Total Posts:73|Showing Posts:1-30|Last Page
Jump to topic:

New Ranking System - Official

J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 10:48:45 PM
Posted: 5 years ago
I think we all know that the current leaderboard is flawed. The only factor it takes into account is the total number of wins, with no consideration for win-loss ratios or the quality of opponents faced. There's been a lot of griping over the years, but with Phil as the admin, we could sooner expect hell to freeze over before any substantive changes took place on this site. This thread will be for general and open discussion of new proposals. But first, let me outline my own plan:

The Elo Rating System

The Elo system was developed as a way to rate the relative skill of professional chess players. Since then, it's been adapted to other games and sports including professional football, basketball, baseball, and tennis. The average professional chess player has a rating of around 2,500. A typical club player is usually around 1,500.

When an Elo system is implemented, everyone begins with a default rating of 1,300. If you win a debate, your rating goes up; if you lose a debate your rating goes down. Beating someone with a significantly rating higher than yours will cause your score to increase more than beating someone with an equal or lesser rating would. Losing to someone with a significantly higher rating will have a minimal effect on either of your respective ratings. The differential varies proportionately.

Proposed Changes to Site Format

The all-time leaderboard would list every site member who has participated in at least one debate and generate rankings starting with the very first debate ever conducted. A quarterly and a yearly ranking system would be introduced listing the top 20 and the top 50 respectively. These would be listed somewhere on the main page.

Benefits of the Proposal

Because the best way to raise your ranking would be to engage other high-quality opponents, there would be less of an incentive for senior members to scoop up debates created by n00bs for the easy wins. This would increase overall debate quality and give new users time to get adjusted to the site.

And with that, I yield the floor.
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:00:37 PM
Posted: 5 years ago
I think we must ask how important is each factor, total wins, win%, and people that you win against.
"Wanting Red Rhino Pill to have gender"
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:07:16 PM
Posted: 5 years ago
I'm thinking a point system.

If you win a debate, you get 20 points + 1 point for every win your opponent has (at the time of the win, or always fluctuating, either way). Take the total number of points squared, and divide by 20*X (where x = the total number of debates you've completed).

This means that a win against a veteran is worth more then a win against a newbie. However there are still points to be had against the newbies.

This basic formula can be reworked with different numbers (say 10 points +1 for each win, or 15 or whatever).
"Wanting Red Rhino Pill to have gender"
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:14:02 PM
Posted: 5 years ago
At 12/26/2010 11:06:52 PM, Zetsubou wrote:
At 12/26/2010 10:59:06 PM, OreEle wrote:
win^2/loses is a simple one.
Hit's you too hard if you lose.

Personal opinion. I think that going 20/24 (16.67 points) is far superior to 25/45 (13.89), but not as good as 30/37 (24.32).

Loses actually don't hurt all that much when you've done more debates (since squares grow so much faster then linear).

You'll find that if you are 53/81 (65.4% winning), that is 34.69 points, which is better then someone that is 34/34 (undefeated). And someone that is 20/20 can go 9 for their next 20 and still grow in rankings.
"Wanting Red Rhino Pill to have gender"
Kleptin
Posts: 5,095
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:17:48 PM
Posted: 5 years ago
I think the whole concept of a leaderboard should be scrapped completely.
: At 5/2/2010 2:43:54 PM, innomen wrote:
It isn't about finding a theory, philosophy or doctrine and thinking it's the answer, but a practical application of one's experiences that is the answer.

: At 10/28/2010 2:40:07 PM, jharry wrote: I have already been given the greatest Gift that anyone could ever hope for [Life], I would consider myself selfish if I expected anything more.
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:19:36 PM
Posted: 5 years ago
At 12/26/2010 10:59:06 PM, OreEle wrote:
win^2/loses is a simple one.

But that doesn't take into account the relative quality of opponents faced. If I'm 23-4-17, but I debate primarily against people like theLwerd, RoyLatham, TheSkeptic, and Puck, I ought to have a higher rating than someone who's 30-0-0, but only debates noobs.
Zetsubou
Posts: 4,933
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:20:17 PM
Posted: 5 years ago
At 12/26/2010 11:14:02 PM, OreEle wrote:
At 12/26/2010 11:06:52 PM, Zetsubou wrote:
At 12/26/2010 10:59:06 PM, OreEle wrote:
win^2/loses is a simple one.
Hit's you too hard if you lose.

Personal opinion. I think that going 20/24 (16.67 points) is far superior to 25/45 (13.89), but not as good as 30/37 (24.32).

Loses actually don't hurt all that much when you've done more debates (since squares grow so much faster then linear).

You'll find that if you are 53/81 (65.4% winning), that is 34.69 points, which is better then someone that is 34/34 (undefeated). And someone that is 20/20 can go 9 for their next 20 and still grow in rankings.
I'll graph it.
'sup DDO -- july 2013
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:21:50 PM
Posted: 5 years ago
At 12/26/2010 11:19:36 PM, J.Kenyon wrote:
At 12/26/2010 10:59:06 PM, OreEle wrote:
win^2/loses is a simple one.

But that doesn't take into account the relative quality of opponents faced. If I'm 23-4-17, but I debate primarily against people like theLwerd, RoyLatham, TheSkeptic, and Puck, I ought to have a higher rating than someone who's 30-0-0, but only debates noobs.

See the point system one that I did afterwards
"Wanting Red Rhino Pill to have gender"
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:22:35 PM
Posted: 5 years ago
At 12/26/2010 11:07:16 PM, OreEle wrote:
I'm thinking a point system.

If you win a debate, you get 20 points + 1 point for every win your opponent has (at the time of the win, or always fluctuating, either way). Take the total number of points squared, and divide by 20*X (where x = the total number of debates you've completed).

This means that a win against a veteran is worth more then a win against a newbie. However there are still points to be had against the newbies.

This basic formula can be reworked with different numbers (say 10 points +1 for each win, or 15 or whatever).

That still places way too much emphasis on total number of wins. An Elo system is the closest you can get to rating debate skill on an objective scale.
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:23:09 PM
Posted: 5 years ago
At 12/26/2010 11:20:17 PM, Zetsubou wrote:
At 12/26/2010 11:14:02 PM, OreEle wrote:
At 12/26/2010 11:06:52 PM, Zetsubou wrote:
At 12/26/2010 10:59:06 PM, OreEle wrote:
win^2/loses is a simple one.
Hit's you too hard if you lose.

Personal opinion. I think that going 20/24 (16.67 points) is far superior to 25/45 (13.89), but not as good as 30/37 (24.32).

Loses actually don't hurt all that much when you've done more debates (since squares grow so much faster then linear).

You'll find that if you are 53/81 (65.4% winning), that is 34.69 points, which is better then someone that is 34/34 (undefeated). And someone that is 20/20 can go 9 for their next 20 and still grow in rankings.
I'll graph it.

If you want to take power away from total wins, and push it towards win% and people faced, do the same equation, but take the ^2 out of the equation.
"Wanting Red Rhino Pill to have gender"
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:31:06 PM
Posted: 5 years ago
At 12/26/2010 11:07:16 PM, OreEle wrote:
I'm thinking a point system.

If you win a debate, you get 20 points + 1 point for every win your opponent has (at the time of the win, or always fluctuating, either way). Take the total number of points squared, and divide by 20*X (where x = the total number of debates you've completed).

This means that a win against a veteran is worth more then a win against a newbie. However there are still points to be had against the newbies.

This basic formula can be reworked with different numbers (say 10 points +1 for each win, or 15 or whatever).

Plus, your system modifies the number of points awarded based on total wins, rather than on the debaters respective point totals, which doesn't make any sense to me. Additionally, it doesn't penalize someone for losing a debates. There's no way to fall in the rankings; it emphasizes seniority too much. The only way for a new user to get his ranking up is to do lots and lots of debates. An Elo system directly compares skill.
bluesteel
Posts: 12,301
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:43:19 PM
Posted: 5 years ago
At 12/26/2010 10:48:45 PM, J.Kenyon wrote:
I think we all know that the current leaderboard is flawed. The only factor it takes into account is the total number of wins, with no consideration for win-loss ratios or the quality of opponents faced. There's been a lot of griping over the years, but with Phil as the admin, we could sooner expect hell to freeze over before any substantive changes took place on this site. This thread will be for general and open discussion of new proposals. But first, let me outline my own plan:

The Elo Rating System

The Elo system was developed as a way to rate the relative skill of professional chess players. Since then, it's been adapted to other games and sports including professional football, basketball, baseball, and tennis. The average professional chess player has a rating of around 2,500. A typical club player is usually around 1,500.

When an Elo system is implemented, everyone begins with a default rating of 1,300. If you win a debate, your rating goes up; if you lose a debate your rating goes down. Beating someone with a significantly rating higher than yours will cause your score to increase more than beating someone with an equal or lesser rating would. Losing to someone with a significantly higher rating will have a minimal effect on either of your respective ratings. The differential varies proportionately.

Proposed Changes to Site Format

The all-time leaderboard would list every site member who has participated in at least one debate and generate rankings starting with the very first debate ever conducted. A quarterly and a yearly ranking system would be introduced listing the top 20 and the top 50 respectively. These would be listed somewhere on the main page.

Benefits of the Proposal

Because the best way to raise your ranking would be to engage other high-quality opponents, there would be less of an incentive for senior members to scoop up debates created by n00bs for the easy wins. This would increase overall debate quality and give new users time to get adjusted to the site.

And with that, I yield the floor.

I really like this proposal is some ways, but not others. I'd like to see more high quality debates on this site, but regular users rarely instigate open challenges, which limits opportunities. There is no tournament system like in chess (where if you win, you are guaranteed higher quality matches), although I'd like to see something like this on this site (also with switch side debating, where people don't just endorse resolutions/sides they agree with; maybe also a monthly topic that everyone uses like in competitive debate?).

I'm also a little skeptical because getting vote bombed on certain debates could really sink your rating (losing to someone who has a really bad record), and also, 90% of users on this site engage in less than 3 debates, so beating someone who is 2-0 really ups your rating, but beating someone in their first debate, and then having them lose two rounds in a row drops your rating? I dunno maybe this isn't true; don't understand perfectly how the system works. J.Kenyon?
You can't reason someone out of a position they didn't reason themselves into - Jonathan Swift (paraphrase)
m93samman
Posts: 2,685
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:43:55 PM
Posted: 5 years ago
I'm for the Elo. Seems most basic, easiest to implement (because we don't have to formulate anything majorly time consuming), and most accurate.

VOTES

Elo- 2 (J.Kenyon, m93samman)
Others- 0
: At 4/15/2011 5:29:37 PM, CosmicAlfonzo wrote:
: Pascal's wager is for poosies.
:
: I mean that sincerly, because it's basically an argument from poooosie.
:
: I'm pretty sure that's like a fallacy.. Argument ad Pussium or something like that.
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:47:47 PM
Posted: 5 years ago
At 12/26/2010 11:43:55 PM, m93samman wrote:
I'm for the Elo. Seems most basic, easiest to implement (because we don't have to formulate anything majorly time consuming), and most accurate.

VOTES

Elo- 2 (J.Kenyon, m93samman)
Others- 0


Since the Elo doesn't have a basic mathematical formula, but a formula for how much change should happen after a given win or loss, can it really be back applied to our current wins and losses?
"Wanting Red Rhino Pill to have gender"
m93samman
Posts: 2,685
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:56:53 PM
Posted: 5 years ago
At 12/26/2010 11:47:47 PM, OreEle wrote:
At 12/26/2010 11:43:55 PM, m93samman wrote:
I'm for the Elo. Seems most basic, easiest to implement (because we don't have to formulate anything majorly time consuming), and most accurate.

VOTES

Elo- 2 (J.Kenyon, m93samman)
Others- 0


Since the Elo doesn't have a basic mathematical formula, but a formula for how much change should happen after a given win or loss, can it really be back applied to our current wins and losses?

It could if it was implemented in the chronological sense that J.Kenyon mentioned. Begin with the very first debate.
: At 4/15/2011 5:29:37 PM, CosmicAlfonzo wrote:
: Pascal's wager is for poosies.
:
: I mean that sincerly, because it's basically an argument from poooosie.
:
: I'm pretty sure that's like a fallacy.. Argument ad Pussium or something like that.
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/26/2010 11:58:22 PM
Posted: 5 years ago
At 12/26/2010 11:47:47 PM, OreEle wrote:
Since the Elo doesn't have a basic mathematical formula, but a formula for how much change should happen after a given win or loss, can it really be back applied to our current wins and losses?

I don't see why not. An Elo script would go through the database in chronological order beginning with the first ever debate.
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 12:02:46 AM
Posted: 5 years ago
At 12/26/2010 11:43:19 PM, bluesteel wrote:
I really like this proposal is some ways, but not others. I'd like to see more high quality debates on this site, but regular users rarely instigate open challenges, which limits opportunities. There is no tournament system like in chess (where if you win, you are guaranteed higher quality matches), although I'd like to see something like this on this site (also with switch side debating, where people don't just endorse resolutions/sides they agree with; maybe also a monthly topic that everyone uses like in competitive debate?).

I'm also a little skeptical because getting vote bombed on certain debates could really sink your rating (losing to someone who has a really bad record), and also, 90% of users on this site engage in less than 3 debates, so beating someone who is 2-0 really ups your rating, but beating someone in their first debate, and then having them lose two rounds in a row drops your rating? I dunno maybe this isn't true; don't understand perfectly how the system works. J.Kenyon?

The way I understand it, Elo ratings aren't modified retroactively. If you beat someone rated 1,900, but later on he drops to 1,200, it doesn't affect you adversely. Vote bombing is a separate issue, but I see how it would be more important if we were to implement an Elo system. We could implement something like requiring a minimum number of debates be completed before allowing someone to vote, or create an option to only allow people with a certain ranking to vote on your debate.
tvellalott
Posts: 10,864
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 12:22:31 AM
Posted: 5 years ago
Mr. Kenyon has my full support with this system. My only concern was that it would be difficult to implement retroactively, but that's been addressed and isn't a problem.
"Caitlyn Jenner is an incredibly brave and stunningly beautiful woman."

Muh threads
Using mafia tactics in real-life: http://www.debate.org...
6 years of DDO: http://www.debate.org...
bluesteel
Posts: 12,301
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 12:23:42 AM
Posted: 5 years ago
At 12/27/2010 12:02:46 AM, J.Kenyon wrote:
At 12/26/2010 11:43:19 PM, bluesteel wrote:
I really like this proposal is some ways, but not others. I'd like to see more high quality debates on this site, but regular users rarely instigate open challenges, which limits opportunities. There is no tournament system like in chess (where if you win, you are guaranteed higher quality matches), although I'd like to see something like this on this site (also with switch side debating, where people don't just endorse resolutions/sides they agree with; maybe also a monthly topic that everyone uses like in competitive debate?).

I'm also a little skeptical because getting vote bombed on certain debates could really sink your rating (losing to someone who has a really bad record), and also, 90% of users on this site engage in less than 3 debates, so beating someone who is 2-0 really ups your rating, but beating someone in their first debate, and then having them lose two rounds in a row drops your rating? I dunno maybe this isn't true; don't understand perfectly how the system works. J.Kenyon?

The way I understand it, Elo ratings aren't modified retroactively. If you beat someone rated 1,900, but later on he drops to 1,200, it doesn't affect you adversely. Vote bombing is a separate issue, but I see how it would be more important if we were to implement an Elo system. We could implement something like requiring a minimum number of debates be completed before allowing someone to vote, or create an option to only allow people with a certain ranking to vote on your debate.

What about legitimately losing to a really good n00b? Wouldn't that screw your rating then, even if the new user's record later went up to 40-0? Seems like top users would only have a pool of 5 or so people they'd be willing to challenge. If theLwerd loses her IP debate to LaissezFaire, for example, this system would over-penalize her because LF's rating won't match his skill level until he has more debates. Or even more extreme, challenging annhasle (who only has done 4 debates) would be really risky, even though she is good.

I dunno, I get the feeling that the limited size of the DDO user pool and the small number of debates the average user engages in would lead to too much variability for a rating system adapted for people who engage in many many matches.
You can't reason someone out of a position they didn't reason themselves into - Jonathan Swift (paraphrase)
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 1:10:08 AM
Posted: 5 years ago
At 12/27/2010 12:23:42 AM, bluesteel wrote:
What about legitimately losing to a really good n00b? Wouldn't that screw your rating then, even if the new user's record later went up to 40-0?

Well, with the Elo system, everyone starts off with a score of 1,300 anyway. Remember, the average club chess player is usually around a 1,500 while pros are ~2,500. The only way one loss could seriously impact one's ranking is if a highly rated debater like RoyLatham loses to someone with a horrible rating; someone with a record of like 2-27, all the losses coming against similarly low-rated debaters.

Seems like top users would only have a pool of 5 or so people they'd be willing to challenge. If theLwerd loses her IP debate to LaissezFaire, for example, this system would over-penalize her because LF's rating won't match his skill level until he has more debates. Or even more extreme, challenging annhasle (who only has done 4 debates) would be really risky, even though she is good.

LF already had 13 or 14 wins under his belt and just 1 loss when he challenged her, so he'd probably have a decent ranking. Even if L is 2,800 like Kasparaov, LF would have been at least above 2,000. The chances of taking a bad loss are the same for everybody; unless someone goes on a really bad losing streak, it would tend to balance out pretty evenly.

I dunno, I get the feeling that the limited size of the DDO user pool and the small number of debates the average user engages in would lead to too much variability for a rating system adapted for people who engage in many many matches.

We could always test it out first to see what the leaderboard would look like.
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 1:25:04 AM
Posted: 5 years ago
At 12/27/2010 1:10:08 AM, J.Kenyon wrote:
At 12/27/2010 12:23:42 AM, bluesteel wrote:
What about legitimately losing to a really good n00b? Wouldn't that screw your rating then, even if the new user's record later went up to 40-0?

Well, with the Elo system, everyone starts off with a score of 1,300 anyway. Remember, the average club chess player is usually around a 1,500 while pros are ~2,500. The only way one loss could seriously impact one's ranking is if a highly rated debater like RoyLatham loses to someone with a horrible rating; someone with a record of like 2-27, all the losses coming against similarly low-rated debaters.

Seems like top users would only have a pool of 5 or so people they'd be willing to challenge. If theLwerd loses her IP debate to LaissezFaire, for example, this system would over-penalize her because LF's rating won't match his skill level until he has more debates. Or even more extreme, challenging annhasle (who only has done 4 debates) would be really risky, even though she is good.

LF already had 13 or 14 wins under his belt and just 1 loss when he challenged her, so he'd probably have a decent ranking. Even if L is 2,800 like Kasparaov, LF would have been at least above 2,000. The chances of taking a bad loss are the same for everybody; unless someone goes on a really bad losing streak, it would tend to balance out pretty evenly.

I dunno, I get the feeling that the limited size of the DDO user pool and the small number of debates the average user engages in would lead to too much variability for a rating system adapted for people who engage in many many matches.

We could always test it out first to see what the leaderboard would look like.

True, we could test out a variety of different methods to see how the leader board would look.
"Wanting Red Rhino Pill to have gender"
bluesteel
Posts: 12,301
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 1:44:16 AM
Posted: 5 years ago
At 12/27/2010 1:10:08 AM, J.Kenyon wrote:
At 12/27/2010 12:23:42 AM, bluesteel wrote:
What about legitimately losing to a really good n00b? Wouldn't that screw your rating then, even if the new user's record later went up to 40-0?

Well, with the Elo system, everyone starts off with a score of 1,300 anyway. Remember, the average club chess player is usually around a 1,500 while pros are ~2,500. The only way one loss could seriously impact one's ranking is if a highly rated debater like RoyLatham loses to someone with a horrible rating; someone with a record of like 2-27, all the losses coming against similarly low-rated debaters.

Seems like top users would only have a pool of 5 or so people they'd be willing to challenge. If theLwerd loses her IP debate to LaissezFaire, for example, this system would over-penalize her because LF's rating won't match his skill level until he has more debates. Or even more extreme, challenging annhasle (who only has done 4 debates) would be really risky, even though she is good.

LF already had 13 or 14 wins under his belt and just 1 loss when he challenged her, so he'd probably have a decent ranking. Even if L is 2,800 like Kasparaov, LF would have been at least above 2,000. The chances of taking a bad loss are the same for everybody; unless someone goes on a really bad losing streak, it would tend to balance out pretty evenly.

What would ann be ranked? Do rankings improve relatively quickly?

I dunno, I get the feeling that the limited size of the DDO user pool and the small number of debates the average user engages in would lead to too much variability for a rating system adapted for people who engage in many many matches.

We could always test it out first to see what the leaderboard would look like.

Ok, I think you've won me over.
You can't reason someone out of a position they didn't reason themselves into - Jonathan Swift (paraphrase)
Puck
Posts: 6,457
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 1:58:43 AM
Posted: 5 years ago
If you insist on a system where rank is meaningful, why implement a ranking system that basically undermines 2 years of ranking? The only way it would be fair to someone like theLwerd would be retroactively calculating one's position based upon all past debates up to the systems implementation. Otherwise this is just a reset button.
Puck
Posts: 6,457
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 2:00:27 AM
Posted: 5 years ago
Personally I'd be happier without a ranking system, or, simply have a way of notating tournament position per tournament you are in - since that may be a draw card for some.
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 2:09:33 AM
Posted: 5 years ago
At 12/27/2010 1:58:43 AM, Puck wrote:
If you insist on a system where rank is meaningful, why implement a ranking system that basically undermines 2 years of ranking? The only way it would be fair to someone like theLwerd would be retroactively calculating one's position based upon all past debates up to the systems implementation. Otherwise this is just a reset button.

I agree completely, I addressed that in my post. I wouldn't support using a new system if it couldn't be implemented retroactively.
J.Kenyon
Posts: 4,194
Add as Friend
Challenge to a Debate
Send a Message
12/27/2010 2:27:08 AM
Posted: 5 years ago
At 12/27/2010 1:44:16 AM, bluesteel wrote:
What would ann be ranked? Do rankings improve relatively quickly?

Well, she's beaten Yvette (16-2, not counting her loss to ann) and Zetsubou (10-5 leading up to his loss to ann), so it would be quite a bit higher than average. It probably wouldn't reflect her actual skill, since her rating would be calculated from only four debates, but it would be a lot closer to it than the current system places her.

The other point I wanted to make is about this debate: http://www.debate.org...

If theLwerd were to lose, her rating (if not her ranking) would drop a bit. Sieben, on the other hand, would rise significantly since it was his first debate and it's against the top ranked debater. Obviously, it's a good thing that new debaters would be able to quickly climb in the rankings; Sieben is currently ranked 190 when it's clear he should be much higher. I also think it would be a good thing to have the leaderboard shaken up a bit from time to time. As things currently are, it's unlikely that anyone will ever come close to displacing L, or even break the top 5. It would make things a bit more exciting, knowing that your ranking is at stake in a certain debate. Plus, since the rankings would be more meaningful, people would be more likely to strive to reach the top.