The Military Academy

[Aqualung71]

It would be interesting to get some views from the other panel members.

[alexman]

Yes!
As well as from the rest of the AU community!

[Risa]

30 turns are too long. AI won't get enough armies by the time to show differences, and those few armies may be wasted by putting in weak units. Meanwhile human can accumulate more armies by generating MGLs, and can save armies for right thing.

I suggest 20 turns interval, if not shorter. It should be at least 15 turns, though, otherwise MiliTrad beelining will be too powerful.

[lockstep]

Re: removing the victorious army requirement and the 'improves army value' flag from the MA - I'm (obviously) for it.

Re: army-spawning MA - as long as the C3C AI won't build armies, I strongly second alexman's suggestion (even if it takes away some strategic choice for the human player). Spawning interval should be 20 or 25 turns.

[Krill]

Personally, I'd agree with almost anything that made the AI use armies, so long as we can still keep on using them, and (at least) in the same capacity in PTW and before.

[JimiD]

I think the current proposals are good, and would favour keeping armies separate from governments (MA can be linked back to a government if this was wanted).

I still argue for decreasing the 'multiplication' bonus to the Attack values that armies receive:

If they are still too powerful in the game (if the AI will not attack them) getting even one army earlier than the AI, though MGL or a beeline for MA, provides an opportunity for exploitative tactics by a human, and potentially a game ender in a MP game (limited experience here).

ie Armies should be good, but not invicible (or the equivilent), and form a part of a larger strategy when at war, not all of it.

ps good work alexman in rationalising this thread down to a proposal

[ducki]

The AI building an Army(modded)

Alright, I have to suggest "too small a sample" on your test again, Alexman. Or perhaps, "not enough information". I dunno. Maybe too many little nooks and crannies in the code for any test to be good enough to be "sure" of anything.(I'm not sure this screenshot means the AI will finish or if it will end up being a prebuild.) No offense intended, alexman.

I modded AU 1.04 to allow the Palace to enable the building of armies. (You have to also edit a txt file.) This works as confirmed by checking my build queue once I have enough Cities to support an Army.

So I'm playing a game to see what happens. I just started a war with Korea and decided to get an ally. I built an Embassy in Beijing and guess what?

[ducki]

To clarify:
I'm not impugning alexman's test one bit. I often wonder where folks that do so much editing and debug-testing find time to actually play. All I am suggesting is that perhaps there is too much unknown about build-queues and the governor and preferences and such that there may be some odd reason the AI does not tend to build Armies.

I also am not suggesting that I've found the silver bullet. This mod involves making the Palace a Small Wonder(to set the flag to enable army building) as well as editing the PediaIcons.txt file - a Wonder needs a splash graphic - so it may or may not be suitable for our "minimal" changes philosophy.

Additionally, I don't know why China is building an Army and I wouldn't even pretend to know where to look for reasons. This early in the game, 400 shields takes a long time, so he's making a Wonder-level commitment here. There's also no way to know what sort of units will end up in the Army. Past tests that involved starting the AI with an Army showed they will load a single warrior and fortify in the capitol; however, this may be simply because there is an Army and a "free", non-fortified unit on the same square and a need for defense.

All I know is, he is building an Army and it seems like a small change.

[Aqualung71]

Good pick-up ducki. And I tend to agree with you - there's too much we don't know to have any real sense of what the AI will do. The AI does some strange stuff.

For example....look at your screenshot. I'm no programmer, but my logical process goes something like this: "If government = despotism AND terrain = grassland AND no useful bonus resource in despotism THEN don't irrigate"....is that really difficult to programme? Yet, China lost a shield irrigating that grassland!

[alexman]

Good job ducki, but I'm reasonably sure that China is building an Army in Beijing because they lost a Wonder race.

I have seen the AI switch to an Army to avoid wasting the shields, but I still have not seen them build one from scratch.

[ducki]

That's entirely possible. I've been dreading the trudge through Ancient and Middle Ages to get to the point that 400 shields might not be such a huge investment just to see if any Armies would get built and therefore wasn't paying close attention to the AIs at this point.

I'll be glad to post a sav for anyone that wants it and I'll try to find an older one to see if that is what happened.

[alexman]

By the way, you don't need to make the Palace allow army builds. The AI builds the Military Academy just fine (if the victorious Army requirement is removed), and it has the same effect. The problem is that it doesn't build Armies except by a Wonder cascade.

So I think we can place this proposal under consideration:

Yes/No: Make all the following changes to the Military Academy:

• Remove victorious Army requirement.
• Remove 'increased army value' flag.
• Remove the ability to build Armies for all civilizations.
• Add ability to spawn an Army every 25 turns.
• Reduce shield cost of Army to 1.

Voting in 1 week, and we will take into consideration that the AI does build Armies in the rare case when it loses the race to a Wonder in the Military Academy city.

[ducki]

I knew that bit, alexman, and I didn't doubt that from your test. The point of my test was solely "If the AI can build Armies, will it build Armies ever?"

I'm inclined to agree with the wonder cascade theory but can't confirm (for my own peace of mind) until tonight when I can check a previous save if I have one(or do wonder initiations show up on Replay?).

At any rate, I wish we could find a solution that encouraged the AI to build Armies instead of removing the building of Armies altogether. I still think this will significantly alter both the balance and the nature/feel of the late game, but lacking a better proposal or hard evidence to support a lesser change, I withdraw my opposition to this proposal until something less, in my opinion, rebalancing is found.

[ducki]

I did double check and the Chinese army in my screenshot is, indeed, a cascade-result.

That said, I wonder if this is necessarily a bad thing. I'm going to think out loud, so bear with me.

1. Who loses out more often on a cascade, the human or the AIs(taken as a group, not any individual)?
-The AIs do.

2. If the AI doesn't have a cascade target, they lose a buttload of shields, exacerbating their bad decision effects.

3. If Armies are available very early, the main cascades - Late Ancient, Early and Middle Middle Ages - might be less painful for the AI.

4. If Armies are available very early, would the human player generally change the way they play?
-I don't think so. The uber-warmongers are going to likely get a MGL for an Army anyways and not be very likely to tie up a productive(very productive) city on a 400-shield Army. The builder would much rather have a wonder at that price, even if it's not his first choice. The hybrid might be conflicted and might end up with an Army as a cascade-target, but probably wouldn't tie up a city that could be cranking out improvements and units(remember, we're talking Ancient and Middle Ages, pre-Mil Trad).

5. The reason the AI doesn't choose to build armies after MilTrad might be due to the fact that there is too much other stuff that is a priority that late in the game. If the opportunity is presented earlier, even in the form of a cascade-target, we might see AI Armies, give the human some actual interest in the decision of whether to pursue Armies or Wonders, and not impact the game as much as it would seem at first glance.

6. To catch the Cascade-rich Eras, Army-building ability would need to be tied to a building that comes very early, can only be built in one town, and is guaranteed to be in one of the AIs top production towns. This means Palace or Forbidden Palace. Palace requires a PediaIcons.txt change, FP might be too late for the AI.

7. This leaves open the possibility of keeping the Military Academy as is. I won't expand on the potential virtues of having a second Army-building town or potential side-effects just now.


So, at first glance, it seems very radical, but when examining the actual effect it would have on the human player's decisions and experience as well as the potential for seeing AI armies, I think it is no more radical or unbalancing than the existing proposal.

Discuss. Rebutt. Ridicule. (But not just reactionary. I really think this is not so radical as your first reaction may be, so let it sink in.)

[bongo]

Much of your reasoning relies on the assumption that pre-c3c, humans would only spend one leader for armies and build additional armies with the military academy.

In c3c, the warmongerer can get armies as a by-product of fighting, and save the shields for wonders. 'Normal' wars should generate more than enough MGL to keep a human warmongerer happy.

However, humans often build large armies during peace as a preparation for war. Having two army-producing cities instead of one would be a tremendous advantage to a human.

Maybe we should just move the 'build-army' flag to the palace? It would give peaceful humans easier access to armies, but b]cause AIs to get more armies 'by accident'.[/b](which would be a good thing)