Well, I'm about to run 1,000 battles of evenly matched forces. Shouldn't take more than about 5 hours (I have to sleep 1 second for each battle to avoid loop problems and another 0.1 per turn in the battle).

Forces:

Army A
Troops: 1,000
ATT: 3
DEF: 1

Army B
Troops: 1,000
ATT: 1
DEF: 3


I'll let you know how it goes in an edit.


Interestingly enough, Army A won 10% more often than Army B, a full 100 battles more. You would think that it would be closer to 50% (+/- maybe 1% or 2%). *shrug*.

Anyway, of note is the fact that on average battles took 50 turns!!! That seems a bit much, so I delved. The problem is diminishing returns. I capped the damage minimum at 1 and that was reached in quite a few battles. Perhaps damage should be based on the beginning size and not the current size?

I see your point, there definately should be some kind of fail safe for when it gets to a certain point, like 100 troops each. I think it should be scaled down to 10 maximum, 1 minimum deaths per attack past 100 or so... Or something simular.

On another stance this is really really helpful, because if I go for something simular to this style of fighting, I can then tell from your little statistics what a good average amount of attacks for say a damage proc should go off...

For instance - one of my races has the ability to poison troops in an army on an attack. Kills 1-10 troops per month for 3 months for every 100... Lets just say that's what it does.

So now I can say well hey if the average battle lasts 10 attacks, then lets give them a chance that in 10 attacks there is a chance one of those will activate the poison damage proc...

You know, you're not going to convince me that pure determinism is best any more than I'm going to convince you that a little randomness is not bad thing. Why keep beating a dead horse?

I never said pure determinism is always best(well I may have but my opinions generally change over time after more thourough thought and testing on the subject :P) since in some cases you lose the feel of the game like deciding hit locations since you can take the average of all the possible outcomes to make it deterministic but you lose the ability to describe out how the attack took place which is important in an RPG* but not in a competative stratagy game where you're trying to create a setup that allows people to beat each other with skill rather than luck and the descriptive nature doesn't become as important(if any especially for people like me :)). Regardless I'm not trying to proove my point to you since I knew from the start that wouldn't change as most people are pretty adamant about the views they express. I continue to argue to get my point out to people who haven't made their decision yet so they get a detailed version of what my side brings as well as yours.

* By RPG I mean the kind that actually involves roleplaying which I feel compelled to point out since everyone has their own definition and would probably assume the wrong one :P.

Personally, I'm a little affronted that you completely ignored the whole point of the post

Well as far as ignoring the statistics yes I did and I'm sorry about that so I made sure to point out later why they're pretty meaningless. Though if you're refering to the point of why you think the randomness is good all I manged to get out of rereading your posts is vague opinions.

"RTS games like the one he's describing are dull if it's purely deterministic"

Well it's hard to argue with why you believe what you believe since I don't really know why. So it's not really that I ignored this but rather had nothing to say about it. That and since my view is so radically different from yours it even makes it harder for me to see why you think that way. However when I say something like randomness kills chains of stratagy it's much easier for you to point out something like 'well that just requires players to come up with contengency plans'. Which is by the way the whole last post I made in which you completly missed the point.

2.) "There's no feeling of risk involved"
Well I did miss this one so allow me to retort. There is a feel of risk involved in playing a deterministic game though a different kind of risk. Depending on the level of randomness in a game you're feeling the risk more from the random number generator rather than the opposing player. In a deterministic game like chess or checkers when you're up against a player who is roughly as good as you there is a lot of risk simply because you don't know what the player is thinking or planning(if you did you're evidently either both begginers or you're much better than the opposing player). Though you may already know the outcome of each individual battle the result as a whole of the game is largely unknown and thus the risk is involved not on each fight but rather the moves you make since they could be helping or hurting your stratagy or your enemy's.

If I seem to be missing a point or overlooking something please point it out rather than say I missed it. I love to argue and to defend my points of view so I have no problem with you stating them. And unlike most arguements this is ontopic and in the right forum.

Are you really that intolerant of someone's opinion on what makes a game enjoyable?

Yeah I'm quite an intollerant person. But I figured people would realize this especially after all the anime debates and various other arguments I've been in. I hold nothing back when defending my point of view or pointing it out for the indecisive people.

Is your idea of a good game the only/best one?

Well best for me atleast which is good enough for me. Of course if I'm out to sell a game my views are entirely different :).

If that is your thought process then I wish you luck in your endeavours.

And if you don't think that people should be able to elaborate the opposing side you're going to have some fun in life if you haven't already cause I'm sure I'm not the only person with a differing opinion that will express it.

Now, instead of renewing the zealous attack on the "horrors" of randomness and what it's done to gaming (I get a real "kids these days, don't know what they want" feel from it), why not respond to the rest of the post?

Well I was under the assumption that the root of this thread was about providing a formula. That's what I did I gave a generic commonly used formula that can be easily expanded on by just adding in modifiers. When I see a post in which there is confusion about my formula I respond to it and if someone points out a view I don't like I point out why I don't like it and what I think is the better solution.

It was intended to give a good indication of how the mechanics would work in the long run.

This doesn't work since the longer the run is the more it looks like the deterministic solution. The randomness comes in to play in the short since it breaks up chains of stratagy(since most people especially me tend to have planned out stratagies that involve more than charging headfirst to exchange blows) which is why it isn't good in a stratagy game. Simulations like these aren't really great for demonstrating the effects of the randomness. To properly see how much it'll trash the game you need a simple prototype to test with actual people attempting to play rather than getting an approximation of the deterministic approach through many runs of the random system on fixed test scheme.

Overall, the larger force of untrained rabble-rousers wins the day (with heavy losses).

But if it doesn't and the large force was a desperate last measure that would have succeeded in the average case you just made someone lose because of a random roll and not their own error. Now despite it may have been a good stratagy the player might think it was his fault(unless you provide a good break down of the rolls and everything so he can see it wasn't his fault and that he didn't think about some random advantage the other guy had that made him lose). This is why it isn't a good idea in a competative multiplayer stratagy game.

Hopefully, someone will comment about how long the battles usually took. Were they too long, too short?

This probably depends more on the game itself which we don't know much about :P. If this is on a large scale with many many units you might just want to have it autocycle through turns until one side is dead like in Civ2 so the length doesn't really matter so much unless it takes a long to calculate. However in something at a more tactical level I'd say it would probably best to stay at around 4 rounds on average for equal sized forces but this number could still vary a lot depending on how you want to set this up. So it's probably best to have a tweakable constant to find the right settings through playtesting.

How about the longest battle. That took quite a few turns. Should an attempt be made to keep battles from going that long? All of this is useful info for the game maker.

Well hopefully one side would retreat or get reinforcments before a battle continues on that long :P.

Finally, should I run the sim with other values? Units closer in size?

As I noted before the sim isn't a great tool for determining how things will actually play out since you can obtain marginally close data by doing a little bit of statistical analysis(unless there is a heavy bas in the BYOND random number generator) :P. Best to just build a quick and dirty prototype for the game and play it to see how well things are working out.

There was no randomness except the chance to hit. The simulation was run purely on your specifications, or at least how I interperted them.

Ahh. Well yeah but I already ackoledeged the fact that most people don't think like me and didn't point out how to make the results deterministic which I guess is a bit hippocritical :P. If I were to do it I would just take the statistical average result and make it the final result rather than resorting to random odds.

We'll see how much of a difference it makes.

The more times you run it the less it'll make and the amount of difference it'll make in the short term depends on just plain old luck.

Jon Snow wrote:

so wait... what exactly do you think is the best amount for each side to lose then?


I'm going with the formulae mentioned above and I will put both your names in the credits for helping me out... I was just wondering though...

sure you have a 25/10 army against a 1/1 army then there's just about no chance the 1/1 can hit the 25/10 even if it's really skilled... That makes sense and is reasonable. A highly unarmed army should not beable to take down a dragon for instance.

Personally, I think that damage and accuracy should be independent of each other. There are a couple of ways to do this.

1. Keep the accuracy formula and add a dmg and arm value to units.
2. Forgo actual accuracy and go pure determinism. Calculate the chance to hit and use that as a percentage of the base damage.

An example of #2.
Army A is 25/10 and has 1000 troops
Army B is 1/1 and has 100 troops.

Just for display purposes, lets use the mean of the troop sizes for base damage. 10% of 1100 = 110 damage.

For argument's sake, Army B attacks first. Their chance to hit is 1/26 which is about 4% or 0.04. 110 * 0.04 = 4. Army B does 4 damage per attack to Army B.

Army A swings at Army B now. Their chance to hit is 96%, 110 * 0.96 = 106. Army B is destroyed by an extra 6 troops.

if the attacker fails to deal damage does the defender deal damage instead?
Or does the attacker get a try at dealing damage and that's that?
OR do they both get a chance to deal damage? Sure this can be dependant on the type of unit I can understand that, but as the normal and basic thing for average units... what should it be?

It really depends on how you are planning on doing this. Do units stay locked in combat trading blows? Or do they fight on their own initiative? Ie: I send 2 units to harass 1 unit. Would you have them all fight at once, or in order of attack? I prefer the latter. It doesn't punish agressive play nor defensive.

If you prefer the idea of a pitched battle where fighting occurs on both sides of battle (more realistic, but broken unless you use other factors) then have the units deal damage simultaneously.

By other factors I mean that a unit fighting on two fronts would have to split it's effectiveness between the two.

Another thing, if it's 10% damage, like they said they would whittle eachother down... and I don't wanna add too much of a randomization factor...

So I was thinking then what about 15%, and say give 1-3% extra damage randomly (to satisfy the randomness in you and to not keep the amounts as bland numbers)...

but is the % going to be based of the means of the army's troop amounts, or is it going to be based off the higher armies troop count... OR would it be 15% of the attacker...

No, I actually agree with Theodis on this one. After running the sims, random damage would make battles drag on longer. I can see adding the +/- 3% damage if you're using the pure determinism method above (although it would lose the "pure" aspect). I would go with either 10% of the mean troop size or 25% of the unit's size. You also need to choose if the formula uses the current unit sizes or the original unit sizes.





All said and done I'd prefer an RTS that emulates a small part of reality (ie: some randomness). I think that the to-hit formula does this just fine. I would also like to see it somewhat more complex on a unit level than size, attack, and defense. I like to see units that are good for some things but bad for others. A fast moving, lightly armed and armored unit is great for harassing but will break like a piece of plywood if hit head on. Slow moving highly armed and armored units are wonderful for holding positions but heaven help you if you find that you need them elsewhere.

I posted something further up about potential uses of equipment for theses purposes. Perhaps instead of adding to ATT or DEF, they add to DMG or ARM and affect the unit's speed.