How Do They Do That?

[Tora]

File under "One rule for the AI, another for the Human Player".

One of my cardinals came across an aptly named Byzantine general, Prince Vlasios the Mad, besieging rebel held Smyrna. Nothing unusual in that except it was with just 2 units of heavy cavalry, a ladder and a ram, miraculously occupying the fort with the same force the following go, 2 turns before the defenders had to sally or surrender.
I took a screenshot of the assaulting force's chart but Imageshack is out to lunch or something at the moment.

Is there some way I can dismount my heavy cavalry or train my horses to climb ladders and use battering rams?

[katank]

Maybe the besieged sallied out voluntarily and were annihilated? Either that or it's really a bug.

[Quillan]

No, I don't think it was. I think that's the auto-calc at work. I believe what the game does is first check if you possess a valid method of assault: spy who opened the gate, artillery capable of breaching a gate or a wall, or constructed siege equipment. If you do, it lets you assault. Only then, if you choose to fight the battle on the map does it worry about the difference between infantry and cavalry. I've noticed that it only places as much equipment on the map as you have infantry units. You can still construct siege towers, ladders and rams with an all-cavalry besieging army. It's just that when you get to the battlemap it won't be there. I've done that a few times and hired mercenaries on the final turn to make the assault.

However, I've also noticed that the autocalc results are way off what they would be in a real assault when cavalry are involved. I can autocalc a battle and win with minimal losses where I'd be slaughtered trying to take the city, just because of the calculations. I think what happened is he constructed the gear, the game let him assault because he had the gear, and the raw strength of his army outclassed the defenders and therefore he took the city.

Want to confirm it? Try it with an all-cav army. Lay siege, construct rams, and don't fight on the map, just use the auto-calculate function. I bet it works for you too.

[BlackAxe3001]

I have had the same thing happen. It has worked for me.

One other thing that is kind of odd with auto-calc. I was sieging this castle and there was 13 defenders(yes only 13 men...), peasant archers, but I didn't want to loose hundreds of men to towers, so I used auto-calc and didn't lose a man taking the castle.

[Orda Khan]

It is one of the frustrating quirks of the game, there are many

......Orda

[katank]

Was this changed in the patch? I definitely couldn't assault even with autocalc with an all-cav army prepatch.

[sapi]

The human certainly can't assault with an all cavalry army (and autocalc is by far the best way to decide a siege, as it treats it as a normal field battle where the defenders can get away)

[OMGLAZERS]

The human certainly can't assault with an all cavalry army (and autocalc is by far the best way to decide a siege, as it treats it as a normal field battle where the defenders can get away)

If by best you mean, best way to win, yes it is.

It's also the best way to exploit the crappy Autocalcing :P

[sapi]

Correct ;)

I can't stand fighting offensive siege battles anyway thanks to the horrendous pathfinding.

[katank]

When you win, you take far fewer casualties (sometimes 0) through autocalc. However, some siege assaults that are lost via autocalc can usually be won by you manually.

[PureFodder]

Is auto calc better? If you don't have any artillery maybe, but with one or two artillery pieces, a bunch of archers and heavy cav even larger more powerful armies can be slain if you do it yourself. The dumb as a brick AI will always stick their cav and infantry next the the big hole you blow in the wall and leave archers in the town square. They refuse to engage the archers that are killing them from 2ft outside the big hole in the wall. Once they all die blow another hole in the wall and any remaining infantry and cav will wander to the new hole. Kill all these, leaving their archer to be mopped up by your heavy cav.

[katank]

Well, I typically blitz and don't have enough arty in my early armies to do such AI exploits.

[Arkatsson]

I remember a crusade was active for Jerusalem and I joined. So I'm sailing to the Middle East while watching the Egyptian Crown Prince with no other units attack a city with half a stack. I double click on him next turn and he has rams and ladders! Next turn he takes the city..... I've seen the bug too. Really stupid bug why didn't the creators see this?

[Murfios]

Happy New Year.


-Murfios

[Oaty]

Before getting on auto calc it has been improved vs RTW (IMO). It is much better at calculating field battles but makes assaults to easily favored for assaults.

The problem in RTW was 1 lone elite rebel unit could wreak havoc on autocalculate where it was obvious if you had fought out the battle yourself you'd easily annilate them, but hitting auto would lead to rediculous results on the campaign map.

A great example is how (RTW) 1 lone naked fanatic unit would shred a superior army to pieces. 6 hasatati vs 1 naked fanatic and you would lose 3 hastati units or close to it. Where even if it was you commanding the rebels you would be lucky to destroy more than 1 units worth of men.

So I do believe autocalc has greatly improved those sensleess battles but from my observatons does not seem to take walls into account

[Tora]

Autocalc doesn't seem to take those unreachable positions on mountains into account either - thankfully.
Just had a battle where 3 units of Santiago Knights were unleashed on a unit of Egyptian javelinmen who inexplicably had parked themselves smack in the middle of the Santiagos. At the first attack they retreated to a mountain stronghold which proved unreachable, a complete ring of red Xs around them, 6 knights lost in the process. Restarted the game and used autocalc, resulting in a victory with the loss of just one knight

[katank]

Note: autocalc still seems to heavily favor units that have multiple HP. Late game as the Turks, I would send around small 1/2 stacks of hashashim and autocalc every battle against the rebels. They absolutely wipe the floor with rebels and lose very few men doing so.

[Kraggenmor]

In my experience: If the Balance of power bar is anywhere near the midpoint autocalc = loss. Regardless of whether its a battle afield or a siege.

I only autocalc if I have a decided advantage or disadvantage and even then, I've often been dealt clear defeats through auto calcing with a distinct balance of power advantage in a siege.

[katank]

I just about never autocalc close battles. For sieges, I think you need at least a 5:3 advantage in order to do better with autocalc than manual. Any less and you often lose.

I think the whole point is for your to fight out most battles. Autocalc is for annoying cleanup. If one always autocalcs, then one is better off playing Civ.

[Kraggenmor]

If one always autocalcs, then one is better off playing Civ.

Agreed but, when you're on hour 16 of your third straight day of playing for 16 hours it would be nice to have a reliable auto calc to fall back on.

[katank]

In that case, going outside and getting some exercise or something is probably in order. Reliable autocalc is needed for the OMFG yet another rebel stack part.

If you are on the third day, you shouldn't have all that many close battles left. Can't you afford to build like 6 stacks of random troops and autocalc? Simply due to vast production advantage?

[dopp]

Turn off rebel spawns if they cause you so much anguish. I hate the way they pop up without any message or warning I only realize they are there when my income starts dropping dramatically.

[katank]

They don't cause me so much anguish. I'd rather they are in the game. However, many of the battles presented by them are simply tedious.

[Kraggenmor]

In that case, going outside and getting some exercise or something is probably in order.

I tried that once but that giant, hideous bright thing out there made my skin smoke.

If you are on the third day, you shouldn't have all that many close battles left. Can't you afford to build like 6 stacks of random troops and autocalc? Simply due to vast production advantage?

Assuming one was spending most of that time winning and expanding as opposed to getting beaten and stubbornly restarting then, yes.

[Goofball]

I tried that once but that giant, hideous bright thing out there made my skin smoke.

That may well be the funniest thing I have ever hear said at the Org.

Well done and thanks for making coffee shoot out of my nose...

[Kraggenmor]

That may well be the funniest thing I have ever hear said at the Org.

Well done and thanks for making coffee shoot out of my nose...

You're welcome, and thank you! Glad to have helped start your day off.

[DukeKent]

How does the computer determine the autocalc odds?

[katank]

Autocalc odds are some function of number of troops, stat of troops, valor of troops, general's stars etc. I don't know the exact numbers off the top of my head.

However, it doesn't seem to take towers into account for siege assaults. You will likely get away with far lower casualties when autocalcing those vs. fighting em out.

From my experience, for cost effective autocalcing, lots of cheap melee troops (peasants, militias, woodsmen etc.) commanded by a high star general is the best approach.

[Atreides]

From my experience, for cost effective autocalcing, lots of cheap melee troops (peasants, militias, woodsmen etc.) commanded by a high star general is the best approach.

It looks like a correction observation. Maybe the tekst with a 8 a 9 star general is correct, it was something like: 'he could even win an empire with just an angry mob'...

Fun.

[Carl]

So much misinformation in this thread....

First CA has confirmed that the battles in auto-calc are AI vs AI v=battles in a simplified and non-graphical version of the main battle engine.

Their are a few issues involved though.

Now we've never been told exactly how the engine is simplified but I have a few guesses based on a particular comment.

That comment is that animations are not taken into account in the auto-calc engine. That leads me to believe that the engine does not deal with things on a man by man basis, but on a unit by unit basis.

Tests by someone else on this forum also show that the type of siege equipment used in auto-calc does mater. Hence auto-calc CANNOT be treating it as a Field battle.

It's my belief that the unit by unit basis is part of the cause of easy auto-calced sieges. In effect if just one man makes it to the top of the wall, the entirety of both units, both at the top and bottom of the wall, are considered to be fighting so the advantages that come from having more men than the attackers fighting at the top of the wall simply aren't present.

In addition I think it's pretty clear to me that the effects of tower arrows are not taken into account which makes it easier still. The new manning mechanism for towers is probably at fault, the auto-calc engine is probably the same/similar to the one from RTW and as a result it can't make the new tower code work and treats all towers as inactive.

Lastly, the AI can get perfect charges and such in cities, as a result, since it is now controlling your Cav in auto-calc it can get perfect charges inside cities that you would never get. Although I think it is still using the RTW charge mechanism based on some results i've seen.


Regarding the power bar. All the power bar takes into account is the relative total attack, defense and numbers of each unit in the armies involved. As a result it will give you good odds when you haven't a hope. For example 20 Byzantine infantry (11 attack, 18 defense, 48 men, sword and shield unit), vs. 20 Papal Guard (12 attack, 16 defense 60 men, Spear unit) would give 4:5 odds on the power bar in favor of the papal Guard. In reality the Byzantine Infantry would massacre the Papal Guard because the Papal Guard suffer penalties vs. infantry that the bar doesn't take into account.