<!--quoteo(post=1674183:date=Mar 24 2008, 11:39 PM:name=kazan)--><div class='quotetop'>QUOTE(kazan @ Mar 24 2008, 11:39 PM) <a href="index.php?act=findpost&pid=1674183"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->You're playing a game called Natural Selection.<!--QuoteEnd--></div><!--QuoteEEnd-->
<!--quoteo(post=1674014:date=Mar 24 2008, 12:16 AM:name=Radix)--><div class='quotetop'>QUOTE(Radix @ Mar 24 2008, 12:16 AM) <a href="index.php?act=findpost&pid=1674014"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Because using teamwork to counter a single skilled rambo is <b>nothing</b> like a bossfight that the skilled player is explicitly asking for by soloing and trying to find someone who knows how to play the game well enough to beat them.<!--QuoteEnd--></div><!--QuoteEEnd--> Exactly. I couldn't have put it any better.
<!--quoteo(post=1674183:date=Mar 24 2008, 11:39 PM:name=kazan)--><div class='quotetop'>QUOTE(kazan @ Mar 24 2008, 11:39 PM) <a href="index.php?act=findpost&pid=1674183"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->You're playing a game called Natural Selection.<!--QuoteEnd--></div><!--QuoteEEnd-->
locallyunsceneFeeder of TrollsJoin Date: 2002-12-25Member: 11528Members, Constellation
edited March 2008
<!--quoteo(post=1673872:date=Mar 21 2008, 04:50 PM:name=Termy58)--><div class='quotetop'>QUOTE(Termy58 @ Mar 21 2008, 04:50 PM) <a href="index.php?act=findpost&pid=1673872"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Also I believe L4D is making the same system, and Red Orchestra has Steam Friends Matchmaking now, as part of the <a href="http://www.steampowered.com/steamworks/" target="_blank">steam works system.</a><!--QuoteEnd--></div><!--QuoteEEnd--> Hmm, I missed this the first time around. Steam provides a matchmaking system all ready? The question of what is measured for game skill becomes more important then.
Edit: The example used for L4D is game hours. Is stat tracking like that found in TF2 accessible to non internal developers?
<!--quoteo(post=1674183:date=Mar 25 2008, 03:39 AM:name=kazan)--><div class='quotetop'>QUOTE(kazan @ Mar 25 2008, 03:39 AM) <a href="index.php?act=findpost&pid=1674183"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->You're playing a game called Natural Selection.<!--QuoteEnd--></div><!--QuoteEEnd-->
Actually no ones playing anything right now, and probably wont be for another 3 years.
<!--quoteo(post=1674235:date=Mar 25 2008, 07:46 PM:name=locallyunscene)--><div class='quotetop'>QUOTE(locallyunscene @ Mar 25 2008, 07:46 PM) <a href="index.php?act=findpost&pid=1674235"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Hmm, I missed this the first time around. Steam provides a matchmaking system all ready? The question of what is measured for game skill becomes more important then.
Edit: The example used for L4D is game hours. Is stat tracking like that found in TF2 accessible to non internal developers?<!--QuoteEnd--></div><!--QuoteEEnd-->Current matchmaking is a bunch of crap, read 'match organising' as opposed to 'matching player to player'.
The current 'matchmaking' stuff shown on Steamworks is simply to do with having a game lobby with chat, team organisation, etc.
I think the possibility for using player stats for matchmaking is there, but currently Valve haven't said anything publicly about whether they have this worked out or in development.
<!--quoteo(post=1674560:date=Mar 28 2008, 10:39 AM:name=haymo)--><div class='quotetop'>QUOTE(haymo @ Mar 28 2008, 10:39 AM) <a href="index.php?act=findpost&pid=1674560"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->yes but the general people who are new or think they have a clue will pick their relative self-nominated skill level<!--QuoteEnd--></div><!--QuoteEEnd-->
You don't seem to understand how this works, so let me put it in a different context:
In Doom 3 you can pick one of several difficulties, each one assumes an averagely competent player, and is relative to who that player is. If you're an extremely good shot and can kite and dodge well, you're free to play on expert level difficulty settings <b>to give yourself more of a challenge</b>.
That's the purpose of implementing a difficulty filter (which is best accomplished through matchmaking based on twitch skill) - to give a player that same option that they're used to from single player: "How much effort do I want to spend playing the game right now? Oh I just got off work, let's do some light easy servers for the time being.... Later on after I've had a few beers I'll jump up to intermediate. Maybe this weekend I'll play on expert."
It's just giving the player more ways to enjoy the game. I'm sure that you could make a defensible argument for static difficulty settings, but it just doesn't seem optimal to me when you can give a player a perfectly tailored playing experience with a roughly similar amount of effort.
<!--coloro:lime--><span style="color:lime"><!--/coloro-->I should note here that I'd be in favor of "unlocking" new difficulty settings ala doom as well, instead of having players vacillate through different difficulty locks on different days when they were in between skill levels. So that if a player who has played the game for over 2 weeks of ingame time (for instance) unlocks an advanced server level, I don't think he should be pushed back because he starts losing immediately at the new difficulty setting. Some permanence would be good, but I think that players just starting out *should* have to unlock those new servers before they're allowed to play on them.<!--colorc--></span><!--/colorc-->
The player base for NS2 will be substantially larger, especially the new player base, so this issue may be alleviated automatically. It's only when a game's base dwindles that you get into problems. Only the hardcore players are left and new/casual players just get slaughtered. Same thing happened with StarCraft and other games. Not much you can do about that; there just aren't enough mediocre players left to go around to even things out.
Choosable(static) skill levels would not necessarily help anything. There will still be skilled turdbags who want to play on the easy servers to pwn noobs.
Rankings would be nice, but it's a ton of work. You have to think about if you would rather the effort be spent elsewhere. Also, in a complex team-based game like NS, what is the metric for rankings? This isn't deathmatch. But how does building as gorge imply skill? Only if it helps your team... makes me tired thinking about it. Rankings would be distorted. But I could imagine a 2 or 3 tiered system where really bad players would always be in the low bracket, and really good players would always be in the higher bracket, no matter their playing style. It's possible. Maybe based on team win-loss(drop) ratio together with ranking within the team? Or how about the top 25% of players on the winning team get points, lowest 25% on the losing team lose points? Something like that. You'd have to deal with stacking though. But with rankings, maybe teams would be formed automatically. But what if you want to play with your friend? Yeh, it's complicated.
Well, as Radix said, you'd base it on twitch skill, since that's the key factor or skill in the game. Building as a gorge is <b>helpful</b> to the team, but it is not necessarily <b>skilful</b>. And with commanding, it's hard to say if it takes skill or if it's simply a matter of remembering how to approach things (tech trees etc.) and knowing where and when to direct marines in response to some very concrete situations. There really isn't much strategy involved in commanding, which makes the Real Time Strategy aspect of the "RTS-FPS hybrid" title, somewhat of a misnomer.
<!--quoteo(post=1674584:date=Mar 28 2008, 03:52 PM:name=juice)--><div class='quotetop'>QUOTE(juice @ Mar 28 2008, 03:52 PM) <a href="index.php?act=findpost&pid=1674584"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->But I could imagine a 2 or 3 tiered system where really bad players would always be in the low bracket, and really good players would always be in the higher bracket, no matter their playing style.<!--QuoteEnd--></div><!--QuoteEEnd--> You do know there's a <i>reason</i> why really bad players are 'really bad' at games, right?
Naturally players should become better with time.. Unfortunately, there are a lot of <i>very</i> special people in the world.
<!--quoteo(post=1674638:date=Mar 29 2008, 04:02 PM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Mar 29 2008, 04:02 PM) <a href="index.php?act=findpost&pid=1674638"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->What would be the defining criterion for determining an individual's skill level?<!--QuoteEnd--></div><!--QuoteEEnd-->
<!--quoteo(post=1673928:date=Mar 22 2008, 11:24 AM:name=Radix)--><div class='quotetop'>QUOTE(Radix @ Mar 22 2008, 11:24 AM) <a href="index.php?act=findpost&pid=1673928"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Click the arrow for my reply.<!--QuoteEnd--></div><!--QuoteEEnd-->
<!--quoteo(post=1674671:date=Mar 30 2008, 02:59 AM:name=Harimau)--><div class='quotetop'>QUOTE(Harimau @ Mar 30 2008, 02:59 AM) <a href="index.php?act=findpost&pid=1674671"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->If you mean twitch skill, then it would mostly have to do with their deaths/kills+assists, or damage.<!--QuoteEnd--></div><!--QuoteEEnd-->
How do you go about doing that, especially if most kills in NS are not at all equal?
<!--quoteo(post=1673685:date=Mar 19 2008, 10:25 PM:name=whocareswc)--><div class='quotetop'>QUOTE(whocareswc @ Mar 19 2008, 10:25 PM) <a href="index.php?act=findpost&pid=1673685"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->TF2 is a game for non gamers...<!--QuoteEnd--></div><!--QuoteEEnd-->
This is one of those statements that pretty much screams 'ignore anything I said after this'
<!--quoteo(post=1674716:date=Mar 31 2008, 09:35 AM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Mar 31 2008, 09:35 AM) <a href="index.php?act=findpost&pid=1674716"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->How do you go about doing that, especially if most kills in NS are not at all equal?<!--QuoteEnd--></div><!--QuoteEEnd--> In statistics, you call it 'weighting'.
<!--quoteo(post=1674716:date=Mar 30 2008, 08:35 PM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Mar 30 2008, 08:35 PM) <a href="index.php?act=findpost&pid=1674716"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->How do you go about doing that, especially if most kills in NS are not at all equal?<!--QuoteEnd--></div><!--QuoteEEnd-->
Way to skip over my post that answers both of yours. <img src="style_emoticons/<#EMO_DIR#>/marine.gif" style="vertical-align:middle" emoid="::marine::" border="0" alt="marine.gif" />
Sarisel.::' ( O ) ';:-. .-.:;' ( O ) '::.Join Date: 2003-07-30Member: 18557Members, Constellation
Twitch-skill to kill a walking skulk versus a leaping skulk? A lerk that flies straight at you versus one that pancakes? A walker fade versus a blinking fade? How do you quantify twitch skill on a server when the skill of the lifeforms that make up a person's skill statistics vary so much in skill themselves?
If you decide to weigh the kills on the stats of the person being killed, then that leaves the door open for statistical deviations due to variations on how well a person <i>decides</i> to play at a particular time. If you decide to weigh the lifeform itself, then that leaves the door open to problems due to inconsistencies in skill levels between players. If you try to use both at once, you will get confounding data.
locallyunsceneFeeder of TrollsJoin Date: 2002-12-25Member: 11528Members, Constellation
<!--quoteo(post=1674794:date=Mar 31 2008, 10:55 PM:name=Sarisel)--><div class='quotetop'>QUOTE(Sarisel @ Mar 31 2008, 10:55 PM) <a href="index.php?act=findpost&pid=1674794"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Twitch-skill to kill a walking skulk versus a leaping skulk? A lerk that flies straight at you versus one that pancakes? A walker fade versus a blinking fade? How do you quantify twitch skill on a server when the skill of the lifeforms that make up a person's skill statistics vary so much in skill themselves?
If you decide to weigh the kills on the stats of the person being killed, then that leaves the door open for statistical deviations due to variations on how well a person <i>decides</i> to play at a particular time. If you decide to weigh the lifeform itself, then that leaves the door open to problems due to inconsistencies in skill levels between players. If you try to use both at once, you will get confounding data.<!--QuoteEnd--></div><!--QuoteEEnd--> No system is perfect, people are people. That's not a very good excuse for not having one IMO. It's still useful even if it's only a rough approximation. If the lerk is pancaking then he probably has a pretty high skill level and that's accounted for. If he doesn't for some reason, well that's one kill out of a couple hundred/thousand. If something becomes significantly skewed then that's a flaw in the design, but that's a flaw in weighting not the system as a whole.
Sarisel.::' ( O ) ';:-. .-.:;' ( O ) '::.Join Date: 2003-07-30Member: 18557Members, Constellation
<!--quoteo(post=1674817:date=Apr 1 2008, 12:36 PM:name=locallyunscene)--><div class='quotetop'>QUOTE(locallyunscene @ Apr 1 2008, 12:36 PM) <a href="index.php?act=findpost&pid=1674817"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->No system is perfect, people are people. That's not a very good excuse for not having one IMO. It's still useful even if it's only a rough approximation. If the lerk is pancaking then he probably has a pretty high skill level and that's accounted for. If he doesn't for some reason, well that's one kill out of a couple hundred/thousand. If something becomes significantly skewed then that's a flaw in the design, but that's a flaw in weighting not the system as a whole.<!--QuoteEnd--></div><!--QuoteEEnd-->
I'm not really making an excuse for not having a skill-weighing system - just pointing out that it isn't as simple to make one that works well, unlike what some would think. And we're just talking about twitch skill here - not the overall skill - I believe. Just because a lerk or skulk might have good twitch skill (or even an aimbot) that allows for really good tracking, doesn't mean that they're good at movement, planning, awareness, etc. Likewise, let's say a marine is very good at aiming but is bad at situational awareness - here twitch skill is trumped. Again - I'm not saying that these imperfections mean that you can't have a skill-weighing system. You can, but it will be flawed on many levels. It will give at most a general idea overall of a player's capability. If this is what is desired, then that's fine. Just don't expect very much from it in terms of balancing or decreasing elitist gameplay.
locallyunsceneFeeder of TrollsJoin Date: 2002-12-25Member: 11528Members, Constellation
<!--quoteo(post=1674828:date=Apr 1 2008, 02:03 PM:name=Sarisel)--><div class='quotetop'>QUOTE(Sarisel @ Apr 1 2008, 02:03 PM) <a href="index.php?act=findpost&pid=1674828"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I'm not really making an excuse for not having a skill-weighing system - just pointing out that it isn't as simple to make one that works well, unlike what some would think. And we're just talking about twitch skill here - not the overall skill - I believe. Just because a lerk or skulk might have good twitch skill (or even an aimbot) that allows for really good tracking, doesn't mean that they're good at movement, planning, awareness, etc. Likewise, let's say a marine is very good at aiming but is bad at situational awareness - here twitch skill is trumped. Again - I'm not saying that these imperfections mean that you can't have a skill-weighing system. You can, but it will be flawed on many levels. It will give at most a general idea overall of a player's capability. If this is what is desired, then that's fine. Just don't expect very much from it in terms of balancing or decreasing elitist gameplay.<!--QuoteEnd--></div><!--QuoteEEnd--> I don't think the statistical anomalies of the system will outweigh the benefit or even be that close. It wouldn't be significantly statistically "flawed on many levels". "decreasing elitist gameplay" is somewhat misleading since the goal of the system isn't to change gameplay styles, but to group people with similar gameplay styles. High K-D will be grouped with high K-D and if one player has really good situational awareness and another has really good movement skills maybe they'll learn something from each other.
I think the anomalies would actually be more prevalent, and the system would have no statistical validity. Using a simple K:D ratio DOES NOT necessarily rate one's individual value to a given a team in a given server.
Sarisel.::' ( O ) ';:-. .-.:;' ( O ) '::.Join Date: 2003-07-30Member: 18557Members, Constellation
<!--quoteo(post=1674835:date=Apr 1 2008, 03:16 PM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Apr 1 2008, 03:16 PM) <a href="index.php?act=findpost&pid=1674835"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I think the anomalies would actually be more prevalent, and the system would have no statistical validity. Using a simple K:D ratio DOES NOT necessarily rate one's individual value to a given a team in a given server.<!--QuoteEnd--></div><!--QuoteEEnd-->
+1, there would have to be controls implemented for types of games (different game modes will probably exist via lua), maps that are played upon (ex. a custom map versus a normal map versus a training map), and more variables to track what's actually going on in a game (how an individual is contributing). It goes way beyond K:D and beyond K:D adjusted for K:D of victims and their lifeforms.
<!--quoteo(post=1674832:date=Apr 1 2008, 02:39 PM:name=locallyunscene)--><div class='quotetop'>QUOTE(locallyunscene @ Apr 1 2008, 02:39 PM) <a href="index.php?act=findpost&pid=1674832"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I don't think the statistical anomalies of the system will outweigh the benefit or even be that close. It wouldn't be significantly statistically "flawed on many levels". "decreasing elitist gameplay" is somewhat misleading since the goal of the system isn't to change gameplay styles, but to group people with similar gameplay styles. High K-D will be grouped with high K-D and if one player has really good situational awareness and another has really good movement skills maybe they'll learn something from each other.<!--QuoteEnd--></div><!--QuoteEEnd-->
I talked about elitist gameplay because that's what the original topic was about. I don't really understand why you would want to group people with similar gameplay styles - by "grouping" do you mean distributing them equally between the two teams? So that you have roughly equivalent teams? In that case you probably meant that high K-D players will be balanced on both sides. And if so, on what basis would you compare the value of movement skills versus situational awareness versus twitch skill, if you have 20 players that have different levels of achievements in these qualities?
I suspect that NS2 is going to be made in a way that makes individual excellence much more limited than in NS1, in favour of teamwork (even though I might not agree with this). In this case, you'll have a bunch of average K:D stats that won't be of very much use to you - whereas individual contributions to teamwork, which will make and break victories, will not be measurable.
locallyunsceneFeeder of TrollsJoin Date: 2002-12-25Member: 11528Members, Constellation
<!--quoteo(post=1674840:date=Apr 1 2008, 04:04 PM:name=Sarisel)--><div class='quotetop'>QUOTE(Sarisel @ Apr 1 2008, 04:04 PM) <a href="index.php?act=findpost&pid=1674840"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I talked about elitist gameplay because that's what the original topic was about. I don't really understand why you would want to group people with similar gameplay styles - by "grouping" do you mean distributing them equally between the two teams? So that you have roughly equivalent teams? In that case you probably meant that high K-D players will be balanced on both sides. And if so, on what basis would you compare the value of movement skills versus situational awareness versus twitch skill, if you have 20 players that have different levels of achievements in these qualities?<!--QuoteEnd--></div><!--QuoteEEnd--> Sorry, I wasn't very clear. There are two uses for a skill ranking system being discussed, matchmaking and autobalance. Matchmaking relates to the elitist gameplay by getting a group of players with about the same skill level together either through tiered servers or traditional matchmaking. The goal is to provide a place for new players to learn the ropes before they get trashed by a CAL/ANSL fade.
Autobalance applies more to servers with a varied skill base. I wouldn't account for situational awareness, movement skills, game knowledge etcetera because I don't know how you would measure that(possibly with the acheivement system <a href="http://www.unknownworlds.com/ns2/forums/index.php?showtopic=104013" target="_blank">here</a> but that's an aside). It would be a measure of overall effectiveness based on K:D and maybe assists. Someone with awesome movement skills but poor tracking could have the same score as someone with the reverse abilities. I don't see that as being a problem, since they're equally effective. It's just a way to setup roughly equal teams for admins.
You have to use the essential results of twitch skill (such as k:d) combined with a player's advantages as the acid test of the value of a player. You can't use a player's mood to determine how skilled they are. This is unfortunate, but if a matchmaking system is a reasonable suggestion (the devs seem to think it is) then you have to make some assumptions, such as "a player is trying to win"
I realize that's not always the case, but it seems like a reasonable assumption for statistical purposes.
In regard to the walking vs leaping skulk, I've posted several times about the use of environmental data to determine the value of a kill. Search my recent posts if you want more details, I'm getting tired of repeating myself.
As am I as you keep repeating the same point that gets disputed.
What about the general skill level that fluctuates from server to server. You are assuming ALL servers are equal, which is hardly the case at all.
You'd have to get server operators to contribute their statistics, log the steamIDs and have an extremely complex algorithm, and then hope the god that your underlying assumptions for said statistical analysis are true.
It can be done, but the devs can spend their time doing much more useful coding then to come up with a system that will most likely be broken.
Even if it is successful, given the amount of uncontrollable variables, the statistical validity of it would be next to nothing.
<!--quoteo(post=1674866:date=Apr 2 2008, 07:02 AM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Apr 2 2008, 07:02 AM) <a href="index.php?act=findpost&pid=1674866"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Even if it is successful, given the amount of uncontrollable variables, the statistical validity of it would be next to nothing.<!--QuoteEnd--></div><!--QuoteEEnd--> That would be akin to saying "Given the amount of uncontrollable variables <b>(people)</b> in regards to social or economic observations, the statistical validity of any observed patterns would be next to nothing."
<!--quoteo(post=1674866:date=Apr 2 2008, 07:02 AM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Apr 2 2008, 07:02 AM) <a href="index.php?act=findpost&pid=1674866"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->As am I as you keep repeating the same point that gets disputed.<!--QuoteEnd--></div><!--QuoteEEnd--><!--quoteo(post=1674236:date=Mar 26 2008, 05:07 AM:name=Radix)--><div class='quotetop'>QUOTE(Radix @ Mar 26 2008, 05:07 AM) <a href="index.php?act=findpost&pid=1674236"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I'm sure they're focusing on that because I said that K:D should be the acid test of a good player, which I think it should.
Underwhelmed, my thesis was (and is) that <b>the objective with matchmaking is to eliminate the frustration of stacking prohibitively good and prohibitively bad players against one another</b>, so it's not that I'm really trying to quantify all of the player's "skills" such as situational awareness and general knowledge with K:D ratio, but rather it's a proposed attempt to negate the first-person rage factor of getting killed over and over.
New players aren't smart enough to realize that they don't understand the game (unless it has no real tactics to speak of, in which case it's a moot point), so that element of play shouldn't bother them very much - but they are smart enough to realize "OH F*** I JUST DIED THIS GAME IS BAD."
I'm not trying to actually balance every element of the game with matchmaking, <b>just to filter players into games where they'll enjoy themselves until they can get over the hurdle of learning the game enough to progress to higher levels of play.</b>
My overarching objective with the proposal is fun. Others are free to suggest systems that are optimized to any number of different ends, but I still hold that matchmaking with twitch skill as the acid test is the best overall option.<!--QuoteEnd--></div><!--QuoteEEnd-->
<!--quoteo(post=1674866:date=Apr 2 2008, 07:02 AM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Apr 2 2008, 07:02 AM) <a href="index.php?act=findpost&pid=1674866"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->What about the general skill level that fluctuates from server to server. You are assuming ALL servers are equal, which is hardly the case at all.<!--QuoteEnd--></div><!--QuoteEEnd--> I'm really not sure what you're trying to say here. Do you mean the general twitch skill level on servers? We're talking about <b>ranked servers</b>, here, aren't we? Under this proposal, skilled players will play with skilled players, and unskilled players with unskilled players. Yes, of course there's going to be an imbalance at the beginning as the expert players have to make their way to the expert servers. That's to be expected. Though I personally believe that if you're a newbie and you want to play expert, knowing full well that it's an expert server, then there's no reason you shouldn't be able to. ie. restrict the servers with ranks below you but not above you. Just add a multiplier if you're a newbie playing in an expert server. (Where this would fail is if you had an expert server filled with newbies. How often do you think that'll happen though?) On second thought, the multiplier isn't even necessary. - OR allow server operators to choose which ranks to restrict (even none at all), in case they want an expert-ONLY server.
Comments
Perfect. Just perfect.
Exactly.
I couldn't have put it any better.
Yes but Flayra is playing a game called Business.
Hmm, I missed this the first time around. Steam provides a matchmaking system all ready? The question of what is measured for game skill becomes more important then.
Edit: The example used for L4D is game hours. Is stat tracking like that found in TF2 accessible to non internal developers?
Actually no ones playing anything right now, and probably wont be for another 3 years.
Edit: The example used for L4D is game hours. Is stat tracking like that found in TF2 accessible to non internal developers?<!--QuoteEnd--></div><!--QuoteEEnd-->Current matchmaking is a bunch of crap, read 'match organising' as opposed to 'matching player to player'.
The current 'matchmaking' stuff shown on Steamworks is simply to do with having a game lobby with chat, team organisation, etc.
I think the possibility for using player stats for matchmaking is there, but currently Valve haven't said anything publicly about whether they have this worked out or in development.
You don't seem to understand how this works, so let me put it in a different context:
In Doom 3 you can pick one of several difficulties, each one assumes an averagely competent player, and is relative to who that player is. If you're an extremely good shot and can kite and dodge well, you're free to play on expert level difficulty settings <b>to give yourself more of a challenge</b>.
That's the purpose of implementing a difficulty filter (which is best accomplished through matchmaking based on twitch skill) - to give a player that same option that they're used to from single player: "How much effort do I want to spend playing the game right now? Oh I just got off work, let's do some light easy servers for the time being.... Later on after I've had a few beers I'll jump up to intermediate. Maybe this weekend I'll play on expert."
It's just giving the player more ways to enjoy the game. I'm sure that you could make a defensible argument for static difficulty settings, but it just doesn't seem optimal to me when you can give a player a perfectly tailored playing experience with a roughly similar amount of effort.
<!--coloro:lime--><span style="color:lime"><!--/coloro-->I should note here that I'd be in favor of "unlocking" new difficulty settings ala doom as well, instead of having players vacillate through different difficulty locks on different days when they were in between skill levels. So that if a player who has played the game for over 2 weeks of ingame time (for instance) unlocks an advanced server level, I don't think he should be pushed back because he starts losing immediately at the new difficulty setting. Some permanence would be good, but I think that players just starting out *should* have to unlock those new servers before they're allowed to play on them.<!--colorc--></span><!--/colorc-->
Choosable(static) skill levels would not necessarily help anything. There will still be skilled turdbags who want to play on the easy servers to pwn noobs.
Rankings would be nice, but it's a ton of work. You have to think about if you would rather the effort be spent elsewhere. Also, in a complex team-based game like NS, what is the metric for rankings? This isn't deathmatch. But how does building as gorge imply skill? Only if it helps your team... makes me tired thinking about it. Rankings would be distorted. But I could imagine a 2 or 3 tiered system where really bad players would always be in the low bracket, and really good players would always be in the higher bracket, no matter their playing style. It's possible. Maybe based on team win-loss(drop) ratio together with ranking within the team? Or how about the top 25% of players on the winning team get points, lowest 25% on the losing team lose points? Something like that. You'd have to deal with stacking though. But with rankings, maybe teams would be formed automatically. But what if you want to play with your friend? Yeh, it's complicated.
And with commanding, it's hard to say if it takes skill or if it's simply a matter of remembering how to approach things (tech trees etc.) and knowing where and when to direct marines in response to some very concrete situations. There really isn't much strategy involved in commanding, which makes the Real Time Strategy aspect of the "RTS-FPS hybrid" title, somewhat of a misnomer.
You do know there's a <i>reason</i> why really bad players are 'really bad' at games, right?
Naturally players should become better with time.. Unfortunately, there are a lot of <i>very</i> special people in the world.
<!--quoteo(post=1673928:date=Mar 22 2008, 11:24 AM:name=Radix)--><div class='quotetop'>QUOTE(Radix @ Mar 22 2008, 11:24 AM) <a href="index.php?act=findpost&pid=1673928"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Click the arrow for my reply.<!--QuoteEnd--></div><!--QuoteEEnd-->
How do you go about doing that, especially if most kills in NS are not at all equal?
This is one of those statements that pretty much screams 'ignore anything I said after this'
In statistics, you call it 'weighting'.
Way to skip over my post that answers both of yours. <img src="style_emoticons/<#EMO_DIR#>/marine.gif" style="vertical-align:middle" emoid="::marine::" border="0" alt="marine.gif" />
If you decide to weigh the kills on the stats of the person being killed, then that leaves the door open for statistical deviations due to variations on how well a person <i>decides</i> to play at a particular time. If you decide to weigh the lifeform itself, then that leaves the door open to problems due to inconsistencies in skill levels between players. If you try to use both at once, you will get confounding data.
If you decide to weigh the kills on the stats of the person being killed, then that leaves the door open for statistical deviations due to variations on how well a person <i>decides</i> to play at a particular time. If you decide to weigh the lifeform itself, then that leaves the door open to problems due to inconsistencies in skill levels between players. If you try to use both at once, you will get confounding data.<!--QuoteEnd--></div><!--QuoteEEnd-->
No system is perfect, people are people. That's not a very good excuse for not having one IMO. It's still useful even if it's only a rough approximation. If the lerk is pancaking then he probably has a pretty high skill level and that's accounted for. If he doesn't for some reason, well that's one kill out of a couple hundred/thousand. If something becomes significantly skewed then that's a flaw in the design, but that's a flaw in weighting not the system as a whole.
I'm not really making an excuse for not having a skill-weighing system - just pointing out that it isn't as simple to make one that works well, unlike what some would think. And we're just talking about twitch skill here - not the overall skill - I believe. Just because a lerk or skulk might have good twitch skill (or even an aimbot) that allows for really good tracking, doesn't mean that they're good at movement, planning, awareness, etc. Likewise, let's say a marine is very good at aiming but is bad at situational awareness - here twitch skill is trumped. Again - I'm not saying that these imperfections mean that you can't have a skill-weighing system. You can, but it will be flawed on many levels. It will give at most a general idea overall of a player's capability. If this is what is desired, then that's fine. Just don't expect very much from it in terms of balancing or decreasing elitist gameplay.
I don't think the statistical anomalies of the system will outweigh the benefit or even be that close. It wouldn't be significantly statistically "flawed on many levels". "decreasing elitist gameplay" is somewhat misleading since the goal of the system isn't to change gameplay styles, but to group people with similar gameplay styles. High K-D will be grouped with high K-D and if one player has really good situational awareness and another has really good movement skills maybe they'll learn something from each other.
+1, there would have to be controls implemented for types of games (different game modes will probably exist via lua), maps that are played upon (ex. a custom map versus a normal map versus a training map), and more variables to track what's actually going on in a game (how an individual is contributing). It goes way beyond K:D and beyond K:D adjusted for K:D of victims and their lifeforms.
<!--quoteo(post=1674832:date=Apr 1 2008, 02:39 PM:name=locallyunscene)--><div class='quotetop'>QUOTE(locallyunscene @ Apr 1 2008, 02:39 PM) <a href="index.php?act=findpost&pid=1674832"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I don't think the statistical anomalies of the system will outweigh the benefit or even be that close. It wouldn't be significantly statistically "flawed on many levels". "decreasing elitist gameplay" is somewhat misleading since the goal of the system isn't to change gameplay styles, but to group people with similar gameplay styles. High K-D will be grouped with high K-D and if one player has really good situational awareness and another has really good movement skills maybe they'll learn something from each other.<!--QuoteEnd--></div><!--QuoteEEnd-->
I talked about elitist gameplay because that's what the original topic was about. I don't really understand why you would want to group people with similar gameplay styles - by "grouping" do you mean distributing them equally between the two teams? So that you have roughly equivalent teams? In that case you probably meant that high K-D players will be balanced on both sides. And if so, on what basis would you compare the value of movement skills versus situational awareness versus twitch skill, if you have 20 players that have different levels of achievements in these qualities?
I suspect that NS2 is going to be made in a way that makes individual excellence much more limited than in NS1, in favour of teamwork (even though I might not agree with this). In this case, you'll have a bunch of average K:D stats that won't be of very much use to you - whereas individual contributions to teamwork, which will make and break victories, will not be measurable.
Sorry, I wasn't very clear. There are two uses for a skill ranking system being discussed, matchmaking and autobalance. Matchmaking relates to the elitist gameplay by getting a group of players with about the same skill level together either through tiered servers or traditional matchmaking. The goal is to provide a place for new players to learn the ropes before they get trashed by a CAL/ANSL fade.
Autobalance applies more to servers with a varied skill base. I wouldn't account for situational awareness, movement skills, game knowledge etcetera because I don't know how you would measure that(possibly with the acheivement system <a href="http://www.unknownworlds.com/ns2/forums/index.php?showtopic=104013" target="_blank">here</a> but that's an aside). It would be a measure of overall effectiveness based on K:D and maybe assists. Someone with awesome movement skills but poor tracking could have the same score as someone with the reverse abilities. I don't see that as being a problem, since they're equally effective. It's just a way to setup roughly equal teams for admins.
I realize that's not always the case, but it seems like a reasonable assumption for statistical purposes.
In regard to the walking vs leaping skulk, I've posted several times about the use of environmental data to determine the value of a kill. Search my recent posts if you want more details, I'm getting tired of repeating myself.
What about the general skill level that fluctuates from server to server. You are assuming ALL servers are equal, which is hardly the case at all.
You'd have to get server operators to contribute their statistics, log the steamIDs and have an extremely complex algorithm, and then hope the god that your underlying assumptions for said statistical analysis are true.
It can be done, but the devs can spend their time doing much more useful coding then to come up with a system that will most likely be broken.
Even if it is successful, given the amount of uncontrollable variables, the statistical validity of it would be next to nothing.
That would be akin to saying "Given the amount of uncontrollable variables <b>(people)</b> in regards to social or economic observations, the statistical validity of any observed patterns would be next to nothing."
<!--quoteo(post=1674866:date=Apr 2 2008, 07:02 AM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Apr 2 2008, 07:02 AM) <a href="index.php?act=findpost&pid=1674866"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->As am I as you keep repeating the same point that gets disputed.<!--QuoteEnd--></div><!--QuoteEEnd--><!--quoteo(post=1674236:date=Mar 26 2008, 05:07 AM:name=Radix)--><div class='quotetop'>QUOTE(Radix @ Mar 26 2008, 05:07 AM) <a href="index.php?act=findpost&pid=1674236"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I'm sure they're focusing on that because I said that K:D should be the acid test of a good player, which I think it should.
Underwhelmed, my thesis was (and is) that <b>the objective with matchmaking is to eliminate the frustration of stacking prohibitively good and prohibitively bad players against one another</b>, so it's not that I'm really trying to quantify all of the player's "skills" such as situational awareness and general knowledge with K:D ratio, but rather it's a proposed attempt to negate the first-person rage factor of getting killed over and over.
New players aren't smart enough to realize that they don't understand the game (unless it has no real tactics to speak of, in which case it's a moot point), so that element of play shouldn't bother them very much - but they are smart enough to realize "OH F*** I JUST DIED THIS GAME IS BAD."
I'm not trying to actually balance every element of the game with matchmaking, <b>just to filter players into games where they'll enjoy themselves until they can get over the hurdle of learning the game enough to progress to higher levels of play.</b>
My overarching objective with the proposal is fun. Others are free to suggest systems that are optimized to any number of different ends, but I still hold that matchmaking with twitch skill as the acid test is the best overall option.<!--QuoteEnd--></div><!--QuoteEEnd-->
<!--quoteo(post=1674866:date=Apr 2 2008, 07:02 AM:name=Firewater)--><div class='quotetop'>QUOTE(Firewater @ Apr 2 2008, 07:02 AM) <a href="index.php?act=findpost&pid=1674866"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->What about the general skill level that fluctuates from server to server. You are assuming ALL servers are equal, which is hardly the case at all.<!--QuoteEnd--></div><!--QuoteEEnd-->
I'm really not sure what you're trying to say here. Do you mean the general twitch skill level on servers? We're talking about <b>ranked servers</b>, here, aren't we? Under this proposal, skilled players will play with skilled players, and unskilled players with unskilled players. Yes, of course there's going to be an imbalance at the beginning as the expert players have to make their way to the expert servers. That's to be expected.
Though I personally believe that if you're a newbie and you want to play expert, knowing full well that it's an expert server, then there's no reason you shouldn't be able to. ie. restrict the servers with ranks below you but not above you. Just add a multiplier if you're a newbie playing in an expert server. (Where this would fail is if you had an expert server filled with newbies. How often do you think that'll happen though?) On second thought, the multiplier isn't even necessary.
- OR allow server operators to choose which ranks to restrict (even none at all), in case they want an expert-ONLY server.