Oh, its too easy to game the system. At least for the last several rounds, I'm pretty sure most of the contributors just go around and downvote everyone elses submissions, which really screws up the vote. Having up and down votes does make it a little chaotic, and its pretty hard to really judge whether the final vote is truly accurate or not. Add to all of that the simple fact that votes get locked in, and there have been some cases where I voted early, but wished I had not...
That just leaves an inaccurate voting record in place.
What we really need are poll topics.
But, I think that is decidedly against the nature of SE.
@jrista Aren't the factors you're talking about worse with small sample sizes? The more picture submissions and the more voting we get, the more this stuff evens itself out, right?
@DLambert Not really...some people are honorable, and don't vote everything else down. It just creates a messy skew, and isn't really a fair voting system for that kind of thing.
@jrista Yeah, I agree there are people who game the system. I don't feel strongly enough about this to haul out my statistics books, but it seems like given a large enough voting population, though, this would start to become pretty insignificant relative to "real" votes. If we've got weeks where two pics are neck-and-neck, this could be a factor, but if there's any sort of gap at all between #1 and #2, I guess I wouldn't sweat it.
2
For the record, I voted "up" on two pics last week, and voted "down" on zero. That's pretty typical - I think I usually vote up about twice as many as I vote down.
@ElendilTheTall There are a couple I think are good, and a couple I think are bad, but nothing that stands out as a clear winner yet. I'm waiting to vote to see if one grows on me. ;-)
I can't get over the people who keep asking for "larger versions". We're not voting on the larger versions, people! What point is there, exactly, in voting for (or against) a photo that looks great at a totally different size than the intended display size?
2
Gosh, I'm really sorry about the quality -- it looks much better in the version you're not going to look at all week.
<rant />
@ElendilTheTall BTW, yours survived resizing pretty well, I think.
So, just about two minutes after I went ahead and awarded the bounty on my Q, I get a banner popping up to say, "Your bounty on question 'X' expires in the next 24 hours." Thanks for that.
"Its [sic] a nice cityscape, but not sure if I'd like to look at it for a whole week.".... "It's a nice frog, but not sure if I'd like to look at it for a whole week." "Its a nice squirrel, but not sure if I'd like to look at it for a whole week." "Its a nice street, but not sure if I'd like to look at it for a whole week."
Oooh! Did you see Jeff Atwood's comment on the gear reviews post?
"of course it would be fantastic if this community gained that sort of recognition, but we needn't wait for that to happen before embarking on this experiment using SE, Inc funding."
@ElendilTheTall I think SE is a big site, but I'm curious how P-SE stacks up against other photo sites. The vote counts we're getting on questions sure don't give the impression that people are busting the doors down.
Doesn't do anyone any good to have the best review in the world that nobody's going to see.
(and yes, I know it's not nobody - just trying to make a point)
For the record, I think reviews could be one of those missing pieces, though, that drives even more traffic to the site, so with some luck (and good management), it'll be a self-reinforcing cycle of growth.
@rfusca @jrista: the company behind it is located not too far from our HQ, and being a bunch of photo nerds we invited him over for lunch. He showed off the prototype for us.
@DLambert Well, from a statistics standpoint, we would need a LOT more than we have now. When your talking hundreds to just a few thousand, thats usually considered a small pool. A couple thousand participants is usually the bare minimum to get a marginally accurate statistical result. Statistical accuracy is ultimately derived from extremely large pools of participants.
Take some of the newer online music sites for example. They have tens of thousands to millions of participants, each of whom have the ability to vote on music quality, genera, similarity to other tracks, etc. The statistical deviation of such an endeavor with say only 10,000 people choosing those factors, vs. 10 million people choosing those factors, is huge...10 million people will produce a far more accurate aggregate result.
In the case of our PotW contest, we don't even have 100 participants, let alone 1000.
kinda goes along the lines of what I brought up in the other post of 'how would we be different/better' - but he just doesn't think a blog review would be better than anybody else's at all from the sound of it
@rfusca It depends on the number of things being rated. You wouldn't get accurate statistics with music tracks with only a couple thousand.
For PtoW, you would if we had a few thousand.
When it comes to say political polls, the bare minimum pool size is about 1500 (a lot seem to involve around 1800 participants), and the results are not 95% statistically accurate.
@JayLancePhotography Thank god you have the ability to cancel stars now. ;)
well obviously it depend on what confidence interval you pick. But if you want like 95% with an accuracy of like +/- 2 of your measure, then you only need a couple thousand to measure a couple of million
I was discussing the music rating and relating issue with some old colleagues of mine a while ago.
With a pool of only a few thousand, we determined that the relationships that people developed between music tracks (i.e. Band A Song 1 is very similar to Band D Song 4, both have a genera of X, and both have M, N, O types of musical qualities) were very poor.
I think we determined that we needed at least 50,000 participants to get sufficiently accurate results.
however, if the participants were only picking genera, a few thousand would do fine
wtf is up with people being so damn touchy lately?! Have a little backbone. If you get into a disagreement, make your case, and move on. Don't go crying home.
So, not even considering the votes, I think the POTW comments are demonstrating that we wouldn't work as a photo critique site.
And I'm not even complaining about the comment on mine (although I think it a funny thing to say). Overall, that's some really poor commentary. The bees are fuzzy! Nothing is off-center properly. I don't like the corners.
Well... It's just not that interesting right now... What I'm struggling with is that there are three things you could do, push in, pull out, change the angle, or some combination of those... Since I don't know the location you're shooting at, it's hard to know which of those would yield an improvement...
man, @Elendil, you've been getting slammed with your photo this week....I wonder what peoples issue is with it. I think the exposure is spot on for the type of shot...
A challenge with this style of photography (same with @ElendilTheTall's Stone Dragon in the potw thread) is that you're starting with someone else's artwork. Then, you have to decide if it's your intention to document, to make a derivative new artwork, or to make something transformative.
I think the least interesting part of the picture is the wall around the sculpture, so i'd be thinking in terms of 'how can I zoom in such that I can completely eliminate the wall.'
The round lines of the shield and the highlights on the torso guide the viewer quickly to the face, and then it's natural to follow the eyes. But they go nowhere.
@francesco you can also right click the new layer, choose add layer mask, click ok, then paint with a black brush to remove parts of the layer non-destructively
