The Big Question - How Does The AI Choose Which Units To Build?
- Thread starter Ozymandias
- Start date
ozymandias,
I have heard that artillery still are sitting in the cities...
You have been a computer-pro.
Is this a problem that is so hard to solve? (For Firaxis)
Otherwise I do not understand why its still not
possible to use in offensives.
It sure have great importance for WW2 and Modern scenarios.
Best Regards
Rocoteh
I have heard that artillery still are sitting in the cities...
You have been a computer-pro.
Is this a problem that is so hard to solve? (For Firaxis)
Otherwise I do not understand why its still not
possible to use in offensives.
It sure have great importance for WW2 and Modern scenarios.
Best Regards
Rocoteh
Hi Rocoteh,
Yeah, I've read that the much-heralded, big improvement in artillery performance is that it will now bombard troops more than improvements -- and some already think (after a very few games!) that this gives the human player a serious advantage especially re: taking cities.
Insofar as hypothetical programming goes, I think it would be very simple to construct an algorothim along the lines of:
1."Whenever possible, for offense, create a stack of two offensive, one defensive, and one support (artillery) units" and reverse the ratio of the first two for defensive stacks.
2. "If necessary (i.e., enemy is approaching a strategic resource tile) proceed with forces at hand."
3. "If constructing larger forces, make multiples of the forces in (1) above, again be ready to leave under circumstances as in (2)."
This would also simply and realistically simulate the construction of combined force armies -- the categories could even be expanded to add a "fast" unit if available, etc., and to weigh the value of all units for an offensive or defensive stack (remember how far we got on THAT one recently
. )-- So, in short, as a programmer, grognard, and one-time wargame designer, my "professional" opinion is that whatever algorithm(s) is/are in place for artillery is needlessly moronic, plain and simple.
I'm beginning to think that a "hybrid" of the Civ2 and Civ3 approaches might be required in almost every case.
I still would suggest that, for ACW, as Longstreet realized, it was an historical moment where the advantage had passed to the defender. Were I designing an ACW mod under the present circumstances, I might consider "Napoleonic" cannon as integral to the infantry and use this as a defensive (range=0) benefit -- Just a thought -- trying to "make lemonade out of lemons" as the saying goes
.All The Best,
Oz
ozymandias,
There is one feature I do not understand why it not
was included with Conquests:
The ability to set a maximum for numbers of units
that can be moved by rail each turn.
This was standard 40 years ago during the Avalon Hill era.
Best Regards
Rocoteh
There is one feature I do not understand why it not
was included with Conquests:
The ability to set a maximum for numbers of units
that can be moved by rail each turn.
This was standard 40 years ago during the Avalon Hill era.
Best Regards
Rocoteh
The Last Conformist
Irresistibly Attractive
While I too would have loved a such limit on RR travel, there would be serious amounts of bookkeeping involved, not too mention that pathing would be suddenly alot harder harder.
(This assuming you mean that every piece of rail could only be used by X units/turn. If you mean that your empire would have fixed capacity of Y units/per turn travelling by train anywhere, it would be alot simpler to implement. Not being around 40 yrs ago, I do not know what was standard then!)
(This assuming you mean that every piece of rail could only be used by X units/turn. If you mean that your empire would have fixed capacity of Y units/per turn travelling by train anywhere, it would be alot simpler to implement. Not being around 40 yrs ago, I do not know what was standard then!)
The Last Conformist
Irresistibly Attractive
What's truly absurd wrt AI artillery utilization is that while it (in PTW at least) does not use the bombard capability of its ships very effectively, it does use it offensively. So the code for using artillery offensively must be "there", so to speak - it's just not used with land-based artillery.
Clarification:
I mean that each empire should have a limit with regard
to how many units that can be transported by rail each turn.
If you take ACW as an example: Union rail cap was 30 000 men
per month.
If you take Op. Barbarossa June 1941 as an example:
Germany had to convert all track. Soviet had a hard choice:
Use the railway cap. for troop-movements or evacuate
factory-resources to Sibiria.
This issue was not a big problem as long as we only had
the "old" CIVIII where 1 Turn = 1 Year.
Now with focus on scenarios where 1 Turn can be 1 week
its a problem.
With regard to "old times" Avalon Hill-era, S.P.I.-era and
so on: Yes, each nation normally had a fixed rail-cap.
Regards
Rocoteh
I mean that each empire should have a limit with regard
to how many units that can be transported by rail each turn.
If you take ACW as an example: Union rail cap was 30 000 men
per month.
If you take Op. Barbarossa June 1941 as an example:
Germany had to convert all track. Soviet had a hard choice:
Use the railway cap. for troop-movements or evacuate
factory-resources to Sibiria.
This issue was not a big problem as long as we only had
the "old" CIVIII where 1 Turn = 1 Year.
Now with focus on scenarios where 1 Turn can be 1 week
its a problem.
With regard to "old times" Avalon Hill-era, S.P.I.-era and
so on: Yes, each nation normally had a fixed rail-cap.
Regards
Rocoteh
The rail cap unit is just the tip of the iceberg (and I agree, it's almost certainly for unit pathing -- and if the AI had to make critical decisions re: what to RR and what not to, that would be one tough algorithm to construct! -- let alone probably excruciatingly time-consuming to execute!)).
The rest of the iceberg is: what scale is Civ? Taking the modern era and a 256x256 map, are units corps? divisions? armies? "Ideally", this would affect (1) Rail capacity (2) Stacking limits, probably with an eye towards how many men and machines can be deployed along a tile side (3) the effects of stacking (overstacking due to retreats, etc.)
Believe it or not, this is one where I understand why it's been done this way --
This does not mean, friend Rocoteh, that I have not been giving your concerns some serious thought! -- It occurs to me that, during the ACW, paved roads were almost non-existent outside cities (there was a fairly well developed canal system, but that was outside every theater of operations, in the north) and that dirt roads, at the scale we're talking about, were probably ubiquitous within each tile (James Dunnigan took this approach with the game "1914" when he realized that practically, literally, every hex would contain a railroad, so he factored in strategic movement in another fashion) ...
... All of which leads to: railroads were critical in the ACW, "regular" roads were not! Personally, I might consider dispensing with Civ railroads altogether, considering roads railroads, and setting their movement to, oh, 15 tiles per MF or whatever.
Realistic; efficient -- and you can even substitute RR graphics for road graphics, and consider using Mountains for uncrossable terrain and Hills for crossable, thereby highlighting -- with roads and uncrossable mountains -- e.g. how valuable the Shenandoah Valley actually was.
All The Best,
Oz
The rest of the iceberg is: what scale is Civ? Taking the modern era and a 256x256 map, are units corps? divisions? armies? "Ideally", this would affect (1) Rail capacity (2) Stacking limits, probably with an eye towards how many men and machines can be deployed along a tile side (3) the effects of stacking (overstacking due to retreats, etc.)
Believe it or not, this is one where I understand why it's been done this way --
This does not mean, friend Rocoteh, that I have not been giving your concerns some serious thought! -- It occurs to me that, during the ACW, paved roads were almost non-existent outside cities (there was a fairly well developed canal system, but that was outside every theater of operations, in the north) and that dirt roads, at the scale we're talking about, were probably ubiquitous within each tile (James Dunnigan took this approach with the game "1914" when he realized that practically, literally, every hex would contain a railroad, so he factored in strategic movement in another fashion) ...
... All of which leads to: railroads were critical in the ACW, "regular" roads were not! Personally, I might consider dispensing with Civ railroads altogether, considering roads railroads, and setting their movement to, oh, 15 tiles per MF or whatever.
Realistic; efficient -- and you can even substitute RR graphics for road graphics, and consider using Mountains for uncrossable terrain and Hills for crossable, thereby highlighting -- with roads and uncrossable mountains -- e.g. how valuable the Shenandoah Valley actually was.
All The Best,
Oz
ozymandias,
This time we can only agree on one thing:
You are computer-pro and I am self-learned on computers.
If you are right with regard to "the tip of the iceberg" and
so on, how could Talonsoft 3 years ago (on a low budget)
present their Century Of Warfare, with restrictions to
rail cap. , and with an excellent AI?
In some of their scenarios scale is 5km, in others 25 km
and so on.
With regard to 1914: I am sitting here with a copy of it
in front of me ... It has railroads. France rail-cap. is 2
corps each turn. Germany has rail-cap. 3 corps each turn...
I guess you are thinking on one of Dunnigans other productions.
Then on ACW: I am sorry: Your suggestion would result in
an unhistorical "railroad-blitz". I repeat Union rail-cap was
30 000 men each month. CSA-cap was lower and did fall
during the whole Civil War. When they lost Tennessee -62,
they also lost all their production of locomotives.
Best Regards
Rocoteh
This time we can only agree on one thing:
You are computer-pro and I am self-learned on computers.
If you are right with regard to "the tip of the iceberg" and
so on, how could Talonsoft 3 years ago (on a low budget)
present their Century Of Warfare, with restrictions to
rail cap. , and with an excellent AI?
In some of their scenarios scale is 5km, in others 25 km
and so on.
With regard to 1914: I am sitting here with a copy of it
in front of me ... It has railroads. France rail-cap. is 2
corps each turn. Germany has rail-cap. 3 corps each turn...
I guess you are thinking on one of Dunnigans other productions.
Then on ACW: I am sorry: Your suggestion would result in
an unhistorical "railroad-blitz". I repeat Union rail-cap was
30 000 men each month. CSA-cap was lower and did fall
during the whole Civil War. When they lost Tennessee -62,
they also lost all their production of locomotives.
Best Regards
Rocoteh
The Last Conformist
Irresistibly Attractive
I think he means we should have "railroads" that are, game-mechanically, infact just roads with RR graphics. Which couldn't possibly result in a scenario terribly different from the current one (normal or ACWR).
Hmm ... I'm unfamiliar with "Century Of Warfare" (1) it sounds like you recommend it ... ? and (2) the scale was Civ-like?
Apologies if I goofed on the 1914 title -- rather than dig out my ancient copy I'll certainly take your word on it! -- But there was at least one WWI strategic / operational game which took this approach, for the reason I cite ... c'est la guerre ...
I'm afraid I'm at a loss as to help ACW -- "15" BTW was just a random number. As for loss of rail capacity, perhaps all Southern workers should be built at a special facility in Tennessee; when/if it falls, no more new units to build railroads etc. could be produced, and as the earlier production become casualties of war the decay is at least simulated ... ?
*sigh* I hate it when Civ leaves us between the proverbial rock and the hard place ...
Best Regards,
Oz
Apologies if I goofed on the 1914 title -- rather than dig out my ancient copy I'll certainly take your word on it!
-- But there was at least one WWI strategic / operational game which took this approach, for the reason I cite ... c'est la guerre ... I'm afraid I'm at a loss as to help ACW -- "15" BTW was just a random number. As for loss of rail capacity, perhaps all Southern workers should be built at a special facility in Tennessee; when/if it falls, no more new units to build railroads etc. could be produced, and as the earlier production become casualties of war the decay is at least simulated ... ?
*sigh* I hate it when Civ leaves us between the proverbial rock and the hard place ...
Best Regards,
Oz
ozymandias,
Century of Warfare (out of print) was designed by Norm Koger.
You can (could) play scenarios like the October War and
scenarios WW2 on global scale (even scenarios on Ancient
wars have been made despite the title).
I never come with critique against people who are not
in place an can not defend their point of position.
...but let us say that Talonsoft were bought by people
who gave profit priority instead of high-quality wargames
with low sales.....
Edit: Yes, I think you are thinking of Dunnigans WWI game
where 1 Turn = 3 months. Some historian: Peter Englund,
later did a revision with an extended map called:
War to end Wars.
Best Regards
Rocoteh
Century of Warfare (out of print) was designed by Norm Koger.
You can (could) play scenarios like the October War and
scenarios WW2 on global scale (even scenarios on Ancient
wars have been made despite the title).
I never come with critique against people who are not
in place an can not defend their point of position.
...but let us say that Talonsoft were bought by people
who gave profit priority instead of high-quality wargames
with low sales.....
Edit: Yes, I think you are thinking of Dunnigans WWI game
where 1 Turn = 3 months. Some historian: Peter Englund,
later did a revision with an extended map called:
War to end Wars.
Best Regards
Rocoteh
Not yet my friend -- real life caught up with me over the past week or so. I'm actually planning on (*shudder*) playing an unmodded full-length Conquests to get a real feel for what if anything has been done re: production from their "static" algorithm's POV -- although I must say, in my heart of hearts, that I believe the unit-generating improvements (shades of Rise Of Nations ...) will ultimately be our salvation ...
Best Regards,
Oz
I am excited about using inprovments to build units. In my scenario there are so many units, and with being able to use this new feature you can get very good balanced production.
If you put a Great Wonder in say London, then have all of your units creating buildings require that Great Wonder in London to build, then you can space these building in the tech tree and the British would be the only ones to build these units.
Glad we will be seeing all kinds of powerfull armies, navies and air forces now.
If you put a Great Wonder in say London, then have all of your units creating buildings require that Great Wonder in London to build, then you can space these building in the tech tree and the British would be the only ones to build these units.
Glad we will be seeing all kinds of powerfull armies, navies and air forces now.
The Last Conformist
Irresistibly Attractive
Oh, I overlooked that particular figure, and was thus assuming ~3 tiles/MF.
ozymandias,
The production-system and overall impression of
Conquests are better than I thought it would be.
At the same time I think we must admit that CIV is
reaching the limits of realism.
You and I and many other Grognards would pay for more
realism: Supply,weather,troop-quality,leadership, tactical
combat display and so on.
....but here I can say I understand Firaxis if they decide
to not go further in realism.
You are an old S.P.I.-Grognard. There are several theorys
about the reasons to the decline and fall of S.P.I.
Some years ago Greg Costikyan presented some of these
theorys in "A Farewell to Hexes".
My main theory is that S.PI. responded to the demand
from the hard-core players, who wanted more and more
complex games. The silent majority did not agree and
S.P.I. fell.
Thus I understand if Firaxis want to avoid this trap.
As I mentioned in an earlier post NO War/Strategy game
can match Century of Warfare in realism and excellent AI.
.....but its out of print.... No profit.
Best Regards
Rocoteh
The production-system and overall impression of
Conquests are better than I thought it would be.
At the same time I think we must admit that CIV is
reaching the limits of realism.
You and I and many other Grognards would pay for more
realism: Supply,weather,troop-quality,leadership, tactical
combat display and so on.
....but here I can say I understand Firaxis if they decide
to not go further in realism.
You are an old S.P.I.-Grognard. There are several theorys
about the reasons to the decline and fall of S.P.I.
Some years ago Greg Costikyan presented some of these
theorys in "A Farewell to Hexes".
My main theory is that S.PI. responded to the demand
from the hard-core players, who wanted more and more
complex games. The silent majority did not agree and
S.P.I. fell.
Thus I understand if Firaxis want to avoid this trap.
As I mentioned in an earlier post NO War/Strategy game
can match Century of Warfare in realism and excellent AI.
.....but its out of print.... No profit.
Best Regards
Rocoteh
