Free Allegiance Forums

yes you are proposing one system. if you are not please shut the $#@! up.

Evincar wrote:QUOTE (Evincar @ Dec 13 2007, 10:29 PM) yes you are proposing one system. if you are not please shut the $#@! up.

/laugh.gif" style="vertical-align:middle" emoid=":lol:" border="0" alt="laugh.gif" />

No, I'm not. And, I don't have to shut up because you want me too. Seriously, tough nuts if you don't like my idea. /tongue.gif" style="vertical-align:middle" emoid=":P" border="0" alt="tongue.gif" />

you are doing a comparison between two groups of people based on 'something'. if that something is not what you are proposing (your age based rank system) then stop trolling this. if it is, then you are quite retarded.

Oh, personal attack plus one.

*dies of deppression*

/tongue.gif" style="vertical-align:middle" emoid=":P" border="0" alt="tongue.gif" />

you made a poor attempt to derail the argument. are you basing your comparison on anything else than thin air?

jgbaxter wrote:QUOTE (jgbaxter @ Dec 13 2007, 11:11 PM) And Terra, I've worked with statistics for over 10 years, so, whatever.

Wiping your ass with old printouts doesn't count.

Raveen wrote:QUOTE (Raveen @ Dec 13 2007, 10:08 PM) Can you post up something to support your position in an analytical style like my post so that I can at least get an idea of why AllegAge would measure experience where HELO etc have not?

I take it that the answer to this question is "no" then?

I wonder if that's because there simply isn't an analysis that would support using AllegAge.

I've tried very hard to keep my temper and debate this subject in an open and constructive way. I felt I'd made some progress by first of all getting the idea that experience is what Bax wants to measure and then showing that HELO already does that. I rather thought we might reach an understanding but unfortunately Bax chose to flame and be flamed in turn rather than enter a sensible debate. Shame but there you go.

Oh and as for the perceived slights by RT, well provocation goes a long way doesn't it Bax? Particularly amusing is you taking the high ground towards a squad that you jumped from before you were kicked for being a jerk.

spidicles please don't lock this, I love it when people I don't like go out of their way to demonstrate just how stupid they are.

Also, I think Raveen secretly belongs in XT.

The history of team sports has always ranked individuals based on individual performance. (ie, stats) There is usually a modification to the ranks based on the team performance, but primary rank is determined by stats. Stats as in , goals scored, goals defended, assists, etc. Some of the greatest players of American football have been on teams with losing records. Yet, these players have been recognized as the best. Not because of their teams win-loss record but because of their individual great performance in spite of the team loss.

No matter how you cut math, basing individual ranks on the team performance alone will produce inaccurate results. Feel really bad for the greatest players in NFL history who under Elo/Helo and now TrueSkill would be mere foot notes in sports history instead of being recognized as the great player they were.

Steve Largeant, by any means one of the top 100 greatest NFL players of all time. http://en.wikipedia.org/wiki/Steve_Largent. Under Elo,Helo,Trueskill ranking Steve Largeant would have been a nobody because his team had a losing record during his years played. http://www.nflteamhistory.com/nfl_teams/se...ar_results.html There are many other great players in the history of team sports with similar records. A pity all the great players would be ranked in the gutter because of a TrueSkill based system.

The number of people (and very smart math people) that care about the accuracy of ranking individuals of team sports far exceeds the number of people that care about Allegiance ranks or ranking on Microsoft Live service with TrueSkill. I would venture to say that up to a multi-million TIMES more people have worked on making individual rankings accurate for team sports then the number of people working on TrueSkill.

Microsoft Research is simply that, research. Does not mean Mircosoft's Trueskill is accurate for individuals in team games. It means Microsoft is experimenting for specific team games. I put my money and faith in the hundreds of years of ranking individual team members based on individual performance and not team record. I hear folks hear putting in absolute faith in a system just because the system has Microsoft name on it. Wow, just wow.
Following the logic being developed here...the hundreds of years and millions of people that have been ranking individuals of team sports are wrong.

Bottom line as I see it, all three systems were/are being pushed forward so that medicore performers can have higher rankings just for being on a winning team. Pity the great players of all time such as Steve Largeant who proved greatness by his actions as recorded by stats and not because he got drafted by a winning team. When I brought up this issue about real sports to Pook. Pook's logic was, "Well Allegiance is not real sports" is called refusing to face the cold hard logic of Elo/Helo/Trueskill is wrong for ranking individuals in team based games. Allegiance and real team sports are not different enough to warrant radical different ranking systems. How about showing and talking about how/why Allegiance is so different from real sports to warrant a different ranking system then coming off with a reply that does not address the issue? I just don't get how such smart people like you guys doing this development are overlooking such very simple logic.

Go with the millions of math folks and go with the history of individual rankings for team sports instead of the very few people performing an experiment called Trueskill at Microsoft. I don't care if Albert Einstein developed Trueskill, Trueskill is wrong in theory and here is a situation (steve largent) where Trueskill is wrong in fact. Tis not about complicated math, this is about simple logic. Elo,Helo and Trueskill rewards mediocre players for being on a winning team and punishes great players for being on a losing team. Pook argument to this last sentence was, "Well Frag, you got on the wrong team". Talk about encouraging stacking.

Show the math that will put someone like Steve Largeant in the top 100 rankings of all time for the NFL based on Trueskill. If nothing else, show some math about why Allegiance is so different from the team sports of history to warrant this departure from the proven and accepted way of ranking.

Elo,Helo and now Trueskill are being experimented with and forced upon the Allegiance community for two reasons:

1. Scouts whores do not get enough points under the original Allegiance points system.
2. The admin was a scout whore.

Highly doubt your development effort would get the data from ASGS games if your effort was based around ranking players based on player performance instead of team performance. I know of a couple efforts for Allegiance that were going in the right direction of rankings based on stats, but support from ASGS seemed to be denied. The "D" word was tossed about for very good reason. Your getting support (with ACGS data, etc) because because Trueskill is based on the same principle(elo) as his effort.

Working on improving the collecting of scouting data and rewarding scouting actions with stats is the path to achieve the accurate ranking we all desire. Many vets have said this. This system of rewarding average players for being on a winning team and penalizing great players (such as Steve Largeant) for being on the losing team is wrong in principal, in fact and in history. Propose Trueskill to the real world of team sports and you will be ridiculed as using math to hide flawed logic. Millions of mathematicians work on ranking individuals of team sports...millions. All those mathematicians proven incorrect by a few at Microsoft and a few in an unknown community called Allegiance. Simple logic is being overlooked here in development of rankings for Allegiance in favor of complicated math.

We can answer a lot of the questions about what the community really wants by a vote. Vote on what type of ranking system. Vote on if no ranking system period. Hold up a few doctors of math and Microsoft as proof of a system? I hold up hundreds of years history and millions of mathematicians of a proven system. Both are just opinions.

You don't know the masses opinion until a valid vote is put forth. The "D" word is tossed around for very valid reasons. Huge decisions that affect the whole community in very negative (Elo,Helo, Trueskill IMO) ways are imposed without the will of the community.

Wants of the few or wants of the masses...

A few points:

1. Organized sports has the problem that players don't move teams very frequently, so it's very possible to play on a losing team all your career. In allegiance, every team is a new team, just about -- yes, there's some correlation in that certain players like to fly together, but it's much easier to tell in allegiance that if the team you play on never wins, you can't blame the team, because the only common factor is YOU. This is one big reason why TrueSkill wouldn't work very well in sports -- every game is the same players on the team, each member has the same win/loss record, and TrueSkill only cares about wins and losses, so every member of the team would get the same rating.

2. Boiling everything down to one number is hard. In sports, there is never an objective ranking. You can compare individual stats -- who is the best scorer, who is the best free throw shooter, who's the best defender -- but comparing across these boundaries it's impossible to avoid the subjective element. You have to take a poll. That's what they do in college football every week. I don't think we want to do that for allegiance though. Is there more skill in being a scout whore, int whore, miner whore, turret whore, bomber whore? People's opinions will differ. The only way to be objective about it is to look at results: wins and losses.

3. I'm not saying that it wouldn't be cool to keep some of these stats. I'd like to go and see fifty leaderboards, who's the best nan, who's the best miner killer, who's the best nan killer, who spots the most sh*t, who picks up the most tech etc. etc. It'd be cool, because maybe I'll never be a great all-around player, but I could work my way on a specific area that I'm good at and get recognition for it (pick up Retro Booster every game FTW!). But I just don't think it should contribute the RANK, that bottom-line number that aggregates all of those things together, because any formula to put that together is ad hoc and arbitrary.

4. TrueSkill isn't very complicated math. You say that it is, as if there was some wizardry behind it, but conceptually it's quite simple. Everyone has an innate skill, and some people's skill is better known than others. The tricky part is proving that it's fair, and that it's zero-sum, so that there's no ranking inflation that goes on over time. But that's already done so you don't have to worry about it. :-)