Comparing scores across victory conditions

[Silu]

Greetings to old and new XotMers alike

We all know that Civ's built-in win scoring formula, something quite familiar to any XotM player, doesn't exactly treat all win conditions equally. It very much values population and land area, making domination and conquest the wins that rack up the highest scores.

Just for my own enjoyment I made a makeshift way to normalize scores by win condition: parsing all the winning BotM (only) entries (medalists + fastest wins), and averaging the scores by win condition, and from those calculated factors that can be used to multiply your score with to roughly compare games with varying win conditions. Thought I'd share the results with CFC.

Maybe this is done already, but if not, I'm sure there's at least one other stat geek out there who can get a small kick out of this

The results:

Win year:
Conquest : 89 results with average: 1461 AD
Diplomatic : 79 results with average: 1771 AD
Cultural : 86 results with average: 1770 AD
Domination : 147 results with average: 1669 AD
Time : 2 results with average: 2050 AD
Religious : 57 results with average: 1303 AD
Spaceship : 93 results with average: 1891 AD

Points:
Conquest : 89 results with average: 137175
Diplomatic : 79 results with average: 66157
Cultural : 86 results with average: 39179
Domination : 147 results with average: 181283
Time : 2 results with average: 19489
Religious : 57 results with average: 65374
Spaceship : 93 results with average: 73439

SVCCF (the Silu Victory Condition Correction Factor ):
Conquest : 1.32
Diplomatic : 2.74
Cultural : 4.63
Domination : 1.0
Time : 9.30
Religious : 2.77
Spaceship : 2.47

TL;DR: Multiply your score with the factor above corresponding to your win condition above to roughly compare wins with varying conditions

 

[WastinTime]

I applaud your effort, but it won't work to equalize scores. The score milkers just happen to choose Domination because it's the most obvious/convenient choice if your going for large pop and land. However, those same scores could just as easily be achieved by simply ending the game with a religious vote, diplo, or even culture.

I know you're not suggesting we use your multipliers, but if we did, I guarantee all the score milkers would use culture. Make sense?

Anyway, it was still interesting data. Thx

 

[Silu]

I applaud your effort, but it won't work to equalize scores. The score milkers just happen to choose Domination because it's the most obvious/convenient choice if your going for large pop and land. However, those same scores could just as easily be achieved by simply ending the game with a religious vote, diplo, or even culture.

I know you're not suggesting we use your multipliers, but if we did, I guarantee all the score milkers would use culture. Make sense?

Anyway, it was still interesting data. Thx

Hmm, I'm not so sure it would be that simple. A huge part of a good score is a very early finish, which is impossible when combining two completely orthogonal development paths such as Domination and Culture. It would certainly make for some interesting "hybrid" win condition approaches though

I think the best milky way would be to go for the famed Diplomation win (either with AP or UN), since there's a lot of synergy via winning with vassal votes. I agree the factors don't work if the player aims to score with them in advance - factors for that would require a different calculation which would be far more complex and probably impossible (though I'd guess they'd be around half of these EDIT: and by half i mean square root). Would be fun to see the gold medal go equally to each win condition though

I'm also interested how Time would do, albeit with such a minuscule sample size it's hard to know if the factor is even in the right ballpark.

 

[WastinTime]

I totally agree it would be nice if gold medal wasn't always a score milking domination, and I assume that's the point you are trying to get across.
That's the main reason why I never played GOTM until very recently.

 

[DynamicSpirit]

The issue of the score favouring military wins has come up from time to time in the past, and there have been various discussions on whether the scoring system is fair (and a lot of energy spent on lots of different proposals for changing it, which almost always fell down on the problem that you could never get universal agreement on what is most fair, and almost any criteria can be subject to milking).

I personally don't pay a lot of attention to score*, and I think there's a fair consensus that your victory date, as compared to the victory dates of other people who played to the same victory condition, is a better measure of how well you played. You can (almost) see that comparison in the results of each game by sorting the results by date (although that still mixes victory conditions).

Interesting approach, Silu.

* Not at all connected with the fact that I normally play for spacerace - and that almost never gives a high score

 

[neilmeister]

Yes, I am the same. Score has never concerned me, I am always trying to achieve the fastest victory date.

You could say that score is given undue prominence, by awarding medals based on it, but I think that is a different argument.

 

[Htadus]

If you really want to make the medal a real value make it a gauntlet style score game.

Rotate the VC in some sensible manner, say alphabetical. Then give the medal to the highest score of that VC. Simple. No need to make it a Challenger start. So the medal process will be fair.

 

[shakabrade]

The only problem I see with this scoring format, is that it is stupid to do anything but sushi domination. I concur with Neilmeister on fastest finish. Although, in my private random uncooked games, I am satisfied with just winning somehow, and most of the game is about surviving and maintaining tech parity. I'd like to see more of that in GOTM games, pure survival, not some sampos and starts with every resource in game.

Regarding Suli's correction factor, for example: how can your factor distinguish between almost domination/conquest/nuclear winter UN, and real diplomatic UN victory (where you have a lot of friends)? Heavy abuses are possible.

If GOTM games, in which score is relevant, were played always on the same map, then average score and even some correction factor would have some meaning. But that UN exploit I mentioned would still be possible.

The best for making GOTM better I can think of, is simply ignoring score, and concentrating on fastest finish dates. It would be nice having all victories conditions competitions separately, in GOTM manner (same map for everyone). But that would be too much work, I guess. It seems that I am getting more and more offtopic, so I'll stop here.

 

[Silu]

No one denies that win date is a very good metric for judging game success. Simple, pretty comparable and not very prone to abusive milking - it works well for what it is. What I was looking for was another metric that could supplement that one.

For example, if I play a regular game from a great start and go for cultural (aiming for fastest win date), I'm often interested of what would the (relatively poor) ending score roughly translate to had I gone for domination - and that's where a factor such as the one in the OP might prove handy. Not when aiming for a high score with the factor from the get-go.

Didn't bother with converting the win dates to turns at first, but this talk about win dates sparked my interest. Using these turn numbers as a bias one could perhaps compare the win dates across win conditions a little better as well, again using Domination as the "standard" (one would add the number after the average winning turn to their win date to get a "standardized" date):

Conquest : turn 209, +35
Diplomatic : turn 266, -22
Cultural : turn 266, -22
Domination : turn 244, 0
Religious : turn 190, +54
Spaceship : turn 317, -73

The idea to use a constant bias here instead of a factor here is pulled from my stetson

For the record the main reasons I posted this on the GotM forums in the first place were 1) the results are based on BotM data and 2) this forum is one of the few places where people pay attention to the Civ4 standard scoring system. Not because of any serious wish to change the evaluation criteria for the XotM medalists

Regarding Silu's correction factor, for example: how can your factor distinguish between almost domination/conquest/nuclear winter UN, and real diplomatic UN victory (where you have a lot of friends)? Heavy abuses are possible.

Why isn't diplomation a real diplomatic win? The win condition doesn't heed the way in which it was obtained. I don't consider a peaceful diplo win any "purer" than a diplomation. And yes, the factor is very open to abuse due to not having developed it to withstand abuse

Though, it's an interesting thought that the hypothetical optimal scoring algorithm would award "peacefulness" as well (naturally along with other things the standard algorithm pays no heed to, such as culture, relations and city/land development outside population).

 

[da_Vinci]

This gets to the unanswerable question "What defines the ideal play of a particular map?" If one could answer that objectively, then we'd have an ideal criterion for medals.

However, ideal can be variously defined as biggest (base score), best (most advanced [space], most cultured [culture], most well liked [real diplo]), meanest (conquest ) or fastest biggest (domination?).

So what date and/or score of culture win "out-ideals" what date and/or score of domination win? I doubt any consensus could be had.

So given these limitations, maybe the system we have is already the best compromise?

dV