AI Scenarios: What, Why and How
Moderators: wdolson, Don Bowen, mogami
-
el cid again
- Posts: 16983
- Joined: Mon Oct 10, 2005 4:40 pm
AI Scenarios: What, Why and How
RHS as an entity is concieved for PBEM players (and also for FTF players) - and as in all forms of WITP - can only be
at its best in these modes of play.
However, for two different reasons, there is a major requirement for AI controlled play:
1) Players are daunted by the complexity of WITP as such, and more so by the more complex RHS system,
and they want to practice vs AI before risking humiliation vs a human opponent
2) Players sometimes can not have a human opponent for various practical reasons - but still want to play.
I. At first, it was hoped that RHSEOS could serve both players wanting a "Japan Enhansed Scenario" and also an AI controlled
Japan. Technical considerations made it clear AI could never play RHSEOS - even to the extent AI can play WITP in general.
For example, AI wasted most of the ships not otherwise wasted already by its awful programming trying to send them to or
from river ports to or from ocean ports.
II. So, second, RHSAIO was devised as a modified RHSEOS - in which technical problems of this sort were addressed. That worked,
kind of, but
a) It did not help players wanting to learn about CVO type scenarios
b) hard code does not understand the EOS Central Pacific Campaign and only gets the first phase of it right - so the Japanese end up
with a lot of wasted resources - which the Allies can eat piecemail.
This means that RHSAIO is only suitable for training - for familiarizing players with RHSEOS - and not for serious games.
III. So, third, RHSCAIO was devised - based mainly on RHSCVO instead of RHSEOS. This also works, kind of, but
it is still only suitable to familiarize players with CVO (or BBO) type scenarios. AI is not good enough and it is too weak a player to
present a challenging game.
IV. So, fourth, RHSMAIO was devised - based on RHSAIO - but with forces allocated to the Central Pacific Campaign redirected Southward.
Now we have a scenario in which Japan is so strong - and focuses its foreces where the Allies are so weak - that a serious challenge is
at last presented (although only at first: about mid war we do not know - and ALL forms of WITP fall apart late in 1944 if AI bosses Japan).
at its best in these modes of play.
However, for two different reasons, there is a major requirement for AI controlled play:
1) Players are daunted by the complexity of WITP as such, and more so by the more complex RHS system,
and they want to practice vs AI before risking humiliation vs a human opponent
2) Players sometimes can not have a human opponent for various practical reasons - but still want to play.
I. At first, it was hoped that RHSEOS could serve both players wanting a "Japan Enhansed Scenario" and also an AI controlled
Japan. Technical considerations made it clear AI could never play RHSEOS - even to the extent AI can play WITP in general.
For example, AI wasted most of the ships not otherwise wasted already by its awful programming trying to send them to or
from river ports to or from ocean ports.
II. So, second, RHSAIO was devised as a modified RHSEOS - in which technical problems of this sort were addressed. That worked,
kind of, but
a) It did not help players wanting to learn about CVO type scenarios
b) hard code does not understand the EOS Central Pacific Campaign and only gets the first phase of it right - so the Japanese end up
with a lot of wasted resources - which the Allies can eat piecemail.
This means that RHSAIO is only suitable for training - for familiarizing players with RHSEOS - and not for serious games.
III. So, third, RHSCAIO was devised - based mainly on RHSCVO instead of RHSEOS. This also works, kind of, but
it is still only suitable to familiarize players with CVO (or BBO) type scenarios. AI is not good enough and it is too weak a player to
present a challenging game.
IV. So, fourth, RHSMAIO was devised - based on RHSAIO - but with forces allocated to the Central Pacific Campaign redirected Southward.
Now we have a scenario in which Japan is so strong - and focuses its foreces where the Allies are so weak - that a serious challenge is
at last presented (although only at first: about mid war we do not know - and ALL forms of WITP fall apart late in 1944 if AI bosses Japan).
-
el cid again
- Posts: 16983
- Joined: Mon Oct 10, 2005 4:40 pm
Late war problems with AI control (ALL forms of WITP)
Between 1 Aug and 1 Nov 1944, AI begins to assign Japanese air groups as kamakazes. By the end of this period, ALL Japanese air groups
are in this status (except on the first day they appear - and they convert in one day). This means Japan has no fighters able to fly
CAP or intercept missions, or escort missions. It means Japan has no bombers able to fly land or sea type missions of any sort -
other than kamakaze attacks - from which no aircraft return. It means Japan has no recon planes, no ASW planes, no transports, - and that it
in effect has no air forces. This is not hiatorical, but it also is not optional. It is only a problem if AI controls Japan. But IF AI controls
Japan, it always is mandatory.
The only solutions to this problem are:
1) stop play about 1 August 1944
2) Find a human player for an interesting late war situation from 1 August 1944
3) Play the Japanese yourself from 1 August 1944 - and mentally "switch sides" and "forget" what you know about the Allies
- once every turn.
4) Play the Japanese yourself in a minimialist sense, using rules: for example
Tell the game you are playing face to face instead of Computer Controlled Japan
Write a number (i.e. one per day) of units you will pick to turn into kamakaze units each turn
and use random numbers (die generated) to select the unit (i.e. roll three percentile dice - one color for hundreds,
one color for tens, one color for ones) - and let this determine which unit converts to kamakaze status each day. If
the unit was already selected, ignore the result. If the unit is at a base or ship with another kamakaze unit,
ignore the result. Do nothing else - end the Japanese turn - and do the Allied turn. If a unit has a number without
an air unit in that slot, ignore the result.
are in this status (except on the first day they appear - and they convert in one day). This means Japan has no fighters able to fly
CAP or intercept missions, or escort missions. It means Japan has no bombers able to fly land or sea type missions of any sort -
other than kamakaze attacks - from which no aircraft return. It means Japan has no recon planes, no ASW planes, no transports, - and that it
in effect has no air forces. This is not hiatorical, but it also is not optional. It is only a problem if AI controls Japan. But IF AI controls
Japan, it always is mandatory.
The only solutions to this problem are:
1) stop play about 1 August 1944
2) Find a human player for an interesting late war situation from 1 August 1944
3) Play the Japanese yourself from 1 August 1944 - and mentally "switch sides" and "forget" what you know about the Allies
- once every turn.
4) Play the Japanese yourself in a minimialist sense, using rules: for example
Tell the game you are playing face to face instead of Computer Controlled Japan
Write a number (i.e. one per day) of units you will pick to turn into kamakaze units each turn
and use random numbers (die generated) to select the unit (i.e. roll three percentile dice - one color for hundreds,
one color for tens, one color for ones) - and let this determine which unit converts to kamakaze status each day. If
the unit was already selected, ignore the result. If the unit is at a base or ship with another kamakaze unit,
ignore the result. Do nothing else - end the Japanese turn - and do the Allied turn. If a unit has a number without
an air unit in that slot, ignore the result.
-
el cid again
- Posts: 16983
- Joined: Mon Oct 10, 2005 4:40 pm
Starting an RHS AI scenario
RHS requires that aircraft plants in R&D status have repairs turned off. When the game starts - and periodically thereafter -
go to each location with aircraft plants and see if any plants have R&D after the type name? If so, turn repair for that
aircraft to no if it is yes. Turn it to yes if it is no. This will insure that initial production is not wrong in many cases, and that
production ramps up in a close to reasonable way.
go to each location with aircraft plants and see if any plants have R&D after the type name? If so, turn repair for that
aircraft to no if it is yes. Turn it to yes if it is no. This will insure that initial production is not wrong in many cases, and that
production ramps up in a close to reasonable way.
-
Buck Beach
- Posts: 1974
- Joined: Sun Jun 25, 2000 8:00 am
- Location: Upland,CA,USA
RE: Late war problems with AI control (ALL forms of WITP)
ORIGINAL: el cid again
4) Play the Japanese yourself in a minimialist sense, using rules: for example
Tell the game you are playing face to face instead of Computer Controlled Japan
Write a number (i.e. one per day) of units you will pick to turn into kamakaze units each turn
and use random numbers (die generated) to select the unit (i.e. roll three percentile dice - one color for hundreds,
one color for tens, one color for ones) - and let this determine which unit converts to kamakaze status each day. If
the unit was already selected, ignore the result. If the unit is at a base or ship with another kamakaze unit,
ignore the result. Do nothing else - end the Japanese turn - and do the Allied turn. If a unit has a number without
an air unit in that slot, ignore the result.
Anytime I have tried to intercede in a game in this manner, I have found that the computer will not plan any moves for that turn. If only done occasionally it doesn't seem to important. However, if it had to be done daily to prevent the auto Kamikaze conversion no further computer orders would be generated for the remainder of the game.
-
Buck Beach
- Posts: 1974
- Joined: Sun Jun 25, 2000 8:00 am
- Location: Upland,CA,USA
RE: Starting an RHS AI scenario
ORIGINAL: el cid again
RHS requires that aircraft plants in R&D status have repairs turned off. When the game starts - and periodically thereafter -
go to each location with aircraft plants and see if any plants have R&D after the type name? If so, turn repair for that
aircraft to no if it is yes. Turn it to yes if it is no. This will insure that initial production is not wrong in many cases, and that
production ramps up in a close to reasonable way.
I am not quite sure I understand. I do initially turn all the R&D factories to "No". When should they be turned back to "Yes"? Also, are you only talking about the Allied side factories?
-
el cid again
- Posts: 16983
- Joined: Mon Oct 10, 2005 4:40 pm
RE: Late war problems with AI control (ALL forms of WITP)
ORIGINAL: Buck Beach
ORIGINAL: el cid again
4) Play the Japanese yourself in a minimialist sense, using rules: for example
Tell the game you are playing face to face instead of Computer Controlled Japan
Write a number (i.e. one per day) of units you will pick to turn into kamakaze units each turn
and use random numbers (die generated) to select the unit (i.e. roll three percentile dice - one color for hundreds,
one color for tens, one color for ones) - and let this determine which unit converts to kamakaze status each day. If
the unit was already selected, ignore the result. If the unit is at a base or ship with another kamakaze unit,
ignore the result. Do nothing else - end the Japanese turn - and do the Allied turn. If a unit has a number without
an air unit in that slot, ignore the result.
Anytime I have tried to intercede in a game in this manner, I have found that the computer will not plan any moves for that turn. If only done occasionally it doesn't seem to important. However, if it had to be done daily to prevent the auto Kamikaze conversion no further computer orders would be generated for the remainder of the game.
You are correct. However, losing all the air forces is a game killer. No orders is better than that. I think other solutions are better: find a human player or play both sides in detail yourself. This is a fundamental design flaw but I don't think they will ever fix it - although it would be easy to do and might be in the final WITP patch - I will be surprised if it is.
-
el cid again
- Posts: 16983
- Joined: Mon Oct 10, 2005 4:40 pm
RE: Starting an RHS AI scenario
ORIGINAL: Buck Beach
ORIGINAL: el cid again
RHS requires that aircraft plants in R&D status have repairs turned off. When the game starts - and periodically thereafter -
go to each location with aircraft plants and see if any plants have R&D after the type name? If so, turn repair for that
aircraft to no if it is yes. Turn it to yes if it is no. This will insure that initial production is not wrong in many cases, and that
production ramps up in a close to reasonable way.
I am not quite sure I understand. I do initially turn all the R&D factories to "No". When should they be turned back to "Yes"? Also, are you only talking about the Allied side factories?
No. BOTH sides factories should be no if a plane (or in Japan engine) is in R&D status (three engines are).
They should be turned back to yes when the plane loses its R&D status - so production will then ramp up.
For AI we must ignore the engines - since no one will turn them on. For planes you can tell because R&D is in the plane name
in production reports.
There are many other things that should be set which are not set in the editor. Not setting these things means Japan is very
inefficient. And the Allies must either spend a lot of time setting things or be inefficient too - at start. Things like building up
everything everywhere - every level of build costs 1000 points of supply per level for each different thing for example.
Task forces will not follow each other. Stuff like that.
-
Buck Beach
- Posts: 1974
- Joined: Sun Jun 25, 2000 8:00 am
- Location: Upland,CA,USA
RE: Starting an RHS AI scenario
ORIGINAL: el cid again
There are many other things that should be set which are not set in the editor. Not setting these things means Japan is very
inefficient. And the Allies must either spend a lot of time setting things or be inefficient too - at start. Things like building up
everything everywhere - every level of build costs 1000 points of supply per level for each different thing for example.
Task forces will not follow each other. Stuff like that.
I have always limited the expansion of bases and facilities from the Game Options menu but had no idea that this would apply to the computer player. I know for a fact that Japan facilities do expand in my games from following the turns as they play. I would like other suggestions.
Now, another issue that tweaking the a/c production factories has surfaced and I am puzzled here. There are at least 3 Allied a/c on 12/07/1942 that are listed as r&d that are shown with zero production points accumulating. They are the A-20G and P-38J at Los Angeles and the B24J at both San Diego and Salt Lake City. After the 12/7 turn on 12/08 these show the factories still have zero although the 12/07 turn the factories were toggled to yes. Questions: 1) why are they still zero after the turns execution 2) If they are zero and the switch is toggled to No, how will they ever get out of r&d status? 3) Why are some a/c r&d factories set at different active numbers from zero up to 12?
I realized my not knowing the game mechanics may be the problem here.
Added after I re-read the manual's a/c production section. Will future a/c not accumulating r&d ever be produced?
-
el cid again
- Posts: 16983
- Joined: Mon Oct 10, 2005 4:40 pm
RE: Starting an RHS AI scenario
ORIGINAL: Buck Beach
ORIGINAL: el cid again
There are many other things that should be set which are not set in the editor. Not setting these things means Japan is very
inefficient. And the Allies must either spend a lot of time setting things or be inefficient too - at start. Things like building up
everything everywhere - every level of build costs 1000 points of supply per level for each different thing for example.
Task forces will not follow each other. Stuff like that.
I have always limited the expansion of bases and facilities from the Game Options menu but had no idea that this would apply to the computer player. I know for a fact that Japan facilities do expand in my games from following the turns as they play. I would like other suggestions.
REPLY: AI seems to have one setting: expand is good. So it always sets everything to expand - period. This thread IS a suggestion: IF I do turn one for Japan - which you as Allies never get to do anyway - THEN I can set some (many) places NOT to expand - and AI does honor such settings. This means Japan won't waste supply points grossly - and has more for useful stuff.
Now, another issue that tweaking the a/c production factories has surfaced and I am puzzled here. There are at least 3 Allied a/c on 12/07/1942 that are listed as r&d that are shown with zero production points accumulating. They are the A-20G and P-38J at Los Angeles and the B24J at both San Diego and Salt Lake City. After the 12/7 turn on 12/08 these show the factories still have zero although the 12/07 turn the factories were toggled to yes. Questions: 1) why are they still zero after the turns execution 2) If they are zero and the switch is toggled to No, how will they ever get out of r&d status? 3) Why are some a/c r&d factories set at different active numbers from zero up to 12?
I realized my not knowing the game mechanics may be the problem here.
Added after I re-read the manual's a/c production section. Will future a/c not accumulating r&d ever be produced?
A-20 G start of production 1 43.
B-24J start of production 9 43.
P-38 J start of production 9 43.
1) NONE of these should be set to "yes" in 1941 or 1942 - or in 1943 until 1 / 9 / 43 for the last two.
2) IF you set "yes" you will get TOO MUCH production when production begins.
Wether the number repaired goes up depends on supply points in the hex - but it happens sooner or later - unless you have consumed
too much supply in that hex. We do NOT want that - because we want the planes to start at 0 or 1 - when production begins. If you do
not honor this, you have defeated the work to make production ramp up - and you may as well just set a daily rate - which comes in at
that rate when production begins (as in stock and all non RHS mods). This reform was not my idea - but was advocated by several for
over a year before I did the (monsterously large) labor to change it. Lots of work to get this right - and lots of compromises too - it is not
an ideal system. But the reformed system has two advantages: many types (not all) ramp up; and many types sunset (so you don't keep producing ancient junk until 1946).
-
Buck Beach
- Posts: 1974
- Joined: Sun Jun 25, 2000 8:00 am
- Location: Upland,CA,USA
RE: Starting an RHS AI scenario
ORIGINAL: el cid again
I have always limited the expansion of bases and facilities from the Game Options menu but had no idea that this would apply to the computer player. I know for a fact that Japan facilities do expand in my games from following the turns as they play. I would like other suggestions.
REPLY: AI seems to have one setting: expand is good. So it always sets everything to expand - period. This thread IS a suggestion: IF I do turn one for Japan - which you as Allies never get to do anyway - THEN I can set some (many) places NOT to expand - and AI does honor such settings. This means Japan won't waste supply points grossly - and has more for useful stuff.
By all means this would be great then.