Topic: Slightly increasing atlantean and imperial attack to make defensive promotion more viable

in the high level play displayed in the latest turns of the 2016 tournament, a few players started to skip the highest level defensive promotion for their soldiers (healtpoints 4 and 3 for empire and barbarians, defence 2 for atlanteans respectively). A bit of math shows the reason: an atlantean or imperial soldier with full attack promotion will kill a fully promoted opponent in 5 hits. lacking the highest level defensive promotion, they still need 5 hits to kill. For example, atlanteans deal at most 48 damages, against an empire with healtpoints 3 (193 hp, 5% defence) they need 193 / (48 * 0.95) = 4.23 so five hits.

The only exception to this is against barbarian foes, as barbarians have higher attack and will kill a partially unpromoted soldier with 4 hits - sometimes they kill even a fully promoted opponent with 4 hits, a feat given them to balance their weaker evasion.

And high level humans will only use a soldier with full or no attack promotion, as there are micromanaging stragtegies to ensure that all promotions go to full and a fully promoted soldier is more cost-effective than a partially promoted one. So, as it is right now, there is no significant benefit to getting the most expensive defensive promotion when fighting against two of the three tribes. This doesn't look good.

I made a bit of math, and I determined that to kill any partially trained soldier with 4 hits, a damage of 50.9 is needed. To make the promotion appealing, it doesn't even need to happen every time, just often enough; an average damage around that value would suffice, while not disturbing the balance significantly.

Therefore, I propose that the attack gain per level of imperial and atlantean soldiers is increased by 1.2.

This will have no effect at attack level 0, but it will boost level 4 attack at 49.8 to 51.8 for imperials and 48.8 to 52.8 for atlanteans. Still no chance to kill with 4 hits a fully promoted soldier, you need over 55 damages per hit even against the squishier barbarians. The only real difference would be a greater chance for fully trained soldiers to kill unpromoted opponents in 3 hits rather than 4, but i think we can all agree if you are sending unpromoted soldiers against fully promoted ones you are going to lose anyway, and given how expensive fully trained soldiers are, it should stay that way.

In general, this small change will make all the healt and defence promotions more useful, as they will increase the survivability against a fully promoted opponent from 3 to 5 hits, while right now they go from 4 to 5, and have a hard time justifying their cost.

Top

Quote

-1

Barbarian soldiers are still the weakest.

Please don't change too much.

I don't see the problem.

So in my opinion, you make things very complicated.

Anyway, it's rare that there are just fights of fully promoted soldiers.

Wanted to save the world, then I got widetracked

Top

Quote

king_of_nowhere wrote:

in the high level play displayed in the latest turns of the 2016 tournament, a few players started to skip the highest level defensive promotion for their soldiers (healtpoints 4 and 3 for empire and barbarians, defence 2 for atlanteans respectively).

Yes, that's right

there is no significant benefit to getting the most expensive defensive promotion

I am trying to confirm or deny that

Therefore, I propose that the attack gain per level of imperial and atlantean soldiers is increased by 1.2.

What do you mean? We will change max and min attack by 1.2 more? So there will be:

13.0+9.2*level    instead of    13.0+8.0*level
AND
12.0+9.2*level    instead of    12.0+8.0*level

for min_attack?

For me it is way too much for barbarians

12.0+8.5*level

But I will prove it

This will have no effect at attack level 0, but it will boost level 4 attack at 49.8 to 51.8 for imperials and 48.8 to 52.8 for atlanteans.

Your proposal will change (mean attack) from 46 to 50.8 for Empire and 46 to 50.8 for Atlanteans.
I guess that this will cause huge imbalance for Barbarians level 10 soldiers.

Short study

I've made 10^5 (100'000) simulations of fights between soldiers (https://github.com/einstein13/wl_soldiers). The percentage is how many wins have attackers (left) to the defenders (top). Higher percentage means better soldier :).
Soldiers:

• bar_09 : 'attack':5, 'defense':0, 'evade':2, 'health':2
• bar_10 : 'attack':5, 'defense':0, 'evade':2, 'health':3
• emp_09 : 'attack':4, 'defense':0, 'evade':2, 'health':3
• emp_10 : 'attack':4, 'defense':0, 'evade':2, 'health':4
• atl_09 : 'attack':4, 'defense':1, 'evade':2, 'health':1
• atl_10 : 'attack':4, 'defense':2, 'evade':2, 'health':1

All values has measurement uncertainty of 0.4%, so "54.6%" and "54.7%" can be the same value.

Current situation

Proposal + 1.2 per level

Conclusion

There is new imbalance between Barbarians and other tribes.
Also new research (10^6 or even more simulations) is needed to have more accuracy here.

einstein13
calculations & maps packages: http://wuatek.no-ip.org/~rak/widelands/
backup website files: http://kartezjusz.ddns.net/upload/widelands/

Top

Quote

A bit longer study

The same situation, just 10^6 simulations

Current state

King's proposal

Uncertainty is 0.1%

My points

What I need from you is to think about barbarians too. If you add attack bonus for Empire and Atlanteans, you will give less chance for Barbarians. They need tweaking too. Give me any specific rules and I will make simulations to prove your point.

I agree that making more difference between lvl 9 and lvl 10 soldier will be better, but I disagree that we should leave only +1.2 attack for E & A.

Last thing is about lvl2 and lvl0 soldiers. I need to check if they have any chance for winning against lvl10. I remember that in B18 I was able to defend/attack with my lvl2 soldiers in early game. Even if other player had a few lvl7 soldiers. Swarm of lvl2 soldiers should be better than one or two lvl10 soldiers. If we change that balance, we will add benefit for very advanced players who are able to produce first lvl10 soldier in a short time. Other players will be in bad situation

EDIT:

Maybe I made mistake about Barbarians? Their lvl 10 soldiers are almost the same as before. But their lvl9 soldiers are very weak against lvl10 soldiers from other tribes: atl_10 56.8% -> 72.7% / 50.5% -> 34.5%, emp_10 55.3% -> 71.7% / 52.3% -> 35.6%.

einstein13
calculations & maps packages: http://wuatek.no-ip.org/~rak/widelands/
backup website files: http://kartezjusz.ddns.net/upload/widelands/

Top

Quote

I am really enjoying this discussion and the analysis!! Here is how I read the situation right now:

• I agree that king makes a good point that highest defense upgrades are not (always) viable, but should be to keep the game interesting. I agree.
• The proposed changes as simulated by einstein13 in his last post do address this. However, they weaken high level barbarians quite substanially.

I agree that we want to address the viability issue, but hopefully without affecting lvl9 barbarians as strongly.

einstein, what do you use to make those tables? Also, could you show (+- %) in the same cell for new values easily?

Top

Quote