Experience Documented

[_Flin_]

I have similar issue's with the way you can trade in a PZ IV and buy a new Tiger 2 for half price, while if you attempt to maintain your experience, and upgrade, it costs the full 900. This makes it much more cost effective to go with green with your tank force as well.
So, far from being just theory i think there is something at the heart of the experience debate.

I guess the thing is that there is no "best" way. Going all-green isn't perfect, because your troops aren't as good and you have useless prestige on the bank instead of in the field. Neither is going all-elite, because you just can't afford it.

My solution was to choose a few elite units and upgrade them, and green the rest. Exception for Artillery and Air Force, which I elite on principle. There is just nothing like a 3-star overstrength Level Bomber descending on a Battleship. Or a battery of 3 2-star overstrength artilleries, allowing to take a fortified city in one turn.

[rezaf]

I don't disagree, just don't see a good solution yet. Infantry might use a bigger bonus to their initiative of 2, but tanks should not get a bigger bonus to their close defense of 2. How can we distinguish such special cases, without adding class-specific exceptions, something I would very much like to avoid?

I thought we were currently discussing initiative?

As for defense, stock setting is 1/10, so a unit with close defense 2 will get 5*0.2 or 1 point of additional CD, but only at 5 full stars.
With my 2/20 experiments, it would get 5*0.4 or 2 additional CD, but, again, only at 5 full stars.
With 4.9 XP, it'd only gain a single point of CD - more than nothing, but not spectacularly big, either.
Units with less than 3 full stars wouldn't get any boni.
I really fail to see how this would be a huge problem.
_____
rezaf

[Molve]

Interesting demonstration, but I feel a vital aspect is missing:

Showing actual combat results.

So far, the entire conclusion is based on the combat predictor. Unless the core combat coding has changed to drastically reduce outlier results, what looks like a good 0-10 result in theory, often ends up as a 4-6 result in practice. And for a overstrength 15 unit, losing four points means taking a huge loss. (Don't hold me to these numbers, but doesn't the final points above 12 cost as much if not more than all the points between 1 and 10?! Meaning that losing four points of Strength in this case would mean the prestige equivalent of losing 9+ points of strength of a regular unit)

Yes, an overstrength unit can withstand damage from an unlucky outcome better than a regular unit. ONCE. In practice, I feel this ability isn't worth the effort and expenditure. Blind luck cuts through any attempt at "roleplaying" where your favorite units enjoy a slow and steady climb towards excellence. At least the campaigns have slowed down the rate you gain XP.

But at least; let's discuss real combat outcomes. Don't stop the analysis before you click that button!

The combat predictor might show 0-10 all day long, but that doesn't help your wrecked units... you will want to use your units when that 0-10 really turns out to be 0-10, or at least 0-7 or 1-9 (a single point can be recuperated without an extended resupply period which effectively would remove that unit from the rest of the scenario).

[Rudankort]

I guess the thing is that there is no "best" way. Going all-green isn't perfect, because your troops aren't as good and you have useless prestige on the bank instead of in the field. Neither is going all-elite, because you just can't afford it.

This was exactly our design goal. One could argue that elite replacements and overstrength are too expensive, but this is where you can direct the excess of prestige you have, to make your army even stronger, but not critically stronger (which would unbalance the game).

[Levada]

Thanks Flin for your detailed reply, very helpful.

[_Flin_]

Yes, an overstrength unit can withstand damage from an unlucky outcome better than a regular unit. ONCE. In practice, I feel this ability isn't worth the effort and expenditure. Blind luck cuts through any attempt at "roleplaying" where your favorite units enjoy a slow and steady climb towards excellence. At least the campaigns have slowed down the rate you gain XP.

I do not agree. Overstrenghtening elite units is actually a cumulative thing.

- you get more "dice roll" and can kill more
- for these dice you have better attack values and better initiative
- Due to the higher initiative, less units counterattack.

All these bonusses add up. And on top of that you have a limited amount of units to deploy. So the only way to add additional strength to your force is to use overstrengthening.

Especially when it comes to dogfights, this is so valuable, and the lesser units (10 strength, 0 experience), are not even close to an overstrength unit (12 strength, 2 stars). Not only will the OS unit get a 5-1 instead of a 4-2, but it will stay out in the field 3 times longer before having to fly back and skip a turn for reinforcements. Which in itself is a huge tempo advantage, that ripples through the whole scenario and quickly adds up to 2-3 turns.

I always overstrenghten air-units and artillery, and always elite replace them. My level bombers sometimes don't get reinforced for 2-3 scenarios. Assuming 3 stars overstrength and an attack value of 10, having 26 with a combined attack of 336 vs. 20 with a combined attack of 200 is a difference of 68%. And that is an amount of additional power that is unbelievable mighty and makes a very recognizable impact on the opponents prestige.

[Molve]

Please don't turn this thread into an argument, Flint.

I do not feel the combat predictor is accurate enough to be useful as "actual documentation behind the arguments".

Getting a 0-10 prediction doesn't prove anything, if the actual outcome isn't 0-10 or very close to it. As soon as you're getting your first 3-7, that changes everything: losing the 15th, 14th, and 13th points of strength is a completely different cost than losing your 10th, 9th and 8th point of strength.

This thread is not meant to discuss whether experience is useful enough, only to provide the facts allowing people to reach their own conclusions. The topic of overstrengthing is further peripheral to this issue.

So let's focus on the topic at hand - which needs to look at the actual results from the (sometimes very swingy) combat code; and not the (overly simplistic) predictor used so far.

Best Regards,
Molve

[Kerensky]

Getting a 0-10 prediction doesn't prove anything, if the actual outcome isn't 0-10 or very close to it.

Guess I'll just repost this again then.

[Molve]

That's a nice start.

Now provide 9 (or ideally 99) more samples for this particular battle, and 10 (or a hundred) samples for all the other cases you were investigating, and we have some real data to discuss.

Specifically, how often out of 10 tries does that Panther G of yours come out of the battle with a chunk of its Strength 15 bitten right off, despite the predictor saying 0-10 all the time? (To take only one example)

Such data would be most illuminating (but quite cumbersome to collect).

Best Regards,
Molve.

[Rudankort]

I do not feel the combat predictor is accurate enough to be useful as "actual documentation behind the arguments".

I can tell you that in the very beginning I did a lot of tests to determine if battle predictor is good enough to draw conclusions or we need a more advanced estimation in order to balance the game. Clearly, a single combat cannot prove anything, I considered using something like 10000 combats and averaging the results (internally of course, this feature would not be available in the released game). But I concluded that we did not need this and the predictor was adequate enough. It does hide some nuances, but very few of them. Essentially, the biggest problem with the predictions is that they are rounded. If a 10-strength unit is attacking, any kill chance from 16 to 24 percent will be rounded to 2 kills. Still, it is the best estimation you can express in integer numbers. Feel free to use kill chances instead, they are shown in the extended combat info, but, as you see in the example Kerensky has posted, the result will be very close (66% from 15 strength is very close to 10).

Any other deviations are random and depend on luck. Yes, you can be unlucky in a given combat, but you can be too lucky as well. This is not something we need to consider in balancing.

If you insist that the predictor is so bad for this discussion, please post some data to prove your point.

[_Flin_]

Please don't turn this thread into an argument, Flint.

Well, that's what forum discussions are for, aren't they?

And, btw, I did not argue against your stating that actual results are a better basis to document the experience engine. This is certainly correct. Having the same combats repeated a thousand times and then showing the data is better than the prediction.

What I argued, however, was your conclusion that an experienced and overstrengthed unit has an advantage only once in a scenario and is not worth the effort.

And, since this thread is about experience, I wanted to point the cumulative effects of experience, which have not been mentioned yet. Because not only does experience lead to higher attack and defense, therefore killing more opponents and suffering less casualties. But additionally to higher initiative, leading to less killed or suppressed opponents shooting back (at least for units with already high initiative. For infantry, not so much). And on top of that, overstrengthening leads to more dice rolls. All of that together leads to massive advantages.

If you want people to reach their own conclusions, provide data and information yourself, instead of arguing with people who do.

[_Flin_]

Aggregated Attack values for a 10 unit with base-attack 10
0 Star: 100 - 0%
1 Star: 110 - 10%
2 Star: 120 - 20%
3 Star: 130 - 30%
4 Star: 140 - 40%
5 Star: 150 - 50%

Aggregated Attack values for an overstrength unit with base-attack 10:
0 Star: 100 - 0%
1 Star: 121 - 21%
2 Star: 144 - 44%
3 Star: 169 - 69%
4 Star: 196 - 96%
5 Star: 225 - 125%

So experience alone is linear, over-strengthing is exponential.
On top of that: Initiative advantage (less shooting back). Tempo advantage due to less reinforce (even higher due to higher defense of elite units).

[Fimconte]

Is the 125% on 15 strength also taking into account that it comes with +5 attack (ie. can't overstrength without stars outside editor)?

[rezaf]

If you insist that the predictor is so bad for this discussion, please post some data to prove your point.

I provided some test results in the RNG thread, and extreme results (mostly unfavorable) were far too likely.
You can check out the exact numbers in that thread.
Once I can muster the energy, I'll make some more statistics.

I really really really cannot understand why - at 1.04 - you STILL haven't provided an option to reseed the RNG upon loading a savegame. It'll cost you all of three minutes unless your code is a complete disaster, and it'll shut up many folks (like me) right from the get-go, because they'll just save-scumm their way around disastrous results and be done with it.
Cmon, be a nice guy and expend those three minutes for us, would you?
_____
rezaf

[Rudankort]

I provided some test results in the RNG thread, and extreme results (mostly unfavorable) were far too likely.
You can check out the exact numbers in that thread.
Once I can muster the energy, I'll make some more statistics.

I already checked my RNG, and on a big number of tests (that's what Molve requested above) it is almost perfectly uniform. I know that sometimes it can generate long sequences of poor (or good) rolls, but in the context of predictor discussion this is completely irrelevant. Any quirks of RNG average out over time.

(This is not to say there is nothing to fix, in background I'm thinking how to improve the situation with RNG. Maybe some more advanced algorithm will work better, there is a bunch of different pseudorandom algorithms out there, but the problem is, it is hard to even formulate criteria for an ideal RNG in a game like this.)

I really really really cannot understand why - at 1.04 - you STILL haven't provided an option to reseed the RNG upon loading a savegame. It'll cost you all of three minutes unless your code is a complete disaster, and it'll shut up many folks (like me) right from the get-go, because they'll just save-scumm their way around disastrous results and be done with it.
Cmon, be a nice guy and expend those three minutes for us, would you?

I'm very reluctant to add options which 99% of players will not care about or even understand. And if you want this option only to workaround RNG issues, this makes me even more reluctant. Disastrous results are part of this game. With changed RNG they might happen more rarely, but they will still happen.

But I can add a cheat code for you and "many folks like you", let's call it "rng off" or something, so that you can avoid disastrous results as much as you want. But this style of gameplay will be classified the way I see it - not an "officially supported option", but a "cheat". Let me know if you are interested.

[rezaf]

I'm not interested in a "rng off" cheat. I don't want to have no randomness, neither am I opposed to horrible combat results per se.
It's ok that they can happen. 95% of the time, I'll shrug them off.
I'm not a minmaxer by any stretch of imagination.

Storing the seed in the savegame AT ALL is nonsensical in my personal opinion. What good does it do? All it does is punch people who were going to save-scumm in the face, nobody else is even going to notice it. It's a small dose of sadism right there.
I linked to a podcast somewhere in which Brian Reynolds (of Firaxis fame) tells the story of how they had the same brilliant idea in Civ2 - and how, in hindsight, he thinks it was a horrible idea to begin with. You should listen to it.

Reluctance to add new features is nice, but this has NO DOWNSIDE and takes MINUTES to implement. It can't break any gameplay mechanic or have unwanted side effects. Sorry, I don't get the reluctance in the slightest. It doesn't compute.

How about an option in the gamerules file, reseed_upon_reload = 0. Anyone who cares and understands can set it to 1, default setting 0, problem solved.
_____
rezaf

[Rudankort]

Storing the seed in the savegame AT ALL is nonsensical in my personal opinion. What good does it do? All it does is punch people who were going to save-scumm in the face, nobody else is even going to notice it. It's a small dose of sadism right there.
I linked to a podcast somewhere in which Brian Reynolds (of Firaxis fame) tells the story of how they had the same brilliant idea in Civ2 - and how, in hindsight, he thinks it was a horrible idea to begin with. You should listen to it.

If you send me a link, I will listen to it. Although it's strange, you have listened to that podcast yourselves, didn't you, but you cannot convince me that reseeding is the best approach. Nor a few other guys who complained about this could give any serious arguments. I agree that it does not make a world of a difference either way, but the game relies on existing behavior in a few aspects (like replays in MP). If you want me to change that code and run the risk of breaking MP (both new games and the ones which are already in progress), give me your arguments. Not "everybody else does that" and "it is sadistic to 1% of people who want to save-cheat". I gave people a whole bunch of nice, easy to use cheat codes instead of that save/load nightmare, and you still call me a sadist? Come on.

PS. This is a bit offtopic in this thread too. We better move to a different one if you want to continue discussing this...

[Longasc]

(The RNG / Seed debate is related, but not exactly on topic, guys)

How are the results for 2-3 stars vs 0 and mid-war units?
Or say 3 Stars vs 1 Star XP

Panther G and Me 262 are late war/game units with very good stats.


But how much would a difference of 2-3 XP stars help a Panzer IVG/H/J vs a T-34/41 or /43?
Maybe Kerensky has some time at the weekend to run a similar series with Shermans and T-34 and other more common mid-war medium tanks!
I think that's also what rezaf was talking about.

This said, I like the experience system as it is. It's not the end of a world when a 2-3 star unit dies as the 0 star units are not totally useless in the late war period, as it was often the case in Panzer General.

[deducter]

But how much would a difference of 2-3 XP stars help a Panzer IVG/H/J vs a T-34/41 or /43?
Maybe Kerensky has some time at the weekend to run a similar series with Shermans and T-34 and other more common mid-war medium tanks! Smile
I think that's also what rezaf was talking about.

From what I observed, a massive amount.

[rezaf]

I'll see if I can find the link and send it over to you.

Though, I fear you're set in your opinion about this and will not change it nomatter what I or anyone else says.
I don't really know what else to say, it'll take you a couple of minutes - you didn't deny that - but your flat out refuse to do it.
You have already spent more time reading and replying to posts explaining that you're reluctant to do it than it would have cost you to just implement it. End of story, I guess. I cannot convince you because you don't want to be convinced.

At least the MP-stuff-relies-on-it is a good reason, though, honestly, would it be a problem to make reseeding NOT work in multiplayer games or when starting a replay? I know some programming myself, but it's somewhat difficult to predict, as it depends on how you have implemented this.
Still, that's the only reason storing the seed in the savegame makes ANY sense. The part about sadism wasn't my idea, it came up in that podcast I mentioned, though it might have been phrased differently.
Brian explains that they got this idea specifically because they'd noticed people getting around unfavorable combat results by reloading their savegame and they wanted to prevent this.

There's a difference between opening up the cheat menu and giving your own units extra strength or yourself extra prestige or whatever and reloading a game to prevent a disastrous result two times over the course of an entire campaign.

About this being the wrong thread to discuss this - maybe. The topic came up in a number of threads before, but you participated in none of them - gotta pick you up where you stand, no point hoping you'd show up magically somehow.
_____
rezaf