PSR

Racer1 · December 2009

What does everyone think of an official PSR (player skill rating) system? Servers who opt-in would have their player's stats sent to an official UW server. This score could then be used to help auto-balance teams, beyond just the number of players. Each player class (e.g. Marine, Skulk, Commander, etc) would earn points based on the characteristics of that class and those would be aggregated into a PSR.

I admit there is no perfect way to measure a player's ability. However, that doesn't mean we shouldn't try -- especially if it <u>helps</u> to balance the teams. It is at least better than "kills per death", which is only a fraction of what makes a good NS player. As to exactly how PSR is calculated will have to wait on NS2 gameplay details we do not yet have.

RobB · December 2009

No. You can't make a system that's able to compare players fair without taking every and single action in account.
We get that suggestion a lot here, but it's still a bad idea.

SentrySteve · December 2009

<div class='quotetop'>QUOTE (Racer1 @ Dec 17 2009, 10:20 AM) <a href="index.php?act=findpost&pid=1744111"><{POST_SNAPBACK}></a></div><div class='quotemain'>I admit there is no perfect way to measure a player's ability. However, that doesn't mean we shouldn't try</div>

You're right in your two points I quoted above. PSR is far from perfect in such a team based game (it's horrible, actually) but we should still try some system. The system I think would work best is Firewater's suggestion which is to allow admins to self-label their server (newbie, casual, advanced/competitive) so that players can quickly find the type of server they want to play on. That's the only system that could actually work and thankfully it sounds easy to implement.

Alaska · December 2009

take a look at Dystopias stats-system. it's also a teambased mod and the ranking-system is quiet good imho. this would be a good startingpoint for a PSR for ns2. And it is definitely usefull. Awards are fun and keep players with the game - ranks also. And you have a mechanism to rate players and to implement an autobalancing-system (look at dystopia, too)

RobB · December 2009

There is a difference between Dystopia and NS2. Dystopia is Deathmatch only and their System seems quite expensive with all that fancy crap and stuff.

The points here have been made way to often already.

To keep it simple, our actions are to complex
<ul><li>build</li><li>heal</li><li>repair</li><li>defend</li><li>attack</li><li>destroy</li><li>assist</li><li>command</li></ul>

you cant put up a system that keeps track of all that properly without opening the backdoor for exploits.

noisymonk · December 2009

<div class='quotetop'>QUOTE </div><div class='quotemain'>you cant put up a system that keeps track of all that properly without opening the backdoor for exploits</div>

I don't think that's entirely true. While current stat systems aren't perfect, if used properly then they really don't cause any harm. If stats aren't publicly available or used for e-peen points (ie, a list of ranked players that everyone can see, and wants to be at the top of), then there is no need to pimp your own stats by starting your own server and just boosting your score.

Then you ask, 'why bother having stats that aren't publicly available?' Well, for example:

What if you take this information and have 'Recommended Servers' in the Server Browser that would try to match you into a server with players of similar ranking? This doesn't STOP noobs from joining a server full of pros, but it would give them the choice of NOT joining a server full of pros - and being totally dominated, and not enjoy themselves. This also works in reverse by helping pros find some good healthy competition instead of a cake-walk. Sure, this isn't a game changer - but this is one simple addition to the game to make it that much more 'easy to play'.

This is just one example of how stats could be used for things other than e-peen points. NS is a very dynamic game with much more than a K/D ratio to determine 'skill'. And as there isn't currently a way to really measure this skill effectively, it doesn't mean that we can't put some stats to good use.

Pipi · December 2009

<div class='quotetop'>QUOTE (RobB @ Dec 17 2009, 05:24 PM) <a href="index.php?act=findpost&pid=1744143"><{POST_SNAPBACK}></a></div><div class='quotemain'><ul><li>build</li><li>heal</li><li>repair</li><li>defend</li><li>attack</li><li>destroy</li><li>assist</li><li>command</li></ul></div>

There's also communication.. how would you rate that. either by voice or typing, communication in this game is more important than any other games I have played.

I think this system is useless anyways.. not to offend anyone but if it is just to put more strain on server bandwidths, I don' want.

Even if most people whined about server registration in NS1 it still had really good one compared to many games and I would like NS2 to be as good or better so.. cut all the crap we can.

EDIT: Server labels like Steve said .. Pro server, Noob server.. even with a big fancy system people will tend to play on good ping servers no matter what so anyways..

RobB · December 2009

<div class='quotetop'>QUOTE </div><div class='quotemain'>What if you take this information and have 'Recommended Servers' in the Server Browser that would try to match you into a server with players of similar ranking?</div>
Which I'd never use since I will spend some weeks to find a clan whose server / playerbase is acceptable and waste my time there, disregarding any skill discrepancys.

noisymonk · December 2009

<div class='quotetop'>QUOTE </div><div class='quotemain'>Which I'd never use since I will spend some weeks to find a clan whose server / playerbase is acceptable and waste my time there, disregarding any skill discrepancys.</div>

Ah, so if it's not useful to you, then it's not useful to anyone - gotcha. Try to think about the rest of the community, and how this could be helpful to quite a few people.

Edit: And like I said, that's just 1 possible idea. I'm sure that there could be some great uses for some general Stats other than e-peen contests.

RobB · December 2009

I'm speaking for myself, yes. But only to make clear what my motivation is. I can't speak for others, but I know a bunch of people that are acting similar.
They have a portfolio of "favourite servers", and don't want to or don't need to use others (except if all their servers - if that's happening at all - are full).

TrC · December 2009

Yes that would be good idea, as for how I would suggest system that uses average points / min. You can slightly improve the balance that way but as for perfect I doubt its possible. As for casual / average / competitive servers I dare say that some kid is always ruining casual servers so no point in making one.

Seker · December 2009

Dystopia ranking would be horrible for ns.

In their ranking those who play often are higher ranked than those who arent playing often...

spellman23 · December 2009

Or you just implement a bland Elo system.

You win, get more points. You lose, lose some points. Lower aggregate points = less chance for that side to win, but more payout if they do win.

Note some problems such as how to balance buddies joining one team (in theory will work better together), however League of Legends has been working hard and seems to have a nice algorithm for their game to fix this in quick matches.

At the end of the day, it doesn't capture EVERYTHING, but if you help win, you get (in theory and in general) and higher rating. Over infinite number of games (and barring jerks making throwaways to boost their rating) it should give a decent idea of a person's skill. Of course, it's far from perfect. See Heroes of Newerth and League of Legends as they try to grapple with this issue.

digz · December 2009

<div class='quotetop'>QUOTE (spellman23 @ Dec 20 2009, 07:42 AM) <a href="index.php?act=findpost&pid=1744386"><{POST_SNAPBACK}></a></div><div class='quotemain'>Or you just implement a bland Elo system.

You win, get more points. You lose, lose some points. Lower aggregate points = less chance for that side to win, but more payout if they do win.

Note some problems such as how to balance buddies joining one team (in theory will work better together), however League of Legends has been working hard and seems to have a nice algorithm for their game to fix this in quick matches.

At the end of the day, it doesn't capture EVERYTHING, but if you help win, you get (in theory and in general) and higher rating. Over infinite number of games (and barring jerks making throwaways to boost their rating) it should give a decent idea of a person's skill. Of course, it's far from perfect. See Heroes of Newerth and League of Legends as they try to grapple with this issue.</div>

This is the system TG used to balance their NS1 server with a pluggin they wrote, and it was pretty effective.

Chris0132 · December 2009

<div class='quotetop'>QUOTE (RobB @ Dec 17 2009, 03:49 PM) <a href="index.php?act=findpost&pid=1744113"><{POST_SNAPBACK}></a></div><div class='quotemain'>No. You can't make a system that's able to compare players fair without taking every and single action in account.
We get that suggestion a lot here, but it's still a bad idea.</div>

You can make a rough approximation of it fairly easily however, people can learn to read the statistics themselves.

PSR

Comments