AEStats : Stats to Track Your Performance In Jedi Outcast

[ed_silvergun]

Originally posted by KillerBee:

<STRONG>hmm well I have serious doubts over stats pages.. demos are far better</STRONG>

 

I have to agree on that. As interesting as stats pages are, I trust far more what I see being played out before my own eyes. You learn a lot from watching others play, and I love to see other peoples' demos, particularly when the person is far better than me!

[KillerBee]

I doubt it will deter new players. on the grounds they will care about their rank, although the encouragement of newbie bashing because it boosts your position quciker.

 

Its just they doesn't really show game skill at all. I am sure on alot of games poeple ranked in the ks prolly could beat the top players.

Even if it works out who they killed, whats stopping a newbie using the name of a great player? they are really problematic to get to represent true skill.

[Guest Krayt Tion]

One thing jk fans don't have is the experience of a good star wars server community. Sure, there were some guys that claimed to have put up dedicated servers but with the host client relationship in JK there was never really such a thing.

 

Players on your server will get a sense of who the regulars are, who the good players are, what weapons and strategies they like to use and are good with, who the typical campers are and where they like to hide (if the game allows such a thing), who is a good team player and can accomplish objectives, all without stats. Most of those things can be further analyzed with a more in depth look into covered stats. However, the actual "skill" ranking on most of the skill stat assements I've seen has been flakey and out of wack. This is one instance where I believe the stats will never tell the true story, but I still enjoy the other numbers stats can provide.

[Wilhuf]

We would have to look at more than just a single index value to determine a players true skill rating, of course.

 

As we know, these stat programs typically provide several ranking methods, and make several measurements to try to get at those elusive qualities of what makes a player skilled. AeStats, for instance is based on the international Chess ladder ranking system, but also allows the server admin to adjust the ranking method.

 

And of course player skill by weapon, and skill against individual players is usually calculated. These are actually pretty good measures, when considered altogether, of player skill.

 

Stat pages do indeed tell a story about a gamer beyond what a single demo can tell. For instance, stat pages can track player habit and performance over long periods of time. Over time, and as skill levels increase as all gamers become more familiar with game mechanics, stat pages can show whether or not a player has improved or worsened relative to his/her peers. This is something a demo recording simply cannot do.

 

Stat pages can also help players predict how their opponents will behave. E.g. the stat page tells me that my opponent, 'l33t_gamerX' uses the concussion rifle and grip for 80% of his kills. Gamers can take this knowledge with them into the game and use it to adjust their strategy. Again, a demo will only show you one sample of gamer behavior, out of the context of other games that the participants have played.

 

I do belive stat pages can quantify player skill. But I do agree with Krayt that they can not quantify player leadership (I don't believe stat pages attempt to quantify team leadership, in any case). And of course if there are errors in the stat tracking and calculation, then the stats are meaningless.

 

I do know that Aeon has worked through several versions of Aestats over the years to ensure the most reliable stat compilations. It will be fun to see what he comes up with for Jedi Outcast.

 

Should we base our gaming behavior on stats? That is completely up to the individual. If gamers believe the ranking system is representative and meaningful, and matters to them, then by all means, they play to get good stats. To some, playing for the stats is fun.

 

On the other hand, if gamers don't care about the stats, then they just ignore them, and get out there and play for the pure fun of it.

 

[ July 26, 2001: Message edited by: Wilhuf ]

[Wilhuf]

To address Bee's comment about 'whats stopping a newbie using the name of a great player? they are really problematic to get to represent true skill.'

 

I agree that can mess with stats. On the other hand there are ways of addressing the problem.

 

NgStats uses a unique player id and password to prevent name jacking. While multiple players can use the same player names in ngStats, as long as you use a unique password, the statistics will be compiled by unique name/password. Merging of stats because of name jacking doesn't happen in ngstats. E.g. there can be multiple 'killerbee' players, but stats will be compiled separately for each password for killerbee.

 

(Unless of course your password is somehow stolen).

 

I don't know whether Aestats has any measures like ngStats for preventing name jacking. It does have name filtering. It might also have IP filtering, so that if some llama comes in, stealing another player name, his ip would still be detected, filtered out, and his stats rejected from the compilation.

[Guest Krayt Tion]

Many Half-Life stat programs use the following to record stats:

 

The first method and arguably the most accurate, is by wonid. Everytime you play, won.net lets the server know your unique id. The problem is, there is more than one won.net server, and they dont talk to each other. At this point, you can have up to three different wonids, and when you connect, you might be using any one of the three. If you want to make sure you only use one won.net authentication server, go into your halflife\valve directory and open a file called "woncomm.lst". Go to the part that says Auth, and remove all but one of the servers. There are three, West, Central and East. Keep the one closest to your place of residence.

 

Another way to store is by IP. This way works great for LAN games and for LPB servers as many DSL, EDU and cable connections dont reset IPs that often. However, for modemers and some cable users, its not good. You probably wont see this option much on internet servers.

 

The other way is to store by name, which is the most innaccurate way. Its a bit more forgiving in terms of what it will log but your stats arent sticky like they are with wonid and ip.

 

Unfortunately the Half-Life logging system is not perfect, and with less than perfect logs you get less than perfect stats. Ideally all stats work correctly and there shouldn't be a problem. However, sometimes when a person joins a game in the middle of log opening or closing (ie server start or level change), the player does not get closed out of logs properly and shows false numbers in some fields ( ie. kills per hour, time on server). Life sucks anyway

 

Stealing names is not really a problem. The person might be able to use a regular's name if the regular isn't around (if the regular is in the game that person would show up as 'regular(2)' or vise versa); the regular player's name the imposter used will definitely show up in the imposter's Alias list displayed with the rest of his stats in programs like Half-Stats. And if the server admin has spoken out against such things, that player can be banned.

 

Besides, Half-Life server admins can check the logs every day if they'd like, so if another person, read: lamer, deliberately uses a regular server player's nick, the logged IP will tell the real story. With a good server admin that cares about his server community, this player will be banned.

 

We would have to look at more than just a single index value to determine a players true skill rating, of course.

 

I agree, and that's all I was stating. I was only saying that a stat program itself couldn't produce an accurate assessment of skill through the display of its own distinct value for it.

 

Yes, we would look at many things on a stat page to determine our best guess as to who we think the most skilled players are. But we are not the stat program, and our collective judgement of all the player's stats does not get displayed as fact on the stat page. The other ranking methods the stats provide are as inaccurate by themselves for ranking as by that "single index value" that calculates and lists skill, being that they also don't tell the entire story.

 

It would be interesting to see the top 10 or so STATed players vote on who they think the top skilled players on the server are in some kind of Jedi Stat Council. They would be able to look at the collective stats, baring the bunk Skill Ranking, and make a decision. They wouldn't be allowed to vote for themselves, could vote for anyone ranked in stats, and the top three voted for would get a different skill ranking, etc. They would of course have to explain their decisions with a quick blurb in written form. This absolutely wouldn't work on your average server lacking a real community and chalk full of repressed teenagers that would rather vote for their friends or vote for the guy ranked 478th because his name is a giant ASCII penis. Nor would this system be perfect.

 

But you know what, nothing is perfect. If these kinds of issues matter to anyone then choose your server very carefully. If not, as has been stated, just have fun how you like to and go about your business.

 

[ July 26, 2001: Message edited by: Krayt Tion ]

[Wilhuf]

Hmm a stat council is an interesting idea. On the surface, I think I'd rather rely on the impartial formula of a stat ranking program to determine 'the top ten' than a council of wise (who are potentially vulnerable to outside influence) in their decision.

 

It is precisely because the stat program does not care who the player is that it will rank without bias or preference for its gaming buddies.

 

But as you've said, such a stat council would have to back up their rankings with an opinion, which presumably would be founded on statistical reporting from reliable servers. A way of compiling interpreted stats on players across a spectrum of servers, and polling for opinion on player skill. Definitely an interesting idea. Kind of like a council of Howard Cosells.

[KillerBee]

Well as someone without a fixed ip for half the year, I don't think Iding a player with an ip will work too well.

IDing with a unique ID local to your poot would be good, and near fool proof. which solves 1 problem.

 

Telling you what weapon someone uses on average isnt too useful, if I am playing level a which needs a rail, I will use a rail. if I am playing level b that needs a shotgun I use a shot gun. :/

showing that someone in jk uses a sabre 100% of the time, and dies with a sabre in his hand 100% of the time isnt too useful

 

knowing someone is 50% accurate with a rail is half useful, although playing people who dont move that gets distorted..

also without number of shots fired its also useless. ive got 50% and shot twice.. rl stats are always confused by RLing

 

comparing deaths to kills can also be problematic. Ive had games where ive doubled someones kills but done virtually the same ammount of damage.

The ammount of info need to be found to make a real statistical annalysis on one person is huge.. eg accuracy would need to be compared to the other players movement level, I wonder if collecting info on how many seconds the player stayed in a streight line on average would be viable

 

also peoples preformance vary from map to map, so this would needed to be collected..

 

collecting data to compare regulars on a server is kinda pointless seeing the players should know who is better

 

yeah if people like looking at the stats is fine.. but if they then decide to boost their kills/sec by masacring newbies... well.

 

if the data could be used to compile a ladder based on who had beaten who that would be ok.. but It would run into problems if some person had no one he could compare to another person directly or indirectly..

[WD_ToRMeNt]

Yep, there are many possible stats for a FPS. But I don't think that it'll be that big of a deal, I mean I don't plan on playing on just one server anyway.

 

BTW, there is nothing wrong with getting 80% of your kills with grip/conc/destruction. The fighting style used now by the experts has evolved to be the most effective. The only way to counter it is simply to be more proficient in that style of playing.

 

[ July 26, 2001: Message edited by: -WD- ToRMeNt ]

[Guest Krayt Tion]

I wouldn't call the stats useful so much as I would interesting. I don't see much to argue here either. No one honestly studies stats like the map of a war campaign. I think it holds true that most of the things you can learn from stats you will realize in-game way before you ever break into the ranking and take a peak at the stats. The most popular weapon, the most popular weapon on a map, the most popular pistol: the first time you play usually gives a good representation of this.

 

The things that helped me the most in cstrike (and I'll make it simple) were knowing the weapons, knowing the map, and knowing how people like to play the map. Not just for people on my server, but every where. There are plenty of common strategies used and a peak into stats wouldn't have helped me learn any of them.

 

Only time I can imagine actually study stats would in a Clan Match against the my clan opponent, in the bizarre occurance that they might be using weapons on a map that were most strange, like they all used shotguns, so you told you clan to pick mid-long range weapons and keep their distances. But the chances of that happening let alone you seeing their data are slime to none.

 

Btw, with a stat council, people would be able to see the normal stats, Skill rating and all, just fine. They could completely ignore the council's decisions if they wanted to, just as how people like me ignore the skill ranking as part of stats themselves. I'm not saying a Stat Council is a practical means to an end, because like a bad bug patch it introduces other problems with it. It is at least an attempt to balance out the often overly artificial Skill Rating part of the Stat System which lacks the intelligence to encompass all which you would usually attribute to someone of FPS skill. I'd rather take my chances with a bit of corruptable humanity than to accept the Skills portion of the stats.

[Wilhuf]

Yes, actually ngstats records weapon use by specific map. You can find out which weapons are the most effective by map. There are a ton of crosstabulations that ngstats generates. E.g. player efficiency by map, which maps get the most player time online, what is the average latency per each map, etc.

 

The weapons stats do indeed tell us something about how effective weapons are: for instance, ngstats tell us how many times people die while using a weapon, in relation to how many kills they get. The information controls for which map is played (e.g., the flak cannon is 80% effective on map x, but only 50% effective on map y). The stats will also show you how player habits and weapon preferences change over time.

 

 

Bee, I'm not quite sure I understand how target movement information would help identify a player's skill. I would think a simple indicator of relative skill (e.g. player ranked 1 used concussion and grip on player ranked 50, earning a very small increase in rank, or player ranked 50 used a bowcaster to kill player ranked 1 and earned a very high increase in rank).

 

In the case of ngstats, the ranking system is supposed to help gamers identify players who have similar skill, btw. E.g. players ranked 40-50 likely have similar skill. It is also broken down by server, so one can identify players of similar skill who frequent a specific server. Of course, as I said earlier, it would be easy enough to use stats to identify gamers who like to go after easy lower-ranked targets, and only use one method of attack.

 

Also, game stats can quantify how good a player is for different game types. (e.g. 'l33t_gamerx' does pretty well in concussion/grip FFA extravaganzas, winning 60% of bouts, but actually can't hold his own in a medium force CTF game, winning only 25% of bouts.) This helps us get a feel for the overall abilities of the player in different game modes and modifications by quantifying performance (as opposed to taking the usual self-proclaimed 'I rock at full force and no force games' at face value).

 

As I said, it helps you back up your talk about how great you think you are.

 

I don't know whether aestats can deal with ip ranges. I do know that q3 tech allows servers to ban by ip range, so presumably at least the ip information can be obtained and, if aeon gets around to it, modify code to account for ranges. Aestats is not as sophisticated as ngstats (nor should it be).

 

Seeing how much ngstats can actually do, I am hopeful we see zonestats support for JKO.

 

As far as 'what an in-game experience v stats can and cannot tell you' goes, in-game experience cannot tell you an overall ranking for all players. In game experience will give you an impression, but it won't quantify it. After all, the game winner usually is the one who has the highest score at the end of the game. And of course, score is a statistic (actually it's a measurement).

 

In-game experience wont tell you what weapon an opponent is likely to use before starting the game (of course). In game experience wont predict whether a player tends to be a flag capper or a flag defender, while stats will.

 

[ July 26, 2001: Message edited by: Wilhuf ]