It seems to inflate the meaning of what may amount to a negligible difference in skill.

well, the concept is good. Sometimes we look at things too critically but forget to just glance at the bigger picture.

How do you effectively calculate the top 25 players by 'talent' - while minimizing the luck factor ??

Well, with the old system, you could be a great player with a lot of talent. To get to the top you had to play flawlessly and you played a significantly fewer amount of games because you tried to protect your rating.

It was extremely easy to drop from 2nd to 27th in one bad game...

That's the flaw. Shouldnt a talented player slowly drop off the top 25 if they can't defend their position over time? Or should they be penalized because they lose a crucial 5v2 in round one that ruined their game and cost them a 20 place drop...

The top 25 now more accurately represents the talented over the long run. So for instance, player A gets to the tops of Elo rating and wins a game there. That's hard to do, very. Now if that player plays again and gets on a bad luck streak, they drop down. (now when they win they dont add to their rank much cuz their elo is lower - but the scoring system respects the hard work they did before, since getting to 1st wasnt easy)


So the next concern is, well how the hell are we supposed to catch up to the people up top? it seems they have a crazy lead.

Well... you COULD play 400+ games and stay hovering around 100th place

or climb up in Elo rating so that you are #1 (like in the old system) and get massive additions to your rank in doing so


make sense?

@Grunvagr
Yes, I agee with the concept. The issue We raised here was that the 1/x function
"seems to inflate the meaning of what may amount to a negligible difference in skill", to quote no_wolf, as a result of ranking by rating being a relative hierarchy.

a) #1 1920, #2 1915, does this justify #1 getting 1 versus .5?

b) are points earned as #1 at elo 1900 are worth as much as the one earned by a #1 at 2100?

These are corner cases, nobody asked for abandoning the system or having it changed in the next 24hr because of these things. We just pointed them out so they can be discussed.

I personally like the new ranking for the reasons mentioned in a previous post.

"or climb up in Elo rating so that you are #1 (like in the old system) and get massive additions to your rank in doing so "

Yet those who _start_ at the top (especially #1) have a massive headstart simply from being in the right place at the right time. The "top player" score differences are more about that head start than anything to do with skill differences among the top players. That is the biggest problem with the new system, I think. Adding needless complexity to an already complicated scoring system is one thing, but it seems a bit unfair as well.

vengeance, while I see where you are going with your point, consider this

In the old system, highest elo = first.

So now the new rank thing gets put in, and the people at the top with the highest Elo score the big points first. Well... they would be at the top ANYWAY. That's my point.


as for your example with the person at 1950s would be in the top 5, so when the new system gets put in, isn't if fair for that person to be able to score more rank by coming in 3rd, 5th, and getting .33, .5, etc?

basically, it wasn't "easy" to have gotten to a very high elo in the first place - so it's not excessively rewarding anyone in a sense

DealornoDeal -
a) yes. assume someone is in 2nd place (elo) by 10 points. they shoot for first and lose and drop a ton of points, well they lose their chance to score big on rank by adding 1 for being first place in elo, games at the very highest point in elo are extremely risky, since EVERYONE at the table has a lower rating than you so you already go into the game needed 3rd or better to score any points, 4th is serious negatives usually. Does it suck to be shy of the #1 person's elo by one or two points and thus not get the huge boost to your rank, yes (but you'd be 2nd in the old system anyway, a couple points shy of them anyway) tough situation but one you just gotta deal with

b)Yes - and here's why. Getting to 1900 elo (and being tops) when everyone else has 1800s or less is very difficult, getting to 2100 doesnt necessarily imply more skill, just implies youve played more games with 2000 skill players or thereabouts and have climbed to 2100

basically, 1900 or 2100 dont matter - they matter how they RELATE to whoever the top 10 are

(1900 elo, 2nd is 1840, 3rd is 1822)
well, that person has a huge lead and hard worked hard to climb that high cuz who has he played games against? 1700ers and some 1800

(2100 elo, 2nd is 1980, 3rd is 1977)
same thing - its just a matter of climbing above all others


Last point: if you are in first in elo and win a bunch of times (or even get +0 in games) you deserve to pad your 'rank' Because you are playing on tables where EVERYONE has lower elo than you, thus you risk dropping to 2nd elo

(note that THEN, whoever was #2 is now #1 in ELO and if they just play a game and get +0) they add 1 to rank...

"In the old system, highest elo = first.

So now the new rank thing gets put in, and the people at the top with the highest Elo score the big points first. Well... they would be at the top ANYWAY. That's my point.


as for your example with the person at 1950s would be in the top 5, so when the new system gets put in, isn't if fair for that person to be able to score more rank by coming in 3rd, 5th, and getting .33, .5, etc?

basically, it wasn't "easy" to have gotten to a very high elo in the first place"

Yes, but the problem appears when you compare the lucky person who _starts_ in first (at reset), to all the other players who are of about equal skill level. There are a lot of players who have reached first or who will reach first, but the one who happens to do it just at the right time has a huge advantage over the rest of those players in terms of the new scores. The same also goes for every other position on the list, but with less extreme effects because there's a less extreme modifier for those positions.

Think of the people at first place when the ranking started getting a reward for first.

Besides 1 point isn't a whole lot. I'm sure we'll see much higher values from other players.

One point's a lot if you're 3rd or lower...

hey ryan, remember how it showed me as 4th but I was actually in 3rd? it showed the same for rewind, displaying him at 2nd instead of 1st, which he is.

seems some people are off by 1 when they sit down, something to check if u get a chance, thx

Grun, I think i know the problem. Its a floating point issue. The database value doesn't exactly match the in memory value. Easy to fix when I restart next.