Ok. This post should capture most of my considerations on this matter, based on what discussed so far. It is a pretty intense post; so probably better you go make some tea first. Hopefully it is worth the time.
I know you are not comfortable with me making big posts, but I think this post simply covers so many matters, that it gets long.
Besides the preamble on the old options and related legacy matters, I've divided this post into different distinct parts, starting with (.); so you (and anyone) can read each of those like they would have been single distinct posts.
So, I've tested TripleA 1.9.0.0.4379 and I see that what you are going to do is:
-
Remove all current AI bonuses.
-
Have only a percent bonus, restyled as a per player bonus, under the current format:
"Russians Bonus Income Percentage", where "Russians" is the "player" name.
So, first of all, I want to say that I'm under the impression that, under the current system and in the recent past, the "flat" is not clearly less popular than the "percent", amongst AI players, and I believe a significant share of them (that might even be the majority) prefer to use the flat over the percent.
With this said, I, for one, am not a fan of the flat, and I'm fine with having this concept of AI bonus removed.
The main difference I can see is that the flat causes longer dragged on games, thus you can have a flat bonus at a great enough level for the game to be highly challenging, while being much less dramatically decided very early on then a competitive percent, as with a challenging percent either you crush the AI fast (reducing its production rapidly) or the reverse is likely to happen soon.
In particular, I guess the flat should reduce the impact of early luck in the game.
I'm mainly saying that, if you remove the flat, you should expect someone to get upset about it (I may be wrong), because he finds the "flat" a more enjoyable AI bonus. But all good for me, except that I rarely play AI, so I'm not a good reference. If anyone here likes the flat, better he speaks out for himself.
As I said, I doubt anyone will miss the att / def bonuses, instead.
I see that you plan to "eventually remove the 4 old AI properties from game XMLs", and this would be fine for all those games for which such settings are nothing more than a marginal touch.
But I don't think that this applies to any maps having them in the xml, as, in some of those, the mapmaker might have set them for some significant reasons, and may not be around (unlike @Frostion) to readily rework them. In particular, how do you plan to deal with those meant to be AI challenges? I guess this item has a minor importance, at this point, since most maps having the AI bonuses set in the xml were made and balanced for the old removed Moore AI; still, this may kill a few maps for which the AI bonus settings are critical, until anyone will rework them, of course (but we all know that the chance of this happening is very low). I don't really know what those maps are, since I very rarely play AI, and I don't think I've ever played any maps specifically built as AI challenge (I'm not saying I'm not interested in them; I actually wanted to try some, but never did, so far).
While I agree on the general principle of removing pointless properties from any xml, especially editable ones, as it is definitely bad to offer users pointless and ineffective options, this, on the other hand, would make harder to rework those maps, if anyone will want, in the future, because you won't get anymore to see what bonuses those games used to have, thus wondering if they had any, if you remove them (of course, this besides referring to older releases of these same games, but such releases are liable to become less and less worthwhile overtime (think about having now to go back to a release before 1.9)).
I'm just wondering if you are thinking just in terms of the majority of maps for PvP, that get played with AI too, while having no AI bonuses set default, but overlooking the minority of PvAI maps, and maybe a few hybrid cases (some @Frostion maps come to mind, like the Primeval version of Age of Tribes, that are not an issue, because he is around to fix them, but this does not apply to all maps in the repository), that have the AI bonuses spelled out. For example, also some of the PvP maps have some default AI bonuses, meant so to kick in just in case someone puts the AI on, without thinking about giving the bonuses, or as a mapmaker's suggestion about how to play an enjoyable game with AI.
Here I'm talking mostly in theory, as I don't really know these kind of maps (I rarely play with AI), but I'm sure there are at least a few of them in the repository. Also, sometimes maps come bundled with AI challenge mods of the main version, or have anyways versions for PvP and versions for PvAI, that might rely on old bonuses, to some extent.
Whatever the case, in the moment in which you remove all old AI properties from the xml, you lose the values they were set at, which I tend to agree is a trivial (virtually unimportant) matter in the majority of cases, but likely not all of them.
It doesn't concern me personally, as I've no maps in the repository, nor I'm interested in taking ownerships of or making changes to or anyway work on or even just play any maps having relevant AI bonuses.
I also tend to think that maps should not stop improvements at the engine level (nor anyone who improves the engine should be supposed to rework all maps having any problems because of this!), but rather getting flagged as broken, waiting for anyone to take them over and rework them.
So, no problem for me, as I think these improvement justify breaking some maps, that are always free for anyone to rework and retune, with the new better system.
The main thing I'm warning is that this is going to be a very slow process (or not done at all), thus this will probably end up with having some maps more or less broken in the repository, for the time being.
So, I think we can divide maps having AI bonuses set at default different from 0 in the xml into 2 main categories:
- PvP regular maps that have also a default "suggested" AI bonus, likely but not necessarily tailored for noobs;
- AI challenge maps that (among probably other things, like a purposely unbalanced setups) use AI bonuses as a main element of the game, plus any other kind of maps that make a not marginal or optional use of AI bonuses (for example, Age of Tribes : Primeval).
For the category number 1, let's take as example "Pact of Steel 2", no less. For what follows, I will take Pact of Steel 2 as the example of a full PvP map that happens to have some AI bonuses arbitrarily set for it, just in case someone wants to play it with AI, while surely not being meant to be played with AI.
You can see that you have "AI Bonus Income Percentage" = 20.
This is clearly a level set for noobs, that might not think of setting it themselves, before giving the map a run with AI, so to give some decent challenge for the casual player, I'm guessing (just trying to enter the mapmaker's mind, here).
I would personally not have it, and I disagree with this decision of having AI bonus set in normally full PvP maps, for some reasons, I'm not going to explain, to keep the post from being even more long (all minor considerations).
For the category number 2, I don't think any examples are needed (but there are a few). Also, I believe all AI challenges TripleA has were made for optimum gameplay with Moore AI, that it is not anymore available, and I don't think many, or even any, of them has been remade for the latest AI (I'm not sure, here), which is anyway continuously under development; so it would not really be feasible to rebalance them, at least not closely.
This might reduce the importance of the matter, since one might argue that all AI challenge games were already broken when Moore AI was removed (I don't really know, since I've never played any AI challenges).
I would defer to the players and the makers of the AI challenge games, as well as other games meant to be played with AI, any judgments on how all this would impact on them, as I know little of what I'm talking about, here.
Of course, there may be AI challenge games that do not rely on any AI bonuses, but just on their own setup / triggers / etc. and, in such cases, they are not part of the second category and the matter would be off topic (I don't really know, but I guess all those are already broken for the latest engine, if they were made for Moore AI and meant to be really challenging).
What are the conclusions?
For the category number 1 (Pact of Steel 2 having a percent of 20, Napoleonic Empires having a flat of 8, etc.) my personal opinion is in support of automatically mass-deleting such options from all maps for good, substituting them with nothing (in the xml), which is what you intend to do (and I would have suggested this anyway!).
For the category number 2, I'm surely not sure, but I guess leaving them as they are, even doing nothing, might be better (very confusing for the users either ways!), so to give a referring parameter, about how the map worked in the past (before being broken), in case anyone wants to rework them, especially in case either these deletions or the deletion of Moore AI will or did much damage them. However, this is hard to say, because leaving ineffective bonuses is liable to cause an endless series of bug reports, in the future, from people that will ask again and again why such options appear doing nothing, and it is going to just waste people time to figure out that they are indeed ineffective. Most likely, keeping the disabled properties in them, for future reference, would be a sensible move only if these maps are removed from download list too (I'm not suggesting this, since removing a map from download list is really a big step).
Admittedly, I don't really know about what to do for category number 2; so my suggestion is really just limited to 1.
Another option (I'm not encouraging nor discouraging), especially for finding a compromise for the second category, you might want to take into consideration, is not removing any AI Bonuses, but just deprecating and keeping supporting them only for maps having them set in the xml (removing only the current feature of having them automatically showing if not set). In this case, I would say it would be sensible to proceed removing all AI bonuses from category number 1, anyhow, especially to keep the options cleaner, and nothing from category number 2, of course. This would depend mostly on what is the consideration and the importance you assign towards all those maps having AI bonuses specifically set (at a value different from 0), and how much those options are important for the same games. Myself, I just don't know, since I know very little about those few maps and their popularity.
Of course, a main item to start with is how to discriminate maps between category number 1 and 2 (if you decide to threat them differently in any ways). The only sensible way I can see to do such a thing (if possible) is making a mass search of all maps having bonuses set in the xml, at any default different from 0, and posting the list here, telling that all these maps are going to have those properties deleted; then waiting for people to contribute their opinion in sorting them out. But I don't actually know how well this would work.
Of course, I'm not requesting or saying you have to do any of the above, nor I'm really personally concerned about any particular maps, as these are just general thoughts, and if you want just to go ahead mass deleting all options from both the engine and the maps, that is fine for me.
As a clarification, my category number 1 does not comprise maps having AI bonuses at 0 default only, set in the game xml; the main reason being the customisation of the "max" and "min" assigned to the option. Of course, no problems removing all AI bonus properties from these.
Now, let's go back on what it is proposed to be added, rather than on what it is proposed to be removed (I've already lingered far too long!), and the related legacy issues for maps' games.
.Starting Resources Multiplication
First of all, an important matter and a whole point I just missed (damn), when putting forward my suggestions, which may well be considered a regression we all overlooked!
In the moment in which I suggested and we agreed to move the bonus from multiplying the PUs stock before purchase to multiplying the PUs flux (the main reason here being not multiplying saved income), we are not anymore multiplying starting income! This is a functionality lost in the old one, where the income everyone had at start game would have been multiplied accordingly, right before the first purchase for that player.
I think this is quite an important matter, especially with reference to the traditional games, where the starting PUs are equal to the starting productions, thus it is nonsensical that players would get more units to buy on round 2 onwards, for the same total production, while not having any bonus on their purchases on round 1. This would have a continuity-breaking feeling as, in the normal games, the PUs the players start with is supposed to be the PUs they would have collected in the round before round 1 (most games don't start at the start of the referring war, but sometimes in the course of it, usually 1942).
So, I think you should add this point as one of the ones at the starting post of the Topic (a point I would have surely added myself, amongst the ones you pasted, had I not just overlooked it totally!):
- Having it multiplying all starting resources, assigned to the respective players (in resource initialize) (particularly important when we multiply placement abilities too).
This is particularly important relatively to the other point about the multiplication of placement capabilities, otherwise, in standard games, the placement multiplication would be not justifiable for the first turn of the benefitted players. Probably not much of a big deal, practically, but it would be a nonsensical element, in the system, making it unrefined.
Moreover, if all starting PUs are multiplied, I think this would be generally more sound for the general impact of the AI bonuses, and lowering the perceived need of assigning both the percent bonus and a bid, too, as the starting PUs multiplication would partially cover the popular balancing-through-bid concept, making the AI bonus alone more self-sufficient, at least as a matter of the feel of it.
Of course, this should apply to other resources, than PUs, just as much as the regular multiplication would.
I think this is practically a regression from the old system, that has to be corrected by assuring the multiplication of all PUs (and any resources) assigned at start game (in the resource initialize).
I actually now think that the main, or maybe only reason, why the old AI bonuses worked that strange way of multiplying what you have before purchase (instead of what you collect) was exactly to assure multiplying the starting income too! I'm not even sure if moving the multiplication from stock to flux (which I suggested) can be considered a net improvement, if we lose this dynamic.
A warning, just in case, is to avoid multiplying the PUs earned when not using some of the assigned bid (the bid getting saved should not be multiplied; only the resources in resources initialize).
.Per Player Bonuses
Personally (as it can be inferred by the fact that it was not part of my initial main suggestions), I hardly feel the need of being able of assign bonuses per player, and I would even question if that is worth the trouble of having to set multiple each time, in case you want all AI players having the same bonus. It may be considerably annoying when you play a new map, because you have to recall all the unfamiliar player names that are in the alliance supposed to be played by the AI.
To feel it, do this:
Select "Domination 1901" with AI, taking the "-Br.Sp.It.O.D" alliance, with the other 2 alliances to the AI.
Then, click on Map Options and assign a 25% bonus to all but the ones you intend to play yourself.
While I can see the point of having a per player flat bonus, I don't really see any major reasons for having per player percentage bonus; as long as there are no issues, inducing me to do otherwise, I think I would always want to just give all AI the same bonus percentage. I can't imagine any reasons, with the AI bonuses not causing issues, in which I would want to give +20% to some players and +30% to some others, instead of +25% to all.
Even in case the game is a traditional two sides one, and I'm taking only some of a side, thus playing with some AI allies against all AI enemies, I think it would still make the most sense to give the same percent bonus to all, comprising the AI players fighting on my side (as per how the current bonus work), as that bonus is meant to be a balanced representation of how the AI is inefficient in strategising, thus it should be applied to AI friend and foe alike.
But, while the arguments against having a per player AI bonus are rather minor, and maybe the biggest one is that, in the moment in which you have to multiply all, this would pretty much forcefully restrict the AI bonus to be of only one kind or, at the very most, two kinds (but probably two is already too much), I'm not seeing here in this topic any real strong argument / example / user case about why the AI bonus are good to be made per player specific.
Does anyone have any clear example in which you would feel the need of setting the AI bonus individually and differently, instead of giving +25% or whatever to just all AI players?
I'm mostly curiously waiting for someone (that it is more into AI gameplay than I am) to give me an example; not being negative on the matter for the hell of it, just a bit curious, since everyone agreed without much discussions, so I feel like everyone is seeing something obvious I'm missing.
But, sure, I guess setting bonus per player may have a bunch of good uses, and having to set the bonus per player should be bearable enough, even if you want just the same for all AI.
@Black_Elk said in AI Bonus Settings Revamp:
Sounds good. I'd say by far the most important would be the ability to assign the bonus on a per nation basis.
I'm not opposing your views but, mostly out of mere curiosity, can you explain why this is "by far the most important"? I'm not really making an argument against it; I'm just mostly curious about what exactly cases I'm overlooking, maybe other people might be wondering too.
But, at this point, we have to realise we have gone off-topic!
We are not talking anymore about AI bonuses at all!
Now these bonuses are all not-AI options, albeit made with AI gameplay in mind (just like you can give some bid to the AI; this does not make the bid itself an "AI bonus"!).
Just like with the bid, now these options can be used for whatever, related or not related with AI, just like other not-AI options.
For example, a mapmaker may decide that Player1 is a more productive civilization than Player2, thus able to get more income out of the same land...
Of course, I'm being just formal here, and I surely realise that these bonuses will be almost always used only for AI players, but they are not anymore "AI Bonus" themselves; just options that you may or may not use for any AI related reasons (like other ones).
.Similar Income Multiplication Property
About this, I want to put briefly out a warning, here, about the co-existence of this option (as per "pact_of_steel_2.xml"):
<!-- Multiply PUs will multiply all PUs gained or lost during a turn. It will not yet multiply costs of units or any other costs, or starting PUs -->
<property name="Multiply PUs" value="1" editable="false">
<number min="1" max="10"/>
</property>
What you are creating are, for now, a per-player, percent and positive-only version of the above option.
This is just a side note, and my suggestion is to leave this "Multiply PUs" option just as it is (keeping it, of course), without any changes.
I've tested it, and your new option changes the income based on the flux after it has been multiplied by the above option, already, as it should be (for example, if "Russians Bonus Income Percentage"=50 and "Multiply PUs"=2, Russians will get 3 times the money, which surely seems the most correct behaviour, to me).
So, all good here, it seems that your option works fine with the existent "Multiply PUs", which is a rarely used, but good to have, property.
"Multiply PUs" is a handy well rounded option, that I've rarely tested, but I recall that it worked totally fine.
I'm thinking to use it for a map with maintainment costs, since when you have maintainment costs, you need much more income per placement, to have some reasonable placement limits (thus I've in mind to use it at "6", so that a production 1 territories gives placement 1 but 6 PUs). The alternative would be setting different unit placement and income production for each single territory.
This was just a notice about the existence of this similar option, and a confirmation that it appears working just fine with your new ones. So, all good, here.
Eventually, your options will become more and more dissimilar from this one, the more you add features (in particular multiplying other resources), while this option should stay as it is.
.Placement Multiplication
Let's retake the matter of the placement abilities multiplication. Not actually sure if you got convinced about that, and will go on implementing it (I surely still suggest it), but, to answer some considerations on the matter:
@Black_Elk said in AI Bonus Settings Revamp:
I'm not entirely sure I understand the increased placement thing. Do you mean increasing the production capacity of territories under the AI's control? Like a territory that would normally build 10 units could then produce 15 units if the AI setting was +50%?
Yep.
And, as you can see, this needs to be coupled up with multiplying the starting PUs too (a needed improvement regardless), otherwise, on round 1, all those players will have more placement capabilities for no reasons at all (you can place 15, instead of 10, but what you can spend on round 1 is the same!).
@Black_Elk said in AI Bonus Settings Revamp:
I can imagine production increases being problematic for the hitpoint spam, since the AI already does a pretty good job of piling up infantry haha, so I'd prefer the ability to separate those two options and control for them if possible.
In my mind, this is rather a good argument in favour of having the placement increase default or maybe even forcefully bundled.
However, I'm always a fan of customisation, so I'm totally with you that I would like this being optional, as well.
Since it should be much less distortive for the AI purchase decisions if income and placement increase of the same amount, I think this will cut down, in a way or another, the problem of having an AI fine tuned for no-cheats working with them.
If the increase income without increase placement would make the AI purchases qualitatively worse, this is obviously a good reason for counterbalancing it by increasing placement.
If the increase income without increase placement would make the AI purchases qualitatively better (like in the case that the AI is buying too much fodder, without cheats), this would be showing that the AI has problems in making normal purchases (as, of course, it is supposed to be fine tuned for no cheats situations), and might be even more of a reason for not trying to reduce cheat-driven placement limits distortions, as the AI purchase itself should rather be fine tuned with respect to no cheats, than having the cheats changing it for the better, as, if they do so, this means that there must be some problems with the purchase decisions (if you are thinking to v5, I don't really know how much going from 5 to 6 in cost should reduce the armour purchases, but in v5 the matter is surely made worse off, for the armour, by the fact that the artillery is still cost 4; I've no experience, but it looks like that should quite dramatically reduce your armour purchases, from v3 or v4).
As long as the AI is meant for no cheats (and aims at minimising the need of them), the less the distortions (for the better or worse) the cheats cause, the more such games are worthwhile references for analysing the AI itself.
Ideally, it should never happen that the bonus income modifiers qualitatively change the AI purchases for the better (well, assuming the AI is meant to be fine-tuned for no cheats).
If you think that, currently, the AI buys way too much fodder (thus making the games less challenging / interesting), then you should hope / suggest the AI being changed as to buy more of other units, not needing having it due to cheat-driven placement limits, I think. Of course, this is just in my mind, but I guess this is it.
I'm not sure if anyone here already know this, but "the ability to separate those two options and control for them" is maybe proved possible by the fact that this is exactly what Dynamix did (maybe), as you can see back in 1.6.1.4:
I said maybe, since I never actually playtested if it even worked.
Just to avoid misunderstandings, I'm not pushing for something similar to this. In particular, I disagree with having the bonuses as something modifiable in the in-game menu, during the course of the game, and I think we all agree with this (especially since an AI player may happen to play right away, after starting a new game).
I'm just showing this here as a curiosity and in case it might be of help for you; not sure if you already know that Dynamix had this option. If you don't, now you do.
Myself, I suggest the placement multiplication being handled with a general true / false (boolean) property, applying to all players having the income multiplication (not only for a specific player), for having it when true, and being off as default (not because I believe it should not normally be used (all the contrary, actually, as I've made clear), but because it makes sense it is not bundled as default, unless the name of the income bonuses get changed to directly clarify that they are both income and placement bonuses, which, coupled with the fact that we have the player name too, would probably make them a tad too long), appearing in the options of all games (even when not in the xml), right after the related bonuses, that, when ticked, makes you multiply the placement abilities of the same amount as the bonus income percentage, as correctly understood by @Black_Elk.
I think that if you choose to have the income multiplication on a per player basis, then probably this excludes having the ability to set it for each player, as this would arguably make for too many options.
To sum it up, my suggestion is either a boolean applying to all percent bonuses (when selected) or having it automatically and surely; but, in this second case, the bonuses should be renamed as "Income-Placement", instead of just "Income".
.Other Resources Multiplication
My opinion on resources is not totally defined.
Surely, I believe the first step would be to expand the options as to multiply any resources, just like the PUs.
I think this should be good enough, for any regularly intended uses of resources, or almost so, as I don't believe there should be any important exceptions.
Still, resources are so extensive, beyond being just a parcelization of income, that I would surely encourage making a way for mapmakers to decide what resources are meant to be duplicated like PUs and what others should not be duplicated, as existing for roles diverse from income.
Age of Tribes is a borderline case, as you might argue that using units as tech "flags" for triggers is basically a (good) hack, which is justified mainly by how limited the true tech is.
In my opinion, unless Frostion really wants to keep the dynamic that buying a tech takes up a placement spot (I don't really see the reason for that), this should be ideally solved by making the actual tech phase able to handle a tech system like Age of Tribes, which seems a totally reasonable one, and I guess a few tweaks away from what is currently possible.
However, there may be other cases, I am sure, in which you would want some resources to be multiplied and some others not; so, if you want to give a way for mapmakers to exclude specific resources from being multiplied, that would be fairly good.
Since (as said) these bonuses are not AI bonuses, an option for setting what resources can be multiplied by these new properties should surely stay all inside the xml itself, and not be assumed or defined as anything AI related at all, as any mapmaker could use them for reasons having nothing to do with AI (non in any specific "map AI configuration", that I surely believe, instead, should not be part of any games' xml).
Anyhow, I believe there should be at least one exception to normally multiplying all resources: never multiply tech tokens or anything that works for anything else but the purchase phase (currently, "techTokens" is the only resource of this kind in existence that is fully supported, but "VPs" should not be multiplied, either). So, as of now, just remember to exclude "techTokens" and "VPs", upon expanding these option to all resources (even tho techTokens are usually bought, you might want a map with player starting also with some techTokens, instead of only PUs, to represent development started in previous rounds (before round 1); in this case, I think the PUs should be multiplied, but not the techTokens).
