HOF Challenge Series II Discussion

[Dynamic]

Game 2 has less dependence from Level, so I try less Base Delta (may be it should be even less):

 

[Dynamic]

For Game 3 after analizing the HOF table better fit Base Delta = 10 (Epic Delta = 15):

 

[Dynamic]

There are too few non-Pangea Domination games but it looks like for Game 4 Base Delta is 14 (Epic Delta = 21):

 

[Dynamic]

And too hard to calculate Delta for Game 1 because of NCR and NV make Conquest on high levels too cruel (fisrt conquest half of the world then accumulate huge force and attack all other AI simultaneously in hope for conquest before culture borders expand). May be Base Delta could be like 20 (30 for Epic Speed).

 

[Denniz]

I understand that the above data is very preliminary, but I feel compelled to state the obvious as follows with regard to the Religious Victory column and the last sentence above ...

The Religious Victory column in the above table seems to be saying that Settler level Wins are later than Deity level Wins, so that Deity Games should suffer a turn penalty (rather than a bonus) with respect to Settler Games.

I don't agreed. It is more likely that the best Settler level Religious Victories are lame (not any where near as early as a motivated, experienced Player could make them). Your early Settler level Religious Victory Win Dates are extremely suspect. What possible advantage can a Deity level Religious Game have over a Settler level Religious Game? The Technology race is to Complete Writing and Monotheism followed by Mathematics or Alphabet and all Technology trading is done solely to improve Diplomacy as opposed to increasing Tech rate via trading. The best Deity level Religious Game is over before the AI's Economy recovers from its initial REX and the AI Technology level is often far behind. In many Religious Games, the Player need only give the AI's Theology and a few Technologies they are missing and possibly Alphabet too to gain the needed Diplomatic trading bonus (+4).

Conclusion: Settler level best Religious Victory Dates are much later than they should be. This results in a bogus Difficulty level turn adjustment curve for Religious Victories that gives a turn bonus to all difficulty levels that are above the Deity difficulty level. Thus, not only must the Deity level Religious Victory Player beat every lower level's Win Turn, he must do so by a margin that exceeds the turn difference in the above table. Although this certainly is a mathematical possibility, at any practical level this is pure nonsense (extremely low signal to noise; extremely high noise to signal; actually signal = 0 with respect to the Settler Win Dates).

I hope we can soon get some solid data to fix this anomaly.

Sun Tzu Wu

That is the way the curve was trending. It seems possible that with Deity the AI spreads region quicker. Don't forget Deity techs faster too. Do I expect it to hold up? Maybe not...

 

[Denniz]

For example, using Linear formula for Game 6:

I appreciate you putting so much time into this.

How are you calculating the base delta? I don't understand the speed factor since all the games are the same speed. Game 6 is an example of where linear doesn't necessarily work. Your 960AD Settler game is ~50 turns faster than your Prince game yet it blows past it with ~80 turns of adjustment. The Immortal game picks up ~140 turns. The adjustments seem to be too high. There is only 129 turns between the best settler and the best deity HOF games.

What I have been trying to do is use the relative difference between best games to put a cap on the adjustments. The best games for the difficulty levels between should be on line/curve between those two points. I don't believe that the differences between difficulty level doesn't ramp up linearly.

 

[Denniz]

Yup... if scoring in the series is important to you, and you can't win the game at Deity, then Settler is the level to go for. And here I've been crashing to desktop trying to get an emperor-immortal penalty in game 5. What an idiot I must be. Either that or I must not care too much about having my immortal level games score lower than a string of settler games I'd have no trouble crushing.


Perhaps I should go do it at settler just to see what makes that level so challenging. Just for giggles, anyhow.

I am not sure that the above is a fair statement. It is not my intent to penalize the levels between Settler and Deity. The adjustment is intented to level the difficulties. You should still have to play a really good game, regardless of level, to get to the top.

 

[Dynamic]

I appreciate you putting so much time into this.

How are you calculating the base delta? I don't understand the speed factor since all the games are the same speed. Game 6 is an example of where linear doesn't necessarily work. Your 960AD Settler game is ~50 turns faster than your Prince game yet it blows past it with ~80 turns of adjustment. The Immortal game picks up ~140 turns. The adjustments seem to be too high. There is only 129 turns between the best settler and the best deity HOF games.

What I have been trying to do is use the relative difference between best games to put a cap on the adjustments. The best games for the difficulty levels between should be on line/curve between those two points. I don't believe that the differences between difficulty level doesn't ramp up linearly.

1) You pay too much attention to Speed in "My" Formula. Every Game uses only one speed, I just don't remove coefficients for other speeds from the shots.
2) There is 67 turns between my Prince and Settler games (not 50), but really I played good Prince game but average Settler (I know how improve it), so I set handicap for Settler.
3) Most of Deity HOF games are OCC+PA so we can't use it. There is a 1760AD Ironhead game (Classical) - it's not exactly that we need but I think it's right date.
So I approximated the line to this 3 games (Settler, Prince and Deity) and additional check Warlord and Emperor levels - it's not far away from line. I playing mostly Space Colony games so I feel it and can say that 1704AD for Emperor can be easly bit with Game 6 settings, not sure about Warlords, but also close.

 

[Dynamic]

For other games I've also reviewed HOF tables via Ad-Hoc Query with fussy settings and add my feeling and experience for searching the right date and right Delta.

 

[Denniz]

That is the problem with either system. It is only as good as the sample of games and interpertation.

I took another stab at the curve. I created two more curves for comparison. I like the "More Linear" curve so I switched to that curve. The total ajdustment is the same as before just how it is distributed along the different difficutlies changes.

The graph shows a comparison of the curves assuming a 100 point max adjustment between Settler and Deity. The other pic shows the adjustments for each game for each curve. "Exp Scale" was the first one I posted above. "Half linear" was an attempt to smooth the back half of the curve.

If we can come up with good dates for my spreadsheet, I think the curve is good enough. It is just a matter of what the best dates would be.

 

[Dynamic]

Denniz, why do you think the Curve has this sigh of derivative 2? I think you should try to change concave to bulge.

P.S. It's better reflect the difference between highest levels.

 

[Dynamic]

But I still think that Linear Formula is closer to the real values than any Curve...

 

[Dynamic]

Sun Tzu Wu, please, play one Settler Religious game for correct Denniz's Formula for religion...

 

[Sun Tzu Wu]

That is the way the curve was trending. It seems possible that with Deity the AI spreads region quicker. Don't forget Deity techs faster too. Do I expect it to hold up? Maybe not...

Yes, the Deity AI does REX faster, but that does not necessarily help the Player win earlier.

Yes, the Deity AI eventually Researches faster, but for the Religious Victory, it rarely helps the Player in Technologies trades, because AI rarely has his economy recovered before the Game ends. This factor would apply more to Space Colony and The United Nations Diplomatic Victory where the recovered AI Economy can really help the Player with Research speed via numerous advantageous trades. Also, with early Religious Victory wins any Technology trading is done a few turns prior to the RL DV resolution, and the Player is more interested in loop-sided trades to gain trading diplomatic bonus than getting any Technology back in trade. In the vast majority of my best Religious Leader Diplomatic Victories, I almost always just give away Technologies to the AI in exchange for Diplomatic Trading (+4) Bonus.

The Turn Delta for Religious Leader Diplomatic Victory should definitely not be positive and certainly not 54 turns in favor of Settler level over Diety. For example, the curent Game 5 winner, compared to a hypothetical Deity level Game that ties it:

Settler Turn 96 Win = Diety Turn 42 Win

This example shows that the current Religious Victory Difficulty level Turn delta of +54 in favor of Settler over Diety is either a joke or based on seriously invalid data. By turn 42, a good Deity player may be about 10-15t from completing The Oracle and getting Theology; after that there is building 8 Missionaries, The Apostolic Palace and then waiting 10t for the RL DV resolution.

The Turn Delta for Game 5, at least with respect to Deity level is way off the chart (+54 -> Settler t96 = Deity t42).

If you don't fix it, I will try to win a Settler Game on Turn 55. The equivalent Deity Win would be a Turn 1 Win. Would that be enough proof that a +54 Turn adjustment is too big?

Sun Tzu Wu

 

[Sir Bugsy]

The Mods need to put a caveat in the game descriptions that the earliest finish wins based on an adjustment for difficulty level.

I have to say that while I agree with this methodolgy, it is changing the rules in the middle of the contest. Sort of like at halftime of the Super Bowl changing the scoring so that a touchdown counts as 10 points rather than 6. Kind of makes the efforts in the first half less important.

If I had known there was going to be an adjustment, I would have played at different levels. My enthusiasm for the challenge has suddenly taken a nose dive.

 

[shulec]

The Mods need to put a caveat in the game descriptions that the earliest finish wins based on an adjustment for difficulty level.

I have to say that while I agree with this methodolgy, it is changing the rules in the middle of the contest. Sort of like at halftime of the Super Bowl changing the scoring so that a touchdown counts as 10 points rather than 6. Kind of makes the efforts in the first half less important.

If I had known there was going to be an adjustment, I would have played at different levels. My enthusiasm for the challenge has suddenly taken a nose dive.

Unfortunately, you didn't see the original announcement for Challenge Series II on the CivFantaic home page. It was announced that the game rankings would be adjusted for difficulty after a sufficient sample size was available. This announcement was the first post in this thread. There was quite a bit of discussion about this early in the thread and there is probably a good amount of discussion yet to come.

 

[Sir Bugsy]

Unfortunately, you didn't see the original announcement for Challenge Series II on the CivFantaic home page. It was announced that the game rankings would be adjusted for difficulty after a sufficient sample size was available. This announcement was the first post in this thread. There was quite a bit of discussion about this early in the thread and there is probably a good amount of discussion yet to come.

Some of us do not frequent the forums or the home page very often anymore. I just assumed that the profile placed on the HOF site was the entire story. Shame on me for assuming I would get the entire story from my CFC colleagues at the appropriate place.

 

[kcd_swede]

I am not sure that the above is a fair statement. It is not my intent to penalize the levels between Settler and Deity. The adjustment is intented to level the difficulties. You should still have to play a really good game, regardless of level, to get to the top.

I was just suprised that the turn adjustment could actually in what I perceive to be the wrong direction for game 5.

But don't worry, I won't complain whatever finish comparison scheme is finally chosen. I'll play the games at a level I think is challenging (or that I can finish before my computer crashes), no matter what. I'm not seriously planning on any settler submissions in this challenge.

I realize that you'd have to create a very illogical and bizarre scoring scheme for me to win the challenge.

 

[Denniz]

Denniz, why do you think the Curve has this sigh of derivative 2? I think you should try to change concave to bulge.

P.S. It's better reflect the difference between highest levels.

I am not using the power of 2 curve any more.

But I still think that Linear Formula is closer to the real values than any Curve...

Linear equates to a uniform increase in difficulty. A curve represents a ramping up of difficulty. A bulge would be saying that the middle difficulties get increasingly difficult relative to the either end.

The bulge would fit the raw data. The question is whether that is a valid conclusion or is it an artifact from more players playing mostly low or high difficulty games.

P.S. I removed the Game 5 adjustments pending better information.

 

[Dynamic]

I can try Warlord and Emperor levels for Game6 for better data for the Formula. I don't want to play Deity, but if will not be the other way to calibrate the Formula I will try.

P.S. But I spend 1-2 weeks per every game so the results will not be soon.