Is Turkey too powerful?

Forum rules
This forum is limited to topics relating to the game Diplomacy only. Other posts or topics will be relocated to the correct forum category or deleted. Please be respectful and follow our normal site rules at http://www.webdiplomacy.net/rules.php.

Post a reply

Confirmation code
Enter the code exactly as it appears. All letters are case insensitive.
Smilies
:points: :-D :eyeroll: :neutral: :nmr: :razz: :raging: :-) ;) :( :sick: :o :? 8-) :x :shock: :lol: :cry: :evil: :?: :smirk: :!:
View more smilies

BBCode is ON
[img] is ON
[flash] is OFF
[url] is OFF
Smilies are ON

Topic review
   

If you wish to attach one or more files enter the details below.

Expand view Topic review: Is Turkey too powerful?

Re: Is Turkey too powerful?

by Bladerunners » Tue May 16, 2023 7:48 am

It seems to me that Italy automatically in bad position by doing so as Turkey and/or Russia profit much faster than Italy can

Re: Is Turkey too powerful?

by Aristocrat » Tue May 16, 2023 5:43 am

Bladerunners wrote:
Wed May 10, 2023 6:51 am
The developers made turkey extra powerful as they for some unfathomable reason have Italy attack Austria in the vast majority of games (over 80%) which is unrealistic to how diplomacy normally played both in regular and gunboat
I think the data for gunboat games shows that Italy attacking Austria isn't really a bad idea, and might actually be the optimal strategy (at least with almost every Austria opening Alb/Ser/Gal). Which is why bots like to do it so much. In press it is a different story because Austria and Russia can ally but no one seems to be able to make that happen in gunboat.

Re: Is Turkey too powerful?

by Bladerunners » Wed May 10, 2023 6:51 am

The developers made turkey extra powerful as they for some unfathomable reason have Italy attack Austria in the vast majority of games (over 80%) which is unrealistic to how diplomacy normally played both in regular and gunboat

Re: Is Turkey too powerful?

by DougJoe » Thu Mar 02, 2023 6:57 am

Bladerunners wrote:
Thu Mar 02, 2023 12:41 am
"Vastly" higher than 12% openings of an attack on Austria ... says 75%?? which is very different than a typical game. A couple of mods have said in the past that bots programmed to learn ... but they can't be as this is normally such a disastrous opening for Italy (not always - but normally)
How are you coming up with 75%?

Re: Is Turkey too powerful?

by Bladerunners » Thu Mar 02, 2023 12:41 am

"Vastly" higher than 12% openings of an attack on Austria ... says 75%?? which is very different than a typical game. A couple of mods have said in the past that bots programmed to learn ... but they can't be as this is normally such a disastrous opening for Italy (not always - but normally)

Re: Is Turkey too powerful?

by DougJoe » Mon Feb 27, 2023 5:44 am

ShipOfTheseus wrote:
Sun Feb 26, 2023 7:03 pm
DougJoe wrote:
Sun Feb 26, 2023 6:21 am
Oh, and those games are from PlayDiplomacy, which, AFAIK, doesn't have bots... So those are (theoretically) human-only games. That might make sense that there's such a difference in the percentages. I don't what games the AI bots were trained on here, but it would be interesting to see the difference in Italy play between low/high level players.
That was the question -- are the bots playing differently than humans? So the dataset I looked at was human players. Are you seeing more than 12% or so VEN>TRI in the bot games?
No. As I thought I said before, I'm seeing Ven-Tri a little *less* than 1 in 8. I see much, much more of Ven->Tyo, Rom->Ven with the bots than with the human data you posted.

Re: Is Turkey too powerful?

by ShipOfTheseus » Sun Feb 26, 2023 7:03 pm

DougJoe wrote:
Sun Feb 26, 2023 6:21 am
Oh, and those games are from PlayDiplomacy, which, AFAIK, doesn't have bots... So those are (theoretically) human-only games. That might make sense that there's such a difference in the percentages. I don't what games the AI bots were trained on here, but it would be interesting to see the difference in Italy play between low/high level players.
That was the question -- are the bots playing differently than humans? So the dataset I looked at was human players. Are you seeing more than 12% or so VEN>TRI in the bot games?

Re: Is Turkey too powerful?

by JECE » Sun Feb 26, 2023 3:44 pm

THC wrote:
Sun Feb 26, 2023 3:09 pm
As always, diplomacy is the key. There's a distinct lack of diplomacy in many online games, I've found, because everyone's out to do well for themselves rather than stopping others.
And the topic of the thread is about AI play, which may be even less interested in preventing other players from doing well.

Re: Is Turkey too powerful?

by THC » Sun Feb 26, 2023 3:09 pm

No power is "too powerful". There are trends in Diplomacy and some powers do well for a time because players tend to play in certain ways, and as a good player begins to find a good way to play one power, others take it up.

What doesn't really happen is that players look for ways to prevent other powers from doing well. It tends to be more a case of how can I succeed. Which means that, if you don't manage to succeed, there's no back-up plan.

It wasn't always the case. Playdip's figures show Turkey leading the field; if you look at DBN's figures for the DBNI, Turkey is bottom of the field. Why? Because the DBNI features the elite players who play to stop someone else from doing well.

As always, diplomacy is the key. There's a distinct lack of diplomacy in many online games, I've found, because everyone's out to do well for themselves rather than stopping others.

Re: Is Turkey too powerful?

by DougJoe » Sun Feb 26, 2023 6:21 am

ShipOfTheseus wrote:
Sun Feb 26, 2023 5:41 am
The dataset I found is here. Admittedly, it's six years old, so there may be new trends in way the game is played.

https://data.world/maxstrange/diplomacyboardgame
Oh, and those games are from PlayDiplomacy, which, AFAIK, doesn't have bots... So those are (theoretically) human-only games. That might make sense that there's such a difference in the percentages. I don't what games the AI bots were trained on here, but it would be interesting to see the difference in Italy play between low/high level players.

Re: Is Turkey too powerful?

by ShipOfTheseus » Sun Feb 26, 2023 5:41 am

The dataset I found is here. Admittedly, it's six years old, so there may be new trends in way the game is played.

https://data.world/maxstrange/diplomacyboardgame

Re: Is Turkey too powerful?

by DougJoe » Sun Feb 26, 2023 5:29 am

Here's my current data on the 348 Classic bot games I have played where I wasn't Italy:

Code: Select all

A Ven->Tri.       A Rom->Ven. F Nap->ION. :  21    6.0%
A Ven->Tri.       A Rom->Apu. F Nap->ION. :  12    3.4%

A Ven->Tri                                :  33    9.5%

A Ven H.          A Rom->Apu. F Nap->ION. :  69   19.5%
A Ven H.          A Rom->Nap. F Nap->ION. :  14    4.0%
A Ven H.                                  :  83   23.0%

A Ven->Tyo.       A Rom->Ven. F Nap->ION. : 149   42.8%
A Ven->Tyo.       A Rom->Apu. F Nap->ION. :  21    6.0%
A Ven->Tyo                                : 170   48.9%

A Ven->Pie.       A Rom->Ven. F Nap->ION. :  32    9.2%
A Ven->Pie.       A Rom->Ven. F Nap->TYS. :   7    2.0%
A Ven->Pie.       A Rom->Apu. F Nap->ION. :   3    0.1%
A Ven-Pie                                    42   12.1%
 
A Ven S H to Tri. A Rom->Apu. F Nap->ION. :  20    5.7%

A Ven->Tri *or* A Ven->Tyo, A Rom-Ven     : 182   52.3%
I find the difference in the statistics interesting.

In my games, I'm seeing Ven-Tri not quite 10% of the time, so a little less than in your data ShipOfTheseus, but...

Bot Italy's favorite opening by far is Ven-Tyo,Rom-Ven,Nap-ION, which is has played 43% of the time. It's Italy's favorite by almost twice as much as the two unique moves sets with Ven H. IMHO, Ven-Tyo/Rom-Ven are just as hostile to Austria as an immediate move to Tri. I'm pretty sure the vast majority of the times Italy played that opening it followed up with an attack on Austria in one way or another (although, to be honest, I don't have the data on Autumn '01 to know what Italy did in all 149 of those cases, my move parsing code can only do S01 right now). Compared to your numbers, that's roughly twice what you have computed for Ven-Tyo. Also interesting is that, in my games, Ven-Pie is played about half as much as the data you are looking at.

...and Italy moved to Tri *or* played Ven-Tyo, Rom-Ven over 50% of the time.

Re: Is Turkey too powerful?

by DougJoe » Sun Feb 26, 2023 1:24 am

ShipOfTheseus wrote:
Sat Feb 25, 2023 6:40 am
Bladerunners wrote:
Thu Feb 23, 2023 8:14 am
Design flaw by developers of programming Italy to attack Austria a high percentage of the time makes Turkey very powerful in lots of bots games...
I've been doing some analysis of a diplomacy moves dataset that includes over 20,000 games -- and 12% of the games, the first move from VEN is to TRI.

38% VEN hold
25% VEN>PIE
21% VEN>TYR
12% VEN>TRI
2% VEN>APU
1% VEN>TUS

Are you seeing it more often than 1 out of 8 times in the bots games?
That's a big dataset, where/how did you get it?

I haven't done a refresh on my analysis numbers in a while, I should do that.

Re: Is Turkey too powerful?

by ShipOfTheseus » Sat Feb 25, 2023 6:40 am

Bladerunners wrote:
Thu Feb 23, 2023 8:14 am
Design flaw by developers of programming Italy to attack Austria a high percentage of the time makes Turkey very powerful in lots of bots games...
I've been doing some analysis of a diplomacy moves dataset that includes over 20,000 games -- and 12% of the games, the first move from VEN is to TRI.

38% VEN hold
25% VEN>PIE
21% VEN>TYR
12% VEN>TRI
2% VEN>APU
1% VEN>TUS

Are you seeing it more often than 1 out of 8 times in the bots games?

Re: Is Turkey too powerful?

by Bladerunners » Thu Feb 23, 2023 8:14 am

Design flaw by developers of programming Italy to attack Austria a high percentage of the time makes Turkey very powerful in lots of bots games...

Re: Is Turkey too powerful?

by DougJoe » Sun Dec 18, 2022 12:37 am

Trigfea63 wrote:
Tue Dec 13, 2022 1:46 am
VinceK65 wrote:
Mon Dec 12, 2022 6:23 pm
I am new to the site and wonder if I can find somewhere succes rates of all powers in practice-games with bots (all games played on webdip).
It's an interesting question. As far as I know, no one has done that analysis for the 100,000+ classic bot games played on webdip. (Yes, that's not a typo.) I believe the most comprehensive bot-game statistical analysis is the one done by DougJoe anayzing his own ~350 bot games along various metrics:

viewtopic.php?f=3&t=2220&start=20#p304270
I've been sort of slowing down a bit lately, still stuck in round 53.

I hadn't really looked at the numbers in a while but I just realized that the bot has more solos as Germany than I do. (He's gotten two in this round while I haven't had one in quite a few attempts with the Germans...)
VinceK65 wrote:
Mon Dec 12, 2022 6:23 pm
I played 3 games with bots the last few days and I think the games helped a lot with understanding longer term consequences of your actions (f.e. Building a F or A in a certain stage, what the consequences are of the strength of other powers on the board, etc).
I would agree that playing the WebDip bots is good for practicing tactics and getting a feel for the game but I believe that it is a different experience than playing humans in either a press or no-press game.

Re: Is Turkey too powerful?

by Trigfea63 » Tue Dec 13, 2022 1:46 am

VinceK65 wrote:
Mon Dec 12, 2022 6:23 pm
I am new to the site and wonder if I can find somewhere succes rates of all powers in practice-games with bots (all games played on webdip).
It's an interesting question. As far as I know, no one has done that analysis for the 100,000+ classic bot games played on webdip. (Yes, that's not a typo.) I believe the most comprehensive bot-game statistical analysis is the one done by DougJoe anayzing his own ~350 bot games along various metrics:

viewtopic.php?f=3&t=2220&start=20#p304270

Re: Is Turkey too powerful?

by VinceK65 » Mon Dec 12, 2022 6:23 pm

I am new to the site and wonder if I can find somewhere succes rates of all powers in practice-games with bots (all games played on webdip).

I played 3 games with bots the last few days and I think the games helped a lot with understanding longer term consequences of your actions (f.e. Building a F or A in a certain stage, what the consequences are of the strength of other powers on the board, etc).

Re: Is Turkey too powerful?90

by Severinius de Monzambano » Sat Nov 12, 2022 4:55 am

The Ottoman Empire is powerful, but France will always be the most powerful country under the current conditions and rules. It all comes from the tiniest decision.

If in Spring 1901, the Ottoman fleet was in Constantinople instead of Ankara, it spell doom for all others. The Ottoman Fleet being in Ankara plays on the irrational impulse in games to use a piece on the first turn simply because its use is possible. This nearly (save for a planned bounce) always antagonizes Russia, which makes the playing field more even.

If you switch the Fleet’s S1901 location to Constantinople, it still allows for that Black Sea tension between the Ottomans and the Russians. (Unlike starting the Fleet in Smyrna, that only creates incentives to attack Austria, and more crucially Italy, quickly.) But it also creates the possibility of attacking Italy, a journey that is the mythic backdrop of the Iliad (Aeneas escapes Troy to find Rome) and rooted in historical experience (Mehmed II’s (unsuccessful) Venetian conquest.)

Turkey is very powerful, so long as its player is diplomatically, tactically, and strategically adept enough to take advantage. But that ever-so-slight decision to move the Fleet to the interior at the game’s start prevents Turkey from being too powerful.

France, on the other hand, is extraordinarily over-powered. Having the Iberian pennisula uncontested from the start gives France plus a likely Belgium gives France such an incredible lead at the end-of-1901 that is reflected in the solo/draw statistics.

Barring any changes to the map or to the rules, the only ways players can combat overpowered France is an incredibly aggressive Italy, allied with Austria, who devotes Italy’s units to take Marseilles in 1901 and then the Iberian Pennisula in 1902. Even though that may be rational, it’ll always feel uncomfortable to players.

Re: Is Turkey too powerful?

by DougJoe » Fri Sep 16, 2022 1:39 am

Jamiet99uk wrote:
Thu Sep 15, 2022 8:28 am
Jamiet99uk wrote:
Tue Sep 13, 2022 10:26 pm
Do well with Turkey in the first three years:

Wow! Turkey is so powerful!


Do poorly with Turkey in the first three years:

Dammit! Turkey is so underpowered! I'm just hemmed into this corner, there is nothing I can do!!
To expand on this in a way that I hope is actually useful:

Turkey is one of those powers that can be tricky to play in the early game, but can become powerful if it survives to a reasonable position in the mid-game. Unlike Austria, Turkey is not dependent on having at least one good ally to *survive* in the first few years, but without at least one ally, it is very difficult for Turkey to break out of their corner.

Outside of a bot game, in real play, Turkey's alliance choices, and the competence and reliability of the chosen ally, are especially crucial.

Partner with a competent Austria? Good idea - contain Italy and hit Russia in the belly.

Partner with a clever Italy? Lots of options here, especially as in this one you can disguise your intentions for several moves.

Partner with a reliable Russian? The most obvious (and initially safest) choice, except that half the board will scream "juggernaut!" and attempt to rally against you.

Turkey has lots of early choices to make, but the most important work for Turkey happens in press, not in your Spring 1901 moves.
Agreed! My comments above were specific to my experiences with bot games. Full press games are a whole different world.

Top