Rankings
Radix
Join Date: 2005-01-10 Member: 34654Members, Constellation
Join Date: 2005-01-10 Member: 34654Members, Constellation
If this has been suggested before, please feel free to post a redirect to the original, because I didn't catch it the first time around. It seems like it should have been at one point or another. Thanks.
The beauty of console gaming is that it pits appropriately-skilled players against one another <i>most of the time</i>.
Internet pubs don't have that. Because of this, there needs to be a way to at least show the level of skill in a given server to prospectively joining players. <!--coloro:red--><span style="color:red"><!--/coloro-->At the very least, warn players as to the par difficulty level of the server they're joining.<!--colorc--></span><!--/colorc-->
You always want to give players the opportunity to play against opponents who are better than they are, but not in such a stark (and more importantly, <i>unexpected</i>) contrast as what happens now between Joe Pub who joins a server full of veterans.
Ideally, instead of locking players out by an arbitrarily-defined "vet server/nub server" ranking (as has been suggested before), there would be a way to distinguish servers by their players. You keep scores, which are securely linked to steam IDs and through that information you can construct a globalized "high scores table" and more importantly, you can display an icon next to a server (or perhaps next to an individual player, if they desired - I would prefer optional anonymity for this) for how skilled the total players on that server are.
I believe this would solve many of the potential community problems inherent in the system between great players and new players. Great players could find appropriate competition, and new players wouldn't always be yelling "I'm being repressed!"
Kudos to anyone who gets the reference. What do you all think?
The beauty of console gaming is that it pits appropriately-skilled players against one another <i>most of the time</i>.
Internet pubs don't have that. Because of this, there needs to be a way to at least show the level of skill in a given server to prospectively joining players. <!--coloro:red--><span style="color:red"><!--/coloro-->At the very least, warn players as to the par difficulty level of the server they're joining.<!--colorc--></span><!--/colorc-->
You always want to give players the opportunity to play against opponents who are better than they are, but not in such a stark (and more importantly, <i>unexpected</i>) contrast as what happens now between Joe Pub who joins a server full of veterans.
Ideally, instead of locking players out by an arbitrarily-defined "vet server/nub server" ranking (as has been suggested before), there would be a way to distinguish servers by their players. You keep scores, which are securely linked to steam IDs and through that information you can construct a globalized "high scores table" and more importantly, you can display an icon next to a server (or perhaps next to an individual player, if they desired - I would prefer optional anonymity for this) for how skilled the total players on that server are.
I believe this would solve many of the potential community problems inherent in the system between great players and new players. Great players could find appropriate competition, and new players wouldn't always be yelling "I'm being repressed!"
Kudos to anyone who gets the reference. What do you all think?
Comments
I like your idea of having an icon ranking on the servers themselves, based on the stats of the players present. I would say don't give the individual players an icon, or you might encourage team stacking/stat postituting.
Even if you suck, you have something to strive for. Sorta like an MMO, but strictly skill based. In NS, it'd be a lot harder to pad stats, unless you just... 'baited' everyone..everywhere everytime--and even then, you'd have to aim to get the kills/squad pts or whatever.
Rankings pwn imho, if done w/ a fair amount of intelligence.
<!--QuoteEnd--></div><!--QuoteEEnd-->
He said that because there's a word filter here.
For me to feel at all good about a ranking of any kind, you need to first understand how the game play works. This way you identify fun and cool things, things that are harder to quantify than just Kills to Death ratios. Then, if this is possible, I would suggest the following:
Make ranking badges and medals of honor in particular categories. Or phermones or psedo-pods of honor in the Kharaa's case.
Make them for team work, fun, and especially randomly silly ones. May I suggest such categories as "Long Distance Runner", "Excited Engineer", and "Helpful Healer" ... and just about anything else you can think of that fit with trying to reward players for those things you want to see alot of in Natural Selection. If all you do is count "kills", geuss what players will most likely have a focus for.
Then, if you need a team or server skill ranking, do your arithmetic and determine the average. (For those in need of a refresher, that is add up the units and divide by the number of each data of units = average)
Better still, it doesn't encourage ANY bad behavior, though it doesn't specifically encourage good behavior like teamplay etc. But it does encourage people to play more.
The difference between somebody who has played 10, 50, 200 and 500 times s certainly big. This is good at separating new players from experienced players, but it won't work at all at separating players who have past a certain number of games played, when amount of playing means less.
One for badges for doing certain things well.
One for overall performance (kills, deaths, things welded/healed/dropped/taken out/whatever).
These two rankings (badges and ranking) would be displayed when one looked at the scoreboard.
Make ranking badges and medals of honor in particular categories. Or phermones or psedo-pods of honor in the Kharaa's case.
<!--QuoteEnd--></div><!--QuoteEEnd-->
this would be cool, especially if the badges or medals were on the actual player model, rather than next to your name in the scoreboard.
if i had earned a medal, i'd love to have that shiny thing pinned to my chest, so my teammates can see it and respect it, and the enemy can see it and fear it.
god knows how they'd manage it though. sure the more medals you got the more you would stand out, but you wouldnt want to be walking around with half a chestplate full of crap...
Hmm, I think this calls for some pros and cons to seeing medals on a model rather or as well as on a "stats" page:
Pro
- Other team mates realize you have experience and could possibly lead, in the field or in the command chair.
- If done well, it looks cool and honorable.
- The other team can possibly recognize your status and target you.
Con
- You don't actually see the medals - unless it is shown on your "stats".
- If done poorly, you look like a silly general.
- The other team can possibly recognize your status and target you.
But the real point I was trying to get at was: Those actions you like in your game, seek to reward them.
As of right now, NS1 does not do this very well IMHO. I was finally using the score table when it showed who had what weapons, very useful for the command chair or finding someone who could weld you. Other than that new usefulness, looking at number of deaths and kills or a non-descriptive, anonamous score points really didn't seem to convey to me the feel of Natural Selection games. Those who weren't as focused on kills just didn't seem to "rank" as high as others, yet that's what happens when you try to work together as a team, selecting different supporting roles.
There are a few details, but building rts and doing other miscellaneous and clerical tasks is something anyone can do, and something a newbie can master fairly quickly. Aiming and ambushing are the most demanding parts of the game (assuming NS2 is like NS) so they take the longest to learn, and are therefore one of the best litmus tests of a good player, since all the other information should come before you master those.
With that said, I do like the idea of quantifying player attributes besides leetness, but I think a heavy emphasis should be placed on the actual player skill (including ability to comm well, which could likely be measured by comparing the success and victory rate of the games where they spent the majority of time in the chair, with the skill ratio between their players and the opposing alien team).
Also, I temporarily forgot this is a half-life <strike>mod</strike> full conversion, so yes, it Definitely needs heuristics beyond K/D ratio.
There are a few details, but building rts and doing other miscellaneous and clerical tasks is something anyone can do, and something a newbie can master fairly quickly. Aiming and ambushing are the most demanding parts of the game (assuming NS2 is like NS) so they take the longest to learn, and are therefore one of the best litmus tests of a good player, since all the other information should come before you master those.
With that said, I do like the idea of quantifying player attributes besides leetness, but I think a heavy emphasis should be placed on the actual player skill (including ability to comm well, which could likely be measured by comparing the success and victory rate of the games where they spent the majority of time in the chair, with the skill ratio between their players and the opposing alien team).
<!--QuoteEnd--></div><!--QuoteEEnd-->
I feel I must disagree with your first paragraph. Mastering building is hardly something any newbie can just pick up, before say, trying to run and gun, something they are more likely to have encountered in other games. Truely mastering building requires an understanding of tagging floors to help commander's choose good places for structures, placing yourself so you are not easily flanked by Kharaa, "touching" an important structure before a Kharaa so it doesn't disappear, using the "c" map to find where a commander placed a structure, stopping building in time to be able to fire your weapon when attacked, knowing to finish important structures sometimes at great risk to ones health for the good of the team, and so on.
Also, I disagree with the first paragraph because teams need all kinds of players, especially ones that are willing to build: seeking out resources, advance points of entry with phases, seige points instead of rushing in... Since teams need supporting roles on top of the point man or the marksman or close quarter specialist - they need "medic welders" and engineer builders and explosive experts and sneaky "ninjas" and recon rangers and resource gatherers and battle hardened commanders ... and so on.
Any system of ranking that doesn't take into account the roles of a team that works together really has no place in NS. There is a definite need to pat new players on the back when they play the game in ways that the developer intend to foster for the good of the community they seek. Heck, there should be a superficial reward (aka badge) for being polite to new players if at all possible! At every possible turn, present a "score" based on what is desired - my lil'take of a part of game design based off a study of some dogs, food, and a bell...
I feel I must disagree with your first paragraph. Mastering building is hardly something any newbie can just pick up, before say, trying to run and gun, something they are more likely to have encountered in other games. Truely mastering building requires an understanding of tagging floors to help commander's choose good places for structures, placing yourself so you are not easily flanked by Kharaa, "touching" an important structure before a Kharaa so it doesn't disappear, using the "c" map to find where a commander placed a structure, stopping building in time to be able to fire your weapon when attacked, knowing to finish important structures sometimes at great risk to ones health for the good of the team, and so on.<!--QuoteEnd--></div><!--QuoteEEnd-->
You're exaggerating the diversity of the building. Sure a good marine knows when to build and when to look around and how to place himself, but those things are the k:d stuff. What needs to be built and where is up to the commander. In ns anyone can go cap west and waste res and its not a big deal who is going to do that. Maybe some marine manages to frag a skulk and other doesn't, but still a builder often has quite little to do vs smart a skulk. Sure the moves you mentioned are useful sometimes, but ns is full of things like that and ns:s will hopefully be even more that way.
<!--quoteo--><div class='quotetop'>QUOTE</div><div class='quotemain'><!--quotec-->
Also, I disagree with the first paragraph because teams need all kinds of players, especially ones that are willing to build: seeking out resources, advance points of entry with phases, seige points instead of rushing in... Since teams need supporting roles on top of the point man or the marksman or close quarter specialist - they need "medic welders" and engineer builders and explosive experts and sneaky "ninjas" and recon rangers and resource gatherers and battle hardened commanders ... and so on.<!--QuoteEnd--></div><!--QuoteEEnd-->
At least I'm happy to cap res and play for the team if I know someone actually makes use of it. Its not much fun spending time capping if the rest of the team is humping the armory and waiting for some magical mystery ha train to pop up from nowhere. Usually there is not much motivation doing team effort nowadays since you can almost certainly tell that half of the team isn't doing even the little you do. I think those happy cappers would start appearing more if everyone at the server is playing at the roughly same skill level. The specific roles have quite little to do with the present ns. Every experienced player know how to weld and when some capping could be done so speaking ninjas and medics is quite off the wall. Its also easier to work as a team if you know the other guy is able to understand your moves and uses the opportunity you've created him. Way too many times I try to decoy, bait, cover and such while watching the teammate not taking advantage of your teamwork.
---
I'm quite sceptical when it comes to ranking. A proper ranking system with big enough playerbase could be a great thing, but I don't know if its even possible to create a working ranking system based on any other factor than k:d, which on the other hand doesn't suit ns after all.
I feel I must disagree with your first paragraph. Mastering building is hardly something any newbie can just pick up, before say, trying to run and gun, something they are more likely to have encountered in other games. Truely mastering building requires an understanding of tagging floors to help commander's choose good places for structures, placing yourself so you are not easily flanked by Kharaa, "touching" an important structure before a Kharaa so it doesn't disappear, using the "c" map to find where a commander placed a structure, stopping building in time to be able to fire your weapon when attacked, knowing to finish important structures sometimes at great risk to ones health for the good of the team, and so on.
Also, I disagree with the first paragraph because teams need all kinds of players, especially ones that are willing to build: seeking out resources, advance points of entry with phases, seige points instead of rushing in... Since teams need supporting roles on top of the point man or the marksman or close quarter specialist - they need "medic welders" and engineer builders and explosive experts and sneaky "ninjas" and recon rangers and resource gatherers and battle hardened commanders ... and so on.
Any system of ranking that doesn't take into account the roles of a team that works together really has no place in NS. There is a definite need to pat new players on the back when they play the game in ways that the developer intend to foster for the good of the community they seek. Heck, there should be a superficial reward (aka badge) for being polite to new players if at all possible! At every possible turn, present a "score" based on what is desired - my lil'take of a part of game design based off a study of some dogs, food, and a bell...
<!--QuoteEnd--></div><!--QuoteEEnd-->
I pat all new players on the back when they press "e" to build.
The low skilled regulars of Community A will be playing against equally low skilled regulars, and will arrive at an average win rate of about 50%. The high skilled regs of Community B will be playing against equally high skilled regs, and will again arrive at an average win rate of about 50%.
New Player C logs on and browses for a community to join. How is the server ranking supposed to distinguish between servers A and B? How will player C know which one is more suited to his skill level? Both servers are full of players who average about a 50% win rate, even though any players from server B who went over to server A would probably clean house. What are you proposing to measure to show the difference?
<img src="http://img519.imageshack.us/img519/9641/nstn2.png" border="0" alt="IPB Image" />
EDIT: Misunderstood your post. I'm not proposing an algorithm, I'm suggesting a feature. I leave the algorithms to people much cooler than me.
Either they stack like he** (ive seen that already) or with a kind of autoteambalance, u would get the sh**z if u want to be a alien, or play with a friend, but the autobalance doesnt allow it.
A simple K:D ratio doesn't reward the supporting roles and will therefore hurt teamplay. Since one of the main ideas behind NS is teamplay, I think this kind of system would hurt the idea of NS.
A more sophisticated approach (taking into account things like building, healing, teamwork, following orders...) will always have some kind of weakness, that can be exploited by some players to increase their stats.
Also, I think the NS game mechanics are far too complex for any algorithm to determine your skill level accurately. (sometimes it's good for a marine to run alone [e.g. ninja pg] but most of the time a marine should stick with his team - there is no way how an algortihm could distinguish between the 2 cases)
If a stats system were incorporated, it would have to decide at the beginning of the round, if the game should be counted, or if it's invalid. There are several reasons why some games shouldn't count to your global stats: too few/many players, unbalanced teams, custom lua scripts, ...
I'm glad that we will have an opportunity to see if a system, like proposed in this thread, will work in a "source" based game. Valve will release <a href="http://www.l4d.com/" target="_blank">"Left 4 Dead"</a> soon (Q2 2007) and with it, Steam will get upgraded. Every Steam user will get his personal stats page, where everybody can see how he plays. Left 4 Dead incorporates badges (like mentioned in this thread) and also features "matchmaking".
Since L4D focuses strongly on teamplay, we will see if a sophisticated stats system will help promote teamwork and also if there are any flaws in the system that can be exploited.
I guess that a stats system in NS2 is not top priotity for the devs, so they can just wait it out and see how well it works wit L4D and decide later.
The demonware solution seems really effective. Look at COD3 matchmaking on the 360 vs Halo2 to see what improvements have been made.
But when playing against other humans, how do you know if that skulk died because he sucks at skulking or whether he did everything right but that marine was just a crackshot?
<!--QuoteEnd--></div><!--QuoteEEnd-->
I believe DoD has a system that moves your stats up or down depending on the ranking of the player that killed/was killed by you.
Isn't L4D exclusively co-op play though?
<!--QuoteEnd--></div><!--QuoteEEnd-->
Not exclusively, players can also join the infested.
But your point is still valid. The stats system in L4D will propably don't be too complex. That's because it has a very simple objective (get from the start to the end of the level without dieing) that can be easily evaluated by an algorithm. Also the other things that will be incorporated in the L4D stats system, like if you heal your teammate, cover them or in general "teamplay" are quite a simple set of rules.
If you plan to do something similar for NS, the system would have to be far more complex, because NS has a variety of different objectives that can not be assessed as easily.
statistics are bad, just let people play, so they manipulate them. some want to have super high ranking to make them think theyre pro and then they get continously owned by a "non-statistical" player with a lower ranking, but a lot more skill. That wouldnt actually help, would it?
a server with a complexe ranking system is already there! its a german one with a non-competitive community called SEK2000 (<a href="http://www.sek2000.de" target="_blank">link</a>).
This servers made me think that rankings actually CAN have a good impact on public play, but they also have a negative one. If you manipulate your statistics or if you play there for a long time and youve become a lot better, the statistic might actually be in favour for you because they arent recent enough to represent your level of skill, that means your old statistical data lowers your overall statistical value which makes your team win more often.
bla bla bla...
However, those statistics become practically stupid in clanwars, since you normally know how got the opponent team and your team is, so you have the best chance to pick an equally skilled opponent.
...a server with a complexe ranking system is already there! its a german one with a non-competitive community called SEK2000 (<a href="http://www.sek2000.de" target="_blank">link</a>).
This servers made me think that rankings actually CAN have a good impact on public play...
<!--QuoteEnd--></div><!--QuoteEEnd-->
You're right, playing on the SEK is a really enjoyable experience. Certainly much better than on the average public server.
But keep in mind that the SEK server is heavily administrated. No player will ever get a chance to exploit some flaw in the ranking system because he will get banned almost immediately.
And I can assure you there are enough flaws in the SEK's ranking system. For example it punishes marine when they walk alone too much. As I mentioned in a previous post, for the most time that's a good thing, but there are cases when a marine should go alone. The algorithm used does not differentiated between these two cases.
Imagine the same system as on the SEK being used on all servers, some of which don't have an admin present all the time. It would definitively ruin the game experience on most of them.
Having servers average the 'ranking' of the connected players to display an average skill amount is probably a VERY good idea (or at least having an option for it).
Having players get a small icon by their name by loosely how uber they are...isn't a bad thing. It encourages them to get better.
Rankings should be taken with a grain of salt, and anyone who has played in any clan matches knows it. However, rankings are normally sorta-kinda accurate. Why are most of the better players (CAL LVL) ranked in the top 100 when GFB2S is ranking them? Why are all the pub noobs 1000+ ranked? ...gee..i wonder.
O sure, blame them for whoring res, but if they didn't ###### res and go higher lifeforms--who would effectively? Pub noobs who dont know how to lerk and fade can still gestate into them and die VERY fast...and still go 3-23. The uber top 100 ranked players on GFB2S at least know WTH to do, and that's why they have the best ratios. It's not easy in NS to go 40-3 nor 50-25 (otherwise everyone would be getting those types of ratios).
Ranking due to percentile is good, exact ranking is bad. Ranking should include common activities such as building, killing peeps, killing buildings, damage to buildings, assist kill if you did 40%+ damage to target, and maybe one or two other things such as welding.
The end, that's all you need and it'll be 'fairly' accurate.
If the server says 10stars uber players play there. 1 star == noobs play there. 5 stars == average. Simple, effective.