Modo44, on 30 April 2014 - 12:31 PM, said:
You are reinventing the wheel. Only Elo tracks all skills that help you win. Literally any other system, however complex, will miss some details and put too much weight on others. It will also likely require more processing power to use. Elo is sound, the issue is with all the randomness surrounding it (matchmaker derping, MWO instability). The launch module is a first step towards reducing that randomness. Given the
discussion so far, many more will follow.
If you say so, like I said...I'm not a coder. I have no idea how difficult or complicated coding something other than stuff in old time land (Power Basic, COBOL, FORTRAN), I haven't got a clue.
What we DO have, or at least what's been shown to me prior to the patch, was a system that was inherently overcomplicated, difficult to change skill "brackets" (insofar as not being stuck with brand new players FOREVER) and not really applicable to the playerbase at any given time other than an area's "Prime Time."
The only way your Elo goes up or down is whether or not you end up on a team that does other than what the Matchmaker predicts. Lose if predicted to win, win if predicted to lose. And, even then, it only changes by +/- 50 at best. The maximum Elo is, I believe, 2900. You start with an 1800 as soon as you're done with your first ten "Derplamore/Nooblering Cadet Matches." The matchmaker is given a spread of 1400 to "fill in slots" in any given match. That means that unless you are the very top tier, you're stuck with brand new players FOREVER...or what feels like it, at least.
All I was suggesting was to use the existing coding that shuffles people to one team or other based on whatever value you're given by the whole Elo process. Whether it's a 2800 or a 4 or A2 or whatever vaule you're given. That coding could be used repeatedly, as a filtering process, and more simplistic values could be used.
Take, for instance, the whole "level" concept. Right now, we do the whole Elo thing...as I explained with my 4th grade writing level

up there. If, instead, the assignment of a "level" based on a quantifiable value and possibly modified based on a secondary quantifiable value, then one would think that the coding to make the choices would be simpler and quicker.
Put the levels in, say 10,000 GXP increments. Still a pretty wide range but....unless you're some kind of prodigy that's invested all of Mommy's credit card into the game....you're going to be playing with people in your GXP range for a minute before you either rise to the next group of people or drop to the next lowest. You'll either be at the head of your group, getting ready to progress to the next tier (let's say 3000GXP increments) or filtering to the next lowest group. Seems rather simple to me.
As far as the "Launch Module" being a first step toward reducing the randomness, let's take a quick look at what we've got here.
Ok...cool, it lets you launch a private match. I've got my time working and the ability to set up a match looks pretty cool. Shame I'm a solo PUGger, I guess. Shame they didn't set up anything that lets someone opening the lobby to allow it to look for random players with random variables like weight class, ecm, etc.
The patch notes state that the primary change to the Matchmaker is the addition of the 3/3/3/3 rule followed by a "slight evening by the Elo process." Whatever the hell that is. As of right now, since the 3/3/3/3 filter isn't even enabled because it was released half-finished, barely tested and overcomplicated (like EVERY OTHER SINGLE THING THEY'VE DONE, EVER)....so....that means our matches are being put together by a Matchmaker using a "slimmed down and more efficient" Elo system (read that as "incomplete, not truly a priority in matchmaking)?
Which is kind of my point. Why is it that everything PGI does seems to be overcomplicated and/or overthought? Sure, I know there are variables that are not addressed in what I suggested, but they seem to be able to add additional filtering codes into stuff...we've seen them pull that off frequently. But they're far from complicated, man.
Edited by Willard Phule, 30 April 2014 - 01:01 PM.