A possible scoring anomaly?

[stormerne]

[Apologies in advance if this has been commented on and discussed before]

When I was finishing GOTM #4, I noticed a considerable anomaly in the scoring rules for GOTM. The Rules and Download page says:

You'll need to finish the game in a valid way: by conquest, spaceship (yours or AI's), or retirement in 2020 AD. You may not retire on any other occasion.

My question is: do you think the same scoring rules should apply to both a "Your Spaceship" finish and an "AI Spaceship finish"? I don't, and I'll explain why using an example of my last GOTM...

My last GOTM wasn't very high scoring. I discovered later that I could have scored perhaps 80% higher. (If I'd discovered that for a high scoring game I'd have been upset about it!) The end of the game went like this...

The space race began and the Sioux were the first to start building. Everyone had nukes so I decided to keep the peace. As soon as I had all the tech to build a spaceship I began. However, the "choice of next tech" sequencing wouldn't give me Fusion Power as a research option, so I kept researching Espionage to give me something to trade for Fusion Power before launch. As soon as I had Espionage, I lowered the science rate to 0%, and put the tax rate at 70% with 30% luxuries.

You know how it's like when you're in the space race, one civ has got a head start and you're playing the AI with its production advantage at higher levels - you do <u>everything</u> to get that spaceship built. This time everything did get done. Most of you will have done this to rush build a spaceship. First you sell off the universities for extra cash, then the libraries, you disband stuff in cities to help with production. To win this race I had to do all of that, plus reduce the military to one unit only per city, sell off granaries, aqueducts and sewage systems AND all city walls in non-critical strategic points. The Sioux launched first by a few years but I got my trade for Fusion Power with the Egyptians and my own spaceship landed 2 years ahead of teh Sioux. I won... Yay!

But why did I do all that? Because it was the <u>honourable</u> thing to do of course! Because I couldn't feel as though I could hand in a game before 2020 which I hadn't won or been destroyed in. But I realised afterwards that if I'd switched to high luxuries instead, celebrated my cities to twice their sizes while the Sioux were busy messing around with spaceships - if I'd have ignored the space race myself altogether, I would have finished with a MUCH higher score and only a 2% penalty for finishing two years later.

Is this really what we want? I think not. My solution to this problem would be to say that a finish by the AI spaceship counts as a finish in 2020 so that the bonus factor is at its minumum. Otherwise, if I get in a similar position again, I shall do the dishonourable thing, let the AI do the work, keep the peace and grow while they land their spaceship.

How about making this part of the rules?

------------------
<IMG SRC="http://www.anglo-saxon.demon.co.uk/stormerne/stormerne.gif" border=0>

 

[Smash]

My thoughts are this:

AI spaceship beats you-you lose
AI conquers you-you lose
time "runs out"(2020)-you lose

Losing games can't win..can they?

 

[stormerne]

Personally, your preferences are also my preferences, but that's not what's in the current version of the rules.

And I think there were some "Retired in 2020" scores in GOTM 3 higher than some other conquests

. And one GOTM 2 entrant with an AI Spaceship finish.

 

[goodbye_mr_bond]

I've wondered the same thing myself. Maybe it would be feasible to have a penalty for AI-won games. Like maybe your civ score is halved before the GOTM score is calculated?

I don't mind 'lost' games showing up in the rankings, but they shouldn't score higher than won games--well, not under normal circumstances anyway.

 

[Matrix]

I agree, Stormerne! The scores of players who ended because the AI built their spaceship should be calculated as if ended in 2020.

But I suppose we should change this from the GOTM V, because people didn't know about it. I mean, <u>you</u> could know, so it's actually your fault you built your spaceship.

------------------
<IMG SRC="http://home.hetnet.nl/~maartencl/tmp/MatrixBW.gif" border=0>
<FONT size="1">Studying chemistry means: having fun, drinking beer, having more fun, drinking more beer, hang above the toilet and have a good night sleep!

And each time Pedro says: "Aaaaaaaaaaaaaaarrggh", I feel so good!!!</FONT s>

[This message has been edited by Matrix (edited May 22, 2001).]

 

[oryx]

All losing games should count as 2020 for finishing data purposes whether retired, AI spaceship, forced retirement. You should only get a reward for finishing early when you actually win by conquest or spaceship.

------------------
"There is no tiddle-taddle nor pibble-pabble in Pompey's camp."

 

[starlifter]

For what it's worth, here is my opinion:

If you lose, then you lose.


The operative word is lose. How can you expect to win if you lose? If you want some consolation points, then don't lose.

The mathematical formula for adjusting the score if you lose is this:

Final score = Game Score X 0.0

(This can be computed individually for each case)

 

[Matrix]

Originally posted by starlifter:
For what it's worth, here is my opinion:

If you lose, then you lose.


The operative word is lose. How can you expect to win if you lose? If you want some consolation points, then don't lose.

The mathematical formula for adjusting the score if you lose is this:

Final score = Game Score X 0.0

(This can be computed individually for each case)

Sorry, but I have to disagree with this one. You could have played quite well for your level, but still the AI launched their space ship. E.g.: the surviving on itself could be a victory for you. If the score becomes zero because the AI space ship arrives before yours, that would be a too big punishment. It's a punishment enough by using 2020 as finishing date, I think.

------------------
<IMG SRC="http://home.hetnet.nl/~maartencl/tmp/MatrixBW.gif" border=0>

[This message has been edited by Matrix (edited June 18, 2001).]

 

[starlifter]

Hypothetically speaking, you are saying that a person who is clearly beaten could still outscore everyone else.

But I do see your point that a loss at the last moment would be crushing to the loser. Softening a GOTM loss does remove some of the thrill of a close space race...

My input is still that loses are loses, FWIW... but whatever method you use is fine with me, since I just like to play the game!

 

[noughmaster]

I totally agree with Stormerne. There are many ways to get a very high point AND gotm score which conflict with most people's understanding of a "win".
With the N^pnp term in the GOTM score, it's better to build to 2019 if you finish late in the game.
Using N^(sqrt pnp) gives a sharper decrease in GOTM score for late finishes, while still maintaining the same ratio as N^pnp for early vs late finishes, thus partly solving the problem brought up by Stormerne.
I've looked at this for N in the range 20 -> 100 (the current GOTM score uses N = 50, which seems reasonable). I'm happy to post comparative data if anyone is interested.

 

[noughmaster]

To Starlifter:
I don't believe an AI spaceship finish should warrant a score of 0. There is more than one object to a game of Civ, so I don't believe the player has necessarily "lost" if they have achieved a lot of other things. I sometimes play Test of Time (though mostly the original civ2 with an altered rules.txt with new units and technologies such as magnetic pulse, energy shields etc.). An AI spaceship is certainly not a loss in this case.
I know TOT is irrelevant to our debate and I agree with the general point. People should have some motivation to build spaceships, so there should be some penalty for an AI finish. Perhaps halving the GOTM score?

 

[starlifter]

To Starlifter:
I don't believe an AI spaceship finish should warrant a score of 0.

As long as we're talking GOTM scores, it is illogical to allow any GOTM loser to place higher that a person who ends the GOTM game in victory. The best (but not only) way to do this is assign a multilpier (0.0) to any GOTM score, whose player did not achieve victory. The player may admire and extol the virtues of his/her Civilization score in the defeat, but for the GOTM score, it is logical to use a 0.0 normalizing factor in order to ensure those that did win their GOTM are not ranked behind even a high-scoring loser.

Without the use of the only logical multiplier (0.0), I'd be willing to bet that lots of Losers will finish ahead of winners that, for instance, succeed in defeating all civs with a mere single city.

There is more than one object to a game of Civ, so I don't believe the player has necessarily "lost" if they have achieved a lot of other things.

One of the legacies in this post-Clinton world is how accepted it has become to view a cut and dry issue as somehow murky. And that's OK.

But... it is interesting to note your verbal maneuveing and pre-conditions used to set up your case that, in essense, a loss can really be a win. And this is in a game whose requirements for victory are known at the outset by all who play! And a game whose rules are implemented/enforced objectively by a machine!

Your use of quotes indicates you acknowledge awarness of the loss conditions in Civilization II, but wish to become a revisionist and somehow define it as a win.

People should have some motivation to build spaceships, so there should be some penalty for an AI finish. Perhaps halving the GOTM score?

So the only real difference is in the magnitude of the adjustment. You're proposing a multiplier of 0.5, and the one I suggested is 0.0. Perhaps there can be two different sections, one entitled "GOTM Winners" that contains all scores which fulfilled the Civilization victory conditions (namely, total conquest or landing the first SS).

The other section can be called "GOTM Losers", and would contain everyone's scores that were not achieved by fulfilling the Civilization II victory conditions. This would render the multiplier question moot. It would also prevent losers from achieving numerically superior scores to winners.

 

[noughmaster]

To Starlifter:
I like the Clinton reference. His presidency has had a little less effect on Australia.

I suspect you have an image of me as a bit of a left wing "no one really fails" social worker type. I'm usually pretty rigorous intellectually

and more right than left socially/politically.
You're right in saying we only disagree on the "loss multiplier". I know the game states that you win by conquest or spaceship and obviously achieving this is a win by any sensible account. If this is not achieved, I don't believe the game has no value at all in terms of ranking against someone who has won. There are degrees of losing and a multiplier of 0 can't take this into account. The GOTM score has to be careful not to let people get away with ignoring spaceship building.
I don't know how interested you are in mathematics, but the development of a formula representing "fair value" for the various achievements in a Civ game is an interesting exercise (see my sqrt suggestion). It might be worthwhile developing a separate thread for the scoring formula. Anyone else interested in this?

 

[starlifter]

I don't know how interested you are in mathematics, but the development of a formula representing "fair value" for the various achievements in a Civ game is an interesting exercise (see my sqrt suggestion). It might be worthwhile developing a separate thread for the scoring formula. Anyone else interested in this?

I've made a few long posts about some computational aspects of the scoring method and global rating before, but I think the topic was not fully understood by all.

I think Matrix has pretty much finished considering other solutions, so the system they have now is probably about as evolved as it's likely to become.

Mathematically, what is needed is a step-function. Ideally, such a function would be continious, yet ensure that the lowest scoring successful finishers (those that won) would finish just ahead of the top scoring losers.

Such functions are relatively trivial to develop, but can be difficult for some folks to really understand. There are also multivariate, non-step treatments, but there would probably be resistance to that, too.

However, if you want to start a thread, I'll be willing to contribute to it.

Anyway, from an academic point of view, I think a comprehensive point system is possible that will not treat those who pursue, say, an OCC strategy as a "Loser" while rewarding a scenario such as stormerne's as a "Winner", relative to the other participants.

 

[noughmaster]

I don't know how interested you are in mathematics, but the development of a formula representing "fair value" for the various achievements in a Civ game is an interesting exercise (see my sqrt suggestion). It might be worthwhile developing a separate thread for the scoring formula. Anyone else interested in this?

If you want to start a thread, I'll be willing to contribute to it.

OK, I've started one at: "GOTM Scoring Formula".

 

[starlifter]

noughmaster,

I'm in your other thread now... I just noticed it today, and I'll continue this stuff there. That massive post I made to it should pretty much get you up to speed with part of the issue (the GR). Matrix might add some stuff, too...