Page 5 of 7

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 26 Feb 2018, 20:47
by NoPunIn10Did
Enriador wrote:
Would this Series include non-Classic maps in the future? In any case SoS runs well with 1900, Versailles and Milan. Not sure about Ancient Mediterranean though.


Right now we're just trying to get a few games off the ground, and I'm limiting it to the classic map since that doesn't require a paid subscription.

Sum-of-squares scaling would be perfectly fine for Milan (which isn't played much anyway), and would be reasonably fine for Ancient Med (since it also has 34 SCs and a victory condition of 18 SCs).

1900 would be a little different (with 39 centers on the map), but probably not in a noticeable way, as the victory condition is still 18 SCs. There will be only rare scenarios where a game ends in a draw with two players at 18 or 19.

Versailles, however, is where scaling differences would start to show up, though again to not any huge degree.

What I meant to refer to in the last post is the variety of variants that aren't currently supported on this site, but could hypothetically be available in the future. You end up with perverse scenarios in which the gain of a single SC, which represents a lower percentage of the overall map, represents a higher percentage of the SoS calculation.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 26 Feb 2018, 22:18
by WHSeward
I think you have it backward, Nopun. SoS will work well on any map because your score is normalized compared to the other competitors. It is C-Dip where the map makes a difference since raw SC count is part of the score.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 26 Feb 2018, 22:58
by NoPunIn10Did
WHSeward wrote:I think you have it backward, Nopun. SoS will work well on any map because your score is normalized compared to the other competitors. It is C-Dip where the map makes a difference since raw SC count is part of the score.


You are correct, and I should clarify a couple things.

C-Diplo would scale poorly too, but other rank-based systems that omit the raw SC component would work fine.

Where SoS breaks down is in comparing disparate maps. While in SoS you're "normalized" compared to your competitors, that starts to break down when comparing disparate maps. The X-squared function just starts to screw everything up. In my previous comment I mentioned that Versailles would present the greatest problem, but it's actually more likely to occur in variants where the percentage of total SCs on the map required for victory is vastly different.

For instance, let's look at three three-player draws where one player is just shy of victory, and the other two split the remaining map approximately equally.

Classic (34 SCs in total, 18 to win)
Player A has 17 SCs
Player B has 9 SCs
Player C has 8 SCs

Total SOS = 289 + 81 + 64 = 434

A = 66.59%
B = 18.66%
C = 14.75%


1900 (39 SCs in total, 18 to win)
Player A has 17 SCs
Player B has 11 SCs
Player C has 11 SCs

Total SOS = 289 + 121 + 121 = 531

A = 54.43%
B = 22.79%
C = 22.79%


Versailles (43 SCs in total, 22 to win)
Player A has 21 SCs
Player B has 11 SCs
Player C has 11 SCs

Total SOS = 441 + 121 + 121 = 683

A = 64.57%
B = 17.72%
C = 17.72%


It is certainly arguable that getting stopped at 17 SCs in 1900 should be worth less than in classic Diplomacy, but by how much? That's a judgement call for any multi-variant ranking system, but my preference would be to drop the SC component entirely.

We're hoping to run C-Diplo games here as a way to see how the community responds to a rank-based system in general, but I don't anticipate seeing C-Diplo as the eventual system of choice.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 26 Feb 2018, 23:43
by Ceasars Geezers
The big key here is enough people to start the game... ;)

It generally sounds like a fun, new style of play. It would be very interesting to see the different incentives and reward structures. Alliances mean more, solos are harder. The game becomes more about convincing people what to do. Right. Now. More immediate. More over reactions.

Did I mention I am excited about the game?

And do we actually have 5 for the 2/1/1? I think with Fuddin out, we are back to four.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 27 Feb 2018, 19:40
by NoPunIn10Did
Ceasars Geezers wrote:
And do we actually have 5 for the 2/1/1? I think with Fuddin out, we are back to four.


We're back up to 5 now.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 27 Feb 2018, 21:35
by dojhar
I'm interested.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 27 Feb 2018, 21:46
by NoPunIn10Did
dojhar wrote:I'm interested.


Let me know when you finish a current game. The minimum bar for joining the Classicists at the "Aspiring" level is one completed game. I'll send you the links for the application process at that point; it's pretty quick.

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 28 Feb 2018, 05:24
by Enriador
NoPunIn10Did wrote:
WHSeward wrote:I think you have it backward, Nopun. SoS will work well on any map because your score is normalized compared to the other competitors. It is C-Dip where the map makes a difference since raw SC count is part of the score.


You are correct, and I should clarify a couple things.

C-Diplo would scale poorly too, but other rank-based systems that omit the raw SC component would work fine.

Where SoS breaks down is in comparing disparate maps. While in SoS you're "normalized" compared to your competitors, that starts to break down when comparing disparate maps. The X-squared function just starts to screw everything up. In my previous comment I mentioned that Versailles would present the greatest problem, but it's actually more likely to occur in variants where the percentage of total SCs on the map required for victory is vastly different.

For instance, let's look at three three-player draws where one player is just shy of victory, and the other two split the remaining map approximately equally.

Classic (34 SCs in total, 18 to win)
Player A has 17 SCs
Player B has 9 SCs
Player C has 8 SCs

Total SOS = 289 + 81 + 64 = 434

A = 66.59%
B = 18.66%
C = 14.75%


1900 (39 SCs in total, 18 to win)
Player A has 17 SCs
Player B has 11 SCs
Player C has 11 SCs

Total SOS = 289 + 121 + 121 = 531

A = 54.43%
B = 22.79%
C = 22.79%


Versailles (43 SCs in total, 22 to win)
Player A has 21 SCs
Player B has 11 SCs
Player C has 11 SCs

Total SOS = 441 + 121 + 121 = 683

A = 64.57%
B = 17.72%
C = 17.72%


It is certainly arguable that getting stopped at 17 SCs in 1900 should be worth less than in classic Diplomacy, but by how much? That's a judgement call for any multi-variant ranking system, but my preference would be to drop the SC component entirely.

We're hoping to run C-Diplo games here as a way to see how the community responds to a rank-based system in general, but I don't anticipate seeing C-Diplo as the eventual system of choice.


Thanks for the explanation! I am actually surprised 1900 only got a 12,16% decrease relative to Classic, especially since 1900 has 14,7% more SCs than Classic. Looks actually quite balanced.

Among all rank-based systems, C-Diplo is certainly my favorite by a wide margin. I won't join now but I'll be following the game!

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 01 Mar 2018, 14:48
by Mr Aedron
Hey!

I'm interested! Sounds like a great structure and more ... diplomacy I guess?

Re: C-Diplo Series: Casual, Play to 1907 (5/7 & 3/7)

PostPosted: 01 Mar 2018, 16:28
by NoPunIn10Did
Mr Aedron wrote:Hey!

I'm interested! Sounds like a great structure and more ... diplomacy I guess?


As soon as you've finished a game, you can join the Classicists at the Aspiring level, and then we can sign you up.