Imrpoved ranking system

[TheMightyOne]

How about you start naming the statisitics which could determine skill?

have you read my post?

imo the rakning shouldnt depend just on the win/lost count. thats not necessary shows your true skill. it should be more like a mix of all the other statistics, like dmg dealt/ressorces shared/etc. so somebody who deals a lot of damage might have a worse rank then a person who is dealing less dmg but is the perfect ally who always helps out by sharing ressources. it means even if youve lost the game your ranking can still improve depending of how well you actually did. if ive dealt more dmg then my opponent but lost the game i might still get more points. besides i think that we should think of a ranking system which wouldnt decrease your rank if youre losing coz this way losing wouldnt "hurt" so much, since thats the main reason why ppl would unplug their internet cable.

[ZellSF]

Damage dealt: What if you start in a spot with less resources? What if your allies leave you to die while they tech?

Resources shared: Good players need to take responsibility and NOT share resources unless it's really needed, on the other side bad players SHOULD share resources they don't know how to use rather than building more and more storage...

Both those statistics really show nothing at all... No amount of analyzing them will. Not to mention what happens in a mod where neither of those statistics exist...

[KDR_11k]

As has been said many, MANY times before: Statwhores suck. Any stat that's tracked will get whores trying to maximize theirs. Maximizing a stat will push people into certain strategies where the stat can be gained more easily and often hurt the overall performance of a team in teamgames or lead to negative behaviour outside of the game (e.g. team stacking, playing only against weak players, etc).

You want a system that tells you which players are good and learns automatically what influences that? Just ask people who they think is good.

[Gota]

Zell,A.K. homo idiotico, said:
"God exists".

Error323 said:
"if you can prove this i would like to see it".

Gota said:
"Error why bother?"

[Error323]

Zell,A.K. homo idiotico, said:
"God exists".

Error323 said:
"if you can prove this i would like to see it".

Gota said:
"Error why bother?"

Indeed, I'm just working on it now.

Those who are interested in the statistics that will be used, contact me in the lobby #xta.

[ZellSF]

Zell,A.K. homo idiotico, said:
"God exists".

Error323 said:
"if you can prove this i would like to see it".

Gota said:
"Error why bother?"

Error323 was the one who claimed that something which most likely doesn't exist does exist. Not me.

[Error323]

Zell,A.K. homo idiotico, said:
"God exists".

Error323 said:
"if you can prove this i would like to see it".

Gota said:
"Error why bother?"

Error323 was the one who claimed that something which most likely doesn't exist does exist. Not me.

zellsf can I contact you in person via some medium plz? I would like to talk not using this forum.

[Tribulexrenamed]

I am satan. Therefore god exists, otherwise I would own the world.


Not really on topic, but i am contributing to this valuable and educational lesson on theology.

[Error323]

http://student.science.uva.nl/~fhuizing/report.pdf

Results are good!!, read the report for details.

[Argh]

Statwhores suck. Any stat that's tracked will get whores trying to maximize theirs.

So... under the current system, we get people whoring for time, to get their stars. Then they find that people won't play them, even though they're really nubs who've played a lot of SP with bots, but they made an "online game" so the time counted, and they smurf so that they can play with nubs like themselves

So... let's put it another way. If all stat-tracking systems inherently suck, but can do useful things for players in terms of some idea of who they're dealing with... isn't the minimal good well worth it, vs. the absence of any good in a system that measures nothing about a person's play at all?

[Regret]

played a lot of SP with bots, but they made an "online game" so the time counted

If you host a game and play alone with bots online, time does not count towards your rank.

[Argh]

Ok, but my point is still that there are various ways to achieve a nubstar- you could just be playing SpeedMetal with friends, with "10 minutes-no-attack" rules, set up a second computer to "play" against for hours... etc.

Furthermore... people bring up stat-whoring. Why is this "bad"? Every serious competitive game has that... why not try to quantify it, and let players decide if the numbers are meaningful? I love FPS servers that tell me how many headshots I'm pulling, for example- it tells me how well I'm playing. And if other players know that number as well, it does no harm. I don't really see a downside. People will boast about stuff... so what? Is it wrong, for people to have something to be proud about, besides equally-meaningless tourneys, etc.? I mean... in the end, we're just playing a game, not proving a theorem or solving world hunger...

[Error323]

I will try to create a widget that implements the eventual neural network model which uses 1 layer, 4 hidden nodes, 3 timeslices (again see report for details). This way, it may be possible to get realtime scores during the game. Also I will give it to Meltrax whois working on the online ranking system.

Anyway, the proof of concept is there.

[Error323]

Ok, but my point is still that there are various ways to achieve a nubstar- you could just be playing SpeedMetal with friends, with "10 minutes-no-attack" rules, set up a second computer to "play" against for hours... etc.

Furthermore... people bring up stat-whoring. Why is this "bad"? Every serious competitive game has that... why not try to quantify it, and let players decide if the numbers are meaningful? I love FPS servers that tell me how many headshots I'm pulling, for example- it tells me how well I'm playing. And if other players know that number as well, it does no harm. I don't really see a downside. People will boast about stuff... so what? Is it wrong, for people to have something to be proud about, besides equally-meaningless tourneys, etc.? I mean... in the end, we're just playing a game, not proving a theorem or solving world hunger...

I couldn't agree more on that. People who don't like it, shouldn't play for rank :) it's just an extra feature.

[Argh]

Well, that sounds like progress. Dunno whether a neural network is really the best way to go about it- whatever heuristic you develop will be great for whatever mod / game you develop it for (assuming it doesn't change) but probably won't be easy to port. I'd think that gathering raw data and letting a third party crunch it, or at the very least keeping the two things separate might be best.

But running it as a Widget's a great idea, then if people think stats are terrible, they can be stat-less. Best of both worlds.

[Regret]

A widget for determining stats haha that's a good one :D

[Error323]

Ok this is going nowhere, I thought it would be interesting for you people to share our findings. But I was wrong...

[imbaczek]

i've read that report, but i don't see anything actually ranking-related - you describe a method and show learning errors, but there's no concrete result shown (ie., an example ranking generated by hand compared to the one output by a nn, together with timeslice data.)

also, you don't say how you select timeslices - e.g. if you take a timeslice at the end of the game, one player's units will be 0. i imagine you don't do that.

otherwise a piece of interesting work, but if something useful should come out of it, you need to publish the generated rankings and perhaps update them for a while to see how the ranking evolves and gather comments.

[Error323]

i've read that report, but i don't see anything actually ranking-related - you describe a method and show learning errors, but there's no concrete result shown (ie., an example ranking generated by hand compared to the one output by a nn, together with timeslice data.)

also, you don't say how you select timeslices - e.g. if you take a timeslice at the end of the game, one player's units will be 0. i imagine you don't do that.

otherwise a piece of interesting work, but if something useful should come out of it, you need to publish the generated rankings and perhaps update them for a while to see how the ranking evolves and gather comments.

lol, all things you comment on are in the report, I suggest you read it more carefully.

[imbaczek]

then maybe point out where, because i can't find them.