...There are many units which are completely useless, for example: Heavy cruisers, Battlecruisers, super heavy tank brigades, military police (due to the broken suppression system) and so on. In addition to these, there are several barely-useful units like multi role fighters, battleships (these get owned by carriers and super-CAGs due to the naval battle mechanics - unless they were changed in a recent patch), engineers (only useful against forts) , militia (too expensive to research, usually), AA (interceptors are much better), AT (tank destroyers or armour are much better) etc. etc. Then we have the ultra-OP units V1 and V2.
BTW: What is wrong with scripting? Just let the AI be scripted until the war starts, and then release it. What has scripting to do with naval invasions? Those seem like just bad coding 2 me.
Heavy Cruisers are 'vital' if your playing a mid-minor with a naval focus, since they are the only way you can really get a fleet out, although they're not very good in my opinion compared to CL, or BC which will leave me with a general focus for either of those given the chance. BC are very useful for a majors and mid-minors again in a naval game. SHARM is pointless, but then it was historically, they really shouldn't be in the game as they are. Same with MP, Paradox doesn't really like the SS and all that, which is what I liken them more to, again they shouldn't really be in the game either from a Politically Correct standpoint if anything. Multi-role are great for interdiction missions in general, TAC and LB/CAS for ground attack, STRAT for Log.Strat bombing. Battleships great for minor navies or nations with Green water operations; Russia, Italy, France, China* Engineers have uses all-round from extra fort defence, to increased movement/reduced combat penalty in rough terrain as well as fort/urban attack. Really engineers are OP in some respects.
Militia too, often they get poo-pooed, but the Manpower heavy nations of Japan, China and Russia can make huge use of them since they are supply unintensive and a good enough to 'hold the line' and can really be worth their weight in garrison duty later preventing whack-a-partisan where cavalry would normally take a week or two to close with them. But militia are cheep enough to stick everywhere. AA guns aren't as good as they could be, but then they weren't 'all that' terribly effective during the wars either in some respects. They could do with tweaks, AT again TD were better in many respects, the downer with AT guns is they take up a brigade slot when normally there was a greater mix of gun types in an army during the time. Tweaks, but not useless, indeed as a minor they can save your very bacon against a larger power.
Agreed with the V-weapons, again they could be better represented.
But this isn't; 'They have no use' it is that they are; 'less useful than [XXX]' and so a gamey player, games the system by building the best they can, because they know the game mechanics work to say make Art more generally useful than AT. Which is the point Darlor makes. It's not bad game balance per se, its bad player balance.
And as Cybvep and others have pointed out, having a political system that would encourage more historical play (I think this would be best done with making politics count for a lot more with the real chance for revolutions and military coups if you don't do what your 'advisors' are suggesting), would be the best way to deal with these unit issues, rather than try and streamline the number of brigades. Indeed I would welcome a more detailed 'division designer' that would use 'hard numbers' of tanks, aircraft and men as a base, linked to a doctrine system. Helping to 'iron out' the ratio of generally useful, to specifically useful unit types.
*Although going navy with China is a bit of cow pat...
----------------------------
There's nothing wrong with either, but its when you try and mix the two types. For instance lets take the Vanilla Skirmish AI from the C&C series of games. It builds a preset force of infantry and tanks at specific 'waves of attack' which get progressively stronger till wave 5, at which point it sends essentially the same wave all the time.
Its great and fit for purposes, because in a typical comp-stomp game, your done with the AI by wave 5, or your just sitting it out to deal with the AI once its exhausted the easy 'resource collection' slowing down its war economy. Fit for purpose.
In the Civilisation series, if every nation had a 'war era' and 'peaceful eras' sending a preset force you'd know exactly when to have armies, of what size, and when to disband them to make the best economy. Not that fit for purpose. So it doesn't use that system, instead it checks off the game state to decide if it needs an army, or can expand economically. The course of events is not programmed to happen in a specific way.
HoI kind of screws it up, because the code modders can't access is how the AI actually fights its battles, we can control what it produces and where, and even its diplomacy via events and the Lua files, but what we can't do is tell it how to use nukes, or subs, or naval invasions. That's all hidden away. Therefore we can't program in a better AI that is able to change its fighting style with respect to what it has at its disposal. We can only optimise what the AI builds such that it fits in line with the scripted diplomacy actions.
This means that an AI nation is always unprepared to fight any war it is thrown into because the AI nation hasn't been able to decide if it can use the army it has, to fight that war, because the AI fighting code isn't linked with the production Lua strongly, and not at all with the diplomacy.
For instance, I can mod Germany to produce an infantry heavy army for 1939, but the AI will always try a blitzkrieg style war that has the divisions in Blitzkrieg mode for a month or so making heavy attacks taking huge losses. Rather than sitting back on defensive. Or I tell Japan to build a tonne of tanks, it will send them to China whatever the terrain and have them get bogged down without supply.
The point is that the AI doesn't check what it has at its disposal and make the right choices.
Because the AI system isn't integrated with a good AI structure or framework and relies on two separate systems for different areas of AI behaviour it doesn't gel well together and leaves us with a clunky game. Not fit for purpose.
If say the AI was scripted up to Danzig, and then free-to-choose that would be just like the start date options sans your nation, therefore instantly making it 'player unbalanced' or 'gamey' to play from 1936 etc. as per above and comment made by another. Opinion undecided here, but the points made are good. There's nothing wrong with this per se, except a good AI will immediately try and change track from Danzig to try and play the 'Metagame' with the other AI, you can put in bias here, but doing that, is exactly taking the route that leads you to Germany doing Blitzkreig attacks with infantry armies, or Japan not producing tanks to fight in the Kharbovsky'kray pocket etc.
It's clunky to try and mix them both. That's the bottom line. I'm sure it could be done very well with some clever forethought, but it would be unique and very difficult to do and above all, either scripted or working on call statements is more likely to end up better for the same level of input.
That was a long post, but I hope it was informative and thought proviking.