Side discussion for the New Scoring System

[Plutonium239]

I plan on playing April's GotM. I have only played a very few games since buying Civ3 early this month. I won't be considering score in my game play since I expect I will probably be in over my head anyway. *crosses fingers, hoping game won't be deity*

Still, I am curious about the scoring system.

BaseScore = (2050 - PlayerScore) * Difficulty

I didn't see anywhere that explained how PlayerScore is calculated. From an excerpt on the linked Scoring Formula Summary page "For this reason, the in-game bonus for all victories is disregarded entirely, and a new method of rating early victories is put in it's place." I can see that it is not the in-game score.

How is PlayerScore determined? Evidently it can't exceed 2050 (unless BaseScore is allowed to be negative) and the higher it is the lower your base score. I'm probably missing something obvious. Is this some kind of player rating, like a handicap?

 

[Aeson]

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.

 

[Plutonium239]

I see. That makes sense. Thanks for the lightning quick reply.

 

[Ribannah]

Originally posted by Aeson
Remember that as long as you are playing well, the score dip from missing the best dates is made up by having longer to increase your base score to make up the difference though.

Yes, but that wasn't the point. My actual finish may be earlier than the earliest actual 20K victory date, therefore getting a LOWER Firaxis score, but because the best date for 20K is set later, the difference made by the Jason score could ALSO be LOWER for me.

Am I correct that the Jason score for any actual game is always higher if, all other things being equal, the 'best date' is set to be later in time?

Of course there is no way whatsoever to know for sure whether the best victory dates should be different or the players who chose that condition simply played better, and chances are that I am on the other end of the stick compared to other victory types.

To avoid such thoughts altogether is why I have always advocated that one should simply give max speed points to the fastest victory per category (or more generally put: derive the 'best dates' from the actual results, rather than map properties).

To illustrate this... a very vague sneak peak at the results we've had submitted so far... in the top 6 results, all 6 victory conditions are represented.

That is a good thing. It also illustrates, however, that the concern you expressed against my idea, that there would be too few contestants to make it work, does not hold. We have a great many excellent players, who don't stick to one and the same favorite victory type.

By the end of the day, personally I don't really play for score no matter what the scoring formula, but having an adequate scoring system supports the idea of the GOTM. Your formula is certainly far better than simply taking Firaxis' in-game score, yet I feel that it can still be improved. But we should probably stick to it for at least several months to see how it goes.

 

[Aeson]

Originally posted by Ribannah
Am I correct that the Jason score for any actual game is always higher if, all other things being equal, the 'best date' is set to be later in time?

If you have multiple victory conditions available, the one with the latest best date will score most at that point. The exception being 2050AD games, which have no date bonus applied. (the difference becoming more and more minimal the longer the game progresses)

We will try to score games based on the best scoring victory condition available in each submission. If anyone plays a game to multiple victory conditions, they should include a note with their submission so we won't miss that.

 

[ButSam]

I don't fully understand the best dates for a victory condition. What happens if you are able to achieve, say, conquest victory 100 turns EARLIER than the best target date?

All-in-all, though, I like this new system you are describing! Keep up the good work!

Sam

 

[Aeson]

Well, 100 turns earlier and you're submission will probably be disallowed due to cheating.

Beating the best date doesn't change anything in the scoring. The best dates can be beaten in many cases, as can the max score... they are just predictions. The faster a person finishes, the less base score they are going to have, but the bigger the date bonus will be. It's pretty well balanced. I think the most any of the best dates have been beaten by so far is the example I gave in the top post of DaveMcW's GOTM15, about 20 turns. His game scored ~10700. 10000 +/- 5% is the range we are shooting for being the best score possible.

 

[Aeson]

Originally posted by Ribannah
The calculator is available now. Even with the Jason formula, your 1880AD 100K victory gets more points than my 1610AD one.

It would be reversed if the 100K cultural victory had a more realistic 'best date'. How can it be only 10 turns behind GOTM18 where the player's civ is religious AND plays pangaea instead of isolated archipel? And even 15 turns ahead of the 'best date' of GOTM15 where one is scientific?

Difficulty level plays it's part. On Regent, the player's tech rate is faster than it is on Monarch, meaning you can start building cultural improvements earlier. The AI's are easier targets, and so expansion to get enough cities for the cultural win is easier.

Scientific and Religious do help the cultural date, but not as much as you'd expect. Part of that is because other traits also help with cultural dates, just indirectly. A militaristic civ can conquer more territory faster, and has more chances at leaders to rush wonders with. Expansionist increases the odds for settlers from huts, and increases tech rate on it's own. Commercial means more commerce to increase tech rate and/or rush more improvements with.

---------

As for why Moonsinger's game scores more than your's, it's due to the base score. She has 3x the pop and territory score that you do, and your 93 turn advantage doesn't quite make up for that.

 

[Ribannah]

Originally posted by Aeson
Difficulty level plays it's part. On Regent, the player's tech rate is faster than it is on Monarch, meaning you can start building cultural improvements earlier.

The reverse is true. On higher levels, the AI researches faster (because it builds faster), and the human player will profit from that.

The AI's are easier targets, and so expansion to get enough cities for the cultural win is easier.

What targets? There were no targets.

Scientific and Religious do help the cultural date, but not as much as you'd expect .... (other traits)

What you are saying about other traits only strengthens what I said. In GOTM#17, one trait was Industrial, which is not that great a help if you need a dense build and to work many coast/sea tiles. The Commercial trait is nice, but will not make up for the extra 170 shields (plus the extra points from early happiness) to be bought per city or the extra 140 plus free tech (and faster research).

As for why Moonsinger's game scores more than your's, it's due to the base score. She has 3x the pop and territory score that you do, and your 93 turn advantage doesn't quite make up for that.

I'm sorry, but once more I cannot agree. Do the calculation again with even a slightly later 'best date' and you'll see for yourself.

To finish in 1610AD requires another approach. There was no time to wage war and conquer land. The few military units I built were disbanded to rush the culture.
That full focus is what one expects to see close to a 'best date'.

Of course all you need to do to prove me wrong is to present an entry where the 100K victory has been reached much sooner.

 

[Aeson]

Originally posted by Ribannah
The reverse is true. On higher levels, the AI researches faster (because it builds faster), and the human player will profit from that.

I specifically said Regent and Monarch. I can research faster on Regent than Monarch, and the AI's don't help much at all in either case. All they really provide is the free era techs. Any Scientific AI will get those techs when I gift them the prerequisites regardless of difficulty level.

Deity, and to some extent Emperor AI's can help overcome the slower player research rate.

I'm sorry, but once more I cannot agree. Do the calculation again with even a slightly later 'best date' and you'll see for yourself.

I know how the formula works. I understand that you like how your score rises if we set the best date to accomodate your game over another. I will not be setting the best dates or changing the modifiers midseason, to avoid conflicts of interests like this.

To finish in 1610AD requires another approach. There was no time to wage war and conquer land. The few military units I built were disbanded to rush the culture.
That full focus is what one expects to see close to a 'best date'.

What targets? There were no targets.

Military targets. Cultural 100k victories are very difficult to balance. You need enough territory, but without causing too much wait for building cultural improvements. The lower the difficulty, the easier territory aquisition gets, both peacefully and otherwise. Avoiding military expansion in a cultural game usually isn't a good idea. The exception would be if you are on a vast landmass with plenty of terrain for peaceful expansion.

If you want to focus on a specific pattern of building, that's your perogative. Just because you choose to play a certain way doesn't mean you have played better than other people who play another though. The formula is set, and you can use your knowlege of it to direct your playstyle towards whatever goal you want to achieve. If your goal is scoring, I'd invite you to analyze the formula and the game you are playing to determine the most efficient course of action to obtain that end.

 

[Ribannah]

Dear Aeson,

If my goal was scoring, and I would have known the 'best dates' in time instead of mid-game, I would probably not have chosen a cultural victory at all. But that is not my goal.

Nor do I 'like how my score rises if you set the best date to accomodate my game over another'.

Please, if you want to discuss this with me, refrain from such childish comments.

My only goal in this discussion is to help improve the scoring formula, which I feel is still unbalanced and favouring, although not as much as before, a milking-like approach. The optimal date may no longer be 2050AD, but delaying the victory just to get a higher score is still very profitable if the 'best dates' are too optimistic.

I have not asked you to adjust the dates for this month.

But you are free to delete my entry if you doubt my motives.

Ribannah,
Please make an effort to bend over backwords to be courteous to the GOTM Staff members who put in literally hundreds of hours each month so you can be free to play the game and just have fun. There is no call for you to label Aeson's reponse to you as "childish comments".

When in doubt, defer to Aeson's judgement. Even if you have no doubt, present your information to him in a way that is helpful, enlightened, and diplomatic and then lets him make the decisions on what is good for the big picture. This is what I do and also what I expect from you.

These last several posts do not serve to cement your position as the recognized leader that you appear to want to be. - cracker
No back talk required

 

[Moonsinger]

Ribannah,

Base on my own study and excellent advices from SirPleb and Cartouche Bee, I realize that the fastest conquest or whatever may not always result in good score. Therefore, in this GOTM, I was very relaxing and wasn't in any hurry to end my game.

 

[Ribannah]

Moonsinger,

I know, and that should be just as rewarding as any other approach if you play it well (which you did ). Even a fully milked game should still be competitive IMHO.

But so should a speed game. And whether it is or not, depends entirely on the 'best date' settings. In another month, and/or for another victory type, they may be too pessimistic, turning the tables once again.

It doesn't need to be that arbitrary. Why don't we define the 'best dates' by what we, the players, make of them? I feel that that's the only objective approach. Nobody can ever complain about the settings, and it also adds an extra element of suspense to the competition.

 

[Bamspeedy]

I will not be setting the best dates or changing the modifiers midseason....

I can understand this, of not changing modifiers for games already in progress. Is a game-in-progressed defined as being 'midseason'? What I mean is, if you find something that convinces you the modifiers should be changed will they be changed for the next game? Like, if you find something today, would it be changed for GOTM19 (GOTM18 'best dates' has already been posted), or much later?

I realize you found some bugs in the modifiers, but I just want a clarification for this GOTM. Earlier you posted the new 'best dates' based on the bug you found (which made domination/conquest much closer to being realistic). But this doesn't match up with what is on the calculator page. Is the calculator page what we should be going by? Your earlier posts says 1200 A.D. for conquest (800 A.D. before you found the bug), but the calculator page says 1150 A.D. Is it 1200, or 1150?

Why don't we define the 'best dates' by what we, the players, make of them?

Like has been stated before, sometimes there isn't many players going for the same victory type. It sounds like in this GOTM, there was enough highly skilled players going for a variety of victory conditions. But that may not always be the case. If only 2 players go for a diplomatic win and the better of the 2 players wins it in 2010 A.D., then 2010 A.D. is a 'best date'? That player would rank so much higher than players who played better, but went for other victory conditions.

I had stated earlier in some other thread, that there may be some cases where some victory conditions would seem unfair for some maps, but that eventually the formulas should balance out, when there is more results/data to look at and feedback from players.

The biggest challenge is adding points while you are striving towards your goal. In the Zulu GOTM, with my quick conquest I finished 3rd based on the Firaxis score. But by Aeson's formula (at that time), I finished 6th or so. My problem was I wasn't adding any more points, I was just looking for speed. I built 6-10 cities ICS style, then nothing but horsemen. I got a quick victory, but I just didn't have the base score (because I didn't build very many cities at all and was low in territory size). I learned you don't want to ignore score just trying to shave a few turns off the finishing date.

 

[Smirk]

Who is Jason?

 

[Aeson]

Ribannah,

I am sorry for the tone of my post, it was unnecessarily harsh.

----------------

Bamspeedy,

The dates on the calculator should be correct. I have been having a lot of trouble keeping numbers (or anything) straight lately. The calculator took about 10 tries before all the bugs, typos, and other mistakes were figured out...thanks mostly to SirPleb.

The scores in the calculator will be very close to the actual results. This is my first javascript project, so I'm not exactly sure how numbers will be rounded. The actual results will be calculated in Excel, and so may be off by a point or two. I just don't know.

As for changing the modifiers, I am looking into adding a set for 'super archipelago', such as this map was. If/When cracker decides to try another map of this sort, hopefully the best dates will be more precise. GOTM18 should play out a lot more like a normal map, even though it's edited as well.

 

[Ribannah]

Originally posted by Aeson
Ribannah,
I am sorry for the tone of my post, it was unnecessarily harsh.

OK, apologies accepted.
Let me add that I don't think the formula should be changed anytime soon, there is no hurry. Maybe once we reach 500 competitors my proposal of basing the 'best dates' on our own efforts will sound more attractive ....
Meanwhile, let's be aware that no scoring formula can ever be perfect.

 

[Aeson]

Originally posted by Smirk
Who is Jason?

Jason the Argonaut.

It's the GOTM Staff's expectation that once the results are posted, we hear a lot of 'golden fleece' puns from those the scoring system cheats out of their rightful gold medals...

FWIW, my vote for naming this formula was the terribly bland, but yet completely applicable "GOTM Scoring System".

 

[Ronald]

Originally posted by Ribannah


To finish in 1610AD requires another approach. There was no time to wage war and conquer land. The few military units I built were disbanded to rush the culture.
That full focus is what one expects to see close to a 'best date'.

Hi Ribannah,

As I understand, the GOTM with the new Jason formula takes the approach, that you need to have as much territory with as many happy people for high base score plus reaching your victory goal relatively close to the set date.

So I would expect a lot of 20k cultural victories to dominate the high scoring games (at least up to emperor):
You can set aside one central city to acquire culture, lots of wars give you a few great leaders to rush some wonders in your super city. You can get in a domination position almost as fast as if you were going for domination, but then you can milk the game untill the 20 k victory is achieved to increase your base score dramatically.

If you want to put more emphasize on the fastest possible finishing date, you should have a look at the tournament sites: There the date counts for almost all the points (90% iirc).

best regards,

Ronald

 

[Ribannah]

Thanks Ronald,

The Tournament has the drawback that I would always have to play emperor/deity.

Besides, I don't want to have speed count for 90% either, at least not for the GOTM. I like to play for speed myself, but I have no intention of forcing everyone else to do the same, and I enjoy reading about alternative roads to victory.

I fully support the idea behind the GOTM that we can all learn from each other's game, top players and new contestants alike.

That requires that all victory approaches should be able to score well, if played well, so one can recognize the pace setters.

The Jason formula (the name of the Argonaut Jason's father was Aeson, btw) is a big step in that direction already, but you should know that we have been discussing the formula for quite a while, so ideas for improving on it have come up.

P.S.: I am still too slow to simultaneously play in more than one competition!