site that are sticklers on titles and didn't like that the photo had a sunset in it while it was titled "MIDNIGHT Mist"??? |
|
|
|
02/23/2005 07:34:06 PM · #108 |
So I propose a simple check, of the people who voted it a 1 how many have ever voted in a challenge before, this would tell us a lot. If it turns out that say 8 of the 10 1 votes were from people who never voted before then what do we say? If it turns out that most of the people who voted it a one registered at about the same time then what?
I will give you a scenario, if it turned out that all 10 one votes were from people who never voted on a challenge before and all 10 registered within the same two hour period would it still seem like a simple matter of artistic taste?
What I am saying is that this pattern is off enough that it deserves to be looked at a bit more.
|
|
|
|
02/23/2005 07:34:20 PM · #109 |
Thats not a sunset.
Its glow from the massive toll area on the Staten Island side. They have huge flood lights in that area, and the light is bouncing off the low level fog and mist.
|
|
|
|
02/23/2005 07:35:31 PM · #110 |
Originally posted by scottwilson:
Originally posted by GeneralE:
That's it's "odd" is in no way evidence of cheating, and that's what you're saying is going on.
Over in the other thread someone was nice enough to analyse the votes, and it was still statistically "normal" which means it's evidence of nothing.
The photo won a ribbon! |
That is very odd is evidence that something is going on, but then we don't have to stop there, there are more tests that can be done to see if something very odd is going on.
As for looking that the Standard deviation, that is a weak test, for example if a photo got only even votes, 2s, 4, 6.. etc the SD could look very normal but still this distribution would not be expected. The same holds true when looking at the 1 vote, 10 1 votes will not have that much impact on the SD, but still it can be way outside the norm for a voting distribution. |
Scott, I'm interested to see what you make of my analysis in the other thread. For anyone interested in a more in-depth look at the statistics of DPC voting, here is the other thread.
|
|
|
|
02/23/2005 07:44:37 PM · #111 |
Originally posted by GeneralE:
That's it's "odd" is in no way evidence of cheating, and that's what you're saying is going on.
|
I wouldn't call it cheating... I would just call it Trolling. Simply people "thinking" they will boost their score by voting a 1 on the photo they think will ribbon. Its childish. and it happens. No way that photo deserved 10 1's. I think the voting system is fair, but even the constitution gets amended once in a while.
|
|
|
|
02/23/2005 07:50:26 PM · #112 |
Originally posted by aerogurl:
.. but even the constitution gets amended once in a while. |
You get my post-of-the-day award :)
|
|
|
|
02/23/2005 07:51:54 PM · #113 |
Perhaps there is a bug in the vote recording mechanism.
This image has not 1 but 2 significant statistical anomalies.
First, it has an extraordinary number of 1s for its placing. You have to go down to around the 200th place to find one with more 1s. Most images above that one are not even close, including the two preying mantis images that certainly could be open for interpretation as meeting the challenge where you would expect more 1s.
Second, most images have 340-350 votes. This image has 10 more votes than the second highest vote total I saw on any other image and I looked at a lot.
One statistical anamoly maybe, but two is to much to ignore. Every vote an image gets is associated with voter. It is possible to list out and report on that data.
This needs to be investigated to prove there are no errors in the system.
|
|
|
|
02/23/2005 07:59:15 PM · #114 |
Originally posted by stdavidson:
Perhaps there is a bug in the vote recording mechanism.
This image has not 1 but 2 significant statistical anomalies.
First, it has an extraordinary number of 1s for its placing. You have to go down to around the 200th place to find one with more 1s. Most images above that one are not even close, including the two preying mantis images that certainly could be open for interpretation as meeting the challenge where you would expect more 1s.
Second, most images have 340-350 votes. This image has 10 more votes than the second highest vote total I saw on any other image and I looked at a lot.
One statistical anamoly maybe, but two is to much to ignore. Every vote an image gets is associated with voter. It is possible to list out and report on that data.
This needs to be investigated to prove there are no errors in the system. |
I agree with you 100%. If the site council were to investigate this matter, we would all benefit in one of two ways: 1. we could dispel any myths associated with the issue at hand or 2. if there were an issue with multiple accounts, falsifying of votes or whatever, it would be discovered.
I'm interested to see if they will investigate this matter.
Deb |
|
|
|
02/23/2005 08:09:14 PM · #115 |
| Hey..so maybe I could of got a blue..Im happy with my red, and if this matter is investigated, then Im all for it. However..I doubt it will, and as long as the majority of the people here at DPC continue their support of one another and continue their constructive comments, then I can't complain. If I would of gotten 20 or more one's. Then I would be real angry. |
|
|
|
02/23/2005 08:10:25 PM · #116 |
Even if it turns out that the non-standard low votes and non-standard number of votes were within the statistical deviation, and no abuse of voting going on, I still think the voting scale could use some tinkering (read: make simpler) to make it fairer.
Why depend on chance on balance out high-average-voters and low-average voters. Why not fix the scale?
|
|
|
|
02/23/2005 08:13:20 PM · #117 |
This is one topic that refuses to go away. It is a riddle in its own right. At one time I got caught up because I saw evidence wherein the low vote affected the outcome. This is not apparant all the time but only when close fraction separate the top five. Now, I factor in these low votes into my calculations.
Often times the suspicion is foul play and in some cases it is. This gives both sides of the argument merit. Personally, I have stopped using one's and two's. I believe the penalty is too high and proves nothing more than the fact that I have the freedom to issue the, "this image stinks so foul that only a 1 or 2 can do it justice."
Yes, I know, others feel justified in attacking the offensive image, but give no reason why a better image is treated the same way.
This notion that your vote can derail a majority is faulty because the system survives and grows stronger with each additional vote. When you are over 300 votes, your vote whether high or low is a drop in the bucket. Either a very high vote or low vote may have a fractional effect to move one image above the other, but the results are never what you expect them and they never follow a fixed pattern. Examples are the less observant voters who are suddenly surprised to see an image they did not acknowledge win. Even if one knew the sequence, no vote after 300 will have a predictable effect.
After examination I simply vote on the merit of the image and let the inevitable course take its stride. Better to devote energy in creating a better image for the next challenge than to be occupied playing futile games.
Message edited by author 2005-02-23 21:28:16. |
|
|
|
02/23/2005 09:40:57 PM · #118 |
| Well congrats rulerzigzag - up until now the only one with the power to polarise the voters to this extent was drJones, and you didn't even need to involve nudity :) |
|
|
|
02/23/2005 09:42:34 PM · #119 |
Originally posted by vfwlkr:
Even if it turns out that the non-standard low votes and non-standard number of votes were within the statistical deviation, and no abuse of voting going on, I still think the voting scale could use some tinkering (read: make simpler) to make it fairer.
Why depend on chance on balance out high-average-voters and low-average voters. Why not fix the scale? |
It isn't broken. What could be fairer then 1 - 10? I've heard proposals that change the system by eliminating 1 and 10 and just having it 2 - 9. I've heard others that want it to go from 1 - 100. And still others that have proposed changing it to a non-numerical system. What in reality would any of these proposals accomplish?
Here, I have yet another proposal to alter the system, (and this is pure sarcasm) why don't we change it so that there are no votes at all and everyone who enters gets a ribbon. That way no one would feel bad about not winning and no one would get there feelings hurt.
.
|
|
|
|
02/23/2005 09:55:48 PM · #120 |
Changing the numbers isn't going to change vindictive voters, or voters that can not think outside their pathetic, narrow little tunnel visioned frame of mind. If you go to a 2 to 9 scale, or a 3 to 7 or whatever you want to do, then they will just continue to vote whatever the lowest number is when a photo doesn't rub them the right way. All you'll change is the actual numerical score.
If we all agree that a 10 is saying to the photographer "Wow, friggin' well done, you can't get better!" or something along those lines, then it would stand to reason that a 1 is basically saying to the photographer that your photo really has NO artistic or technical merit whatsoever and you should have refrained from clicking the shutter.
IF we are truly here at DPC to LEARN and improve ourselves as photographers ( since the big cash prizes and car giveaways haven't happened yet ) then maybe it would be reasonable to require some sort of comment on a "1" vote. That way at least the photographer can understand why there are those out there that think their photo is absolute trash.
But, if you do that, who will vote a 1? Surely my 70's challenge pick is REPLETE with 1's and 2's, with no comments to explain why. If they were forced to do so, would they vote that way or cowar off into a corner?
I'm sorry but to vote a photo a 1 when it clearly has at least some artistic merit and has met the challenge does not prove you have different taste, it proves you are a horse's ass and not much else. If you don't like the photo, or are offended by the photo, or thought there was HUGE room for improvement, there is still a 2,3, or 4 to show your disapproval, along with a comment box to help that photographer see the error of their ways and help them get back on the right track.
Okay that's enough |
|
|
|
02/23/2005 09:56:03 PM · #121 |
Its funny no "one" wants a "one", but every "one" wants to be number "one".
Now one is a strange looking word
B-Þ~~~~~~~
I think we should just draw straws....
Message edited by author 2005-02-23 21:57:41. |
|
|
|
02/23/2005 10:04:39 PM · #122 |
I'd like to propose this one more time - I wasn't sarcastic the first time around. Do you think this may reduce trolling occurence or not?
So, how about this solution:
Assuming that the primary reason trolls exist is to downgrade other images once they see their own being bashed, we could achieve multiple often-discussed goals with one action:
- people would have more time to put comments
- people would be more productive at work
- people would not be able to compare their scores to others'
And you get all that by simply not updating results until the end of the voting period. I know, this would take the fun out for some, but please think about it and try to realize what is more important. I personally do not see any disadvantage in not knowing what my score is from minute to minute.
Ask yourselves how much time do you spend looking at the score - we have threads that deal with this topic. If you don't pay attention to this, then you won't mind not having this feature. If you DO spend a lot of time, then you could benefit from this change.
Thoughts?
-Serge
or tell me to zip it up...
Message edited by author 2005-02-23 22:05:32. |
|
|
|
02/23/2005 10:06:29 PM · #123 |
| Probably already been suggested, but I think better photographers should have more influencial votes. |
|
|
|
02/23/2005 11:53:28 PM · #124 |
Originally posted by srdanz:
I'd like to propose this one more time - I wasn't sarcastic the first time around. Do you think this may reduce trolling occurence or not?
So, how about this solution:
Assuming that the primary reason trolls exist is to downgrade other images once they see their own being bashed, we could achieve multiple often-discussed goals with one action:
- people would have more time to put comments
- people would be more productive at work
- people would not be able to compare their scores to others'
And you get all that by simply not updating results until the end of the voting period. I know, this would take the fun out for some, but please think about it and try to realize what is more important. I personally do not see any disadvantage in not knowing what my score is from minute to minute.
Ask yourselves how much time do you spend looking at the score - we have threads that deal with this topic. If you don't pay attention to this, then you won't mind not having this feature. If you DO spend a lot of time, then you could benefit from this change.
Thoughts?
-Serge
or tell me to zip it up... |
I agree 100%. It would definitely be a little less fun, less immediate, but I do think it would help to address the problem. And while it's quite humorous to think that people would gain a lot more personal time not pressing the update button, it's unfortunately, true! And think of the bandwidth savings to the site.
|
|
|
|
02/23/2005 11:59:59 PM · #125 |
Originally posted by srdanz:
I'd like to propose this one more time - I wasn't sarcastic the first time around. Do you think this may reduce trolling occurence or not?
So, how about this solution:
Assuming that the primary reason trolls exist is to downgrade other images once they see their own being bashed, we could achieve multiple often-discussed goals with one action:
- people would have more time to put comments
- people would be more productive at work
- people would not be able to compare their scores to others'
And you get all that by simply not updating results until the end of the voting period. I know, this would take the fun out for some, but please think about it and try to realize what is more important. I personally do not see any disadvantage in not knowing what my score is from minute to minute.
Ask yourselves how much time do you spend looking at the score - we have threads that deal with this topic. If you don't pay attention to this, then you won't mind not having this feature. If you DO spend a lot of time, then you could benefit from this change.
Thoughts?
-Serge
or tell me to zip it up... |
This is one idea I would not object to.
|
|
