| Author | Thread |
|
|
06/04/2013 01:11:39 AM · #51 |
Originally posted by Ja-9:
Originally posted by Cory:
Originally posted by Mike:
Originally posted by LydiaToo:
I'm so excited that there's another TPL in the works!
I had so much fun last time getting to know folks and learning from them.
Regarding the scoring system...
Since the "5% rule" was designed to make percentage results from varying numbers of entries in challenges more fair, (for instance an entry that scored 87% in a challenge with 200 entries should win over an entry that scored 87% in a challenge with only 45 entries), I like the idea that if two people whose scores are being pitted against each other are from the same challenge, then there is no reason for the 5% rule... since they were both competing on the same, equal ground.
So... I like option #2 since it's the only one that allows for no tie if one entry scores higher than the other in the same challenge.
I am confused though, about whether the percentages for ties varies with the number of entries in differing challenges. For instance, if there are two challenges with approximately the same number of entries, would that have a 0% variance for a tie, while two challenges with 50% more entries in one than the other would have 3% variance for a tie... and then if there were two challenges with double the amount in one than the other, a 5% variance would be given to make a tie? (or similar? I'm simplifying, of course.)
Regarding, the favorite image vote. I had three contenders, two of them from the same person (not on my team). Good idea! |
i believe option two is the one i suggested during the first TPL. if so, what it does it determines the the difference in the # of entries in the two challenges and determines the variance based on that difference, so the greater the disparity in entries the higher the variance percentage. it actually determines the true variance the difference makes, not a predetermined value or a tier and thus is actually the most fair way to equate two challenges with a differing number of entries.
a very bright mathematician cued me into the formula. i'm not sure why this question is even posed as this method is clearly the option that should be used going forward as it eliminates any possibilities of problem in the future. you simply cannot argue against the math. i only hope that it isn't so confusing as to what the approach is and the voters opt for one that seems more tangible. |
Mike is correct, this is a corrected version of his suggestion. (Me and that same brilliant mathematician had a subsequent conversation that led to a small mod)..
I agree as well, that #2 is the preferable system - as it really balances the difference as it should be balanced, based upon the difference in number of entries between challenges, and it doesn't bother with the compensation if the two images are from the same challenge, preventing improper tie situations.
So, if you don't have a clue which way to vote - trust me and just vote #2. |
Can I change my vote??? I can change my vote on the pictures...can I on this too???? |
Sure - just put a !!!! in the image ID field so I know it's an update. |
|
|
|
06/04/2013 02:24:55 AM · #52 |
| Excellent! Just put my name in the hat and got to peruse lots of exceedingly fabulous images while searching for the one that I consider my 'season favourite'. Afternoon well spent. |
|
|
|
06/04/2013 09:40:48 AM · #53 |
Originally posted by bohemka:
Is there a form I can fill out to disassociate myself with the last TPL? Just in case I ever need a job, or somehow end up in politics, I wouldn't want any of my teammates' names coming up. |
Two of your teammates changed their names during the TPL, but even that won't help them. There's no escape!
|
|
|
|
06/04/2013 01:47:33 PM · #54 |
Originally posted by Cory:
Originally posted by Mike:
Originally posted by LydiaToo:
I'm so excited that there's another TPL in the works!
I had so much fun last time getting to know folks and learning from them.
Regarding the scoring system...
Since the "5% rule" was designed to make percentage results from varying numbers of entries in challenges more fair, (for instance an entry that scored 87% in a challenge with 200 entries should win over an entry that scored 87% in a challenge with only 45 entries), I like the idea that if two people whose scores are being pitted against each other are from the same challenge, then there is no reason for the 5% rule... since they were both competing on the same, equal ground.
So... I like option #2 since it's the only one that allows for no tie if one entry scores higher than the other in the same challenge.
I am confused though, about whether the percentages for ties varies with the number of entries in differing challenges. For instance, if there are two challenges with approximately the same number of entries, would that have a 0% variance for a tie, while two challenges with 50% more entries in one than the other would have 3% variance for a tie... and then if there were two challenges with double the amount in one than the other, a 5% variance would be given to make a tie? (or similar? I'm simplifying, of course.)
Regarding, the favorite image vote. I had three contenders, two of them from the same person (not on my team). Good idea! |
i believe option two is the one i suggested during the first TPL. if so, what it does it determines the the difference in the # of entries in the two challenges and determines the variance based on that difference, so the greater the disparity in entries the higher the variance percentage. it actually determines the true variance the difference makes, not a predetermined value or a tier and thus is actually the most fair way to equate two challenges with a differing number of entries.
a very bright mathematician cued me into the formula. i'm not sure why this question is even posed as this method is clearly the option that should be used going forward as it eliminates any possibilities of problem in the future. you simply cannot argue against the math. i only hope that it isn't so confusing as to what the approach is and the voters opt for one that seems more tangible. |
Mike is correct, this is a corrected version of his suggestion. (Me and that same brilliant mathematician had a subsequent conversation that led to a small mod)..
I agree as well, that #2 is the preferable system - as it really balances the difference as it should be balanced, based upon the difference in number of entries between challenges, and it doesn't bother with the compensation if the two images are from the same challenge, preventing improper tie situations.
So, if you don't have a clue which way to vote - trust me and just vote #2. |
this is so funny... i just picked #2 b/c it was in the middle, so i figured it was the "compromise" between two extremes... i'm a scientist, but no way was i competent to wade into to the details...
"There's lies, there's damned lies, and then there's statistics" - attributed to Mark Twain |
|
|
|
06/05/2013 09:42:31 PM · #55 |
If the NBA had a standings system like this ... all teams would be tied for first or second the whole season.
I do not like the whole tie mechanism. Let the standings stand. Use percentage if you must instead of raw score. But dump the tie mechanism.
|
|
|
|
06/05/2013 09:57:17 PM · #56 |
| I am sad to have to sit out the next TPL but the trade-off is a trip that will hopefully get me a heck of a July FS entry. I'm okay with that. There's always next time. |
|
|
|
06/05/2013 11:00:37 PM · #57 |
Originally posted by ciaeagle:
I am sad to have to sit out the next TPL but the trade-off is a trip that will hopefully get me a heck of a July FS entry. I'm okay with that. There's always next time. |
Premature.
We haven't announced a start date yet. :)
If you're lucky you'll have a FS and a hell of a first entry for the TPL Season 2
Message edited by author 2013-06-05 23:01:29. |
|
|
|
06/06/2013 07:31:01 AM · #58 |
Originally posted by Dr.Confuser:
If the NBA had a standings system like this ... all teams would be tied for first or second the whole season.
I do not like the whole tie mechanism. Let the standings stand. Use percentage if you must instead of raw score. But dump the tie mechanism. |
if the TPL was like the NBA we'd be able to pick our own teams and we'd end up with a powerhouse team that no one could compete against.
whats wrong with ties? if you have two statistically exactly equal entries (which i doubt will happen every often) then so be it. Tweaking the rules so you dont end up with ties creates other problems that we had where lesser entries beat out the higher ones.
Make the rule set most fair and let the cards fall where they do. No one will complain because they can't.
also, the NFL has ties and hasn't spun into turmoil. |
|
|
|
06/06/2013 09:07:59 AM · #59 |
Originally posted by Cory:
Originally posted by ciaeagle:
I am sad to have to sit out the next TPL but the trade-off is a trip that will hopefully get me a heck of a July FS entry. I'm okay with that. There's always next time. |
Premature.
We haven't announced a start date yet. :)
If you're lucky you'll have a FS and a hell of a first entry for the TPL Season 2 |
Woot! A glimmer of hope. But I don't see y'all waiting until August to start. I'll keep watch and see how it goes. :) |
|
|
|
06/11/2013 11:09:33 PM · #60 |
 mitalapo brought to our attention a problem with the Scoring system 2.
The bottom line is that the proposed Scoring system 2 can give a better result (as in tie better than loss) for a match if a lower score is used (from a challenge with less entries) than if the higher score is selected (from a challenge with more entries). Below is an example of the situation:
First a note about percentiles that seems to be a recurrent concern. For a challenge with N entries, the position X will have a percentile P given by:
P=(N-X)/(N-1)*100
This is just a scaling of the percentiles between 100% (for the blue) to 0% (for the brown).
So if we consider a hypothetical situation where the three challenges have:
Ch1 10 entries
Ch2 100 entries
Ch3 101 entries
Suppose a match between Player "A" and Player "B" have these results in those challenges:
Player "A" Player "B"
Ch1 2/10 (88.89%) Ch3 9/101 (92%)
Ch2 10/100 (90.91%)
In the System 2 as proposed the results would be:
Player A (Ch2) and Player B (Ch3) would be compared with a N%=0.01, and would lead to a victory of Player "B".
If instead the result of Ch1 is used for Player "A", the N% is 10 and the match results in a tie.
In summary, a lower score (88.89% vs 90.91%) gives a better result (tie vs. loss). This is a direct result of having a different N% for different challenges.
To avoid this situation the tie threshold N% needs to be the same for any two entries compared regardless of the challenges in which they participated.
The other two proposed system give a consistent result:
With the system 1, N% is fix to that of the lowest number of entries (10 entries, N%=11%) and regardless of which of the two images of Player "A" is chosen the match would ended up in a tie.
With the system 3, all entries between a percentile of 83.5% and 94.5% (regardless of the number of entries on each individual challenge) would tie. Then again both comparison on this example would result on a tie.
Thanks to mitalapo for pointing out the problem.
--------
Given this issue we proposed to drop the proposed system 2 and choose between System 1 or 3. If you have already voted you can change the vote in the original voting form here. Just put "!!!" in the Image Id field:
|
|
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: 08/29/2025 11:10:20 AM EDT.
