Statistics

[Blacktiger]

I have noticed the suggestion of statistics have been posted before, but they seem to be oriented primarily towards rating players. What if we started by adding end of game statistics similar to warcraft/starcraft? For example, I did this list fairly quickly.

Unit Stats:
- Units Recruited
- Units Killed
- Damage Dealt
- Damage Recieved
- Experience Earned
- Highest Level Unit

Resource Stats:
- Gold Earned
- Gold Spent
- Gold Lost to Upkeep
- Average Villages Owned vs Total Villages on Map

After we have seen how these statistics actually correlate to how well a player is doing, then we could perhaps use them to adjust a player's rating on the multiplayer server. Thoughts? Other suggestions for statistics that might be helpful?

[Wintermute]

After we have seen how these statistics actually correlate to how well a player is doing, then we could perhaps use them to adjust a player's rating on the multiplayer server. Thoughts? Other suggestions for statistics that might be helpful?

While statistics are enjoyable to many players (I would like to see such statistics), the development team feels strongly that an official ranking system for the multiplayer server is a bad idea. Ranking systems have several drawbacks, most notably IMO:

1. Players not wanting to play games for fear of losses, or because they don't like the atmosphere of competitive games
2. Incentive to cheat. As long as there is a ranking system, there is incentive for players to hack the game to win.
3. The game is constantly changing, what happens to rankings when new versions come out frequently, sometimes with large changes?

That said, there is an [url=While statistics are enjoyable to many players (I would like to see such statistics), the development team feels strongly that an official ranking system for the multiplayer server is a bad idea. Ranking systems have several drawbacks, most notably IMO:

1. Players not wanting to play games for fear of losses, or because they don't like the atmosphere of competitive games
2. Incentive to cheat. As long as there is a ranking system, there is incentive for players to hack the game to win.
3. The game is constantly changing, what happens to rankings when new versions come out frequently, sometimes with large changes?

That said, there is an unofficial ranking system already. It is pretty good, and should suffice for I daresay most players.

unofficial ranking system already. It is pretty good, and should suffice for I daresay most players.

[Turuk]

One thought on this is the sheer number of different games that people play in multiplayer and how skewed these statistics could be. Even if you did indeed limit the ranking to only certain maps or types of games.... That leads to my next thought.

My second thought on this (which was covered broadly by Wintermute), is that there is really no way to limit cheaters without making some changes. Currently you could run two clients and lose against one version of yourself to drive those ratings up, or if you want to have a friend help, just have him use his IP to start a game and turn the other side over to you.

[Blacktiger]

Personally, I feel multiplayer ratings are mostly for fun. If they aren't actually used to determine who you play against or anything then who cares if someone wants to cheat? That said, I am really just interested in some good statistics tracking.

[Turuk]

After we have seen how these statistics actually correlate to how well a player is doing, then we could perhaps use them to adjust a player's rating on the multiplayer server.

Personally, I feel multiplayer ratings are mostly for fun. If they aren't actually used to determine who you play against or anything then who cares if someone wants to cheat? That said, I am really just interested in some good statistics tracking.

Ok, well, which is the way that you want it then? You have offered two slightly conflicting opinions.

The only issue with the argument of having them "just for fun" is that you are asking the developers to come up with a way to track all of those statistics on the MP server to display to the player (be it hard or easy for them to do), for no benefit other than to have a ranking system for the heck of it. Also, what is the actual fun when you never really know if the person you are playing has cheated or not, so the ranking is only relevant to you.... but then it's not proper ranking as determining rank that is influenced by those who cheat would mean your rank is not realistic.

[Blacktiger]

Well, I intended this post more for the game-by-game statistics not the online multiplayer part anyway. I know I have wanted them for awhile myself. I'm actually hoping I can submit a patch to allow tracking statistics myself, just as soon as I figure out how to get the mac build environment set up and start talking with the other developers about it. So what I really wanted to know what what kind of statistics people would like to review at the end of the game.

[Turuk]

Unit Stats:
- Units Recruited
- Units Killed
- Damage Dealt
- Damage Recieved

Right, well you are aware that these statistics are already tracked?

[bert1]

Well, I intended this post more for the game-by-game statistics not the online multiplayer part anyway. I know I have wanted them for awhile myself. I'm actually hoping I can submit a patch to allow tracking statistics myself, just as soon as I figure out how to get the mac build environment set up and start talking with the other developers about it. So what I really wanted to know what what kind of statistics people would like to review at the end of the game.

I'm working on a stat designed to show a winner after a certain time. This is the thread about it.

It's based on Sapient's original idea that you linked to. The base stat is a measure of 'army strength' and then it gets modified according to time of day, who played last, villages held and gold in the bank. It's a tricky time-consuming business (lots of entering numbers into spreadsheets) but I'm analysing a game at several points and it seems to be pretty workable IMO. I'll have some results soon and I'll post them in the original thread.

[Cernunnos]

Turuk wrote :

Right, well you are aware that these statistics are already tracked?

I think, the idea is more like an ending screen or an ending statistiques window pop up...
Victory:
Units killed... etc...

Blacktiger mentionned WC and SC... so that's what i understand.

Anyway, some more stats accessible while you play would be good imo, i remember having suggested to be able to see the gold lost (but search engine is down right now) :

What i want to say with gold lost is that you should know how much gold are the X units you killed and how much gold are the Y units you lost.
Because just kill/lost is not always relevant (for example when there is gobs...)
=>
Units Killed: 11;
Units killed value: 88; (was 11 gobs)
Units lost: 5;
Units lost value: 95 (was 5 clashers).

This idea derives from chess when at each moment you're able to see the pawns killed and the pawns you lost, this is easier with gold but would be eventually even better if you could see the units. (graphically would be more fun, but by text would be largely sufficient):

If this has already been debated and rejected, then please forgive me.

Thanks, bye.

[Turuk]

I think, the idea is more like an ending screen or an ending statistiques window pop up

Right, I am aware of that. If you had quoted my whole post you would see that I was referencing him to the fact that the ones I noted are already tracked, and when a scenario ends, you can indeed view them by hitting the statistics button (even if it does not pop up for you).


Also, bert1 I believe you are focusing on a different idea that establishes a winner after a certain time in a multiplayer game, while he is looking for more single-player statistics.

Well, I intended this post more for the game-by-game statistics not the online multiplayer part anyway.

[Blacktiger]

Right, I am aware of that. If you had quoted my whole post you would see that I was referencing him to the fact that the ones I noted are already tracked, and when a scenario ends, you can indeed view them by hitting the statistics button (even if it does not pop up for you).

You're right, I always forget about that statistics screen and that status table (which tells you how much gold everyone has/is earning at a given time). Probably because it's buried in the menu and I hardly ever look at it. I do still think that some more end of game style statistics (like the average number of villages owned) would be helpful. Cernunnos, as you said, I was thinking that a splash screen showing them after the game ala warcraft or starcraft would be a good place to review them. Although, maybe we could just add them into the statistics screen to make it more comprehensive. That would certainly make it fit into the BFW style.

[Turuk]

Although, maybe we could just add them into the statistics screen to make it more comprehensive. That would certainly make it fit into the BFW style.

Indeed, this is probably a better idea, having it come up at the end would not be any worse if it was tacked on to the other information already displayed (i.e. how much gold the player is carrying over). Though the players would probably prefer an option to choose whether or not this was displayed.

Probably because it's buried in the menu and I hardly ever look at it. I do still think that some more end of game style statistics (like the average number of villages owned) would be helpful.

Hmm, you will have to be careful about what statistics you put in there, or at least how they are computed. Average number of villages owned is fine as long as the percentage per map is averaged, and not just how many villages the player had on average for each scenario, as that is affected by the total villages, which obviously varies by scenario. Even then, some scenarios are designed with villages that the player may never claim such as in a direction the player will not travel for a leader they do not have to kill, or placed far behind the enemy leader. So that may skew the player's idea of how well they did.

[Blacktiger]

[Average number of villages owned is fine as long as the percentage per map is averaged, and not just how many villages the player had on average for each scenario, as that is affected by the total villages, which obviously varies by scenario. Even then, some scenarios are designed with villages that the player may never claim such as in a direction the player will not travel for a leader they do not have to kill, or placed far behind the enemy leader. So that may skew the player's idea of how well they did.

I was assuming that the stat could give the average number of villages the player held with the total number of villages on the map displayed next to it. I'm not sure how useful this statistic could be or not, but generally speaking I would expect a match with two even opponents to give each player an average village count of around half of the villages available for a multiplayer game. Obviously this may be much less useful in a campaign mission. It could be that the average number of villages in a multiplayer game won't vary enough to be helpful, but who knows. The primary reason I thought this would be useful is if the player can compare how they did to the other players, if you consistently hold a lower average than the other players when you lose (for example), then perhaps that is the reason you lost.

[Turuk]

I was assuming that the stat could give the average number of villages the player held with the total number of villages on the map displayed next to it. I'm not sure how useful this statistic could be or not, but generally speaking I would expect a match with two even opponents to give each player an average village count of around half of the villages available for a multiplayer game. Obviously this may be much less useful in a campaign mission. It could be that the average number of villages in a multiplayer game won't vary enough to be helpful, but who knows. The primary reason I thought this would be useful is if the player can compare how they did to the other players, if you consistently hold a lower average than the other players when you lose (for example), then perhaps that is the reason you lost.

Ah, you keep going back from wanting this just for campaign or for MP, so it's hard to track what you want each statistic for. Well, would it not make sense that as you lose and someone is about to kill your leader, with the exception of lucky hits or an assassination by them, they have probably overrun your villages by then anyways?

[Wintermute]

I was assuming that the stat could give the average number of villages the player held with the total number of villages on the map displayed next to it. I'm not sure how useful this statistic could be or not, but generally speaking I would expect a match with two even opponents to give each player an average village count of around half of the villages available for a multiplayer game. Obviously this may be much less useful in a campaign mission. It could be that the average number of villages in a multiplayer game won't vary enough to be helpful, but who knows. The primary reason I thought this would be useful is if the player can compare how they did to the other players, if you consistently hold a lower average than the other players when you lose (for example), then perhaps that is the reason you lost.

Your intentions are obviously good, and I agree with you that tracking more statistics would be nice, but tracking average number of villages is not very useful, IMO. The factors that go into your economy are a lot more complicated than just how many villages you control. Things like upkeep, number of losses, cost of losses all factor in. In addition, when villages are taken really makes an "average village-turns" statistic harder to keep track of. For example: we both have 8 villages, I steal one of yours on my turn, then on your turn (you have 7) you kill my unit, and retake the village. I collect gold on my turn, and I have 8 villages again. At what point in there did the village counting happen? At no point did I ever collect gold for that village and moreover, I lost a unit, while you lost one turn of income from one unit. What if this happens for multiple villages on a given turn? We have to be careful not to count them as 10 -6 when it is really even. Also: I think that most players who are serious enough to care about the statistic already have a *very* good idea if they are behind in villages. In so-called "high-level games" players very often are counting gold (or estimating if they are ahead/behind etc.) on a pretty regular basis - counting villages is a given.

Summary: many players would like to track more statistics, but one of the reasons that they are not implemented currently is that we have to be very careful about what is really a useful statistic. Current EV statistics are also flawed (you can do lots of damage but always miss that critical killing blow, or poison, or whatever), but they are still interesting. Wesnoth is a very complicated game in many ways, and lots of the ideas that get thrown around but it is seldom as straightforward as the proposer suggests. So... keep thinking!