AU Mod: Re-balancing the power of Armies

Aqualung71 · February 21, 2004, 22:23

It's quite clear that the AI is not able to recognise and exploit the hugely increased power of Armies in C3C. While the AI does still build Armies from time to time, it appears to be on an ad-hoc basis; often the unit make-up does not make too much sense, and you may never even see them being used.

By contrast, the human player now uses Armies to an even greater extent, probably builds the Military Academy and Pentagon more often and earlier than in PTW, and will often be able to move quickly through a whole AI continent just with a few Armies and artillery support.

The change in the use of MGL's in C3C has also resulted in the building of many more armies by human players, with often only one used for a FP build and all others for armies. The AI clearly does not follow this strategy.

If it is considered desirable to re-balance the power of Armies in an AU Mod (or to encourage Firaxis to deal with it in a patch if it is too game-changing for AU), perhaps we could start collecting ideas here.

Some things the panel may consider (subject to these being "do-able") would be:

Removing some of the new features such as blitz/extra movement points/faster healing rate;
Promoting earlier AI builds for M.Academy and Pentagon (and possibly Heroic Epic);
Promoting Armies on the AI Build queues;
Giving Armies a limited life ("use-by date").

Ideally, the AI should be encouraged to use battle-generated MGL's to build Armies, but I have no idea whether this can be done. Along the same lines, is there any way we can influence the unit types the AI places in its Armies?

More ideas?

ducki · February 21, 2004, 23:02

Quote:

Originally posted by Thriller
- Promoting earlier AI builds for M.Academy and Pentagon (and possibly Heroic Epic);
...
Ideally, the AI should be encouraged to use battle-generated MGL's to build Armies, but I have no idea whether this can be done.

One solution that was proposed and, I believe, voted down for a change in the AU, was to remove the Requires Army flag from the M.Academy. I won't hide that I am a proponent of this idea.

The biggest barrier to the AI building armies as opposed to the human has little to do with actually deciding what to use a MGL for; instead, I believe the biggest barrier is the way the AI fails to "Get the Most out of your Elites".

We humans tend to prioritize Veterans and coddle our Elites, somewhat manipulating the RNG for our benefit. The AI doesn't and will never generate as many MGLs as the human (over time, multiple games). The AI is, therefore, at a distinct disadvantage for both Armies AND the FP.

Somehow I'd like to see the AI given some sort of flag that allows them to build the Academy without having to generate an MGL and choose to build an Army with it AND send it into battle and win.

Aqualung71 · February 21, 2004, 23:18

Quote:

Originally posted by ducki

One solution that was proposed and, I believe, voted down for a change in the AU, was to remove the Requires Army flag from the M.Academy.

Your argument akes sense to me. If implemented, this should probably also be in conjunction with a "build often" flag for Armies. Then the issue would be how to get the AI to put the right units into Armies, and then to use them properly.

So, are we fighting a losing battle here (no pun intended) by trying to help the AI with Armies? Should we be concentrating more on reducing the attractiveness of Armies for the human player? And therefore take some of the fun out of the game

nbarclay · February 22, 2004, 00:38

There's been some testing that showed that the AI is perfectly willing to squander MGLs on truly stupid rush builds instead of using them for armies. If Firaxis would fix that "feature," AI armies would at least show up occasionally.

Jaybe · February 22, 2004, 01:13

You could always (ahem) remove Hurry Improvements from the Special Abilities section in the Editor. The problem is that in the Editor there is only ONE Leader type listed; I wouldn't know if it applies to just MGL, SGL or both (with hardcoded-only differences).

Yes, I DO appreciate the small wonder (FP) problem ....

Aqualung71 · February 22, 2004, 18:47

Different issue I know, but while I think of it, would there be any interest in seeing a return to the Civ2 rule of losing half your accrued shields for a switch? This would hinder the human's pre-build abilities. Why was this changed in the first place?

Dominae · February 23, 2004, 13:53

One fix for Armies that I would like to see in the AU mod:

Increase the # of cities required to support an Army from 4 to 8.

The reasoning is as follows:

1a. The AI does not build many Armies, so restricting the number it can build has little effect.

1b. The human player does build Armies, so restricting the number of Armies he or she can build has the intended effect of reducing their power.

2a. An AI with few cities (under 8) should not be building an Army via the Miliary Academy. With respect to MGLs, it has been shown that the AI typically rushes improvements with them instead building Armies, so eliminating the already minor chance that a small AI empire will build an Army with a MGL should have little effect.

2b. It is very rare for a human player to have under 8 cities, since he or she has the option of using tight city-spacing. Thus players still get to play with at least one Army and benefit from its effects (Heroic Epic, Military Academy, etc.).

The only argument that I can think of against this change is that Armies are no longer an option in 5CC games. To that I say: boo-hoo. We can always make a seperate mod for those 5CC players out there.

Dominae

vmxa1 · February 23, 2004, 14:43

How about the first one can be as it is now and then go to the increased number of cities for each additional army.

Being able to build an army with less than 8 cities has saved me at deity a few times.

ducki · February 23, 2004, 14:59

Quote:

Originally posted by Dominae
One fix for Armies that I would like to see in the AU mod:

Increase the # of cities required to support an Army from 4 to 8.

Is this a hard number, or does it scale for mapsize like the OCN? Is there an adverse effect on Small and Tiny maps?

Quote:

2a. An AI with few cities (under 8) should not be building an Army via the Miliary Academy. With respect to MGLs, it has been shown that the AI typically rushes improvements with them instead building Armies, so eliminating the already minor chance that a small AI empire will build an Army with a MGL should have little effect.

I recall alexman's test, but I forget if he gave them enough cities to support an army...
Also, even if the AI tends toward improvement rushing, wouldn't further limiting the already seemingly small number of circumstances under which the AI would normally build an Army exacerbate the problem of not having AI armies?

Quote:

2b. It is very rare for a human player to have under 8 cities, since he or she has the option of using tight city-spacing. Thus players still get to play with at least one Army and benefit from its effects (Heroic Epic, Military Academy, etc.).

Some players. Specifically, warmongers with a bit of RNG love. Yes, it's an old argument, but I felt the need to clarify your statement to point out the special-case-ness of it. Sue me.

Quote:

The only argument that I can think of against this change is that Armies are no longer an option in 5CC games. To that I say: boo-hoo. We can always make a seperate mod for those 5CC players out there.

This might fall in with my first question about mapsize. Not sure.

I need to go back and reread alexman's test, but if the AI didn't have enough cities, it now makes sense that it rushed warriors.

Also, I feel that any discussion of Armies is inherently intertwined with discussion of the Military Academy. If we cut the number of buildable armies in half, what does that do to the value of the MilAcad?

Another thought on nerfing army power - what about reducing the number of units they can hold by 1? I think that would be a very subtle change with larger effects than you would initially think - assuming it's editable.

alexman · February 23, 2004, 15:19

The Army transport capacity is editable (just like the trasport capacity of any other unit), and it's a good idea.

Another idea (which I like better) is to make it so the unit spawned from a battle is an Army instead of a Leader. That would force the AI to use it, and with the power of Armies in C3C, the improvement-rushing ability would not be a big change in gameplay.

However, I think that Firaxis will eventually fix the problem by the final patch, so I'm not eager to make big changes in the meantime.

PS: To prove that the AI doesn't use leaders for Armies I reduced the required number of cities for an Army to 1. What, do you think I'm some sort of newbie?

ducki · February 23, 2004, 15:46

Quote:

What, do you think I'm some sort of newbie?

You're not?

Naturally, no, your tests are always high-quality, but it's been a while since I read it, we were all confused as to why the AI would rush a Warrior, and even the most experienced handyman sometimes forgets to check if a "broken" appliance is plugged in.

alexman · February 23, 2004, 16:10

Forget the battle-created army idea. I just tried it and you can't generate an army if you already have one (just like with leaders now), so you would never be able to get a second army from a battle unless the first one dies.

--Alexnewb.

ducki · February 23, 2004, 16:23

We should see if we can get Firaxis to prioritize this issue, and maybe give the AI a priority list of what to do with a MGL.
Army then Small Wonders then BuildOften Improvements then other improvements.
Actually, maybe just Army->SW->hold it/disband outdated army in SW-building city and replace Army.

I think the big problem is, the City Governor is isolated and to have a MGL try to influence the Governor is probably not "possible" - I think I read that somewhere.

Anyway, thanks for trying that out, alexman. I'll have to try out smaller armies after my current game. Assuming I can even play dark enough to get any.

Dominae · February 23, 2004, 19:09

Quote:

Originally posted by ducki
Is this a hard number, or does it scale for mapsize like the OCN? Is there an adverse effect on Small and Tiny maps?

It's a hard number.

Quote:

Also, even if the AI tends toward improvement rushing, wouldn't further limiting the already seemingly small number of circumstances under which the AI would normally build an Army exacerbate the problem of not having AI armies?

If the AI builds an Army with a MGL, say, 25% of the time (sound reasonable?), then you're very lucky indeed to see an AI Army given how "bad" the AI is at generating that MGL in the first place. Even then, take the (say) 5% of games where the AI does generate a MGL, multiply by the chances that the AI does not have eight cities or more, and you can see that the case you're talking about is very rare. I see no problem with just eliminating it.

Quote:

Some players. Specifically, warmongers with a bit of RNG love. Yes, it's an old argument, but I felt the need to clarify your statement to point out the special-case-ness of it.

Yes, some players, specifically the Warmonger and Hybrid players (what does a Builder want with Armies anyway?!). The discussion here is about the power of Armies relative to the AI, not the Military Academy. It may or may not "balance" Armies if Warmongers and Builders alike could build them, but that does not solve the problem that they're just too darn good versus the AI (in fact, such a change would amplify the problem).

Quote:

This might fall in with my first question about mapsize.

Mapsize is a valid concern. However, note that Armies are better the smaller the map; four cities and an Army on a Tiny map and you can conquer the world. Thus I feel it's not a big deal to require the owner of an Army on Tiny or Small maps to invest a non-trivial number of cities. This increases strategic options (i.e. you need to backfill or conquer more land if you want to build that Army).

Quote:

If we cut the number of buildable armies in half, what does that do to the value of the MilAcad?

It reduces it. Perhaps a cost reduction would be in order.

Quote:

Another thought on nerfing army power - what about reducing the number of units they can hold by 1? I think that would be a very subtle change with larger effects than you would initially think - assuming it's editable.

This is indeed a workable solution that would have the desired effect. I dislike it on aesthetic grounds ("Hey look, those two Immortals are teaming up to become the Super-Duper Immortal Duo!!"). I'm not even sure smaller Armies would be that much weaker; the AI would still avoid them, and they would win nearly every battle.

Dominae

Theseus · February 23, 2004, 20:58

/me is naked and shivering in a corner of the room. "NOOO!!! NOOO!!! Leave Armies alone... WAAAAAAA!!!"

Seriously, Armies as they are are a bit much.

* I have asked all over the place for examples of C3C AI civ Armies, and have not seen one confirmed proof... I am operating under the belief, then, that AI civs do not use MGLs for them, period.

* Like alex, I presume that Firaxis will address this in the final patch (and have directly and repeatedly discussed this with a number of Firaxians), and thus am hesitant to go overboard tweaking to much at this time.

* And, as always, I like to go slow.

How about 1 Army per 6 cities?

Aqualung71 · February 23, 2004, 21:31

Seems the consensus of opinion to date is that:
[list=1][*] Armies are too strong for the human, largely because the AI doesn't use them;[*] It would be difficult to encourage the AI to use Armies more often because of the AI's tendency to use MGL's to rush buildings, and more critically, the AI's failure to be able to generate sufficient MGL's in the first place;[*]The way to address the issue is therefore to reduce the power of Armies rather than encouraging the AI to use them more; but[*] Firaxis has already flagged an interest in re-visiting the power of Armies in the next patch, so we should wait to see what they do before deciding on any changes for the AU Mod.[/list=1]

The most practical changes (from an AU Mod perspective, though Firaxis will clearly be able to do more) seem to be:

Increase the city support requirement for Armies above 4;
Reducing the number of units that Armies can carry;
Possible changes to the M.Academy (such as reducing the cost), although these would be less effective and it would benefit the human player just as much or possibly more.

In the interests of achieving a desirable result from the forthcoming patch, should we try to engage Firaxis with some suggestions for Armies?

Aqualung71 · February 23, 2004, 21:33

DP

alexman · February 23, 2004, 22:05

I don't agree that the AI doesn't generate enough MGLs. They usually have many more units than the human, so they compensate for their poor Elite management by sheer number of battles.

So that means that I don't agree with item 3 above. We should reduce the power of Armies only as a last resort, if we can't make the AI use their leaders to make Armies.

I think there is a way to create a separate leader unit in the editor (military leader) that gets generated from battle, and has combined Army/Leader characteristics. The human would use this new leader unit as he does MGLs now, but the AI would use it to load units, since it would have the AI Army strategy, not the AI leader strategy. The downside would be that the AI would be able to have only one of these units at a time, but this is for sure better than what we have now, where the AI has zero Armies.

Theseus · February 23, 2004, 22:28

Quote:

Originally posted by alexman
I think there is a way to create a separate leader unit in the editor (military leader) that gets generated from battle, and has combined Army/Leader characteristics. The human would use this new leader unit as he does MGLs now, but the AI would use it to load units, since it would have the AI Army strategy, not the AI leader strategy. The downside would be that the AI would be able to have only one of these units at a time, but this is for sure better than what we have now, where the AI has zero Armies.

??

So now there would be three Leader-types?

As you say, GOOD, if this gives the AI civs a shot at at least one Army.

PITA, though, for the AU Mod read me, and making sure newer players understand WTF that wacko Leader is, but that is not the end of the world.

And the third Leader would interact with MGLs how (in terms of generation and co-existence)?

alexman · February 23, 2004, 22:57

Right now there is just one leader unit in the editor that acts as a MGL or SGL depending on how it was generated. My thinking is to separate the two leader types in the editor, although I'm not 100% sure it will work yet.

Tall Stranger · February 24, 2004, 13:02

This may be putting the cart before the horse, but I think we also need to get a better sense of what the AI would do with an army. Would it do something stupid, like put an Med Inf, Horseman and Warrrior in it? Would it fill it with defensive units and leave it in rear areas? Or would it put its best attacking units in there and use it offensively?

Has anyone given the AI an army and watched how it uses it?

Theseus · February 24, 2004, 20:50

I had seen a number of pretty dumb AI Armies in PTW (usually veeeery early, when the number of units was quite limited), but for the most part they were darn good.

ducki · February 25, 2004, 00:53

Was that stock or primarily with the AU mod, Theseus?
(Edit: Yes, this is a leading question.

)

Theseus · February 25, 2004, 01:25

Both, actually, not just my own games but paying attention to reports by others.

I don't think that the PTW AU Mod did anything in particular to enourage AI Armies, other than fostering stronger AI civs.

alexman · February 26, 2004, 11:07

Well, it turns out that it's possible to separate military from scientific great leaders:

New unit: Military Great Leader. Appears during combat instead of Leader. Has ability to rush improvements (not Great Wonders), to build Armies, and also has the properties of an Army. The AI uses it as an Army. The human uses it as a normal MGL. The normal leader unit still appears unchanged as a SGL, but it would have to require a non-existant resource (easy to do) so you can't build one.

The problem is that the AI still would be able to have just one of these units at a time, while the human would use them to build potentially multiple Armies. Also, any unit with Army properties has just one movement point when empty, whereas we are used to having leaders with 3 movement points. Finally, it gets confusing for first-time users of the mod if they load units into the MGL instead of building an Army. I don't like the idea after all. It's clumsy.

Instead, I would support the idea to make Armies the combat-generated unit and to give them the finish improvement ability. That way both the AI and the human can have just one Army at a time before the Military Academy. Deciding which units to put into your one-and-only early army would certainly make for an interesting strategic decision.

So what is better: Removing the aspect of Elite management from the game after the point where you already have an Army in the field, or leveling the playing field between human and AI?

ducki · February 26, 2004, 11:32

How about another option - can we have both Armies AND MGLs with the Generated From Combat flag?
If so, what happens? Do they alternate? Would it always be the first one on the list? Would it be random?

I really don't like either of the ideas above, alexman. If we just want the AI to have armies, we could start each AI with an army or let the palace generate 1 army every 3000 years, ensuring the AIs get at least one army at 1000BC - late enough that they don't just put the starting warrior in it(which they do, I'm running a test) and early enough that they could use it in the post-expansion phase.

They still have to decide to put offensive units in and actually use it on offense(they don't seem to with the 4000BC single warrior army) in order to be able to build the Epic and Academy, but then, so does the human. Builders could save it to be used as a strong defender from hell and Warmongers could use to it for either their Sword UU or Knights or hang on to it for late-game conquest.

I'm still running my test on the AI (but not the human) starting with an Army in 4000BC to see if they actually get used once expansion is over, but in the first war, neither AI moved the army from MP duty. Yet.

Tall Stranger · February 26, 2004, 11:34

I strongly oppose this idea, as it takes away one of the fundamental benefits of a warmongering strategy and thus reduces strategic choice. Only being allowed to have 1 army in the critical early-to-mid game period is a massive change to stock and certainly does not retain the "feel" of standard Civ3.

alexman · February 26, 2004, 11:39

Quote:

Originally posted by ducki
How about another option - can we have both Armies AND MGLs with the Generated From Combat flag?

Not possible, sorry.

Quote:

we could start each AI with an army

This is possible only in scenarios, but not in a general mod.

Quote:

let the palace generate 1 army every 3000 years

This is not possible either. You can set it to generate an Army every 80 turns (which is what 1000 BC is from the start), but that would make the Military Academy pointless - a big change from stock.

By the way, my bet is that the AI will use the Army to put whatever units they have available at the time.

alexman · February 26, 2004, 11:43

Quote:

Originally posted by Tall Stranger
I strongly oppose this idea, as it takes away one of the fundamental benefits of a warmongering strategy and thus reduces strategic choice.

The fundamental benefit of warmongering versus building is that you beat up the AI while gaining territory. This doesn't change.

Armies as currently implemented in C3C are a human-only toy that just makes warmongering ridiculously easy, even against technologically and numerically superior AIs. That's what reduces choice!

ducki · February 26, 2004, 11:45

Quote:

This is not possible either. You can set it to generate an Army every 80 turns (which is what 1000 BC is from the start), but that would make the Military Academy pointless - a big change from stock.

How so?