Staffa Staffa

Comments on the Metaverse scoring system

▼

Comments on the Metaverse scoring system

from Stardock Forums
This may be a long post, I tend to be long winded but
I will try this time to be very concise. :)

I recently finished a game that I think demonstrates
some rough edges in the logic for the scoring system.

My score was 23232 using version 1.035. Which is about
10k more then the next highest score im aware of in any
version.

Before I get into how I got this score, I will first
list some balance concerns I have.

Problem 1. It is far to easy to win by influence,the
computer will just roll over and be influenced to non
existence without even once putting up a fight.

Suggested Fixes.
1. Put a cap on the number of systems an empire can
lose through influence before war is automatically
declared(make it depedent on map size). This will
keep influence useful without making it the obscene
steam roller it currently is.

2. Cut in half or even 1/3 the effect influence
resources have. Currently they are way to good for
both the computer and the human. I will come back to
this.

Problem 2. Buying trade goods/wonders/achievements.
It is fairly easy dash to a trade good early in the game
buy it on the long term payment plan, and then trade
it between 4 races for every non military tech I am
missing(which is about 20+ techs at this point, as I
do a super land grab at start and fall way behind in
tech). Then taking these new techs, buy the new
trade goods/achievements/wonders and basicily own them
all and trade them all for obscene rewards. This is
so easy and cheap that it seems highly abusive. I can
even do it so I dont go below -500 funds, by building a
preemptive stash of cash and trading much of this stuff
for cash.

Fix. Simple, prevent buying wonders, trade goods
and achievements. Make sure to slow the computer down
in its own creation of such things to keep it in
relative balance.

Problem 3. Tech victory is worth the most and is also
the easiest way to win. There is some interesting
dynamics that make a tech victory challanging IF the
computer could win via tech. But as it is now, tech
victory should be worth the least number of points, why
it is worth the most, I cant fathom.

Fix. Military worth most, Cultural second, alliance
3rd, tech last.

The first 2 require you to dominate the galaxy, and are
by far the hardest.

Problem 4. Pop calcuation. This is how I got my score
My pop at the end of the game was over 23 trillion and
was still growing at nearly 100 billion a month. I just
got really bored with hitting turn over and over again.
I dont know where the pop would have leveled off at, but
a rough guess would have been about 50 trillion. What
ever time penality there is, it doesnt even come close
to making it worth not running ur pop up(if ur going for
score, for fun, this is not fun).

This was my 2nd masochistic game where I succesfully
took over the galaxy(except 1 or 2 planets left to
prevent the half price military victory) The first
one took me longer(I wasnt as good!) and my pop when I
finally got to the final frontier was 6 trillion.

My 2nd game, I did it quite abit faster and my pop had
just edged past 5 trillion. I went to submit and low
and behold, identical game, but this time I was getting
7k score instead of the nearly 9k i had last time.

I did better(by taking out the galaxy faster and
getting to the tech quicker) but got a much worse score

So I ran up the pop, it quickly became apparent to me
that pop easily dominates the time factor(at least on
a gigantic map).

Crude Fix. Put a turn limit based on difficulty/map
size in which a game can be submitted. This will
prevent running up the pop which is both abusive and
not even fun. It literally takes all the fun out of
competing for good scores.

Better(harder) fix. Try to rebalance the time penality
so is more effective in curbing obscene pops on larger
maps. This can be hard to balance without actually
knowing whats people can achieve(and the reason its
so out of wack right now).

Problem 5. Metaverse is penalizing you for submitting
low point value games. As has been pointed out before,
this is illogical.

If you are scoring every game that everyone ever plays
then you can penalize for bad results because you
penalize everyone. This is how most multiplayer scoring
systems work(the MSN zone, BattleNet, etc) The
metaverse is not of the same type, its a comparision
scoring system for a single player game in which
its optional for a player to submit a game. When its
optional, dont punish people for submitting all their
games.

I am currently ranked 6th with 10 games. If I had
submitted by game on a new alias, I would have been
ranked first with 1 game.

Fix. Several have been suggested, you could simply sort
the scores from highest to lowest and stop
calculating the score when the next game would lower
the overall score.

Ok, back to why innfluence resources are to powerful.

With 3 influence resources built up, I can simply drop
1 billion civillians into a computers home system on
some 12-14 pop world and it will on its own, without
any starbase support, begin converting their HOME
systems.

In my last game I had about 10 influence resources by
the end of the game, it was a joke.

This also works in reverse, if the computer gets all
the influences resources, no amount of early starbases
will stop him from creaping into my sectors either.
By the time you get cultural palace, this can be
stopped, but thats a rather late game tech.

If your curious about my strategy, I will make another
post after I hit submit here :)

I do hope the developers will address the pop explosion
problem before everyone has 20-40k games where they
spend literally more time hitting turn waiting for their
pop to grow then they did actually beating the computer.

That isnt fun for anyone.





1,532 views 66 replies
Reply #51 from Stardock Forums Top
"This would truly reward the person that beats the game in the fewest turns, which is as it should be IMHO. If you take that and then for the metaverse, account for the 10 best games for each player, you accomplish two things. The casual player will stand a chance and two the top rating goes to the person who can beat the computer the quickest."

Agreed.

With your lots of evil games, how about joining G.R.O.S.S.?!!

G.R.O.S.S.: Together we will ruin the Galaxy. :D
Reply #52 from Stardock Forums Top
"Now another solution to the 'problem' would be to vastly devalue their worth as a tradegood."

There is a third solution: Teach the Ai to buy tradegoods!
Galactic Civilizations IV
Reply #53 from Stardock Forums Top
"Any respectful empire would not tolerate obviously "unethical" points."

First, we are not a respectful Empire. :D

Oh, RESPECTABLE you were meaning to say? That´s better, but we would still rather be formidable than merely respectable. :)

And what´s the nonsense about unethical points? Unless you use a cheat, all points are ethical. Mistakes in the scoring system are not the player´s fault, and milking is perfectly acceptable. At Civfanatics, they take great pride in their milking skills, and why not? If I had the patience, I´d do it myself. (But I´m all for a more rational and transparent scoring system, of course.)
Reply #54 from Stardock Forums Top
"Secondly, the difference between a scoring system that can be milked for an effort-free 10000 extra points and a scoring system that can be milked for a +10% boost if you're really trying is significant enough for serious discussion."

Couldn´t agree more. Great post, Cataleptic! With players like you, no wonder the Aussies are a teeny bit difficult a challenge, even for the mighty Empire of G.R.O.S.S. LOL
Reply #55 from Stardock Forums Top
After thinking about it, and reading these messages, I realize that I don't have all the answers, but I'd still like to add a comment or two.

I don't have much of an issue with the PQ instead of population bonus. It hurts the builders, but that's going to happen in just about any case, since our game style most closely matches that of the abusers. Likewise, properly adjusting the diminishing returns of longer games would help as well.

The current method, which as near as I can tell is "if it's a cultural victory, divide final score by 20" doesn't cut it. I think that with at most one exception, a victory should be a victory, the path shouldn't matter, only the end results. What did you have when the game was over? You'd earn points for the kind of activities that each victory condition would require.

Score points for the culture you generated.
Score points for the production cost of ships you destroyed (minus the production cost of ships you lost?)
etc.

The one exception is for the alliance victory. I'd be much more impressed by someone that actually managed to forge an alliance with all 5 AIs than someone that managed to forge an alliance with the one domininant AI, and then defended himself while his ally killed off the universe. Not that I consider that a cheesy win, I've done it, I just don't think you should get that many points for it. Hmmm.... On the other hand, if someone pulls off a cultural victory and manages the same level of alliances, why shouldn't he get the points for what alliances he's got?

Also, I can't think of any method of earning points that shouldn't be lessened by the time it took to earn those points.

On a side note, I do think that the AI needs to be able to detect culture bombs in sectors that it doesn't own. In a game last week, i walked in with 70 constructors and built 4 starbases that were maxed out culturally and defensively. The next turn, I owned the sector culturally and every planet in the sector was flying the skull and crossbones. The AI never got a chance to complain about my buildup in his territory, because it wasn't his territory any more by the time it was his turn. I didn't have a planet within two sectors at the time, so this was definitely an offensive action, and should have raised a red flag or two.

On the other hand, the AI also needs to be able to tell the difference between a culture bomb and normal cultural buildup/defenses. Maybe if I own a planet in the sector, and only build up to the point that I will easily maintain control of that sector, but won't be grabbing control of any sectors that have the AIs planets in them....
Reply #56 from Stardock Forums Top
Well if everyone agrees that current metaverse isn't that bad, let's start 1st Masochistic challenge. I started a tread on this forum http://www.galciv.com/browser/msgboard.asp?id=61242&BoardID=GF
Any suggestions about the starting conditions?

~SDC~
Reply #57 from Stardock Forums Top
Legend:
// alexti (me)
> Ellestar
= Staffa

//The main reason for my position is that any other scoring usually makes one path better than another thus making everybody who wants to score high follow the certain path ignoring much of the depth of the game. To eliminate this problem and allow the player to chooce any path he wishes it is necessary to remove dependency of score from the any parameters at the end of the game.//
>This is wrong because then will be most efficient to
>play games that will took least real time.
In my book it's a good thing. If you want to explore different options and try something new, you won't be penalized scoring-wise. On another hand you don't have to waste time after you've realized you're going to win the game, because you won't be penalized for completing it in a shortest real time possible.


//The second reason is that it also eliminates Ctrl-N exploit. Or rather converts it into difficulty selector.//
>Better respawn - less micro you need to win -> more >wins/less losses -> better score
I'm not exactly sure what you mean, if you're talking about calculating win/loss ratio, how would you know about one's losses?

//How to estimate the starting position.

> And resources too.
Right

= I still think that a victory that occurs
= after total galatic domination is > then a victory
= that occurs when you scrape by on a tech or alliance
= victory with owning very little of the galaxy and your
= scoring method would rate these as equal.
I think that victory is victory. If certain kind of victory is considered undesirable it should be removed from the winning conditions (or maybe rated separately)

= I also think time should be a factor.
I thought about this too, because it is "anti-milking" and helps to save time, but the problem is that it moves accents from playing to reliably win to gambling on quick win. That's something that was common in Civ3. Early gamble on conquest could bring a huge score or a total failure. Basically, the riskier your strategy, the better score you get if it pays off. Besides, some victory methods require more time. On tiny and small, tech victory will always be slower and I feel it shouldn't be ruled out.
Reply #58 from Stardock Forums Top
//The second reason is that it also eliminates Ctrl-N exploit. Or rather converts it into difficulty selector.//
>Better respawn - less micro you need to win -> more >wins/less losses -> better score
-I'm not exactly sure what you mean, if you're talking about calculating win/loss ratio, how would you know about one's losses?

I mean that with worse respawn you need to do more to win, so you'll spend more your time for this victory. Also, any loss = one win less. So, less games played in the same amount of real time (say, a week) and lesser overall score.

~SDC~
Reply #59 from Stardock Forums Top
As further examples of what's wrong with the current scoring formula, consider the game I just finished a bit ago. For all more or less the same scoring factors (pop, tech, etc.) I got 548 points for an alliance victory, 788 points for a military victory, and a whole 38 points for a cultural victory. The only difference is that the military victory netted me a whole *2* additional planets versus the either two, no other factor changed significantly between winning that game any of the 3 ways and yet the scores vary that much because of all their attempts at balancing "unbalanced" victory strategies.
Reply #60 from Stardock Forums Top
Ellestar:
"I mean that with worse respawn you need to do more to win, so you'll spend more your time for this victory. Also, any loss = one win less. So, less games played in the same amount of real time (say, a week) and lesser overall score."

I didn't mention it but I was assuming that the total score is not the total of all games played, but rather total of a few (5 or something like that) best games. Idea is to make pointless playing game after game using the same strategy on the same difficulty. Then to improve your score after few initial games one will need to play more difficult game. As one is getting better and better, it will get harder and harder for him to improve the score. Can you win from one lone planet in the remote corner? - Yes, your score is going up. But to increase score again, after you've proved that you can win reasonably reliably (5 times) from such a start, you would need to win from a lone 14 planet in the remote corner ;)
Reply #61 from Stardock Forums Top
To alexti

Interesting idea, but it will be hard to balance standart difficulty vs. starting location modifier to the metaverse score.

~SDC~
Reply #62 from Stardock Forums Top
CodeMonkey, for the millionth time, no one is saying
in this thread that the fact that the different
victory conditions are worth such bizare amounts
makes sense. Most people, like you, favor the victory
condition to be a non factor in determing score.

Your making points 'against' no one.
Reply #63 from Stardock Forums Top
Ellestar: " To alexti
Interesting idea, but it will be hard to balance standart difficulty vs. starting location modifier to the metaverse score. "

I don't think it's very difficult. First step is to introduce separate difficulty ratings for AI level and for starting location. Let's normalize them to [0;100] range, where 100 is most difficult. Then let's define score as aX+bY where X is AI difficulty and Y is starting location difficulty.
After releasing the game with a=b=1 some submissions will accumulate. At that point select 20 submissions with highest X and 20 with highest Y. Each group will have average X and Y ratings (EX1 and EY1 for the first group and EX2 and EY2 for the second). Then select a and b such that aEX1+bEY1=aEX2+bEY2. Obviuosly, only the b/a ratio actually matters. This will produce pretty good balance, but procedure can be repeated after more games with a new coefficients are submitted.
Reply #64 from Stardock Forums Top
"Your making points 'against' no one."

Last time I checked the subject of this thread was 'Comments on the Metaverse scoring system', not 'Feed Staffa's personal beefs with the Metaverse scoring system'. I care that the devs see my posts; I could care less about you seeing, reading, caring, agreeing, or disagreeing with any of them as I think it's well established that while we have some common ground, the majority of our views for this game are completely opposite.
Reply #65 from Stardock Forums Top
Other then you blaming someone like me for causing
the devs to edit the different victory conditions in
a way that really annoyed you, I am not sure what
ground we dont have in common. I know you disagree
with me on the princaple that you hate me because of
the previous point, but you have never really pointed
out a reason that was relevent to the point.

Making tradegoods unbuyable, bad idea because it would
make the game harder for the players. Well duh, it
makes the game way to easy for the players right now.

Other then that, I dont even think you ever voiced
a disagreement with me, other then your general
disagreement with me as a person.