Defending by running away megathread

[Seravy]

Okay, no one can be that stupid to seriously suggest implementing a "capture territory" system on the middle of nothing where the game does not even show an area to capture. At least not someone who claims he beats Lunatic too easily.
I have to conclude this is indeed a trolling attempt.

f you've done some analysis of that procedure then possibly you've done a schema of what's going on in there in pseudocode, or a state graph, or any type of analysis? Or was it all in your head?

You asked for it! I considered posting that at first but I thought..."No I'll be merciful and won't terrorize this poor guy with THAT screenshot."

Well, here you go :
   
This is the original, without the mod's changes, so the current state of the procedure is quite a bit worse.

(The chance to die is rolled in two separate places btw, that's what I meant by "redundancy".)

if there was a way to make the flee function behave like the retreating exhausted would you be less contrary to the proposal?

Well, doing that is probably the last requirement, it's pointless unless I believe the proposed solution actually improves the game, which I most definitely don't and at this point likely never will - I can't take you seriously anymore after you suggested city defense mechanics for an empty overland tile.

Speaking of, does the same function handle retreating exhausted btw?

Yes, fleeing, retreating, additional units that need room, etc.
wtf am I doing answering seriously to a troll. I think I'm the stupid one here.

[Suriname]

That's just a IDA screenshot, I meant something that you'd done as your own analysis. No?

You have addressed my points with your usual care. What do you think of the concept of game month as the token lost by the AI when the human uses the trick?

Flyers have to count. Other wise a teleporting hero plus a weak ranged unit can take away the city from 9 sky drakes. The drakes will believe they can win due to quick combat strength calculation, and will leave the city to chase down the ranged unit to avoid taking damage from the ranged unit shooting them. The hero teleports in, gets the counter up. The drakes come back to the city after killing the ranged unit. The hero teleports to a corner. Wait out the rest of the battle. Human takes city due to drakes not adding to the count.

Already considered and dealt with by the counter advantage to the defender. Drakes come back early enough - and force the hero to retreat - that the few counter points gained through that are not enough. The initial defender counter value just needs to be high enough to account for the vast majority of cases, and yet not allow for the biggest abuse that currently is easy to achieve.
Note that in any case this mechanic, by not being obvious, won't be so easy to measure and abuse: that's why the "complicated" proposal includes a random value as well. And if that's not possible the value just needs to be higher.

where's the issue?

1. If you're attacking then: Right now: turn 1 to 10, the AI goes towards your bowmen.You can spread the 2 bowmen and avoid it, plink away. It may lose 1 unit, and turn 10: it goes back towards the city. After the proposal nothing changes.

No. Currently in that scenario, the halberdiers never stop chasing - on the defense side against slow ranged units, they have no need to go back to the city on turn 10. So the bowmen can't stop and shoot or they die.

Ok, nothing still changes with the proposal. If there are slow ranged units but also something quick that might abuse this, they go back. So: are you left with any actual counter-example?

[Nelphine]

I don't think there can be a large enough counter. 9 great drakes defend. The human has a horsebowman and 8 unicorns. Due to quick combat calculations the great drakes will always come out and chase the horsebowman - and if they don't the human can bring extra ranged back up and just kill the drakes. The horsebowman runs to a corner, and the drakes come out. The unicorns bounce in. That's 8 marks for the unicorn. Repeat that as many times as you need. The counter would need to be set to ~150 to prevent the unicorns from taking the city. If it's set that high then your cavalry can easily stay long enough to increase it beyond where the attacker can even catch up, and then you're literally right where you are now.

So it hasn't changed anything.

[Nelphine]

Suriname

where's the issue?

1. If you're attacking then: Right now: turn 1 to 10, the AI goes towards your bowmen.You can spread the 2 bowmen and avoid it, plink away. It may lose 1 unit, and turn 10: it goes back towards the city. After the proposal nothing changes.

No. Currently in that scenario, the halberdiers never stop chasing - on the defense side against slow ranged units, they have no need to go back to the city on turn 10. So the bowmen can't stop and shoot or they die.

Ok, nothing still changes with the proposal. If there are slow ranged units but also something quick that might abuse this, they go back. So: are you left with any actual counter-example?

I don't need one, nor a fast unit to support them. Just with those 2 Bowman I can get a fairly high amount of 'counter' increases, to the point that halberdiers MUST stay in the city - at which point my Bowman can just shoot them with no fear of reprise. At that point my bowmen are literally draconian bownwn, and win the game for me. Or you set the base counter so high that it doesn't actually help the cavalry scenario that you think is so bad.

[Suriname]

The hero+8 unicorns example is rare enough (preposterous, is a better word) that it can be ignored. If someone knows the mechanics so well to
do that then he's doing much worse with the human defence side of the trick, that is to say, what's abusable right now. Case in point, my savefiles. And I don't even know the game that well.

And when there's a side with 8 drakes that side has quite some spellcasting skill too. I really want to see you keeping them alive. Casting on the two sides don't cancel each other when you consider the difference in unit defence and resistance values. And no, you can't make your example even more preposterous by postulating buffing on only your side. That example simply doesn't happen.

Your example with the bowmen: no, if your bowmen get inside the walls and remain there then the AI has quite an easy time beating them to death. If you plan to go there, score points, and leave all while the AI moves towards you then you're really good at tactical, that's skill that you have and it should be rewarded. That's not easy, quite the contrary of what I'm bringing up as a problem.

By the way. I don't think you could do that. But even if you could... Skill.

[Nelphine]

Have you looked up the stats on unicorns? Do you regularly play lunatic past 1410? Yes, you absolutely can keep unicorns alive. Yes, I absolutely will have more buffs than my opponents (my current game my main stack has: 72 buffs on it. The ONLY ai that can threaten this is sorcery with dispelling wave and aether binding. MOST ai with 9 great drakes/death knights/archangels, do not have both those 2 spells. I have 700 overland casting skill. I can cast literally scores of buffs per turn on whatever stack is important. The ai, even with 1000+ overland casting skill, doesn't know what stack is important to buff. And NO buff will allow great drakes to kill unicorns, except mystic surge. And even then only if it rolls haste.)

I could easily spread that around to 9-18 buffs for each of half a dozen stacks, and then the AI wouldn't even use dispelling wave even if they had it.

But all of that is irrelevant! Unicorns have great resistance. Put bless on them and by itself, that's plenty to stay alive for 25 turns against 9 great drakes and a chaos wizard with 400 casting skill. 

The whole problem with weaker tier units is that they can't HURT the stronger tier units without engaging and dying. But you give me a way that my units that cost half the amount and 1/10 the research can conquer every ai flier city? Hell yes I'm going to do that.

[Suriname]

But that's your main stack. With it... You're going to win, anyway. If drakes are so easy to deal with with your unicorns just web one and deal with it one at a time. Then, nothing changes, or at most you gain some time in an already won match. Do you hate having more of your time?
Anyway - if you've buffed your unicorns that much, then they possibly costed more mana and skill than the opposing forces. So how are you even surprised that they win?
Yours is another trick, it doesn't take skill either, but you wouldn't be winning that because of the proposal. You'd be winning thanks to a wholly different trick, and other abusable game mechanics. You've even named some of them in your post. I'm not going to bother. It's off topic in this thread anyway.

More importantly. Your example takes place at turn over 9000 with life and buffing. The abuse I am bringing up takes place at turn 40, independent of books or retorts. Do you really not see the relative importance of each?

So. Any example that can work and that is actually an example of abuse of this proposal and not something else?

[Nelphine]

Turn over 9000? No, this is talking about an uncommon summon, with bless, against very rares flying units. Whether you play it often or not the very rare tier is IMPORTANT. You can't have the very rare tier suddenly make the AI lose to uncommons. My particular game with a megabuff stack? I literally don't care about the buffs. They don't do anything for that stack except make me happy when I right click on them. Literally the only buff in the entire stack that matters, isn't a buff, it's a wraithform item for my windwalking hero. And even THAT doesn't really matter because I can just use airships. But the point of stating what I have, is that in the very rare tier that is a real thing. You CAN fully buff entire stacks in one turn in order to make an important attack. But even more important than that - no matter what buffs I give them, those unicorns cannot kill great drakes on anything like a cost effective balance. They can't HOPE to defeat the drakes. But you want to let them conquer the city??

You can focus on early game shenanigans. But the whole goal of your proposal is to make the game harder, which means, make it last longer. If you completely wreck the balance of the later tiers (and what do you think happens when the AI has 9 cockatrices in a city, and you attack with 1 unicorn stacked with 8 elite dwarf units that have such high resistance the cockatrices can't hurt them, but the dwarf units can't do any significant damage to the cockatrices? Use magic to kill one cockatrice, move into the city. Stand there for 20 turns. Now your counter is higher than the cockatrices, win the battle, without killing a second cockatrice. Fliers MUST count or by the uncommon tier you can abuse it) then you've made the game much worse than your perceived dislike for this tactic.

[Suriname]

The AI already loses to commons. But that's again another thread. OK, I see your point, let's keep the possibilities open.
Test the flyers counting and see how that goes, test without... I bet that in the process we can find fixes for those issues.
Quick example: adapting the city leaving algorithm to the new logic by accounting for teleporters, not on the entirety of the stack but on perhaps a few only.

Anything that makes tactical less foreseeable is good, so adding more randomness could do the trick.

... And if anything, just fixing the cavalry running pants on head and keeping the city (but not flying) would already be a good thing.

[Nelphine]

Ok, so agreed that fliers will probably need to count. You might be able to come up with something where fliers don't count, but I find it highly unlikely - fliers are simply too important from uncommon onwards.

So then, next is ranged units. What algorithm do you use to teach the AI when to go out after ranged units, and when to stay in the city? This is already a huge discussion that Seravy has tweaked several times. With your proposal, it becomes much more important. So, I'd like to hear what you think the exact conditions should be.