|
tylerm
May 5, 2004, 3:04 PM
Post #1 of 29
(4295 views)
Shortcut
Registered: Aug 18, 2003
Posts: 153
|
Does anyone know the criteria (if there is one) for getting photos on the front page? I'm just wondering because I would like to get some feedback about my photos...
Tyler
|
|
|
|
|
joe
May 5, 2004, 3:13 PM
Post #2 of 29
(4295 views)
Shortcut
Registered: Jun 22, 2003
Posts: 897
|
hand job.
[billydeewilliams]works every time[/billydee]
|
|
|
|
|
tylerm
May 5, 2004, 3:15 PM
Post #3 of 29
(4295 views)
Shortcut
Registered: Aug 18, 2003
Posts: 153
|
Nice.
|
|
|
|
|
jammer
May 5, 2004, 3:19 PM
Post #4 of 29
(4295 views)
Shortcut
Registered: Jun 25, 2002
Posts: 3472
|
If you want feedback that will help, join the Photographers Critique Forum club. As for criteria for the front page ... :?:
|
|
|
|
|
krillen
May 5, 2004, 3:22 PM
Post #5 of 29
(4295 views)
Shortcut
Registered: Jul 19, 2001
Posts: 4769
|
try doing a search, this has been talked about ENDLESSLY... :roll:
|
|
|
|
|
tylerm
May 5, 2004, 3:25 PM
Post #6 of 29
(4295 views)
Shortcut
Registered: Aug 18, 2003
Posts: 153
|
Thanks, krillen, for stating the obvious. Go waste someone else's time.
|
|
|
|
|
tylerm
May 5, 2004, 3:32 PM
Post #8 of 29
(4295 views)
Shortcut
Registered: Aug 18, 2003
Posts: 153
|
It's sooo much better when I have things handed to me.
Thanks, I'll go to sleep now.
|
|
|
|
|
brittamac
May 5, 2004, 3:33 PM
Post #9 of 29
(4295 views)
Shortcut
Registered: Dec 18, 2003
Posts: 246
|
you're welcome, sleep tight. :wink:
|
|
|
|
|
coldclimb
May 5, 2004, 4:12 PM
Post #10 of 29
(4295 views)
Shortcut
Registered: Jan 14, 2002
Posts: 6909
|
Dude, if what you want is critique of your photos, that is exactly what the photo critique forum is for. And you'll get a lot better responses there than in your photo comments too, believe me. The folks there will give honest constructive criticism to help make you a better photographer.
All you have to do is PM Krillen for membership. :roll: :wink: :lol: :lol: :lol:
|
|
|
|
|
krillen
May 5, 2004, 4:18 PM
Post #11 of 29
(4295 views)
Shortcut
Registered: Jul 19, 2001
Posts: 4769
|
I wouldn't want to waste his time... :roll:
|
|
|
|
|
tylerm
May 5, 2004, 4:27 PM
Post #12 of 29
(4295 views)
Shortcut
Registered: Aug 18, 2003
Posts: 153
|
I give.
|
|
|
|
|
tylerm
May 5, 2004, 4:32 PM
Post #13 of 29
(4295 views)
Shortcut
Registered: Aug 18, 2003
Posts: 153
|
Krillen, why didn't you just say that in the first place?
|
|
|
|
|
popol
May 5, 2004, 9:33 PM
Post #14 of 29
(4295 views)
Shortcut
Registered: Nov 9, 2003
Posts: 390
|
Because it was already said one reply before? 8^) :lol:
|
|
|
|
|
phugganut
May 11, 2004, 2:55 AM
Post #15 of 29
(4295 views)
Shortcut
Registered: Jan 17, 2003
Posts: 648
|
OK I read this thread, the thread linked to this one, and have done a search. I've read much info about how the front page photos are selected, but if it's all true, then I don't think it's working. I mean, if you go to the photos section, there are many that qualify, but there are 10 or 11 photos that are selected over & over again, yet some never get shown. I have no empirical evidence to back this up, but it is obvious to me. So...
Last night I kept refreshing the the main page. Several photos came up multiple times, while others never did. Today I did the same thing, with the same results. So what's up? Is the system not working, or is there some aspect of the process that I'm not straight on.
-Maybe it's the 'who you know, not what you show' thing, or maybe Joe's post was correct.
|
|
|
|
|
phugganut
May 11, 2004, 3:50 AM
Post #16 of 29
(4295 views)
Shortcut
Registered: Jan 17, 2003
Posts: 648
|
BTW, I tried my experiment (haha, that's funny) again, with the same results. Either I'm not up to speed on the whole process or the system isn't working right, probably the former. Maybe a Photo Mod out there can enlighten me.
|
|
|
|
|
roughster
May 11, 2004, 3:56 AM
Post #17 of 29
(4295 views)
Shortcut
Registered: Apr 3, 2002
Posts: 4003
|
This issue comes up time and time again. Because the criteria is also "time" based as well as "vote" based, some times there will be a larger selection of front page photos, and other times there will be a smaller selection.
We "fairly" recently reduced the criteria to allow more photos to show (it used to be the same 10 photos over and over ad-inifinity) but it does seem that we are once again seeing the same pics over and over. I think part of the problem is over all, photos now receive lower marks even the goods ones. This is turn drops many "worthy" photos out of the Front Page selection criteria.
The problem is, how low do we drop criteria before we start seeing butt shots on the front page? I believe that a new "user selected criteria" is being worked on. It would be great if it got implimented.
|
|
|
|
|
phugganut
May 11, 2004, 4:06 AM
Post #18 of 29
(4295 views)
Shortcut
Registered: Jan 17, 2003
Posts: 648
|
Thanks for the input Roughster!
However, the problem is not that the criteria is too high or low, but that there are several pics that meet the criteria but aren't being shown as much as some others that come up all the time. They meet the criteria (as I understand it) just like others do, but don't make it to the front page.
Personally, I would like to see a larger group of photos be in the loop. However, I don't have a problem with the criteria as it is, just as long as all the photos are being treated equally, which I don't think they are.
BTW: it's really not that huge of a deal, and I'm sure that the Mods often have other more important things to do, but it is something that I've noticed quite a bit.
|
|
|
|
|
roughster
May 11, 2004, 4:15 AM
Post #19 of 29
(4295 views)
Shortcut
Registered: Apr 3, 2002
Posts: 4003
|
phugg,
I am in complete agreement with the "see more even if they are of slightly lesser quality" mentality. I was pushing for that very thing when we did the last reduction in criteria.
I also do agree that the random slector seems to favor a select group of pictures. Either Biff/Eric/Tim are going to have to answer that one as it must be code based.
Hopefully the raised awarness surrounding the issue will prompt one of them to take a look!
|
|
|
|
|
coldclimb
May 11, 2004, 4:27 AM
Post #20 of 29
(4295 views)
Shortcut
Registered: Jan 14, 2002
Posts: 6909
|
They're working on a "customize your front page" type of thing, so each user can see whatever rating they want to see on the front page. Personally, I prefer quality over quantity, while Rough hates seeing the same pics over and over again, so we could both have our own settings and be happy. ;) I'm looking forward to this.
|
|
|
|
|
phugganut
May 11, 2004, 4:38 AM
Post #21 of 29
(4295 views)
Shortcut
Registered: Jan 17, 2003
Posts: 648
|
In reply to: They're working on a "customize your front page" type of thing, so each user can see whatever rating they want to see on the front page. Personally, I prefer quality over quantity, while Rough hates seeing the same pics over and over again, so we could both have our own settings and be happy. ;) I'm looking forward to this.
Yes, that's a great idea. But what about the (supposed) random selector?
|
|
|
|
|
tim
May 11, 2004, 5:37 AM
Post #22 of 29
(4295 views)
Shortcut
Registered: Apr 4, 2002
Posts: 4861
|
the setup:
[*:d471e4561b]x = a random number between 1 and 45
[*:d471e4561b]y = a random number between 1 and the # of eligible photos
[*:d471e4561b]z = the position of the photo to be displayed for the next minute (0 < z < y+1)
the query (paraphrased and as pseudocode; rephrased for clarity):
[*:d471e4561b] select a random number (x) of days, between 1 and 45
[*:d471e4561b] select all the photos from the past x days, where the rating is > 7.5, and them order by rating, from highest-rated to lowest-rated
[*:d471e4561b] choose a random set of y photos from the above selection
[*:d471e4561b] choose the z'th photo in the set, and display it for the next 60 seconds.
the extensions:
[*:d471e4561b]allow for a user-specified lower bound (or possibly a photo-ed-specified limit if Eric isn't going to be available to profile the caching mechanism if done per-user) on photo rating (not done, not even written yet as far as I know)
[*:d471e4561b] throw out abnormally high and low votes when calculating the rating (done somewhat heavy-handedly right now, I will try and shoehorn some statistically valid constraints onto Biff's existing implementation tonight while testing #3)
[*:d471e4561b] give more weight to votes from people who have photos that are already ranked very highly (a.k.a. "Taste for Makers") and use the patterns uncovered while compiling these weights to identify abnormal or incestuous voting patterns between pairs or communities of users ("you bomb this guy's photo off the FP, and I'll blow yours"), then discarding patterns that defeat the purpose of the system (namely, to select the most compelling images submitted by our users for showcasing on the Front Page). I wrote this in an email last week and am testing it tonight. The results will tell me a lot about whether this type of weighting can also be used for automating certain tedious aspects of moderation, and whether we can generically trust people with 'good taste' in a specialty, to train an adaptive system (eg. a Bayesian filter) to distinguish good stuff from bad stuff.
In the end there will always be error, slop, and subjectivity in any community-based system. But one of the reasons I decided to pursue my PhD was a conviction that sharper tools existed than are currently used to manage data and assertions. Y'all get to be my guinea pigs, hope you enjoy it ;-)
More on this will be posted as the numbers bear out one way or the other, what is the right course to take. If any of this crapinski sounds interesting to you (eg. if you can code this better than I can, which if you have a Master's or better in statistics, CS, or epidemiology should not be hard), let me know (send me a PM or an email and let's talk).
Note that Biff handles the user interface for this stuff, so if you have something good to say about the process, tell him or Eric. I only want to hear what sucks :lol: (such as, for example, my initial description).
Edit: does it 'work'? Well, obviously it is not perfect or we would not be discussing it. But the methodology has for some time been as outlined above (in 'Setup' and 'Query') and since it seems that more improvements are to be had, you might say that, yes it is random, but no, it is not random enough for our tastes (since we only want the upper extrema of a normal distribution to be displayed on the Front Page).
|
|
|
|
|
phugganut
May 11, 2004, 6:02 AM
Post #23 of 29
(4295 views)
Shortcut
Registered: Jan 17, 2003
Posts: 648
|
Wow. You are obviously MUCH smarter than I am. :shock: I didn't understand but a small bit of that. My limited memory of College Algebra and Behavioral Statistics aren't helping me too much here.
I mean, sorry to be such an ignorant computer 'tard :oops: but does all that mean that the 'random selector' will be working equally?
|
|
|
|
|
roughster
May 11, 2004, 6:38 AM
Post #25 of 29
(4295 views)
Shortcut
Registered: Apr 3, 2002
Posts: 4003
|
Tim,
Wouldn't this:
In reply to: choose a random set of y photos from the above selection
Most likely be the culprit for why we see the same photos over and over again? Is it a linear set? If so, doesn't that more heavily weight and create a greater % chance of seeing "middle" ranked photos inside of the overall total of "qualified" photos?
Why not just eliminate the "random set" and just have it select the zth photo of the entire qualified set?
|
|
|
|
|
|
1
