New ladder system for fortress

[Lucifer]

This isn't Lucifer Land, here's what you actually said:

It is, I am saying he is the only one with good ideas.

Watch out for that wall!

[~*PsYkO*~]

Oooo I love when people prove my points.

[~*PsYkO*~]

So..to get back on topic before I got interupted..

My opinion is this: A lot more bragging will start with good ladders, a new generation of players will emerge...Imagine everyone like skip I think the point system in game is fine enough. I've got a good in-game reference to the ladder systems.

Ok we all know ladder in fortress doesn't mean squat but you can go to stats.xzlclan.com a prevelent stat site for a prevelent high rubber clan you can see they have an at least semi-representative ladder, for the most part top players are in the top positions on the ladder. Now if you want to take a tiny stroll into their high rubber server you will see A LOT more skill talk and bragging etc. I went to play there to practice about a week ago when I was kinda bored at fortress and it is unbelievable how meaner people are than in fortress. And I can just imagine what it would be like if you could swear(since xzl has a no cursing rule)!

[Phytotron]

Um, just for the record, I'm not a developer/programmer, and I do play regularly and frequently (more than I'd like to admit). But, you see, I never play fortress or any of the other popular servers because I think they're bad servers. I remain almost exclusively in Shrunkland. Of course, anyone who is actually familiar with me in any basic sense would know these things.

One word for the rest of psyko's blatherings: incredible. (I intend its classical meaning, not the overused, hyperbolic, junk english meaning of "really great.")

[Z-Man]

Ok, pSyKo says newbie is the one making good suggestions. psYKo also says that a good ladder is a bad thing. newbie is making suggestions about a ladder (he's posting on a ladder thread, and making suggestions involving score). Does that mean now that his suggestions are good because they create a bad ladder system?

Alos, pSYKo, if you actually read mine and Philippe's posts, what we're saying is exactly that a ladder with the suggested rules on fortress is a bad idea because it will affect gameplay in a negative way. In your rare on-topic and non-people-bashing comments, you agree with us

[newbie]

Alos, pSYKo, if you actually read mine and Philippe's posts, what we're saying is exactly that a ladder with the suggested rules on fortress is a bad idea because it will affect gameplay in a negative way.

do you have better 'system'?

it is just a solution for one particular thing like suiciding every round and still being on the road to a better ladder position

i'm sure here are only two solutions: a) no points for rounds; b) different points based on the survival ability (because fortress is about _surviving_ and conquering/defending the zone)

if someone still wants to give points for nothing then at least those points can't be leading points (f.e. 1/5 of the base value)

btw, after all 'killing teammates' is not a convincing argument against what I suggested, because _killing teammates_ can be easily punished

it would be nice to hear about a potential better suggestion than mine instead of writting why something does not work and not offering anything else

[Lucifer]

A couple of things, newbie. First, for most people it's easier to find problems in other people's suggestions than it is to find problems in their own, and second, for most people it's easier to find problems in other people's suggestions than it is to come up with a better idea.

I generally agree after a point that if someone continually says "That's a bad idea" without offering anything in return they should put up or shut up.

However, in software design (or any systems design afaik), it's usually a very productive model to have a person make a suggestion, and then have someone find holes in it. Then the first person revises their suggestion to accomodate the holes, fixing holes he finds on his own, and presents the new and improved version. THen the second person finds more holes. And so on and so forth. Within the context of a single thread, that sort of behavior is very good. It's when it spreads across multiple threads that it gets damned annoying. Anyway, z-man frequently takes the roles of the naysayer not because he doesn't have a better idea but because he wants you to revise yours and make it better, for whatever reason. There are quite a few threads where he says both "That's a bad idea" and "I have a better idea". So I'd tend to suspect he's wanting you to revise yours and see what it looks like.

Now to the topic at hand.

I don't personally think there can be a one size fits all ladder. Different people expect to receive rewards for different things, and the ladder is supposed to generalize all of those rewards into some metric of ability.

The way I see it, you take the traits that you think make a good player, figure out some way to measure these traits, then take regular measurements. You wind up with several functions of time, where each function gives how much the player has shown he can do <something that makes him good>. The composition of all of these functions would be the player's overall skill level. The average value of the composition function would be the player's general skill level, and the value at any point in time of the composition function would be how well the player was playing at that point in time. Of course, for real thoroughness you'd probably have to do some fancy math with a player's measurements against those of his opponents.

So that's the system the way I'd set it up. Now, what do we measure, and how? I'd prefer to take enough measurements that people can't advance on the ladder without becoming good at several things and using those things successfully. The more you require a player to be good at in order to rank on the ladder, the more you deprive him of the option to choose his play with the sole intent of rising on the ladder (and thus abusing the system).

I don't think rewarding dead players is inherently bad, because there are times when a dead player deserves plenty of reward, or even most of it. In a deathmatch play, the player that killed 6 opponents before crashing out of exhaustion deserves more ladder standing than the person who won the round by killing a chatbot. Similarly, rewarding living players is already known to be inaccurate, the example I just gave serves to show that too. In fortress play, the MVP for the round might well be the guy that crashed into the goalie's wall and let in 2 of his teammates, but he's not alive to collect the reward. The problem here is in first deciding what to measure, and second, deciding what to do with it. So I'd like to know what people want to measure. We can worry about how to measure it afterwards, but a brainstorm of what you want to measure would be pretty useful, imo.

[ed]

For the most part the rest: Lucifer, Phytotron, wrtl, ed, philip, z-man all program this game

Not me. I just admin the CT servers, which is why I have more than a passing interest in much needed changes to the ladder system.

I'm sure it will not please everyone. Certain styles of play will always better rewarded better than others. It's up to the server admin how the system is set up, what style of play he/she thinks deserves ladder points.

[Z-Man]

I can't offer a perfect system for a player ladder on a team server, because a perfect system doesn't exist. There will always be manipulation methods. People who don't care about their own score can pull the score of selected other players down. There are completely skill-less ways to climb a bit on the ladder.

If our goal is skill measurement for individual players on a free-for-all team server, I'd say the best we can do is to make sure that for every skillless way to rise some on the ladder, you can improve it with skillful play. For the current "winning team gets points" system, we have that: sure, you get points by suiciding right at the start if your team still wins, but by actually playing to the best of your abilities, you increase the chance that your team wins, hence your score. Switching to the winning team and doing nothing there improves your score; but switching to the winning team and actually helping it is still better.

No, we can't punish team kills. The team kill or suicide display is inaccurate and will always be, because the computer can't read the players minds and see whose fault a death was. And Fortress is not about survival, it is about taking the Fortress no matter whether you survive it or not. There is a reason death is not punished on Bugfarm. As a corollary, it also is not about killing the enemy. If someone comes and suggests to remove the kill score entirely, I'd probably do it.

Of course, there is a better system: get rid of the ladder for individual players and make it a ladder for the teams. That should be possible in 0.3 where we have customizable team names.

[~*PsYkO*~]

I'm saying if there was a ladder put in place newbie is the only one putting forth good ideas that the PLAYERS would like. OVERALL, the ladder is a bad idea...I read your posts z-man AGAIN and I do now catch the underlying ladder is bad for the game idea but our reasons for the ladder being bad or completely different. And again please find a thread where I actually begin the berating...seems to me this one was started by ed so go after him.....and ed sorry if I mistook you for a programmer but when you post codes after codes after codes I got misconstrued

And to add something to newbie's points about fortress is about surviving and conquering the round, how about give points for someone capturing the zone. So ultimately everyone's points at the end of a match = the team's total...Make it kind of like sumo: If one person captures the zone that person gets 10 points, if two people capture it each get 5 points.

[newbie]

z-man, just about what fortress is

i dont know how many hours have you spent practicing and testing fortress on the grid, but my point of view is a lot different than yours

how a team player can take the zone and win without killing all opponents?
he can get in.... but this does not mean that he will win,
he can kill the goalie... but he will not win too
so what he can do? what's the fastest and easiest way to win? the easiest way and the most productive is simply to _stay alive_ in the zone and wait for a teammate

that's the final part, that's how people win, not only by taking the zone but also by staying alive in the zone

edit: this is a good idea too, psyko

[Jonathan]

The whole idea of a ladder is to show how good you are at achieving a goal, or helping your team achieve it. It shouldn't encourage a specific way of achieving that goal, or to stick to certain roles in a team just because they are more profitable in terms of ladder score, while there are also other things that have to be done. Tell me how basing the ladder score largely on deaths does that.

[Z-Man]

Newbie: yes, kills are very helpful in getting the enemy zone conquered, but so are left turns Still, you're right, core dumps are important, that (and the fact that they're about the only thing we can detect) is why they're rewarded with personal points.

Rewarding kills has this potential bad side effect: Say you're in the situation where an enemy player has overwhelmend your defense and you could start an emergency rescue. You know chances for success are slim, say 20%. You also have a 50% chance to kill an opponent that carelessly follows you, smelling sweet victory from the imminent conquest already. What will you do? If you care about your ladder score in the current setup, you'll go for the zone rescue, benefiting your team, because you only get points when your team wins the round. If you care about your personal score in the current setup, you go for the kill. If the ladder somehow involves kills (say, you lose less points for the lost round because the kill shows you made at least an effort), you'll likely go for the kill as well. But what would be the right thing for the team? The right thing for the team would be to do the rescue attempt. It gives on average 2 points (10 for winning the round times the 20% success probability), while the kill only give 1 (50% of 2).

Psyko: revarding the player who ultimately takes the zone is wrong. Taking the zone is a team effort, those players keeping additional defenders away, players holing the defense intentionally or by daring attack maneuvers, they all would deserve points. But we can't possibly detect all kinds of helpful activity, and by rewarding some (by detecting a player passing through the holes you made, for example, that would be feasible), we'd create false expectations. By rewarding a certain set of actions, we'd be saying "Here's a list of good things to do. You get rewards for them. So go ahead, do that, and don't think about other ways you can help. You may be of help indeed, but you won't get a reward." I'd consider this an insult to creative players.

I think Jonathan puts this quite nicely.

Lucifer: No, my naysaying here is the other kind The "No, you can't build a Perpetuum Mobile." kind of naysaying. I'm deeply convinced that in a sufficiently complex team game, you can't measure the skills of individual players if you don't have a set of unbiased reference players around and don't do controlled experiments.
You can measure the individual skill by putting together two roughly equally skilled teams of reference players and put the two players you want to have compared in one of those teams each. Let them play for a while, then let them switch teams, then play another wile. The player that was on the winning team more often must be the better overall player.
Needless to say, we don't have reference players and can't set up such an experiment.

[joda.bot]

hm, I just skimmed most of the topic, so I will only post a sketch now... to remind me and give you more for brainstorming:

Event logging (which creates statistics) + the possiblitity to award all events differently just leaves the problem of identifying events.

Event 1: Player A kills Player B => Player A score +10, or Player A's team +10
Event 2: Player A makes zone spin faster by x => player a score +x

Note: Event2 might be very difficult to log.

In a soccer team there is one player who scored the goal, and then perhaps an assist. Awarding players in the zone is better than no award (IMHO).

[madmax]

...with good and accurate reporting...

Lucifer's post about the ladle reminded me of this.
Any changes? What about stats you can actually view (not only in terms of the ladder)?