New Leaderboard Commentsgreenspun.com : LUSENET : MAME Action Replay : One Thread
Well... I think it should be in the range of a 10 to 20% penalty... I think 15% will do it personally... but who cares right now? You've made a nice new leaderboard Zwaxy... well done!
-- Gameboy9 (firstname.lastname@example.org), June 11, 1999
I think 30% is way too much "bonus" to give to someone who may have only just beaten your score or you couldn't quite beat theirs. Around 10-15% is enough of an incentive to improve your score beyond theirs without overly penalising someone who came oh so close.
Definitely drop any scores who's raw score is < 25% of the top score from leaderboard scoring.
With this new system there is definitely more incentive to improve scores on games where while you may not get a better score than the current high you can greatly improve your points ranking.
Well done Zwaxy.
PS: I favour the look of 10% but maybe 15% is more appropriate. We'll see what the majority think - but count me as definitly thinking anything > 15% is too large a "bonus" for 1st place.
-- beejay (email@example.com), June 11, 1999.
While everyone else is in favor of this, it looks like, I am not for a few reasons. 1) A percentage scoring is automatically biased against games with low scoring, example Balloon Bomber. A good score in that game is around 8-10k, with first place here holding about 14-15k If I remember the score correctly. Now, if your 1,000 pts off the 1st place in this game, you'll lose 7% thereabouts, whereas on a game where scoring is easy to get, Example, Quiz of Dragons(US). The high score on this game is 2.4Million. Now, if your only 1,000pts off the 1st place, you still would statistically have 100% because it would be 99.9....% 2) It discourages competition. Anyone close to first place doesn't real have the push to go on, while they already have 60-70% of the total score they could. 3) It encourages submitting anything that is within 30% of the top score, just to grab a few points. Are we going to begin to reward quanity rather than quality? That's what the currently proposed system would. Okay, enough from me, let's see what else can be done, I actually perfer the 10-3-1 board, as it forces you to get to the top 3, or forces you to remain there, to have any points.... Chris Parsley PS- What are all these tournaments I keep hearing people talk about here... PPS- Coming soon, a one round Track and Field score of over 100,000.
-- Chris Parsley (firstname.lastname@example.org), June 11, 1999.
I agree with Chris Parsley on this one. Also I don't think players should get zero points just because they can't get within 30% of the top score. Sure they should get less but not zero. On some games where the best player can get an exceptionally high score (e.g. Galaga or whatever) I doubt many people could get within 30%. But under the old system of 10,3,1 they could still battle each other for the second and third places and get some leaderboard points for it.
-- Octavian (email@example.com), June 11, 1999.
I don't believe in a points cuttoff if below a certain percent, I do believe the 10%, 15% or 20% adjustment to anyone who is below 1rst, 2nd, or 3rd place would be good. With the exepction of ties, it's wrong to give someone more credit if they have the same score because they did it first. So if you had 300 300 300 299 200 100 scores you'd have 100,100,100, 100-10=90, 66-15=51, 33-20=13. I still don't like all those 100's but it is the best score you can get in a game unless we want to count some other measure, say duration of playback, so they deserve the best score.
-- Chad (firstname.lastname@example.org), June 11, 1999.
Aaahhh, progress! I like this change a lot! The first place bonus % should exist, but be lower (10-20, perhaps settle on 15% for now). However, this is a misnomer because it is really not a bonus, it's a PENALTY to 2nd place scores or lower. I prefer to look at it this way.
As far as the cutoff that's dead wrong! Why have one at all? Really low scores don't contribute much to the bottom line. Furthermore, in the voting thread only 1 person thus far wants to see some sort of cutoff.
-- Pat (email@example.com), June 11, 1999.
I'm now thinking 10% would be too low of a bonus/adjustment/penalty between 1rst and 2nd place, 15% maybe even still too low. Though, i am reiterating that there should be an additional bonus between 2nd/3rd and maybe even 3rd and 4th, or we might get less competition.
Example: 1rst, 15% 2nd, 20% 3rd, 25% 4th. This would actually nullify scores less than 25% ONLY if there were already 3 recordings which were better, which is good for those 4th placers to try to do better to get some points out of the game. If there are three people that can get more than twice as much as you, you should be able to do better to get points. Values of Percenatge penalties are adjustable but have more than one level of penalty percentage. Peter Piper picked a peck of penalty percentages, perhaps.
-- Chad (firstname.lastname@example.org), June 11, 1999.
I was wondering if it would make sense to divide the total score by the number of available games (currently 1400). This would make the larger numbers more manageable for comparisions. NOTE: I'm not suggesting to use average percentage as that would be total_score / games_played. I'm thinking total_score / available_games. Since everyone's score is divided by the same number, it changes nothing in the standings, just makes the numbers more readable.
-- Pat (email@example.com), June 11, 1999.
I think that's a much better leaderboard....1st because gives points according to the score actually reached, and 2nd because encourages more people to submit recording, and get points even not in the top 3.
Personally I wouldn't use the 1st place bonus, but I have the feeling it is preferred from the community....the next step will be to define how much % scores not 1st will be reduced....if any, I think 15% should be ok.
Another great feature Chris has added (just noticing) is that you have the % of any score compared with the 1st, when listing games per player.
The challenge must go on !!!
-- Cicca (firstname.lastname@example.org), June 11, 1999.
Despite the fact that it's shrunk my lead on the leaderboard, I actually like the new scoring system.
Although I'm guilty of doing it too from time to time, I hate seeing someone upload a cheesy score on a game where one person has a kickass first place score, giving them 3 points on the leaderboard. Sure, some people could just upload a score better than that but I'm sure some of us figure it's not worth the trouble just for 3 points. Now, a score much lower than the high doesn't yield much of a reward... I like that.
Plus, it gives more motivation for someone that has the high score on a game to submit even higher scores, since it will decrease the percentage of all the competing scores too...
So, I like it. But now we're going to have to forget about the "2000-point barrier"... it's 20,000 now ;)
-- BBH (email@example.com), June 11, 1999.
I have a very simple idea, if we decide to stick with this "penalty system". We pick a penalty percentage, let's say a nice round number like 10%. Then for each score posted (except 1st place) we take the place that the recording finished and multiply by that 10%. e.g. 2nd place recording=20% penalty, 3rd place=30%, 4th place=40%, etc. This system could support up to 9 scores per game, but I suppose the 9th place score would have to reach 90% of the high score. In a situation like this all 9 scores would be within 10% of each other. Doesn't get more competitive than that!
-- Pat (firstname.lastname@example.org), June 13, 1999.
Wow - that's a lot of responses! I'll try to reply to them...
I think 30% is way too much "bonus" to give to someone who may have only just beaten your score or yu couldn't quite beat theirs
Well - the board was just put up as an experiment. I've left the bonus percentage entirely configurable, so that people can play with it and see what feels best - although I'm guessing that will be different for each player, depending how it affects their placing... :o) Also, bear in mind that in the 'traditional' leaderboard, there's an automatic 70% bonus for 1st place - 2nd place always gets 70% less than 1st place, whatever the score.
Definitely drop any scores who's raw score is < 25% of the top score from leaderboard scoring
That happens automatically (except it's currently 30%), since we take the 'bonus' percentage off their score, and if that makes the score negative then they get no points for it.
I actually perfer the 10-3-1 board, as it forces you to get to the top 3, or forces you to remain there, to have any points
That's a good point. I'm undecided which system is better. I guess I really need to get the leaderboard to show both types of scoring at the same time, but then it will still have to be sorted according to one of the other... it's always possible to make it an option, but which one will be official? :o)
As far as the cutoff that's dead wrong!
But the cutoff is a natural affect of the 'bonus' (or 'penalty' as he calls it). If we're taking 15 points off each score which doesn't get 1st place, what do we do for a score which gets 10% of the top score? It would score 10 points before the penalty, so it scores -15 after the penalty - I'm being kind by only taking off 10, making the score zero, since nobody wants a win negative points! :o)
One alternative I was thinking of was that instead of taking the penalty off the player's score I could scale the player's score. If the penalty percentage - (we really ought to get some terminolofy sorted out here!) - if the PP is 15%, say, then the 1st place gets 100 points, and all other scores getpoints. that way somebody with 99.99% of the highest score gets 100-PP points, and somebody with 0.01% of the highest score gets 0 points, with a smooth transition between the two, and no cutoff. I was thinking of doing this at the time that I last changed the script (Friday morning) but it was like 5am at the time and I couldn't work out the formula I needed to use...
(player.score / highest.score) * (100 - PP)
Chad suggests tiered penalties:
Example: 1st, 15% 2nd, 20% 3rd, 25% 4th. This would actually nullify scores less than 25% ONLY if there were already 3 recordings which were better
Maybe there's some scheme which mixes the 'scaling' idea in my above paragraph with Chad's idea of a tiered penalty scheme, such that people still get points whatever score they contribute, but there's some guaranteed gap between each position, to keep competition alive.
How about something like this (I've shown it for PP (percentage penalties) of 10, 15 and 20. The tables below show the highest position score for a score in each of the top 10 places. For example, the 2nd row of the first table says "2 : 90 : 10" and means that if you're in 2nd place, your highest available score will be 90, and you can gain 10 points by moving up one place. In 3rd place, the highest you can score is 81 points, and you can gain 9 points by moving up to 2nd place, and so on. This way, the percentage of the maximum score lost between each place is a constant (if you see what I mean)...
pp = 10 1 : 100 2 : 90 : 10 3 : 81 : 9 4 : 73 : 8 5 : 66 : 7 6 : 59 : 7 7 : 53 : 6 8 : 48 : 5 9 : 43 : 5 10 : 39 : 4
pp = 15
1 : 100 2 : 85 : 15 3 : 72 : 13 4 : 61 : 11 5 : 52 : 9 6 : 44 : 8 7 : 38 : 7 8 : 32 : 6 9 : 27 : 5 10 : 23 : 4
pp = 20
1 : 100 2 : 80 : 20 3 : 64 : 16 4 : 51 : 13 5 : 41 : 10 6 : 33 : 8 7 : 26 : 7 8 : 21 : 5 9 : 17 : 4 10 : 13 : 3
Pat also suggests that we:
divide the total score by the number of available games (currently 1400)
That sounds reasonable. Another couple of ideas would be to scale the whole table so that the top player always has 100 points (but that's probably not a good idea, 'cos the top player then has nowhere to go - he's always got 100 points...) so to divide all the scores by 100 - so that in effect the top player for each game gets 1 point, not 100. Or maybe I'll just write the last two digits in a smaller font?
Then Pat wrote:
...for each score posted (except 1st place) we take the place that the recording finished and multiply by that 10%. e.g. 2nd place recording=20% penalty...
It's similar to the nasty table I put up above, but my idea works for more scores - it's like a 'compound interest' or 'diminishine returns' thing, where 2nd gets 90%, 3rd gets 90% of 90%, 4th gets 90% of 90% of 90%, etc... So even 50th gets something (although it'll be pretty near to zero by then). And also, I think we should scale the scores using the penalty, not subtract the penalty from them.
I think that's all the messages kind-of replied to. So what now???
traditional: Chris Parsley, Octavian (2)
modern: Gameboy9, BeeJay, Cicca, Chad, Pat, BBH (6)
-- Zwaxy (email@example.com), June 14, 1999.
The scaling formula, (player.score / highest.score) * (100 - PP) slightly modified would do it. Again taking my simple idea of using place information, I would suggest this:
(player.score / highest.score) * (100 - PP * finishing place). e.g. If PP=10%, then a 3rd place score would use 70% of the original score. Again, you're limited to only the top 9 scores, but I don't think this is really an issue. It sure beats only the top 3 counting!
-- Pat (firstname.lastname@example.org), June 14, 1999.
How about this:
(((100 - PP) / 100) ^ (finishing.place - 1)) * (player.score / highest.score) * 100
That gives an exponentially decaying scaling factor, asymptotically aproaching zero.
-- Zwaxy (email@example.com), June 15, 1999.