"The Leaderboard"

[Mr. Kelly R. Flewin]

Why not keep the system we use, but sort by average instead of points?

This will make a leaderboard really showing the top players, and it would
stop the so-called ABC uploaders. At least for players who cares about the
leaderboard, they will think twice before uploading something.

I know this idea goes a long way off the rest of the solutions, and there's
probably a lot of things I haven't thought about that's wrong with it, its
just an idea.

Now I like this idea... but... there's gotta be an easier way to do it... because technically if someone submits 25 1st place scores, they have a 100% average, compared to someone like BBH who's submitted god only knows how many scores and has an average of 64.7%.

That's about the only main problem there... and I don't think it's right, as it awards someone with less submissions but has a far better average.

I get this feeling we could discuss this for a long time and still be where we are now... which is sad...


Kelly
[Who got hooked on Krazy Kreatures]

[LN2]

You get onto the leaderboard as soon as you've uploaded a certain number of recordings, maybe 25.

Yeah, but then you have gamers potentially removing lesser place(but still perhaps great) scores just to raise themselves on that type of leaderboard. That wouldn't be good for MARP.

It almost sounds like out of all of the above perhaps the best and easiest thing would be the current system BUT with the change that any points determined by the current formula to be less than 10....would be changed to 0 points.

That would zero out most(not all, but most) of the ABC submissions just submitting whatever they score playing the game only a couple times....which might be enough to keep them from bothering to submit scores that would get zero LB points.

Yes, in a few cases that might result in 2nd and 3rd etc. places even getting zero points. In those cases I say the person who got the top score has established a high enough score where they deserve to have the 100 point advantage over all other scores for that game.

[Frankie]

You get onto the leaderboard as soon as you've uploaded a certain number of recordings, maybe 25.

Yeah, but then you have gamers potentially removing lesser place(but still perhaps great) scores just to raise themselves on that type of leaderboard. That wouldn't be good for MARP.

Yes, but if you simply disallow and make it impossible to delete top 5 recordings, players will have to fight back, and that way at least the 5 best scores stays online. Of course the editors can delete any top 5 recording if it doesn't playback or violates any rules.

If you make a what you think is a great first place and someone comes along trashing it, you can't delete it as you are still in second place. This means you must fight back to get your perfect average score, or wait until it drops out of the top five to delete the score, which might never happen if the score is really good.

But as Barry and Kelly said, it doesn't really work unless something else is done. The system would be unfair to players like BBH. I see that.

[LordGaz]

SUGGESTION 5!!!!!!!!!!!!!!!!!

Make leaderboard points for a game decay over time!

Somebody gets an 'unbeatable' marathon score on some game for example, I dunno, 1 million on Sarge. Then he/she disappears into the wilderness and never returns to MARP. First off he would be awarded 100 leaderboard points but this would decay over time by a compound 1% per day giving it a 'half-life' of 69 days. After 1 year, the leaderboard award is down to 100*0.99^365 = 2.55pts.

Now after 1 year somebody else comes along and scores 100,000. This would be awarded leaderboard points by the following formula.

L = 100s/S * 0.99^A

where L = Leaderboard points, s=Score submitted, S=First place score and A=Age of the recording in days.

In this example, player 2 would receive 100*0.1 * 0.99^0 = 10 Leaderboard points. Note there is no 15% scaling down for each position beyond 1st place.

Another way of doing it is to scale the scores down by age first, then calculate the leaderboard points from that. Applying this to the ongoing example:

LBScore1 = Score1 * 0.99^365 = 1,000,000 * 0.99^365 = 25,518
LBScore2 = Score2 * 0.99^0 = 100,000 * 0.99^0 = 100,000

Now automatically award the player with the top LBScore 100pts and all others in terms of their percentage of it ie:

Player 1 gets 25,518/100,000 * 100 = 25.5 LB pts.
Player 2 gets 100 LB pts.

Yes I think the second way is better because then there will always be a player with 100 LB pts for all games, but then again...

Of course when viewing scores per game the rankings are still listed by score and not by leaderboard points. This system applies to the leaderboard only so it will become more of a separate thing. Yes no one likes change especially when there's nothing wrong with the current system, but hey what do you want?

This system has all the following advantages:

- Players are encouraged to stay active at MARP if they want to stay high on the leaderboard.
- It benefits the good and active players.
- People who want to upload just for the hell of it will still see some results. If they want to keep their results they will have to 'refresh' their scores .
- Tactical uploading becomes possible. Ie. you bide your time waiting for a score to age to a point where you know are cabable of grabbing the 100 LB pts.
- Top scores still stay in 1st place in the rankings and so still get the same recognition, they just won't score 100 LB pts any more as they get older and 'decay'.

And the following disadvantages:

- The 'purists' won't like it and there are many at MARP. Why should a lower score get more LB pts?
- Players with hundreds of old and gold recordings will suffer in terms of leaderboard points. However, I think this will only serve to help revive MARP as new members join instead of them taking one look and thinking SHIT! how the ***k am I gonna compete here? and never to return. Besides I've heard some of the top ranked players say the leaderboard doesn't really count for anything, so they won't mind anyway.

Gaz.

[Weehawk]

Why not keep the system we use, but sort by average instead of points?

I gotta jump in at this point.

If player "A" submits a first place recording (maybe on an obscure game no one else has played) he has an average of 100. If player "B" plays a hundred games well enough to get 1st and 2nds on them, his average will be lower. Does he deserve less leaderboard points than player "A"? Of course not.

I know a minimum of 25 submissions was suggested, but the same principle applies on a larger scale.

Scoring by average is a bad idea, in my opinion.

[DaviL]

>If player "A" submits a first place recording (maybe on an obscure game no
one else has played) he has an average of 100. If player "B" plays a hundred
games well enough to get 1st and 2nds on them, his average will be lower.
Does he deserve less leaderboard points than player "A"? Of course not.


Mmmm... I play Bongo (ok,I like this crap......) and get 100pts. Someone else play a legend like pac-man and also get 100pts.
There is something strange here.
Special points for some games can be an idea. Games like space invaders,donkey kong, pacman, street fighter 2, track 'n' filed, Gng and other classic can't be compared to Anteater or Mahjong.
They are classics ! Be first is not easy, you need not hours but entire weeks to dominate and can't be enough because everyone want be first !

[The TJT]

Mmmm... I play Bongo (ok,I like this crap......) and get 100pts. Someone else play a legend like pac-man and also get 100pts.
There is something strange here.
Special points for some games can be an idea. Games like space invaders,donkey kong, pacman, street fighter 2, track 'n' filed, Gng and other classic can't be compared to Anteater or Mahjong.
They are classics ! Be first is not easy, you need not hours but entire weeks to dominate and can't be enough because everyone want be first !

Thats why evil proposit...umm...I mean suggestion 4. More lb points for most played games.

I just couldn't keep mouth shut, damn
TJT

[LordGaz]

Thats why evil proposit...umm...I mean suggestion 4. More lb points for most played games.

Good idea in principal and IMO it would have been great if MARP had started out with this system for the leaderboard. But now MARP is what, 6 years young and all these 'classic' titles have been 'burned out'. Everyone has seen what is possible on these games, maybe some complete newcomer to the internet who was an old arcade addict might stumble on MARP and beat one or two of the scores but these players are very few and far between.

Hence suggestion 5! Let the average gamer enjoy these old games once again, make LB points easier to score on games where the 1st place old. Otherwise we'll remain with the problem we are trying to find a solution to, and that is having people upload worthless scores on newly released games for the sake of LB points.

Gaz

[LN2]

Yeah, there is to a degree more value on high scores where many have submitted high scores for the game. How to properly weight that though would be difficult.

Should the weight increase just cuz some ABC submitter submitted a lame score for that game so the total #submissions for that game has increased?

Obviously, there are some scores and records much more obtainable versus others. There isn't much that can be done over that.

I don't think the overall points you have on the leaderboard really matters that much. Almost all in the top 15-20 got there perhaps by submitting lots of scores(submitting at least 150 scores all first place would be required to get into the current top 10), but also having a decent amount of 1st, 2nd, and 3rd place scores.

The total points are cool mainly for reaching certain plateaus. Like can BBH reach 50,000?!?!? Will anyone else ever reach the points that BBH and Phil Lamat have? I think other than the personal plateau whether you have 15,000 or 20,000 or 25,000 points means little. All at that level have made a significant contribution to the MARP replay database.

Lordgaz's suggestion that points decay versus time is bad. If anything, high scores that have stayed high scores enduring a test of time actually deserve more points if anything...not less.

Many of the top scores are scores from 3-4+ years ago. Would you decay them into oblivion?

That means someone submitting an inferior , but fresh score for the game overall would get more leaderboard points.

Perhaps we are active, but not submitting for many games cuz we already have scores submitted.

I doubt BBH keeps trying to improve his scores for the 270+ first place scores he has at MARP. Even if he did, how often and how easily are you going to beat some of your own top scores?

I say no way to decay...ok!

[LN2]

Otherwise we'll remain with the problem we are trying to find a solution to, and that is having people upload worthless scores on newly released games for the sake of LB points.

That's actually only part of that problem....plus, those LB points earned from a "worthless" score would only be temporary. It wouldn't stand any test of time or challenges of others.

The problem is they seem to have a goal of uploading a score for every single rom set. It's just not about the new ones.

[LordGaz]

Many of the top scores are scores from 3-4+ years ago. Would you decay them into oblivion?

That is entirely my point. The scores would not decay into oblivion, a top score would remain ranked one for that game. We are trying to discuss a new system for the leaderboard only and my suggestion separates the leaderboard system from all time rankings for the games.

That means someone submitting an inferior , but fresh score for the game overall would get more leaderboard points.

That was one of the disadvantages I listed but what I am saying is the new leaderboard would be a separate thing. It keeps players active and as time goes by players will see themselves drop in the rankings so they will be forced to compete once again and players will return to MARP time and again.

Gaz

[LN2]

That is entirely my point. The scores would not decay into oblivion, a top score would remain ranked one for that game. We are trying to discuss a new system for the leaderboard only and my suggestion separates the leaderboard system from all time rankings for the games.

If a gamer 3 weeks ago or 3 years ago set a bunch of high scores, IMHO, the one that did that 3 years ago and yet still has those scores as top scores should still be prominent on the leaderboard. To take that away from that gamer just because it was done 3 years ago seems unfair to me.

What is to stop someone then from holding back some replays? ie. I get high score for the game. I kill the game off when I have barely beaten the high score. I might have the skill to far exceed that. Your method discourages my submitting my absolute best I can do. I would hold back...and then every 6 months submit a slightly higher score for it.

That's not right IMHO. Submit your best...let it stand the test of time. You don't want any leaderboard interfering with that aspect of the scores.

I have viewed and always will view the leaderboard as an aside to the MARP scoreboard. It's an extra statistical thing for added "fun" and reference. I don't care if it has flaws. Any stat system has flaws.

If Hank Aaron was a yankee so had to play home games in a much larger field than the Braves had, would he have topped Ruth's HR mark? That's difficult to say. Odds are he wouldn't have. The fact Aaron played 3 or 4 more seasons than Ruth did and had many seasons playing 162 games per season instead of 154 adds up also.

A player batting 0.350 today might be a better or equal hitter to Ted Williams and others that bat 0.400 or higher. We now have relief pitchers used regularly so you have a strong armed pitcher the entire game. That wasn't the case back then. I wonder how much higher their batting averages were in the 6th-9th innings versus the 1st-5th innings. hmmm.

I wonder if today's 50+ HR hitters would hav easily hit 20+ more back then. Pitchers got tired toward end of game..give up easy balls to hit. However, perhaps the ball itself was a little more dead then. These things all affect the stats and what they truly represent.

Each person can study the stats and make their own conclusions.

I am only pointing out that any statistical representation can be analyzed to expose flaws and questions about what those stats even represent.

The MARP leaderboard is no different here. Study the leaderboard, the #firsts, seconds, thirds, the average LB points per submission...the # of 4th or worse etc. make your own conclusions. Order them yourself.

It keeps players active and as time goes by players will see themselves drop in the rankings so they will be forced to compete once again and players will return to MARP time and again.

Yes, but how can they compete? You often might then have them competing against their own older scores just to try and get some LB points!!! At some point, you will have personal best scores you aren't very likely to beat...or even care to try and beat...so you are stuck with nearly zero LB points for a game you are the best at. Is that fair?!?

[LordGaz]

What is to stop someone then from holding back some replays? ie. I get high score for the game. I kill the game off when I have barely beaten the high score. I might have the skill to far exceed that. Your method discourages my submitting my absolute best I can do. I would hold back...and then every 6 months submit a slightly higher score for it.

Hmm, yeah I did consider that, a possible weakness in the system in that your new score you submit 6 months later doesn't beat you're own current submission so you are forced to overwrite your best score with a poorer one if you want to maintain good LB points. I can't see a way around that at the moment .

so you are stuck with nearly zero LB points for a game you are the best at. Is that fair?!?

Yup .

Ok, I've quickly glanced again at TJT's suggestion 4, that is LB points get a 'popularity multiplier'. Let's do it and see how it works and simply make 'P'=the number of submissions, no need for P=10 for the 20 most popular etc.

Gaz

[LN2]

Ok, I've quickly glanced again at TJT's suggestion 4, that is LB points get a 'popularity multiplier'. Let's do it and see how it works and simply make 'P'=the number of submissions, no need for P=10 for the 20 most popular etc.

We need a weight formula for how to handle that versus #submissions for a rom set. It shouldn't be proportional.

Plus, he has this as a popularity multiplier relative to the ranking of that game versus all games?

For one, that requires quite a bit more calculation by the code to determine that for each score and to redetermine it for each and every submission for each and every rom set. That's a lot of calculating.

Also, it's misleading for as many if not more games than other aspects.

ie. I see a game with a few very high scores submitted for it already. I likely will not submit anything if I can't get close to or beat those scores.

On the other hand, if the scores for that game were lower, then I would submit. The basis of submitting or not has nothing to do with the popularity of the game...but just whether I think my score is worthy enough to submit.

You could end up with very popular games that have a low popularity multiplier based on just the #submissions relative to other rom sets.

There are some silly little old late 70s games MAME supports that have lots of scores submitted for it cuz it takes only a few minutes to play the game so you can play it tons of times in 1 evening and scores are limited so you can get close to the top score with enough tries.

I might love another game more, but to reach the top score would require hours of gameplay to even reach the score. That might turn me away...not the popularity of the game.

Also, games that have lots of clones perhaps suffer...cuz that dilutes the popularity of it over many rom sets.

I hope no weighting factors along these lines happens....or if it does...perhaps have something that less LB points are given based on some multiplier until the game has received 5 submissions for the rom set.

5 or more weighting is same for all.
The effect of this would be new scores initially set for rom sets added or changed to working status with a new mame release wouldn't get 100pts for first, but perhaps only 20 pts for first for the first score submitted.

...then with a second score submitted, 40 pts for first...third, 60, etc. until you hit 5 or more where first gets 100. It would be proportional for all other places. ie. if 2 scores, 2nd place is calculated as it is now, but based on a max 40 point system instead of 100. so max 2nd could get if only 2 scores for the game would be 40-15% if they had 99.999% of the top score.

This type of weighting makes sense. It's like an introductory period for a new rom set making scores for it not mean much for LB points until more scores have been submitted for the game.

This would even affect those more obscure rom sets that have perhaps been around for quite a while now but only 1 or 2 have submitted scores for the game.

[Zwaxy]

Aaarrrggghhh!!!

I wish there was a clear way of sorting this out once and for all, but I still can't see it.

Having LB points decay over time is a non starter in my opinion. Just because a score is old in no way devalues it.

Awarding more points for popular games would just result in the game becoming more popular, at the expense of the less popular games. The calculations involved needn't be a problem. The 'hard sums' can be done once per day and the results cached, at little extra cost in terms of CPU time, but I don't think it's a good way to go.

So what are we left with? I quite like the "use the current system, but round anything less than (10) points down to zero" idea. (where the '10' is subject to revision). I also quite liked the "use the current system but make the 15% decay cumulative - 100%, 85%, 70%, 55%, ..." idea. That only awards points for the top 7 scores. That means the site can accomodate no more than 7 ABC uploaders before they start having to fight each other for points. In time every game will have 7 decent scores for it, and the ABC uploaders will give up. Are we willing to wait for that?