Side discussion for the New Scoring System

[Ronald]

Thanks Aeson,

Now the only diffculty I have is to determine the exact number of specialists.
You are mentioning a utility that gives you the exact amount of food. Is this utility available for the public?

Ronald

 

[Aeson]

http://www.civfanatics.net/uploads4/Jas2342on.zip

This is the most up-to-date version, but it hides all the terrain counts. The crossing level needs to be set manually, and the difficulty doesn't come out right either for PTW saves... I never took the time to figure out why. You can set it manually though.

Definitely this shouldn't be used on a GOTM before the map is known (it's not very useful anyways before then), and the numbers it gives out should be considered spoilers.

 

[Ronald]

Hi Aeson,

Thanks very much

 

[jeffelammar]

Rather than start a new thread I thought I'd just resurrect this one.

I was just looking at the scoring bonus curve and I was a bit surprised by the weight it gives to the year of victory.

In particular, I was shocked at how little difference the victory condition makes in the Jason score.

For example in GOTM 20
If you have in game score 5000 in 1790.
Culture20K : 4895
Conquest : 4758

I was a little shocked that with the same score you could hit the 20K date right on and only get 137 extra points over the conquest whose best date was so far behind.

I'm not saying it's wrong, mostly just wondering why it is this way. My gut feel is that the difference in score for hitting the best date vrs missing it by 100 turns should be greater.

Edit: I just realized that this may be a difference between the old and new curves. Is that so?

 

[samildanach]

In particular, I was shocked at how little difference the victory condition makes in the Jason score.

I agree jeffe, this has been bugging me as well and strikes me as being a little unfair. It does appear that even with the new curve that the chosen victory condition has little baring on the Jason score.
It seems that the curve is mainly based on the ability to reach the domination limit quickly while milking and that the victory conditon that you choose to pursue is pretty much irrelevant. Thats my perception of it-maybe I'm wrong.

 

[jeffelammar]

Originally posted by samildanach

Thats my perception of it-maybe I'm wrong.

That's exactly how I feel. I am perfectly willing to find out that I'm wrong, but I would like to understand it.

 

[Aeson]

If I set up a 3 way victory (say conquest, domination, cultural 20k) at 1700AD... should the score vary wildly based on which of the 3 I decide to go with? I played the same game either way, and IMO, it should score the same (or close to it) either way. I shouldn't be forced to finish one way due to the scoring system, and that's doubly important now that the Medal Play has been introduced, giving a specified target victory condition for each game.

Another reason is that the games are modded now. This results in the best date predictions not being as accurate as they should be. The max score predictions remain relatively stable. So using a formula which weights the 'score' portion higher is preferable for overal stability of the formula.

The third reason that comes to mind is that a conscious choice was made to target the scoring towards an all-around type gameplay instead of those in which extreme tradeoffs are made. A submission where all phases of the game are played well should score more than a submission where only one aspect of the game was focused on. The awards still recognize the extremely fast games, even if the tradeoffs made were extreme enough that the score wouldn't compete for a medal.

 

[samildanach]

If I set up a 3 way victory (say conquest, domination, cultural 20k) at 1700AD... should the score vary wildly based on which of the 3 I decide to go with? I played the same game either way, and IMO, it should score the same (or close to it) either way.

I think the problem with the example you have given Aeson is that in order to elect the 20 K victory condition in 1700AD you would have had to have made alot of early game sacrifices for that to be possible.
IMO a player who has the 20K option at 1700 AD has played a much better game than a player who just has the dom or conquest option. If you consider some of the sacrfices that would need to be made to make the 20 K option possible-
1. Joining workers to your 20K city very early in the game-crippling your civ productive capacity, leaving much of your territory underdeveloped apart from the 20 K city,less settlers being produced as pop points are being spent on workers. Probably no worker turns being spent on strategic road development to speed early conquests-thats if you manage to have built any sort of military.
2. In order to get GLs to give you a shot at getting more than one of the ancient wonders at the higher levels you will need to prosecute an early war on the back of an empire whos early development has been stunted for the above reasons- victory isnt nearly as certain for 20 k players than it is for straight dom/conq players.
3. Only having one city producing wonders- the 20 k city- sometimes means losing out on necessary wonders eg. you build the sistine chapel then you get beat to Leonardos as you didnt start the prebuid quickly enough-consequently leaving you without enough money to upgrade your probably quite beleaguered military.

My opinion is that the differences in scores at any given finish date dont reflect the opportuinity cost for players who go for one of the cultural victory types over the straight conq/dom.

 

[Aeson]

I chose 1700AD as it's the 'normal' date for a 20k victory. Certainly tradeoffs need to be made to reach it, but it really isn't that much to hit the best dates usually. DaveMcW's extremely early 1555AD 20k in GOTM19 is an example of where large population/territory to culture tradeoffs are made. Compare it to Moonsinger's GOTM18 1758AD 20k which won the gold medal, where she kept up with the Conquest/Domination players (or very close) while getting to her victory.

The awards are there for those who make the more extreme tradeoffs. The medals are there for those who play well rounded/dominating type games, where the AI is firmly in hand as quickly as possible, and any victory after that can be chosen for very little or no tradeoff at that point. Sometimes games can win at both sides of the competition if played very well.

 

[jeffelammar]

Maybe this is really a difference of the old vrs new curve.

I notice that using the new curve Moonsingers Jason Score would have been 10467 instead of 11879. The new curve appears to make a significant difference in scoring based on victory type.

Overall it looks like the new curve penalizes the slower games more than then old one did.

I guess what this reallly boils down to is your definition of what the scoring system should measure.

Using the examples you used, I would consider Dave's win far more impressive, but Moonsinger's raw in-game score is enough that despite the 43 turn difference. (Admittedly they are different games so the comparison is not completely accurate)

If I set up a 3 way victory (say conquest, domination, cultural 20k) at 1700AD... should the score vary wildly based on which of the 3 I decide to go with? I played the same game either way, and IMO, it should score the same (or close to it) either way. I shouldn't be forced to finish one way due to the scoring system, and that's doubly important now that the Medal Play has been introduced, giving a specified target victory condition for each game.

I would say that the answer is yes. The mere fact that you could have done all three doesn't change the fact that you could have done your conquest or domination 50 turns earlier. To me this is the whole reason for having different base dates.

I would further argue that Medal Play should have nothing to do with GOTM scoring. They are different events. There is overlap in the games, but they shouldn't effect each other.

All that said I'm going to put one more example out there.
In GOTM 20
Culture 20K in 1555 with firaxis 3000 : 5389
Culture 20K in 1700 with firaxis 6000 : 5799

Who Played the better game? It is open to opinion but I would vote for the 1555 Win.

All this said. It all boils down to scoring and a decision appears to have been made that the scoring should reward expansion to the Domintaion limit. Fine, I don't agree, so I'll just play my games and not worry too much about what the score says.

 

[Strong2theEnd]

Would it be possible to weight losses so that "staying alive" longer results in a better score than a quick loss? I've noticed that losing in 2050 AD is no different than losing in 1000 BC given the same in game score. Granted, you should have more time to improve your score by staying alive, but it seems that staying alive longer should give you a better score than losing quickly. (Note - I'm speaking from experience - 6 losses in the last 8 GOTMs!)

 

[hotrod0823]

Yes the key is pushing to the domination limit. Aeson and company showed that in SirPlebs GOTM19 game. You increase your land area and push up the points all the while going for your decided victory as fast as you can.

The player that choses not to push to domination will have a lower in game score and reguardless of the speed at which finished will get a lower Jason score. The milk is still around but just in a different color. Speed, coupled with raw miliatry power will get you the best Jason score.

 

[Aeson]

All that said I'm going to put one more example out there.
In GOTM 20
Culture 20K in 1555 with firaxis 3000 : 5389
Culture 20K in 1700 with firaxis 6000 : 5799

Don't get hung up on the Firaxis score. It's bonus system has been taken out of all calculations for a reason... it's terrible. Based on difficulty level, it can change the 'value' of your two example games quite a lot.

If we assume Emperor level difficulty... That means at 1555AD, the Firaxis date bonus would be 2475, and that would only leave 525 points from territory/population. The 1700 game would have had a bonus of 1750, and base score of 4250... 8x the territory/population score of your 1555 example.

You used GOTM20 for your example, which was Monarch. The Firaxis bonuses are a bit lower, so the territory/population differences aren't as much. 1020 to 4600.

One game is slower, but has several times the territory/population score, while keeping a very good speed... the other is faster, having focused almost totally on speed, sacrificing just about all territory/population to do so. They score relatively closely, which is how I think it should be.

 

[jeffelammar]

Originally posted by Aeson

One game is slower, but has several times the territory/population score, while keeping a very good speed... the other is faster, having focused almost totally on speed, sacrificing just about all territory/population to do so. They score relatively closely, which is how I think it should be.

Good points. I will think about this a bit before I bring up anything more here. Hopefully I can play with the numbers tonight and see how I feel about it.

 

[Aeson]

Would it be possible to weight losses so that "staying alive" longer results in a better score than a quick loss?

I've thought about different ways to do this, but it usually boils down to the possibility of someone milking a game and intentionally losing to get bonus 'survival' points.

The option that circumvents this would be to give survival points to all games, win or lose. Not sure how much survival should count towards score, or how much difference it really makes. If it's too much, then forced Milking again becomes a problem, if it's not enough, then there isn't any real change in how losses will rank anyways.

I just haven't figured out a viable option that yet. If someone comes up with one I would certainly add it to the scoring system.

 

[DaveMcW]

How about:

Losing games receive a survival bonus of Turns*Difficulty to base score.
Winning games receive a survival bonus of 540*Difficulty to base score.

 

[Aeson]

It's a good suggestion.

The max score would need to take it into account, and probably vary by victory condition. If a conquest on Monarch can get an extra 2160 points at a 300AD best date, that modifies the curve quite a bit in comparison to a 1100AD spaceship that would also get the 2160 points.

Also, survival would become much more important on small maps than large (usable tile count). If the max score is ~8k, the 2k survival points make up a much larger percentage than on a game with the max score ~20k. So a % modifier based on the max score/normal max score would be in order.

Then Difficulty would probably want to be different than the ones used in-game. The increase in survival difficulty becomes progressively harder at each difficulty jump.

So maybe something like:

(Turns * Difficulty) * ((540 - PlayerTurn) / (540 - BestTurn)) * MaxScore/10000
(540 * Difficulty) * ((540 - PlayerTurn) / (540 - BestTurn)) * MaxScore/10000

Difficulty:

Chieftain: .25
Warlord: .50
Regent: 1
Monarch: 2
Emperor: 3.5
Deity: 6

 

[AlanH]

Originally posted by Aeson way back in March
PlayerScore is the score that is given in-game. The formula for BaseScore isn't right, it should read: (and this is only done for wins, as no bonus is added for losses)

BaseScore = PlayerScore - ((2050 - PlayerDate) * Difficulty)

This just subtracts the in-game bonus from the in-game final score.

I've only just reached the dizzy heights of actually taking an interest in optimising my score, so I visited the page that describes the Jason scoring formula and fell over the above error. As this was apparent at least four months ago, perhaps it ought to be fixed by now?

 

[jeffelammar]

Originally posted by Aeson
I shouldn't be forced to finish one way due to the scoring system, and that's doubly important now that the Medal Play has been introduced, giving a specified target victory condition for each game.

Quick Question, I thought that the Tournament was scored by earliest victory and that Jason score was not a factor. Am I wrong in this?

 

[Aeson]

Tournament is by fastest finish.