A Modest Proposal to Examine the Japanese Economy

[PaxMondo]

I should add, following Andy's direction from some time back I always play on "Hard".  So my results (no issues) are with setting at "Hard".  A rare "Very Hard" turn thrown in on occasion, also per his suggestion.  [:D]

[DD696]

Numdydar - hey, you have it right there. It was collapsing, did collapse, and I see evidence as I had described that it still collapses. My "definition" of collapsing means that the various resource types are not available in Japan to power the economy and the economy begins to barely hobble along. Probably the crutches are what prevents it from doing so if they do the job they are meant to do.

My intent is to see if the oil, fuel, resources, ect., are being transported by the AI to those places that require them. My initial premise to do this on a hard setting may have been wrong and I perhaps should have started testing using the historical setting. The interesting time will arrive once Japan has expanded her empire from about May 42 on. It will probably be necessary to do these tests using different difficulty settings to see what the differences are.

I realize that there have always been, and always will be, those who wish to burn the books and suppress knowledge. I feel that if my research into this reveals deficiencies, then there may be ways to look for various solutions. As I have stated, I do not expect that changes will be made to the code to correct any shortcomings (I am a very realistic AI player and am very aware of the atmosphere here), however it may show ways that I may be able to improve games I play against the AI through the only avenue available to me, which would be data manipulation (changes outputs, locations and so forth of the resource elements).

It is the only option I see as a viable one.

[Andy Mac]

My tips for improving the AI economy (and play) for modders are as follows

1. Put LI and Resource production in small quantities in lots of bases (in stock resource produced some organic supply which I am convinced was a stock WITP AI crutch) - we removed in AE the resource produces supply and it means that the AI sometimes struggles to supply forward bases. By putting in LI on bases like Kwajalein, Majuro, Tarawa, Rabaul etc etc you enable the AI to generate subbsitence level supply generation on these bases without making it unbombable cheating so a player can still bomb the supply generation but at least the AI will not try to suicidally re supply every small base.

In Ironman the single biggest change I made to teh jap economy was to put 30 - 40 Resource and LI in lots of bases making it less likely that bases will starve without ships and giving the AI a little help.

2. If playing on historic change the outputs of HI and Refineries to create more fuel and supply - again its not free it just allows the AI to produce more for the same.

3. Tokyo only should get some free daily supply/resource/oil/fuel its the only base on the map with an out and out un bombable 'cheat' as so many LCU's/Air and Ships end up in Tokyo you need to give it some ability to generate a supply level no matter what the allies do.

4. Do not add large aircraft factories add small ones paired with a same size engine factory no more than 10 - 20 size as these appear optimal

5. Giving the AI in the HI forts already built to lvl 6 means it stops supply getting sucked into fort building.

6. Add a small Oil source to a HI base no more than 100 Oil centre but its again usefull for allowing the AI to generate some fuel.

The AI will lose a lot more ships than the allies in the early going giving the AI some extra tankers/AK's is a good thing to help it absorb casualties - another good one is to increase the range and capacity of th erespawnable barges as it again give the AI some resiliancy capacity.

There are more tips but if what you are trying to achieve is a stronger AI economy these are the kinds of things I do to make sure the economy keeps ticking over.

The main one is pairing LI and Resource Centres in forward bases as this makes a huge difference to the AI's resiliance which is what I did in Ironman

[DD696]

Andy, any tips are very much appreciated.

I have run the first 91 days thru Tracker at the hard setting. I will not post screenshots nor will I willfully reveal data that appear to be the "crutches", and I will do my utmost not to ask any questions. I haven't played scenario one since CHS came out back in the WitP days, so my familiarity with it is lacking.

It appears to me that I could have had the difficulty setting in the game I was reporting on to have been something other than hard or very hard. From the results I am seeing at the hard setting, I have to assume that I may have had it on historic.

As my goal is to see if the various resources do get moved to where they are needed, it is time to go back and redo this on the historical setting. Probably just make another copy of the game so present data is not lost. If the crutches are not present in the historical game, then this is the data that I am seeking.

[Andy Mac]

p.s. for Allies best crutch is

1. Free Fuel at Sydney in small quantities as AI is bad at stockpiling for operations (or Oil and Refinery if you prefer it to be bombable)
2. Liberal use of convoys device at key locations

e.g. certain key bases are kickoff points for Allied AI scripts having a convoy arrive and dump 300k fuel and supply in those key bases the day or week before the offisive is due to kick off can be a big help not going to list the bases or times in an open thread but there is a list at home of which bases need help on which days

[Andy Mac]

AI does better with allied economy its main issue is lack of forward dumpage when on attack

[janh]

ORIGINAL: PaxMondo
I should add, following Andy's direction from some time back I always play on "Hard".  So my results (no issues) are with setting at "Hard".  A rare "Very Hard" turn thrown in on occasion, also per his suggestion.  [:D]

I admit I play a mix of a few turns hard to beef up supply and allow especially air units to get replacements, followed by historical to fight by more realistic supply rules. Even if the AI struggles with them. I hope "historical" doesn't have other AI crutches removed besides the "always in supply benefit" present at "hard". Not sure about pilot replacements and training, for example, or need for HQ and naval/air support squads at airbases and ports. But I prefer the ability to be able to cut of supplies but pocketing moves, or isolating bases with the navy.

But good tips Andy had there. It would be a neat crutch if there were "invisible production or resource/oil center", i.e. that give benefits and can still be bombed by the player, but don't show on the overview screens and don't give such benefits to human players upon taking that base.

Can LI centers be incorporated into the scen files and become active in ongoing games, or does it need a restart?

[DD696]

Thank you again.

My cat is preventing me from sitting down and doing anything now. Obviously, his loyalty and integrity have been compromised.


Edit: We're starting to make some actual progress here. My goal is to have as good of an AI game as I can. If enough AI players truly had the interest to do so, various ways and methods could be found to improve our gaming experience. We wouldn't all agree on the same scenario we like, but it is easy to modify the location file (the one which contains the base, production and LCU information) and incorporate it into your favorite scenario with whatever changes/mods required to fit. The game can certainly be modded to the benefit of the AI player.

Andy Mac has been a godsend for the AI player, but we can help ourselves as well if we have the desire to do so.

I do.

[Shark7]

ORIGINAL: Andy Mac

My tips for improving the AI economy (and play) for modders are as follows

1. Put LI and Resource production in small quantities in lots of bases (in stock resource produced some organic supply which I am convinced was a stock WITP AI crutch) - we removed in AE the resource produces supply and it means that the AI sometimes struggles to supply forward bases. By putting in LI on bases like Kwajalein, Majuro, Tarawa, Rabaul etc etc you enable the AI to generate subbsitence level supply generation on these bases without making it unbombable cheating so a player can still bomb the supply generation but at least the AI will not try to suicidally re supply every small base.

In Ironman the single biggest change I made to teh jap economy was to put 30 - 40 Resource and LI in lots of bases making it less likely that bases will starve without ships and giving the AI a little help.

2. If playing on historic change the outputs of HI and Refineries to create more fuel and supply - again its not free it just allows the AI to produce more for the same.

3. Tokyo only should get some free daily supply/resource/oil/fuel its the only base on the map with an out and out un bombable 'cheat' as so many LCU's/Air and Ships end up in Tokyo you need to give it some ability to generate a supply level no matter what the allies do.

4. Do not add large aircraft factories add small ones paired with a same size engine factory no more than 10 - 20 size as these appear optimal

5. Giving the AI in the HI forts already built to lvl 6 means it stops supply getting sucked into fort building.

6. Add a small Oil source to a HI base no more than 100 Oil centre but its again usefull for allowing the AI to generate some fuel.

The AI will lose a lot more ships than the allies in the early going giving the AI some extra tankers/AK's is a good thing to help it absorb casualties - another good one is to increase the range and capacity of th erespawnable barges as it again give the AI some resiliancy capacity.

There are more tips but if what you are trying to achieve is a stronger AI economy these are the kinds of things I do to make sure the economy keeps ticking over.

The main one is pairing LI and Resource Centres in forward bases as this makes a huge difference to the AI's resiliance which is what I did in Ironman

I've done something similar (but still needs to be tweaked) in my no treaty mod. I try to balance it so the bases use all the resource they produce.

If you build 20 LI in a base, you need 15 resource so that you produce 300 resource and use 300 resource. This can be increased as needed...IE you can double to 40/30 etc.

I've also tried just putting in an AI crutch fuel booster for bases. Free fuel that can't be destroyed, even if it is only 25/day, it can build up over time.

[DD696]

Testing resumed, with setting on historical using the procedure as described in my first post and up to turn 117 or 4 Apr 42 and the results are so far as expected. Using latest beta q5, and we do know that there was a change made in the last few betas that would apparently affect the AI convoy system. In the game I stopped playing I was up to turn 300 with the J economy stumbling. I plan to continue running this until it becomes obvious whether the economy works or not. In the game the southern resource area is still in the process of being conquered and the mighty Houston still sits defiantly in the heart of the Philippine Islands.

The reason I am testing in this manner rather than testing AI vs AI is that I believe that in order to test something adequately it must be tested in the manner in which it will be used, i.e., myself against the AI even if I am issuing absolutely no orders. There are no submarines to oppose the J convoys. I do not know if there are any programming hooks put in which makes running AI vs AI different than the execution of me vs AI. Another problem with running AI vs AI is that you only have results to view at the time you stop the continuous execution. I have chosen 4 day cycles for now with a save at that point. I don't know at this time if it will be necessary to run single day cycles for more definitive peeks at what is going on with the J side. I wrote my first program for an IBM 1401 computer using SPS back in 1965 and quit my working life as a senior systems analyst, so I do have a vague knowledge of what I am doing.

One thing I am not sure of is whether or not any changes have been made to the execution at the historical setting. Running 4 day turns I cannot verify each days HI numbers, etc., but from what I understand the historical setting is undisturbed and reflects true game data.

There have been more people than myself now stating that the J economy did collapse back in the WitP days. Obviously, if the economy did work then the need for various crutches would not be necessary. The only reason that the economy would collapse would be from a lack of the various resource elements. Either they were not moving around in J as they should have (we know that there has been considerable work done upon improving the movement of various resource items and that it now works far better than in the past) or the needed resource elements were not getting to Japan. Everyone expects Japan's economy to collapse at some point due to hostile action, however it should not just up and die on it's own.

[chesmart]

I tested AE with computer vs computer and reached july 44 without the japanese economy collapsing. My only problem is the AI is he converts to many squadrons to Kamis.

[DD696]

Your difficulty setting is of utmost importance. The established crutches overcome any deficiency in resource element shortages.

Have now run up to 3 July 42, but have not yet run thru tracker. Time for a peek at the J side also.

On a side note, the Mighty Houston was forced to abandon Iloilo on 18 May due to a sneaky overland attack by dasdardy elements of the J empire. It sailed to Brunei where it's sailors regaled the local population with it's tales of mooning all of the passing IJN ships and the pilots of the J empire. It does appear that during June a wandering J carrier force damaged Houston and it now sits in port being attended to by a great multitude of local lasses eager to observe for themselves the superbly tanned arses of the crew of the Houston.

[DD696]

Testing has continued to turn 401. It has been decided that I can post screenshots and not give away any secrets by testing the economy on the historic difficulty setting. This setting was what I was using in my game in which I observed the faltering of the J economy. At the time I posted my concern regarding this I thought that I had been playing on the hard difficulty setting - after so many restarts I failed to remember each and every setting of the game - definitely a case of AIzheimers.

I do have save game files for every turn of my actual game as allied against the AI and I do and did load up the Japanese side into tracker to monitor it's economy. I was playing a personally modified version of Babes 28. I try not to simply go in and out trick the AI or to simply smash it where possible. I do admit that sometimes it is easy and tempting and that I have smacked around AI TF's that insist upon being targets. However, the net effect of that type of activity would be to spoil the game later on but sometimes it is so hard to resist.

I will be posting screenshots of tracker information from that game to show why I was concerned about the economy. I will also be posting screenshots of tracker information from my testing of an unmolested J economy as described in other posts using scenario one. There are differences in numbers, classes and carrying capacity of the ships between the scenarios. I was executing different beta versions of the code and MichaelM did make a change to the cargo handling programming routines in his ongoing efforts to improve the game. Results, as expected, will have differences.

Now I need to learn the art of posting screenshots. I have never done that before as I was always more interested in playing the game than posting pretty pictures.

[DD696]

From the actual game: the global view for Honshu.

[DD696]

Next, the resource type situation for Honshu:

[DD696]

Last one I have is of some of the alerts:

[DD696]

Again, my observation and question: It appears that the J economy is faltering. Should I be concerned?

Edit: At game start I did increase oil production in Japan by about a third if I remember right, and I did add extra oil, fuel and resources (I think). I'd have to go back and check, but regardless of any increases I added, this is what was happening by turn 298.

[DD696]

Now let's look at the test game:

[DD696]

2nd half of global view:

[DD696]

Now the resource history