DC's House of Whacky Variable Testing

[Soapy Frog]

Well, actually, I wasn't thinking in that mode. The starting values placed in the brigade.txt file in DB wouldn't need to have any values other than the negative MP one. Any other effects you want become tech activated, where a given tech WILL do completely different things for the different kinds of units.

Alas it wont...

Tech improvements are applied to brigade types as though they are independent units.

If I give +1 SA to artillery brigades then all different unit types with artillery brigades will get the benefit.

I can't (unfortunately) give +1 SA only to infantry with artillery brigades.

Or am I misunderstanding what you are saying?

 

[unmerged(19545)]

Damn, you are right. The tech will apply a different effect to different unit types, and the weird brigade will be just one of them. Fine if the brigade is only attached to units where the effect is always similar, but having the brigade cancel out an advantage that a unit type didn't actually get is bad.

Oh well, the best laid plans of mice and men...

 

[Soapy Frog]

Yes it is a pity... I am resigned to using models and vastly reducing actual tech stat bonuses (i.e. new techs give you new models which have different stats).

 

[unmerged(19545)]

Yes it is a pity... I am resigned to using models and vastly reducing actual tech stat bonuses (i.e. new techs give you new models which have different stats).

Oh, that's brutal man! I don't know if the cure isn't worse than the problem in that case. The poor ai will be upgrading every division as soon as it's upgraded. I usually play minors because majors are too easy (don't even mention Germany. That's about as much challenge as beating my granny in an arm wrestle) and so I'm too short of MP to build a big army. As a result, I have no choice but to tech up, and the new units coming available would be coming up faster than I could upgrade them.

BTW, have you determined what formula the game uses to calculate cost and time for a unit upgrade? It seems too high for aircraft and tank units, but more reasonable for infantry types. Infantry of course need to be upgraded when you develope tech while some forces are away as an expeditionary force. They don't get the benefits of the tech you developed, and have to be upgraded when you get them back. Makes sense.

 

[jdrou]

BTW, have you determined what formula the game uses to calculate cost and time for a unit upgrade? It seems too high for aircraft and tank units, but more reasonable for infantry types.

My observations:
tank/air upgrade: 3/4 cost, 1/2 time of highest available model, not modified by tech or ministers

infantry upgrade: 1/5 cost, 1/5 time of highest available model, not modified by tech or ministers

 

[unmerged(19545)]

Yeah. Kind of arbitrary huh? So, Soapy, how many techs would give new models, and how many would give improvements to values?

Actually, I suppose you could do all your upgrading at once, when you figured you were high enough on the tech tree not to ever need to do it again.

Y'know, you could make a very good case for not using new models of anything other than ships. When you consider the combat life of aircraft vs pilots, and the rate at which these things were replaced, you could just as well assume that new models of fighters or tanks replaced old models when damaged or destroyed vehicles were replaced individually. Certainly when a squadron went from the P-51B to the P-51D they weren't taken out of action. They just got the new planes delivered as they became available.

Does anyone have hard data (or intesting anecdotes) about divisions changing their model of tank, and how they did it?

 

[Soapy Frog]

Well, with regards to tech, I like to have most techs giving a small incremental boost to one or at most two stats.

Then I have a couple of upgrade required techs that give larger boosts... then finally I have major doctrinal "model" stepping stones, so you might get say 3 successively improving models of infantry throughout the war.

As for tanks well, I outlined what I was doing to tanks in the HSP thread and I have to say the results so far are very cool. Tanks require real finesse to use, and they really burn out on a sustained offensive. Basically each tank unit represents a battalion/regiment/brigade (depending on what country you are) of tanks with the minimum supporting troops, so an "armoured division" would consist of a motorized division plus 4 armoured units (or a 1941 German armoured division would be a motroized division with 2 armoured units)

Anyway if you want a taste of what I am doing check out this thread:
http://forum.paradoxplaza.com/forum/showthread.php?p=3196313#post3196313

 

[Soapy Frog]

Certainly when a squadron went from the P-51B to the P-51D they weren't taken out of action. They just got the new planes delivered as they became available.

Does anyone have hard data (or intesting anecdotes) about divisions changing their model of tank, and how they did it?

Well for sure there has to be significant retraining on the new equipment, if nothing else.

As an example, those units that received the Panther D before Kursk went through significant re-organizationa nd re-training, including maneuvers with the new vehicles and so forth, so I think upgrading equipment should be non-trivial.

 

[unmerged(19545)]

That is certainly true of the German deliberate approach to things. On the other hand, new Soviet tanks were just driven up and delivered. Loose crewmen would be picked up from collection points by "buyers" and stuck in a vehicle they'd never seen before. Of course, the relative effectiveness of the respective tanks was rather lopsided.

I seem to recall Math Guy saying that half of the cost of an armoured division was in the tanks, with the rest being in what would be basically a light motorized division. That being the case, I guess a tank unit, of just tanks, should cost a little less than a motorized division. Or am I barking mad.

 

[Soapy Frog]

Well there are game balance considerations. With no real bottlenecks on tank production except total industrial capacity (and this is true of all units really), and additionally the very tricky need to balance combat capability with price somewhat, the relative cost of tanks in IC will inflate.

I think the real relative costs are a useful baseline (which I certainly work from), but in the end game balance has to take a front seat.

The upgrade model provides another useful effect: it allows for significant variety in the tank types you field at any given. If you simulate the introduction of T-34s by giving an immdediate stats upgrade to all tank units, well then you entire tank fleet just morphed into T34s for the static cost that it took to develop the tech.

If you force upgrades, then people pay a real cost for introducing a new tank type. Also they can introduce it in the quntities they want and at the pace they want... it is a much better system, even if upgrading sometimes seems like it takes too long.

 

[unmerged(19545)]

Well, that makes perfect sense to me. Consider me convinced.

I'd like to keep this thread alive, just because it makes a good collection point for odd notions or to record unexplored avenues. But maybe it should be allowed to die. Are there any other variables in the game we haven't thoroughly examined yet?

These last ones I added to the thread had already been examined by someone, but not added.

 

[Soapy Frog]

Of course this thread should be kept alive!

I cant think off the top of my head what variables need testing... some of the efficiency variable seem wonky... like the "frozen" penalty, whcih seems to pop up at odd times (Paris, mid-July??).

I am trying to create a barbarossa scenario for my mod, which I find is no small undertaking so thats kind of chewing up most of my free time.

 

[unmerged(19545)]

I'll have a look, when I get the chance, at those negative variables, and the impact on the game engine.

In the meantime, I generally find that if game balance over rides actual historical relative costs or values, then something has been modded wrong. The mechanism we or the game is using is giving us a curve that only works for some limited range of the possible values. Curve Fitting is the cardinal sin of all simulations. But what's a fellow to do? We should take up a collection, buy the game from Paradox, and publish the source code. Linux/HoI anyone? I think I'm kidding. Short of sleep y'know.

 

[Soapy Frog]

Its not that something has been modded wrong, its that there components missing entirely from the simulation (in this case production bottlenecks, and means of production costs).

In this case we must fudge to some degree. I'd love to have another way around it, but for the moment I am stumped...