x
ETF2L is looking for new Staff to recruit! Are you interested in supporting the league? Then click here for more details on what we can offer and how to apply! 

Forum

TF2 Player Rankings website - e-peens at the ready

Created 10th January 2015 @ 23:38

Add A Reply Pages: « Previous 1 ... 5 6 7 ... 14 Next »

Sideshow

(ETF2L Donator)
ft.
WiK?

I’m very worried if you’re not trolling InstantMuffin.

I’ll try to make this simple for you. If I know that damage is the biggest factor in the rankings, then I will game the system by only playing in a way that nets me big damage numbers. If I don’t know how the rankings work, I will not attempt to game them as I have no idea how. I will just play the game and see what happens. It’s blindingly obvious that this is the case; an open system would be open to abuse.

Your point that you “backed up with literature” is regarding reliability, not ability to skew results. Yes, if there were several of these ranking systems and the others were more clear, then perhaps Jon would have to divulge more info to get his system respected. But there are not competing systems. And everyone knows how much work, time and thought Jon puts into his stats. So (thankfully) he doesn’t need to reveal information.

RTC

Players are better ranked by interpretations by a human, not statistics generated by a machine. Statistics can be used as evidence that can be used to express an interpretation, but they are absolutely not the interpretation themselves, and I think a lot of people don’t understand this.

I could go on for days explaining all the various variables I could think of, but I’ll give you one example. A Demoman has a bad K/D Ratio, his team loses, and he doesn’t do much damage. This is all statistics can pick up. What statistics can’t pick up is that the Demoman’s sticky traps cost the opposing team a lot of time to capture certain points, that the Demoman had little to no protection, that the Demoman was not healed in the right situations, that the Demo had perfect rollouts, that he had great comms, that the Demoman was prioritized by the enemy team because of his annoyance to their objective etc etc etc. There are so many essential elements to determining how good a player is that cannot be interpreted by simple statistics. As I said before, statistics can be used as evidence to support a theory that a player is of a certain skill level, but the thing is, people are treating them as if they’re more important than human interpretation, which is ludicrous.

I’m not vehemently against the idea of a statistics based system in TF2 at all. I just feel that the ethos of it being a ranking system gives the wrong impression and that the ethos of the project should be modified to support the ethos of it being evidence of whether a player is good or not, not the conclusion of such an interpretation.

(this post was much shorter than I was expecting actually, probably because I didn’t explain all the various examples that were swirling around in my mind).

Kuferl

(TF2Pickup.net Unicorn)
FLANK

Knowing how jon’s system works, i can say that the system could be gamed quite easily if you know how it works. Therefore it is quite needed to keep the important parts secret.
I guess if you find an accurate system yourself feel free to publish it =P

About tf2pickup:
I haven’ touched the matchmaking for quite a while, but for 6s we’re using this system, atleast for calculating each persons rating. The algorythm for balancing teams isn’t perfect yet, but i think it’s not too bad. Guess what most people confuses is that we display avg. div per team, what doesn’t influence the balancing at all.

InstantMuffin

Quoted from Gentleman Jon

[…]
I’m sorry I should have been clearer, I’ve applied secrecy because I know it’s potentially vulnerable and because I also know that I’m using stats in a way that has not entered public discourse, so secrecy is an excellent way to preserve their efficacy. I shouldn’t have abbreviated this to “this process”.

I’m not sure why you would attribute a desire for personal glory to this failure to anticipate your criticism, I’ve been trying to get something implemented in the background for at least 6 months, and running the site is nothing but a headache.

Have you ever considered collecting necessary data plugin-wise?

Kuferl

(TF2Pickup.net Unicorn)
FLANK

Quoted from InstantMuffin

[…]

Have you ever considered collecting necessary data plugin-wise?

What difference would a plugin make to logs? not like logs got literally every stat you can collect already, and improvements coming apparently as well o/

taube

BWB

My little suggestion for the site :

Add a “log-in through steam” button, with which you can check your score just by clicking on your profile, so you don’t have to copy and paste your community number every time you want to check your stats/ranks. It could also have the function of checking your friends stats/ranks aswell.

Just my little idea.

fraac

JOHN
CENATION

RTC, your examples are good for players in teams, but for players in pickups they can quickly get in enough situations with enough different teammates to produce a useful rating.

The best possible system isn’t gameable, I think. You would just find yourself doing what it takes to win. But when you throw away variables to make an efficient system it means you can do less winny things and still rate well.

Gentleman Jon

Quoted from Kuferl

I haven’ touched the matchmaking for quite a while, but for 6s we’re using this system, atleast for calculating each persons rating.

I should point out that this is an early version lacking some of the improvements that have been made over time =D

idk i just looked at the ranking and the players before and behind me and it seems ok and perfectly useable for a pickup/lobby site.
dunno what the problem is i dont think jon ever said this rating is absolutly accurate and without mistakes.

CHERRY

Quoted from RTC

Players are better ranked by interpretations by a human, not statistics generated by a machine. Statistics can be used as evidence that can be used to express an interpretation, but they are absolutely not the interpretation themselves, and I think a lot of people don’t understand this.

Cmon, everything people do, a machine can do too – it’s just a matter of time and resources and it just happens that there are still things that are cheaper done by humans and interpretation of statistics IS NOT one of those things.

Like I don’t even think people are creative – everything is a remix of something. We have started with nature and ended with remixes of remixes of remixes and that may be why we cannot imagine how colors people don’t see would look like or another new taste, like some of us don’t even get existing taste like umami and cannot imagine it by itself.

PS. What Jon provides is not just summarized statistics, the whole ranking is an interpretation of all those various stats and parameters visible on the players’ pages, but not exclusive to them. There are various things in the logs people don’t seem to notice one of them being player’s positions for example :]


Last edited by CHERRY,

MEGAMIDDIE

Quoted from Gentleman Jon

[…]
I can’t wait

This post is great

CHERRY

Oh btw. RTC a lot of what you listed is in fact reflected in the logs, you just need to look better, something people might not be as good in :D

torden

broder
syster

Nice job with this!

For inspiration, check out Ashenfelter’s algorithm for the future value of fine Bordeaux wine. It beat the human counterparts by a landslide.

CHERRY

Quoted from torden

Nice job with this!

For inspiration, check out Ashenfelter’s algorithm for the future value of fine Bordeaux wine. It beat the human counterparts by a landslide.

Link for reference: https://www.betterment.com/resources/investment-strategy/behavioral-finance-investing-strategy/human-vs-algorithm-investing-a-lesson-from-wine-country/

stephen

breakfast jr confirmed 13th best scout eu

Add A Reply Pages: « Previous 1 ... 5 6 7 ... 14 Next »