Spidey's Command List

[Terralthra]

QUOTE (jgbaxter @ Dec 14 2007, 05:34 AM) You still aren't reading simple words dude. /laugh.gif" style="vertical-align:middle" emoid=":lol:" border="0" alt="laugh.gif" />

AllegAge = Experience

No, it really doesn't. Unique days logged in is perhaps the 2nd worst method of counting experience possible, with the absolute worst being "time since first login." Games played is better than AllegAge, by so significant a margin that it's not even really in the same sport, let alone the same ballpark. And that's even setting aside the primary piece of the rationale for using AllegAge, which is the tenuous connection between experience and skill.

QUOTE (jgbaxter @ Dec 14 2007, 05:34 AM) Not skill per se. And in a general fuzzy sense just like ANY RATING SYSTEM will attempt to measure skill, when it will only measure wins.

Skill has a rather obvious and fairly consistent connection with winning (since 'skilled' can be defined as capable of winning, and is defined as such in a rating system); experience does not. Experience has a tenuous and inconsistent relationship with skill, as shown by the multiple examples given in this thread, which you have simply attempted to evade. Until and unless you can answer these (frequently encountered) examples where your system would be obviously and definitely inaccurate, your system isn't even worth trying.

QUOTE (jgbaxter @ Dec 14 2007, 05:34 AM) If you're not getting it still let me draw another fun rough 'n' dirty example. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

Firing ab at base and realizing scouts generally don't carry abs = experience.

Firing ab at base without lock and including speed in your range of 20mps for .1k distance while dodging mines and pu podded scouts = skill.
Knowing what a nan is and how to nan = experience

Knowing you can nan a drone while circling around it between you and the int trying to kill it while naning it and dropping mines and lobbing a df or 2 = skill.

And what is the mathematical relationship between 'number of logins' and your 'experience' definition quoted above? At how many unique logins has someone figured out that scouts don't carry AB? How many unique logins does it take for someone to realize that the scan range of ships is 25% when approached from behind? If you have a two people with 100 unique logins each vs. someone with 150 unique logins, does a person with 50 unique logins make it balanced? These are necessary and relevant questions that need rigourous answers before your idea can even be considered as a viable 'rating system'.

QUOTE (jgbaxter @ Dec 14 2007, 05:34 AM) AllegAge is not based on their join date, but frequency of logins. And I'm not advocating (as I've clearly sad many many times) to use AllegAge solely, rather to use AllegAge (or a similar mechanism at the least) as a component to any ranking system.

Their (#) should last longer then a week or two, it would be nice if people without (#)'s knew what a nan is.

So, if a newbie has a crappy computer that crashes Allegiance every 10 minutes, and rather than fix it, they just reboot and log back in, they'll be an inter. 2 on the first day. Brilliant! The problem with your system is that no matter how it is formulated, it is pathetically easy to find an example of something that renders it completely and utterly worthless. Many vets these days leave themself logged in for quite some time, just going afk and leaving themself in the game lobby, for example.

QUOTE (jgbaxter @ Dec 14 2007, 05:34 AM) TrueSkill has nothing to do with Allegiance as far as I've read. That would be like saying Internet Explorer is great to use with Allegiance as they are both made by the same company. TrueSkill / AllegSkill will likely be better or equal to helo, that's never been my contention otherwise. Simply put that any new ranking system should look at as many facets of the game as possible in order to make the most accurate measure of a players rating.

Given that TrueSkill was created specifically to determine an individual's skill rating based on a series of game outcomes in which said individual was a member of a team, I'd say it's pretty directly suited to rating Allegiance players.

QUOTE (jgbaxter @ Dec 14 2007, 05:34 AM) Of course another issue is this 'rating team of people' have also said that they may use data from old games to generate ratings, in which case AllegSkill is poo.

It's interesting how people who have no real math or statistical background at all are judging a system without any real knowledge of it, while people with direct math, physics, and statistical experience are withholding judgement until they actually can see how the system works.

[HSharp]

QUOTE (jgbaxter @ Dec 13 2007, 07:34 PM) TrueSkill has nothing to do with Allegiance as far as I've read. That would be like saying Internet Explorer is great to use with Allegiance as they are both made by the same company. TrueSkill / AllegSkill will likely be better or equal to helo, that's never been my contention otherwise. Simply put that any new ranking system should look at as many facets of the game as possible in order to make the most accurate measure of a players rating.

Of course another issue is this 'rating team of people' have also said that they may use data from old games to generate ratings, in which case AllegSkill is poo.

And how much did you read???

How is using data from old games going to make AllegSkill poo??
AllegSkill hasn't arrived yet but to criticise it without even doing a little bit of research is stupid. I don't know exactly why 'the rating team of people' are using data from old games but using a lil bit of common sense I would take guesses that it could be to generate test data to see how HELO and AllegSkill compares, it could be to further refine AllegSkill, it could be that starting off everyone as rank 0 might be a little bit silly, it might be to make the current HELO rankings more accurate.

However I fail to see how any of that (or other possible scenarios) make AllegSkill poo (unless somehow faeces is used to power the special server used to power up AllegSkill to calculate data).


Criticising something that has already been done or is currently in place is one thing, criticising something that you really don't know anything about and have not even seen in action yet just means you are really sucky at arguing.

[Raveen]

QUOTE (jgbaxter @ Dec 13 2007, 07:34 PM) You still aren't reading simple words dude. /laugh.gif" style="vertical-align:middle" emoid=":lol:" border="0" alt="laugh.gif" />

AllegAge = Experience

Yep, sorry a misthink there on my part. Let me rephrase myself.

"No it doesn't. AllegAge is a simple function of how many unique days a player has logged in on. That has no relationship to experience whatsoever. Which is why people get frustrated when you suggest that it magically does. There may be a very fuzzy correlation (and I actually doubt it beyond the first 3 months of play) but that's all. Certainly I'd be surprised if there's any link substantial enough to base a ranking system on."

Experience is best measured by games played, not days when someone has logged in. I suppose ideally one would measure it based on minutes played but that's by the by.

Now you also say that a system like HELO does not measure experience. This is not true. Let's simplify things for demostration purposes.

HELO measures games won. Let's say that skill can be defined as what proportion of games played are won. A good player may win 67% of their games played, a bad player 33%. We'll ignore stacking and so on because they add complications that, although very much grounded in the real world Alleg experience, would only confuse the issue in hand. Now then. The way HELO works is to award points for a win and deduct them for a loss. A new player loses less points than they would gain leading to a gradual increase in rank over time.

Now you want a system that takes into account both skill (proportion of games won) and experience (games played). HELO obviously measures the first but by strictly limiting all other factors we see that it clearly measures the second as well. With our ideal players winning 67% and 33% of their games respectively their HELO scores will both rise in a straight line which is proportional to their experience. The rate of point accumulation is defined by the player's skill. The fact that points accumulate is due to experience.

So not only does HELO measure experience (defined mathematically based on your comments, you're never going to not have people play a huge number of games and still not know stuff about the game) it does it more accurately than AllegAge does because it's based on games played rather than unique days logged in.

That is why I think adding AllegAge to HELO is a bad idea. It is both redundant and less accurate than what's already being measured.

AllegSkill will do something similar in terms of measuring experience but that'll have to wait for Baker to explain it because he actually understands it.

[jgbaxter]

Terra, I saw you posted, I didn't read it but thanks for taking the time to post it. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

Rav, I understand what you are saying, and I disagree with you. We're now just going in endless circles, thanks for the discussion. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

QUOTE (HSharp @ Dec 13 2007, 01:49 PM) How is using data from old games going to make AllegSkill poo??

Simple, any system that has it's handlers say it needs AB to work is inherently going to be screwed over and inaccurate if it uses data for games that didn't use AB, and where the information available for people to join the games was also wildly inaccurate (since they were viewing ranks based on one set of information and those ranks are changed NOW but the decisions made THEN are slaved in).

Any new rating system used must use new data, otherwise you are using the system to magically calculate odds of success based on data that is old and inaccurate in and of itself, and is in most cases not how the teams would have formed in the first place had the information been made available at the time.

Of course, there's ways around that and I think Baker might be onto something if ranks are actually not shown to the community it might actually work out to a certain degree... however I seriously doubt it will be an improvement over new data that isn't corrupted by situations that only took place because the Earth was round one day and flat the next,

Ah well, something to think on. /mrgreen.gif" style="vertical-align:middle" emoid=":D" border="0" alt="mrgreen.gif" />

[pkk]

QUOTE (jgbaxter @ Dec 13 2007, 10:46 PM) Any new rating system used must use new data, otherwise you are using the system to magically calculate odds of success based on data that is old and inaccurate in and of itself, and is in most cases not how the teams would have formed in the first place had the information been made available at the time.

Why did we collect 1 GB XML data just on EU GPZ?!

There are a few more games on other servers... not using old data is stupid. /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" />

[factoid]

Well, I'm convinced. If bax disagrees, then that's good enough for me. Terr, Baker, MrC; I've got nothing but respect for you, but it's clear you can take your logic, testing, and statistical methods and shove 'em. It's painfully obvious from bax's disagreement that your ideas are flawed and unworkable.

[Raveen]

QUOTE (jgbaxter @ Dec 13 2007, 09:46 PM) Rav, I understand what you are saying, and I disagree with you. We're now just going in endless circles, thanks for the discussion. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

Ok, where do you disagree? You want experience measured in the ranking system. HELO already measures it and that AllegAge would be a bad way to measure it. So what exactly do you disagree with? Do you still think that AllegAge is the best way to measure experience? If so why?

Do you maybe want experience based progression to be slower? If so what would that actually achieve? You want players graduating from the newbie server to have a basic level of knowledge that you feel they currently don't have. How would enabling them to play more games on the newbie server help? According to your supposition, they don't actually learn this stuff there anyway.

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?

[jgbaxter]

QUOTE (factoid @ Dec 13 2007, 03:03 PM) Well, I'm convinced. If bax disagrees, then that's good enough for me. Terr, Baker, MrC; I've got nothing but respect for you, but it's clear you can take your logic, testing, and statistical methods and shove 'em. It's painfully obvious from bax's disagreement that your ideas are flawed and unworkable.

Wow, many RTs are really into the passive-aggressiveness aren't you, ever since Badger left you guys slide further into the dark. Which sucks since so many of you are also good guys. Ah well, nothing like marching in goose-step shoulder to shoulder right into the abyss. Very nice.

In case anyone doesn't know, I don't have to flaunt my real life intelligence or ability.

And Terra, I've worked with statistics for over 10 years, so, whatever.

I'm out. Too many of you are too close to this in your heart.

[Terralthra]

QUOTE (jgbaxter @ Dec 14 2007, 08:11 AM) And Terra, I've worked with statistics for over 10 years, so, whatever.

If you'd actually worked seriously with statistics for 10+ years, you'd not evade simple and rational questions about your dubious idea. You'd just answer them, as they'd be easy questions to answer if you had over a decade of statistical experience to draw on, and your system has any real viability.

[Cadillac]

I've "worked with" statistics for 4 years, and those of you who know me will just laugh (ok shutup, stop laughing now /tongue.gif" style="vertical-align:middle" emoid=":P" border="0" alt="tongue.gif" />).

But the heart of my point is that it's easy to mislead people by saying "I've working with a for b-years" and give them the wrong connotations.