Topic: A bug or a feature?

I am playing a simulation where there is a large area between me and the AI enemy which was cleverly designed so that few building can be buit there, and all are limited to the smallest. So I build a string of sentries and then a cluster as I reach the border. I can the attack the opposing sentry with about 20 soldiers, one from each of my sentries with one left behind guarding or washing his hair.

A short while later the expeditionary forces return victorious but bloodied, most have some reduction in their health, some are almost dead, now is the time tp press on with my advantage in numbers but if I select the next sentry to attack all the wounded soldiers return to battle, is there a way I can select the ones that did not fight the first battle so the wounded guys can get some rest?

It could be that the buttons for prefering rookies or veterans does something like this though I have not found out what the option actually do yet, but it does seem logical to me that the soldier with the worst health would be the one to stay at home. It seems that the last on in is always the first one out.

On a similar topic, when I atack an AI military object the AI sends assistance from other nearby buildings, when the AI attacks one of my buildings nobody leaves any nearby buildings to help defend, How do I tell soldiers to help defend a nearby building?

Top

Quote

1. I did not want to clutter bug reports with things that are not bugs, but will add something following your suggestion.

2. That was how I thought it might work, thanks for the confirmation.

3. I think in this case it was a sentry I had captured but there was no road connection to the rest of my economy, got to remember to join captured buildings to the rest of my road network.

Top

Quote

Hello, everyone! I found widelands back in may 2012 and I've been lurking ever since. One thing I did notice is that -- thanks to https://wl.widelands.org/wiki/SoldierLevels/ -- it's possible to estimate the power of any soldier with a single number (feel free to guess how the three tribes compare; I'll give their stats a few paragraphs down).

Defense is "you block X%" of the damage, so 100 HP and 75 defense function the same as 400 HP and 0 defense. So you can multiply HP by (100/(100-defense)).
Evade is "you avoid damage X% of the time", so 100 HP and 75 evade is, on average, the same as 400 HP and 0 evade. So you can multiply HP by (100/(100-evade)).

That just leaves Attack and HP. But since "10-12 Attack, 100 HP, 0 defense, 0 evade" and "20-24 Attack, 50 HP, 0 defense, 0 evade" are pretty much the same (they'd both need five hits to take each other out), you can just multiply the two to get a single number.

That means the formula for a soldier's power would be ((MaxAttack+MinAttack)/2)*(100/(100-Defense))*(100/(100-Evade))*MaxHP.

So, how do the three compare? For total min-stat rookies, Atlanteans (2872.34) lead, followed by Empire (2736.84) and then Barbarians (2501.72). With maximum stats, Atlanteans (28668.09) still lead, again followed by Empire (27268.70), while Barbarians (24022.91) lag further behind. But make sure your Atlanteans get gold! Without gold-requiring equipment, they fall to last place (13081.4), behind Barbarians (14914.09) and Empire(16038.78).

So, why am I posting this seemingly off-topic info? Because it can be used to help solve the "wounded soldiers go out to fight" problem! If a soldier is 10000 power (according to the formula) but at 40% HP, he'd be about as useful as a soldier that's 5000 power but 80% health! So my advice is to use a formula something like this to determine who goes out:

(MaxAttack+MinAttack)/2)*(100/(100-Defense))*(100/(100-Evade))*CurHP*(CurHP/MaxHP)

That way, it'll prioritize how healthy a soldier is when choosing who to send out; a badly wounded hero and a slightly wounded rookie may be similar in battle at the moment, but the badly wounded hero'll heal a lot faster!

Top

Quote

The idea makes sense.

I think there an error in the formula, it should read (MaxAttack+MinAttack)/2)*(100/(100-Defense))*(100/(100-Evade))*MaxHP*(CurHP/MaxHP). If you cancel MaxHP and the constant factors, you get (MaxAttack+MinAttack)/((100-Defense)(100-Evade))*CurHP.

Those soldiers are equally strong and die with the same percantage, but the trained soldier is a bigger loss. I would therefore divide this value by the costs the soldier has produced, cumulative for every step of his training. The costs should be something like log = 0.5 (thus wood = 2*log = 1), fish = meat = bread = 1, coal and the ores based on the food cost, since gold is rarer, multiply it by 1.5 or so. These parameters have to be adjusted with care so that the "best" soldier (the soldier a human would send out?) is sent out.

"Only few people know how much one has to know in order to know how little one knows." - Werner Heisenberg

Top

Quote

In the first two paragraphs, I did not introduce anything new, I merely corrected a fault you had made (at least what you wrote did not accord with your formula). In your first formula, you multiply with with MaxHP at the end, while in your second formula, you multiply with CurHP (and the relHP factor you introduced). It makes even sense that the maxHP cancel out, because when a soldier (with given evade, attack and defense) has 100 HP, it does not matter in battle whether he can have 130 or 214, simply because he will never get there. So the formula (MaxAttack+MinAttack)/((100-Defense)(100-Evade))*CurHP describes the soldier's power (in the lack of a better term). Soldiers of equal power have the same chance of winning, even when the distribution on the attributes is different.

As GunChleoc pointed out, a rookie is not such a big loss when he is killed as a hero (with the same strength), therefore some corrections have to be made when deciding which soldier should be sent out.
You did this by dividing by maxHP. I did a more general approch by using the costs a soldier has produced. Imagine two soldiers with the same power and the same maxHP (therefore the same send-out-value following your argumentation), but one is trained in evade (which is cheap) while the other is trained in attack (which needs iron and maybe gold). You would send out one soldier at random, but I would send out the cheaper one (remember that they have the same power, i.e. the same probability to win a battle). I therefore suggest something like efficiency (again in the lack of a better term) = power/f(costs), where f is a function that we need to optimize. (I imagine something like (sum(ware_weight*ware_amount))^cost_weight. For every ware a soldier has consumed, you check how many and multiply them with the ware_weight, which indicates the value of the ware (examples see above, e.g. meat = 1). You add up all those values. cost_weight indicates how important these costs actually are (^ means power, cost_weight = 0: does not matter, infinity: only they count and not the power, i.e. rookies will be sent out).)

To conlude:

• power = (MaxAttack+MinAttack)/((100-Defense)(100-Evade))*CurHP. Soldiers with the same power fight equally well.
• efficiency = power/f(costs). The most efficient soldier (a cheap one who fights relatively well) should be sent out.

Btw, I do not understand your last paragraph:

That way, it'll prioritize how healthy a soldier is when choosing who to send out; a badly wounded hero and a slightly wounded rookie may be similar in battle at the moment, but the badly wounded hero'll heal a lot faster!

"Only few people know how much one has to know in order to know how little one knows." - Werner Heisenberg

Top

Quote