Old GOTM Scoring Discussions

[ButSam]

Here's a thought that could very well put milkers away for good. Throw in your own bonus to the score added on after the GOTMs are submitted to take into account year you finish at as a larger factor. It could be based solely on years less than 2050, which gives a step-function (not quite exponential, but same basic idea) addition to the score. You could say you get 10 additional point above and beyond your original score for each year before 2050 you finish the game. So if you finish in 1450, you get 6000 points automatically added to your final score for GOTM ranking purposes, or something like that. I realize turns space closer and closer as time goes on, but that's just fine--it's a lot harder to win when the turns are 5 years apart than to win when they are 1 year apart, so that would balance out with this bonus. Of course, the bonus only applies if you win...if you lose in 50 AD you don't get 20000 bonus points--sorry

This would also be VERY easy to calculate with an Excel spreadsheet or a simple computer program or something clever, which has to be done anyway before posting results. Just be sure to treat BC dates as negative numbers and it will all work out just fine

I don't feel 10 points per year is too hefty a bonus either, but if the GOTM elite do, how about a 5-points-per-year bonus, or even 1-point-per-year bonus?

How about it? We want to have the FIRST finishers rewarded extra-much. Under this suggestion, those who finished in 2050 wouldn't be penalized any either...but the results should counteract milking and encourage finishing ASAP.

Sam

 

[Taé Shala]

This will only help fast conquerers!

People who like to win by spaceship :alien: or UN will be forced to become warmongers.

So this is a NO from me.

 

[ButSam]

Good point. The top score could then become a factor for each victory, and the peaceful victories could receive 10 points for finishing each year earlier, but the conquest and domination only 1 point for finishing each year earlier to help balance that out.

Sam

 

[Cartouche Bee]

As it has been pointed out many times there is no easily implemented solution to this problem. But no matter how you look at it, the early bonus that the game awards is inadequate, to offset milking. For all intensive purposes though, if that early bonus award were tripled, the results would be a better reflection of achievement for all, under most circumstances. [Except the OCC players of course ]

 

[Creepster]

I don't think there is a happy solution to the scoring problem. My suggestion would be to set up different categories and tables for the GOTM winners. There could be one for Conquest, one for fastest domination, one for OCC, one for highest score(milkers) and so on. This would help to offest the milking by giving people other targets to shoot for.

You probably would not see the same people at the top in the same categories every month. They would most likely be willing to try to be the best in another category for each month.

 

[Beard Rinker]

A scoring formula has been evolving for the tournament that would work well for the GOTM scores. It is based on game score and finish date and produces a normalized score similar to the normalized score produced by Matrix.

A good scoring formula should not dictate the best method of finishing the game. Each player should play the style they are best at to get their best score.

This formula comes close to meeting these goals as a well played game of any finish type will place well. Fast finishes seem to do the best, but milked games can also place well.

For example, SirPleb, Cartouche Bee and Lucky who are number 1, 2 and 3 in the global rankings repectively should place well regardless of the victory condition they choose. Using this formula, they placed as follows:
Player Game Victory Placing
SirPleb 5 Milk 7th
SirPleb 6 Milk 5th
SirPleb 7 Milk 1st
SirPleb 8 Space 6th
C.B. 5 Dom 2nd
C.B. 6 Milk 3rd
C.B. 7 Diplo 10th
C.B. 8 Milk 2nd
Lucky 5 Con 1st
Lucky 6 Milk 6th
Lucky 7 Space 7th
Lucky 8 Milk 7th

Note that when these players did not milk their games, they still placed well. For example, SirPleb's space victory in GOTM 8 gave him a 39th place finish using game score but a 6th place finish using this formula.

The attached spreadsheet contains GOTM 5 - 8 results with the normalized score computed.

Also note this is not the formula currently used in the tournament but a proposed revision to it. It does not contain any artificial weighting of finish date over score.

Edit: The formula has changed slightly and this spreadsheet has been removed. A spreadsheet using the changed formula and containing data from GOTM 1 - 9 is attached to this post.

 

[Cartouche Bee]

Sorry Beard, the formula used in the Tourney sucks!!!!

I stopped playing the tourney cause it was simply ridiculous and many good games submitted were waaaay under rated. They don't need a forumla for the tourney, the date is all that counts.

 

[Creepster]

My only problem with the tournament formulae score is the weighting of the finish time. The fastest finish is what determines the top rankers, and this in not what the GOTM should be about. I believe that currently the finish time is weighted by a factor of ten over the total end score.

The tournament formulae might work out if the weighting were reduced from what it is, but I also believe that Matrix is against changing the scoreing.

 

[Cartouche Bee]

Yes, they had to eventually had to go to the 10-1 ratio in the tourney because once the true results from actual play started coming in, the "score normalization" was skewing the results so much it took that much offset to create the results that they were looking for. If you build on a poor foundation, you end up with a poor result.

We know that fast finishers don't score enough against milkers, so give them more score bonus to narrow the gap. I know that no system will be perfect but working toward a slightly better end result should not be outlawed.

I also know everyone knows I'm just trying to turn the tide against milkers so I can play fast games again.

 

[Beard Rinker]

Cartouche Bee & Creepster

Yes you are right. The currently used tournament formula weights finish date at 10:1 over game score making your game score essentially meaningless. The formula I'm suggesting is a revision to that one where the only weighting of finish over score is a natural one. It uses years left instead of turns left and an equal weighting between score and finish date.

The result is game score is important and milking a game is still a viable strategy. Cartouche Bee, note you would receive a 3rd place finish in GOTM 6 and 2nd place finishes in GOTM 8 and these were both milking games. Milking a game however is usually not the best strategy but it shouldn't be ignored.

Check out the spread sheet a few posts up, you will see that good games placed much higher and milked games placed lower. Higher scores often overcame slightly faster finishes yet milking to 2050 only got the best score in GOTM 7, the deity game.

For example:

• Nathan Barclay placed 26th in GOTM 5. This was a very fast space ship victory that also had a good score for the finish date. He would place 7th using the proposed formula.
• Beard Rinker placed 4th in GOTM 5 with a milked game. Beard would have placed 12th using the proposed formula.
• [pnp]dredd placed 61st in GOTM 6 with the fastest finish cultural victory. [pnp]dredd would have placed 6th using the proposed formula.
• Beard Rinker placed 91st in GOTM 6 with the fastest finish space victory. Beard would have placed 19th with the proposed formula.

With this formula, it appears that finishing with a fast domination or conquest victory yeilds the best score but not by a huge margin. My guess is that to achieve your best score, play the style you are most comfortable with. If fast conquests or domination victories are not your style of play, then you probably wouldn't get your best score with that victory type.

FYI, an analysis and the details of the formula are contained in
this post.

 

[pagh80]

I actually think it is a bad idea to give the fast finisher an extra bonus. It will turn out to be allout conquer games from 90% of the people who plays GOTM.. no spaceship, no cultural and no diplomaticvictory. It is allmost the same to make a rule that makes the only victory condition to be conquer.
You all have to remember that milking is not a bug. Its a way of playing and it should therefore not be punished by milking a game.
I think that the countless hours(often a bit boring) it takes to milk a game should be rewarded with a better score than the people who finish as fast as possible(unless they are really fast)

Right now it is very different how people end the games and there are no winning conditions noone uses. It should stay that way

I think that the people who dont like to milk should stop and play the way they like best... there are many different awards to win beside best score

 

[Beard Rinker]

Originally posted by Cartouche Bee
Yes, they had to eventually had to go to the 10-1 ratio in the tourney because once the true results from actual play started coming in, the "score normalization" was skewing the results so much it took that much offset to create the results that they were looking for. If you build on a poor foundation, you end up with a poor result.

Another difference of this formula compared to the tournament formula is it uses average score and finish date instead of highest score and fastest finish. This prevents the problem of freakishly high scores (your GOTM 8) or fast finishes skewing everyone else's results.

When I build on a poor foundation, I just keep adding duct tape and toothpicks until it works.

 

[pagh80]

Originally posted by Cartouche Bee
[B
I also know everyone knows I'm just trying to turn the tide against milkers so I can play fast games again. [/B]

I know that you know that we knows that you know that we knows you want to play fast games again

 

[Beard Rinker]

Originally posted by pagh80
I actually think it is a bad idea to give the fast finisher an extra bonus. It will turn out to be allout conquer games from 90% of the people who plays GOTM.. no spaceship, no cultural and no diplomaticvictory.

This formula does not give an extra bonus for fastest finish. It gives you a ratio for score and a ratio for finish date based on how you did compared to everyone else. These ratios are combined with equal weighting to give your normalized score. The only extra bonus for fast finish already exists and that is each turn is worth more than 1 year before 1850.

The way it stands now, to get the best score you must milk your game to 2050. There is no alternative strategy for this, other victory types do not even come close. This formula will make it so conquest and domination are probably the best way to get the top score, but not by a wide margin. A well played game that is milked to 2050 will still place very high, perhaps even 1st (GOTM VII). With milking as an option, that opens the door for a fast spaceship, cultural or diplomatic victory as a possible winning strategy also.

 

[Cartouche Bee]

Beard, I have my share of posts in that thread, seemed like anarchy to me.

The problem with these types of schemes, even if one could be agreed on, is that you can't mark the game until you have the results from all the participants. Too boring for everyone, you need to be able to know what your score is when you finish the game, regardless of anybody else's results.

 

[Beard Rinker]

I've read a number of posts recently suggesting/requesting more diversity in GOTM games. It seems to me, the easiest way to add diversity is to use a scoring system where you can place well regardless of the victory condition you choose. Instead of there only being one way to score well, there would then be 6.

The way it stands now you either a) milk your game to oblivion or b) ignore your score and just play it because there is a large pool of people playing the same game. Fastest finish by victory condition doesn't add a great deal of diversity. If you don't get the fastest finish then its simply a 79th place finish or something like that.

Another advantage of the scoring formula I suggested is the scores are then comparable from month to month. If you get a score of 65 one month then 70 the next, it is probably because you played the game better. This is not the case using the game score. A game score from a large emperor game does not compare to a game score from a standard size monarch game. The scoring formula compensates for this and makes the scores comparable.

Originally posted by Cartouche Bee
The problem with these types of schemes, even if one could be agreed on, is that you can't mark the game until you have the results from all the participants. Too boring for everyone, you need to be able to know what your score is when you finish the game, regardless of anybody else's results.

There is little difference in this respect when you use the game score only. You may have your score when you finish your game, but the number is not very meaningfull until you can compare it to others.

 

[Cartouche Bee]

A comprehensive scoring system would have to take more into consideration that just speed and score. Things like Net National Income after corruption could be used to show how well the civilization's infrastructure had been placed on the map that was played. A scoring system with more depth would allow for better parity between play styles and the different victory types achievable.

I'm a firm believer in duct tape, BTW. This allows for interactive modeling. An educated guess can predict the result of a milked game on any given map. Time trials can predict the speed and score that the other types of victories can achieve on these maps. Connecting the dots, reveals alot about the types of modifiers needed for a successful scoring system.

 

[Beard Rinker]

I agree a scoring system should include more than finish date and game score. Culture, technology, production and economy come to mind right away. More essoteric things like turns at peace, reputation and perhaps even average number of tiles per city would be a good addition too.

One big problem with using anything other than game score and finish date is someone has to retrieve this data. Until Firaxis includes these components in the game score or someone develops a utility that extracts them from your final .sav file I don't think using them in a scoring formula is practical.

The formula I've suggested is not perfect, but is a great deal better than using just game score or Matrix's normalized score. Both are simply measures of how well a game was milked.

 

[Cartouche Bee]

Beard, the demographics screen gives quite alot of useful information that could be used for those calculations.

Then the advisors domestic, cultural and military give other info that could be rated further.

Add the infor that mapstat provides and you could do some fairly extensive analysis on the actual situation at end game.

Plug in all the factors and let the spreadsheet do the work.

 

[Beard Rinker]

Originally posted by Cartouche Bee
Plug in all the factors and let the spreadsheet do the work.

How? Will Matrix be willing to check all these screens and run MapStat for every game? I doubt that. Having each player do this is not feasible either. For GOTM 9, there where 13 invalid submissions and a submission now only consists of 4 pieces of information and 1 file. I suspect there are probably allot more errors that Matrix corrects or doesn’t catch.

What is needed before using other statistics for a scoring formula is a utility like MapStat that will process all players .sav files and put the required stats in a text or spreadsheet file.

Once that is achieved the next problem is deciding what statistics should be included in a scoring formula and the weighting of each statistic. Some may be too easily exploited and not usable. There would have to be input from allot of civ players to determine which stats make the most sense to use.

I think the best solution now is to use the formula I suggested, as it is far better than the existing scoring system. It would be months, perhaps never before a utility is available and a more comprehensive formula can be devised.