Historical AI?

Adanac's Strategic level World War I grand campaign game designed by Frank Hunter

Moderator: SeanD

Graf Leinsdorf
Posts: 27
Joined: Thu Sep 27, 2007 8:17 am

Historical AI?

Post by Graf Leinsdorf »

I think that long time wargamers (e.g. former boardgamers) are not so much interested in winning against AI or PBEM, as in seeing the game mechanics so well devised that historical developments are magically simulated, with only some credible and well founded what-ifs reasonably tested.
I think that the current development of this really intelligent and riveting game doesn’t yet meet such grognards expectations, at least as to the conduct of operations in the role of EP by the programmed opponent.
I will focus on the few points which, after some games played as CP against the AI as EP, have slightly baffled my need of a sound historical simulation of World War 1.

1. In one game I concentrated my effort on the Western front, leaving a very tiny veil of units deployed against the Russians. Well, albeit being so feebly opposed, they were able to somewhat threaten German or Austrian territories only after 2 years of war. That seems really too little also for the not so glamorous and well equipped Russian troops, which on the contrary historically did do well (Brusilov offensive teaches) overwhelming the CP lines in Galicia and threatening even to reach the heart of Austria-Hungary and Prussia. We know that Western Allies often istigated the Russians to renew their efforts in order to relieve the pressure they were sustaining in France and Russian generously responded to such demands.

2. In another game I completely neglected the Serbians, leaving against them only a couple of weak Austrian corps and they didn’t dare to undertake any attack against my feeble line. I think that the hard-pressed Western allies would and could have required more collaboration from the part of the sturdy Serbians too!

3. With great astonishment in the May-June 1916 game turn I assisted to the quite un-historical surrender of Britain, I suppose just for the outcome of some naval battles in the North Atlantic and losses in the Western Front. I can’t imagine any development like this in the real Great War, at least until the fall of Paris and that was not the case in my game.

4. In all the games I played as CP, the Ottoman Empire could serve as supine purveyor of raw material and industrial points to Germany as no military threat ever materialized against them to require the build-up of a solid line of troops: no Allenby advance in Palestina, no sustained Russian offensive in the Caucacus and almost nil also in Mesopotamia. Even when once I dared to reach the outskirts of Alexandria with my weak Ottoman Corps the 5 or more strong British corps deployed there remained unmoved without daring any offensive against such feeble line. And the same could be repeated for the sleepy Mesopotamian Front. I wonder moreover whether a simulation of Lawrence and the Arab uprising could not add some flavour and, together with a more sustained British initiative, help to upgrade Ottoman Empire from sheer source of raw materials and industrial points to active partner in the CP alliance.

5. In no game at all played as CP I did see any attempt by the EP to repeat a kind of Gallipoli endeavour. Besides being quite unhistorical, that’s undoubtedly something which deprives the game of a not minor source of fun and variety.

In any case I’d thank very much Frank Hunter for the great game design and for the patient effort of responding to so many users update demands!


Image
Attachments
aus2.jpg
aus2.jpg (12.69 KiB) Viewed 392 times
User avatar
kcole4080
Posts: 31
Joined: Sat Sep 22, 2007 3:14 pm

RE: Historical AI?

Post by kcole4080 »

In my game right now, there is a very powerful Brit army massed in Greece.
The Bulgars have joined the CP, as have the Ottomans, Italy is still neutral, Belgium, Serbia, & France have surrendered.

The Russians have mounted many small attacks against the Germans in Prussia, & have given the A-H a hard time before the mass transfer of German troops east after the fall of France.
they certainly weren't timid, & I'm only in the early summer of 1916 so far.

The Brits countered the Turks easily in the middle east & are taking territory slowly with quite a small force.
The Turks have much of their army tied up in the Caucasus. 
SMK-at-work
Posts: 3396
Joined: Mon Aug 28, 2000 8:00 am
Location: New Zealand

RE: Historical AI?

Post by SMK-at-work »

IIRC the AI plays better as the central powers, as it has fewer decisions to make.
Meum est propisitum in taberna mori
FrankHunter
Posts: 2111
Joined: Fri Mar 26, 2004 6:07 am

RE: Historical AI?

Post by FrankHunter »

Actually the AI plays better as the TE than it does as the CP.

SMK-at-work
Posts: 3396
Joined: Mon Aug 28, 2000 8:00 am
Location: New Zealand

RE: Historical AI?

Post by SMK-at-work »

Oh - I guess I did not RC then. But the TE would have been my next guess!!
Meum est propisitum in taberna mori
hjaco
Posts: 872
Joined: Fri Mar 23, 2007 4:09 pm

RE: Historical AI?

Post by hjaco »

[:D][:D][:D]
Hit them where they aren't
Graf Leinsdorf
Posts: 27
Joined: Thu Sep 27, 2007 8:17 am

RE: Historical AI?

Post by Graf Leinsdorf »

ORIGINAL: Graf Leinsdorf

I think that long time wargamers (e.g. former boardgamers) are not so much interested in winning against AI or PBEM, as in seeing the game mechanics so well devised that historical developments are magically simulated, with only some credible and well founded what-ifs reasonably tested.
I think that the current development of this really intelligent and riveting game doesn’t yet meet such grognards expectations, at least as to the conduct of operations in the role of EP by the programmed opponent.
I will focus on the few points which, after some games played as CP against the AI as EP, have slightly baffled my need of a sound historical simulation of World War 1.

1. In one game I concentrated my effort on the Western front, leaving a very tiny veil of units deployed against the Russians. Well, albeit being so feebly opposed, they were able to somewhat threaten German or Austrian territories only after 2 years of war. That seems really too little also for the not so glamorous and well equipped Russian troops, which on the contrary historically did do well (Brusilov offensive teaches) overwhelming the CP lines in Galicia and threatening even to reach the heart of Austria-Hungary and Prussia. We know that Western Allies often istigated the Russians to renew their efforts in order to relieve the pressure they were sustaining in France and Russian generously responded to such demands.

2. In another game I completely neglected the Serbians, leaving against them only a couple of weak Austrian corps and they didn’t dare to undertake any attack against my feeble line. I think that the hard-pressed Western allies would and could have required more collaboration from the part of the sturdy Serbians too!

3. With great astonishment in the May-June 1916 game turn I assisted to the quite un-historical surrender of Britain, I suppose just for the outcome of some naval battles in the North Atlantic and losses in the Western Front. I can’t imagine any development like this in the real Great War, at least until the fall of Paris and that was not the case in my game.

4. In all the games I played as CP, the Ottoman Empire could serve as supine purveyor of raw material and industrial points to Germany as no military threat ever materialized against them to require the build-up of a solid line of troops: no Allenby advance in Palestina, no sustained Russian offensive in the Caucacus and almost nil also in Mesopotamia. Even when once I dared to reach the outskirts of Alexandria with my weak Ottoman Corps the 5 or more strong British corps deployed there remained unmoved without daring any offensive against such feeble line. And the same could be repeated for the sleepy Mesopotamian Front. I wonder moreover whether a simulation of Lawrence and the Arab uprising could not add some flavour and, together with a more sustained British initiative, help to upgrade Ottoman Empire from sheer source of raw materials and industrial points to active partner in the CP alliance.

5. In no game at all played as CP I did see any attempt by the EP to repeat a kind of Gallipoli endeavour. Besides being quite unhistorical, that’s undoubtedly something which deprives the game of a not minor source of fun and variety.

In any case I’d thank very much Frank Hunter for the great game design and for the patient effort of responding to so many users update demands!


After some more games playing as CP against the AI as EP, I’ve not experienced any difference in the outcomes that in my previous message I complained of being so unhistorical. Rather I could complement my perplexities with further remarks.

6. Saloniki expedition, if undertaken at all by EP (and in my games it has been seldom the case), has always been very timid, posing almost no serious threat to CPs south flank. We know instead that in real Great War the Entente advance was deep, bringing about the collapse of Bulgaria which, according to a Ludendorff remark, “decided the fate of the Quadruple alliance” (namely of CPs).

7. It’s a bit strange that as CP player I’ve closed all my campaigns with a flourishing production and plenty of industrial points and raw materials, due to conquests and docile supplies by the Ottoman Empire (as already said, quite unscathed by British or Russian operations), but mainly to an unrivalled control of the seas by German fleet. We know instead the Germany and Austria were reduced almost to starvation by the Entente blockade and that was one of the main reasons of their surrender.

8. All in all: just since my first game, I was able to win easily against the AI in the role EPs, provided that I followed a “France and Serbia first” option, stubbornly sticking to the Schlieffen setup (and plan). If the programmed opponent hasn’t been able to put up a serious challenge even against a human player not yet experienced in game mechanics, that seems as ample evidence of its inability to simulate historical events: after all CP WERE defeated in the Great War and there were sound historical (namely political, economic and military) reasons for that to happen.
Venator
Posts: 96
Joined: Mon Jan 22, 2007 9:08 pm

RE: Historical AI?

Post by Venator »

The AI is no worse than in most wargames (and better than many). When you consider that a game with no element of chance and with a simple board and few rules - chess - has only recently developed its AI to the point where it is superior to humans, and that with massive resources pumped into so doing, I'm not sure that wargames, which are much more complex in terms of variables can have too much expected of their AI. I think the only fair reflection of the games accuracy as a simulation is player vs player. And from my limited experience of that, I'd say things were a bit too easy for the Triple Entente.
Besides being quite unhistorical

Nothing unhistorical about that at all. There were sound reasons to concentrate on the Western front. The sideshows against the Ottomans need not have happened at all had the 'West First' men got their way earlier.

One thing I would say is that it seems too easy for the TE to knock Turkey out of the game with a first turn DoW and concentration of the BEF and a couple of French corps against Jerusalem and Beirut/Damascus respectively while the Russians send nine corps into the east and drive along the Black Sea coast. The French then land at Smyrna and the BEF at Gallipoli. Given that the human player does not sit around having tea on the beach, Constantinople tends to fall in short order.

Also against the AI I have driven the Russians through Prussia to within three hexes of Berlin before the Germans diverted any units to oppose them. This is not, on the face of it, very wise of the AI...
hjaco
Posts: 872
Joined: Fri Mar 23, 2007 4:09 pm

RE: Historical AI?

Post by hjaco »

Well I for one don't want to play with a straight jacket on being forced into a fully historic course [&:]

Regarding development of AI that takes a lot of time and therefore the end product will not only be delayed but naturally substantial more expansive as well.

Regarding the example with Saloniki I must make a clarification. Absolutely nothing happened from intervention until the primarily Serb offensive in September 1918 and the following collapse was of various other reasons than action on behalf of the Entente who of course took full advantage of the situation.

http://www.oldandsold.com/articles26/wo ... e-47.shtml
Hit them where they aren't
hjaco
Posts: 872
Joined: Fri Mar 23, 2007 4:09 pm

RE: Historical AI?

Post by hjaco »

expansive = expensive [;)]
Hit them where they aren't
Venator
Posts: 96
Joined: Mon Jan 22, 2007 9:08 pm

RE: Historical AI?

Post by Venator »

Well I for one don't want to play with a straight jacket on being forced into a fully historic course
 
 Quite. The ability to do what one pleases within the limits of resources is the great strength of the game. My only concern is that the ability the TE has to hop about using transports is perhaps a bit overpowered.
SMK-at-work
Posts: 3396
Joined: Mon Aug 28, 2000 8:00 am
Location: New Zealand

RE: Historical AI?

Post by SMK-at-work »

There are counters to the Galipoli gambit....for starters you can only invade with 1 corps......Turkey should put a corps on each of Gallipoli, Smyrna and Beirut ASAP, which gives the allies a good chance of being completely wiped out at sea!!

then the allies have to have invested enough transports to ship in a HQ, or else their corps sits there for a whole turn doing nothing....so that's 6 Transports - 2 for each the Nth Atlantic, Eastern and Western Med's.  If you want to land a corps at Smyrna at hte same time then that's 2 more transports for the Med sea zones if French, or 3 if British as you need 1 in the Atlantic as well - so that's 8-9 transports.

Supply in all those coastal cities is appalling, so you'll need at least 1 pref 2 transports on Sea supply in the Eastern Med.

In the mean time the Turks have 4 or 5 Corps within easy strat transport range of Gallipoli - the most dangerous, and it's no great effort for them to get there.

You can also expect the AH fleet to sortie immediately to try to upset shipping in the Eastern Med once you've landed, plus perhaps a few U-boats (yes they reach!!)

And all this time you're missing those troops from the Western Front - you certainly can't afford to do so in 1914 if the CP is attacking there.  If they're attacking Russia then you can probably do so without too much bother.....but then attacking in France is a good option too, and costs less!

And if the CP is attacking Russia then the Turks can easily ignore the Caucasus....that's another 4 corps they can put into the Balkans.

If Turkey is still in the war when Bulgaria enters then Germany can start sending manufactured goods to the Turks....that can get really irritating - the Turks send raw materials to Germany, and the Germans send MG's back.....it costs a bit more in 1.2, but it's still eminnantly do-able, and with 8-10 MG's per turn the Turks get to be quite hard to over-run.  They can rebuild destroyed corps quite easily - sure they aren't going to conquer much, but that's not the point.  1 or 2 German or AH corps in a stack of toerhwise Turkish & Bulgarian units and you can kiss goodbye to your dreams of forcing the straights.

Playnig 1.2 over the weekend until 1919 the British landed 6 corps at Salonika & it took me a whole 2 moves to completely wipe them out!! :)
Meum est propisitum in taberna mori
Venator
Posts: 96
Joined: Mon Jan 22, 2007 9:08 pm

RE: Historical AI?

Post by Venator »

To be clear, my comments were really that the AI does none of those things if it is acting as the CP...
SMK-at-work
Posts: 3396
Joined: Mon Aug 28, 2000 8:00 am
Location: New Zealand

RE: Historical AI?

Post by SMK-at-work »

Yeah - but playing against the AI is still easier than PBEM.
 
Want a game??!
Meum est propisitum in taberna mori
Venator
Posts: 96
Joined: Mon Jan 22, 2007 9:08 pm

RE: Historical AI?

Post by Venator »

Want a game??!
 
 If you like - though I must warn you that I might be a litte erratic in getting my moves done. So long as you don't mind the odd day's wait, I'd be keen.
 
 
User avatar
EUBanana
Posts: 4255
Joined: Tue Sep 30, 2003 3:48 pm
Location: Little England
Contact:

RE: Historical AI?

Post by EUBanana »

I think attacking at Gallipoli is pretty crazy, given you can attack from Egypt much more safely.  The Ottoman Empire doesn't take all that much effort to knock out - so if you wanna go for an Ottoman first TE strategy, why fart about with some dire little landing strip on the Dardanelles??
Image
hjaco
Posts: 872
Joined: Fri Mar 23, 2007 4:09 pm

RE: Historical AI?

Post by hjaco »

Quite so.

Besides its not really necessary to knock out OE but just to prevent a surplus generation of OE raw materials which can be constantly sent to German factories.

Basra and the resource next to it can be taken on turn two and Russia can with the November reinforcements take the 2 raw materials near the Caucasus in March at the latest around which time OE will use the remaining two resources themselves.
Hit them where they aren't
Venator
Posts: 96
Joined: Mon Jan 22, 2007 9:08 pm

RE: Historical AI?

Post by Venator »

My point was that as the AI never seems to defend Beirut, Smyrna or Gallipoli, there is no risk in landing there. It was really linked to my earlier comment that the AI controlled CP often fail to attack through Belgium thus freeing up some French corps who can be used to hop about with the BEF. I'm quite aware that a decent CP player would not make it so easy... But I'm not sure why the AI doesn't take precautions in this regard.
FrankHunter
Posts: 2111
Joined: Fri Mar 26, 2004 6:07 am

RE: Historical AI?

Post by FrankHunter »

Originally the AI was written using basic scripts.  Germany would attack through Belgium because I programmed it to.  I wasn't happy with that because in a game like this there are just too many things I don't foresee.  So I instead wrote the AI I have now but the earlier versions of it was married to a "memory script".  Basically the AI would watch what you did when you were playing a side and record it, every move.  So whatever I did, if it was successful, the AI would eventually use against me.  If I attacked through Belgium and it worked, the AI, when playing Germany, would do the same.  Yet, when confronted with a new situation it would fall back on the more general AI. 

Memory scripting also ran production and naval.

When I write it here it might sound pretty good but in practice I wasn't happy with it.  Even dynamic scripting caused major problems.  The current AI, which doesn't use scripting, actually plays better than the previous one although it has a tendency towards cautiousness.

As for the particular problem of the Ottomans not protecting their ports  it would be simple for me to add a script demanding that that be a top priority.  It would leave the Ottomans short several corps on the front but it would block amphibious moves.

 


James Ward
Posts: 1163
Joined: Tue May 09, 2000 8:00 am
Location: Baltimore, Maryland, USA

RE: Historical AI?

Post by James Ward »

ORIGINAL: FrankHunter

As for the particular problem of the Ottomans not protecting their ports  it would be simple for me to add a script demanding that that be a top priority.  It would leave the Ottomans short several corps on the front but it would block amphibious moves.





Could you add some maximum 5-6 point corps to the Ottomans and use them for garrisions? At least you would have to fight your way ashore.
Post Reply

Return to “Guns of August 1914 - 1918”