Defending by running away megathread

[Nelphine]

Per my examples (I await your reply to them with baited breath), so far, your proposed solution makes the game much more abusable than it currently is, regardless of whether you dislike the tactic in question.

And yes, the AI knows how to do the tactic with cavalry as well. Unfortunately they only know how to judge a battle based on quick combat strength of an army (not of individual units, only the whole army), and they completely cannot account for magic (again, coding restriction). So unless the cavalry is alive, and the human army largely outnumbered them in terms of quick combat strength, the AI literally doesn't know that it needs to run away with the cavalry. But since cavalry are fast they're usually the first units the ai loses, before they realize they might need the tactic.

[Seravy]

Guess I have to repeat myself again : 

I've explicitly stated before that I'm NEVER changing the fleeing mechanic again because it's that hard to change. It's not always 50% by the way, see the game mechanics thread.

But if we can't even agree on the fact that fleeing to hold a city is a problem then I'm not going to bother tbh.

It's not a large enough problem to change the core mechanics of the game and spend several hundred hours on implementing it plus risk the possibility of even more severe abuse surfacing due to the change. Even you admit it's only a worthwhile tactic on small, worthless settlements where there is nothing to lose anyway. (In which case it might even be better to leave the settlement and let the AI hold it for you for a dozen turns - they'll build stuff for you and you can avoid losing units. People don't use the running tactic because it's good - they use it because they hate to lose cities, basic psychology.)

[jsb]

You're just treating me badly because I'm a contrarian asshole. But have you read the yom kippur war pdf? Contrarian assholes are useful.

*Sigh* You're seriously so butt-hurt that Seravy doesn't like YOUR proposal for HIS mod that you start multiple threads about it, whining and throwing superlatives and insults - then you try to lecture him and draw some kind of asinine comparison about leading a game-development to the effing Yom Kippur war?? OK, buddy - maybe time to work on your communication skills instead of blaming everyone else for not realizing what a useful genius you are?

[Suriname]

No, it's not a tactic worth on only a small subset of cities. It's a tactic that is worth to use on your empire, specifically, by planning to stop the enemy at the worthless small settlement. Sure, you implement the trick there, but it matters on the entire realm. So, basic psych - use it because you don't like to lose something - is not relevant here. You use it to avoid risking the capital.

Unless you mean with basic psych that people don't like to lose the game. Then we're back to wraiths.

I'm not making this up, check my savegames and what I've written to Nelphine. I've stopped the enemy at the gates only thanks to this stupid trick, the capital was one city away but never risked anything, so "zeroing" as we say in climbing the difficulty of the climb. Due to this, I can't claim that I've won at lunatic. I've only applied a monkey trick to "beat the game".

Are there any hopes that you'll engage in the arguments I've made? Or are you just pretending to converse?

On fleeing: yes, I've read that. Could you either,
- point me to the relevant thread
- make a short summary here

Now answering to Nelphine.

[Suriname]

Wraiths

It's just an example, I could name 10 other that Seravy fixed when he first started working on this and was still working on this mod for the challenge. I wonder why he's doing it now. But anyway.

Wraiths are worse you say. I say wraiths only worked with many Death books, so this trick is worse. It doesn't matter! Let's say that you are right, ok? Then why don't we want to get rid of another trick, even if it is less worrisome as you say? You'll agree that the direction is the same as removing wraiths: removing a trick. As I've proven it doesn't bring any richness to the games, only detracts from it.

And it really is worse than wraiths, you couldn't win a 3 fronts war with wraiths - you still needed 10s of turns to summon the first one so it could help a single front. But it doesn't matter, if this is false, then it'd still be worthwhile to get rid of the trick. Because it's an easy trick, not a strategy, and because it's fucking ridiculous. It's running away from a city and keeping it too. Have cake and eat it.

One particular:

This tactic allows you to keep a city, assuming the AI doesn't have the correct spells to defeat it.

No, it always allows you to keep the city: AIs target ranged units first. So, you only need to build 2-3 archers. You don't even need barracks, alchemist etc: just the second cheapest unit. See my savegames example, no barracks or alchemist. Where's the richness that Seravy vaunts this trick to bring then? Ah, and then as soon as you have catapults you can upgrade to those, for little production time, and you get more staying power - spell absorption - still without barracks or alchemist. Anyway, again: even if you're right and it's not as effective a trick as wraiths, why not continue on the same path of the wraiths trick removal?

Ever fought Djinn in a node? or air elementals? AI wins LOTS with this tactic. You may consider frontier cities important; I consider nodes in 1403 important.  The human is powerless against this tactic RIGHT NOW.

I'm not following you here, we're talking about cities. What I propose doesn't change nodes, I initially coupled it with node treasure but that was an option, it's not necessarily linked.

Any mention of quick combat doesn't help your case. Both I and Seravy (and everyone else I believe) agree that quick combat is a problem; however, it is a coding space problem and simply cannot be fixed.

Only if the mention is in the sense of changing strategic, but I have done the contrary, I've raised that by changing tactical the two can be brought to be closer to the former, thus reducing the unfairness towards the AI (or the player, if you believe that strategic helps the AI more than this trick helps the player, but that's incorrect as I've shown: beating 3 front early war easy at lunatic)

This proposal therefore helps with the objective of bringing the two closer together, that you like it or not...

Example 1: if fliers can't hold cities, then the most important units in the game, in late game, cannot hold cities. The AI can't deal with this. Regardless of any unit restrictions (and yes, when you both have a ton of skill, units still matter; skill cancels skill, and units end up deciding the victory. As I usually play my lunatic games until 1415ish, I regularly play games where the AI has 600+ combat casting skill) by teleporting into the city on any given turn, I immediately win because you don't want fliers to hold cities. That simply does not work. (And in fact, would stop working as soon as uncommon summons came into play. Flight is only overpowered while the AI can't deal with it - the common tier of the game. By uncommon tier, flight is so common to all players, that it becomes the standard, not the exception, and is no longer overpowered.)

Ehm? Flyers can still hold cities. All they need to do is to attack the units inside the walls and not have the cake of the city and eat it too (keep their hit points), or prevent them from entering (which is quite easy with the stupid gate trick, but that's another thread).

Your example is valid only in the subset of cases where a slow flyer defender is attacked by a fast ground attacker, but I can make that subset exceedingly minor by introducing a defender initial advantage to holding the city (which just makes sense btw so thanks for that). So, the problem happens only when the fast ground attacker holds the city for a number of turns > of the initial slow defender counter advantage (which we can define as we want - say, the defender starts with counter=10), survived not only the spells but also the fight, and then abused its speed. In short, a super rare instance, making it not abuse anymore, as it would actually require skill.

The AI is well equipped to deal with this thanks to the priorities btw. You could in theory bait your theoretical slow flyers out with ranged troops, and while they go after the ranged troops occupy the city. But:
1. the spells target ranged troops first, making this an expensive strategy, so it doesn't really matter because it has a consequence (unlike using the trick in defense)
2. the defender advantage can be set as high as needed, and could change based on the number of troops present in the city to make it relative to troop strengths: defender advantage = k*n=number of attacking troops/number of defending troops.

A good k is 10: meaning that if there's a single unit in attack and 1 in defense, the attacker needs to hold the fort for 10 turns more than the defender needs.

To even make it more sure that the mechanic is not abusable, the counter shouldn't be visible and should be randomised a little: def adv = random (k-1, k+1)*nA/nD. Voila'. We should have a way to see it in test mode, but not once the change goes live.

Before anyone says it's too complicated: come on, it's one counter per side, the defender side with an initial value of def (one single calculation) and the attacker side with an initial value of 0...

I'm editing the initial post to reflect this.

Example 2: currently, if i attack with 2 ranged units, then the ai chases me mercilessly with its 4 melee units. If I stand and shoot, I die. If I run away, then I retreat exhausted having gained absolutely nothing.  Under your proposal, if the ai chases me mercilessly, he will not be in his city.  Therefore, by a little judicious choice of where to run (of which the human will ALWAYS be better), i can either ensure that I conquer the city if the AI is not taught to return to the city; or if the AI IS taught to return to the city to prevent losing it due to 'not defending it', then my ranged units will eventually not be pressured, and can fire all of their attacks on the AI melee units.  Repeat 4 times, and I win the city.  In either case, I win the city in a scenario where currently, I could not win the city from the AI.  (In effect, your suggestion actually makes ALL ranged units into fliers when attacking, which, is FAR worse than the current topic of cavalry running around; even worse, unlike the current running away tactics, the AI can't be taught how to replicate that at all. The AI can't be taught how to make those exact movement choices that will allow it to potentially steal the city with its ranged units; nor can it be taught to understand that it might be helpful NOT to stand and shoot for x number of turns.)

This scenario just doesn't happen. Try to attack a city with just 2 archers and you'll see why. (If the AI is out of mana that's another issue, and you've already beaten it)
Anyhow, if you bring a stack with some staying power, and the AI comes out, then it goes back to the city on turn 10. First of all, if it's - say - 9 rangers, then by that point the AI is going to slaughter half of them with spells, don't forget the priorities. Even then, if you are still worried about the delay - "a little judicious choice of where to run" - then the same fix of a defender counter advantage fixes it.

I have considered your examples and made a minor change to my proposal to account for them, making them almost impossible to happen. Thanks for actually showing me a weakness of the proposal, but please acknowledge that I have considered what you brought up, thought about it, changed my thought and proposal accordingly.

Do you see any other weakness to help me make the proposal better?

[Suriname]

You're just treating me badly because I'm a contrarian asshole. But have you read the yom kippur war pdf? Contrarian assholes are useful.

*Sigh* You're seriously so butt-hurt that Seravy doesn't like YOUR proposal for HIS mod that you start multiple threads about it, whining and throwing superlatives and insults - then you try to lecture him and draw some kind of asinine comparison about leading a game-development to the effing Yom Kippur war?? OK, buddy - maybe time to work on your communication skills instead of blaming everyone else for not realizing what a useful genius you are?

This has entertained me

Read the PDF, it's a good read. It's not about the Yom Kippur war, that's merely its localisation. It's about groupthink instead. Can you see the relation now, oh faithful whiteknight?

[Suriname]

Per my examples (I await your reply to them with baited breath), so far, your proposed solution makes the game much more abusable than it currently is, regardless of whether you dislike the tactic in question.

And yes, the AI knows how to do the tactic with cavalry as well. Unfortunately they only know how to judge a battle based on quick combat strength of an army (not of individual units, only the whole army), and they completely cannot account for magic (again, coding restriction). So unless the cavalry is alive, and the human army largely outnumbered them in terms of quick combat strength, the AI literally doesn't know that it needs to run away with the cavalry. But since cavalry are fast they're usually the first units the ai loses, before they realize they might need the tactic.

Right, forgot this. OK, thanks for the explanation of why the AI doesn't use the cavalry tactic despite knowing about it. I wish it did, because then Seravy would be forced to fix the trick due to all the complaints that that would engender.

[zitro1987]

I forgot to mention the following:

The original game technically has this same problem, but the AI wasn't taught to take advantage of it.

So it appears a gold bonus of reaching 75% destruction may not be feasible with coding limitations. So my proposal of rewarding attacker for wrecking town may never happen. That leaves:
*teach AI to re-prioritize nuke spells from slow ranged to escaping units, if attacking a city with very little defense.
*teach AI to re-emphasize city destruction and not chase a much faster unit, especially if city is big.
*raising the destruction cap (though does not really solve anything if town is small)
*[Maybe] kick defending units that are out of city when exhausted and some trigger of max city destruction and massive difference in strategic strength  (I now say maybe, because player can bring them all back in turns 23-25). In other words, in deeper thought, my idea #1 may not be worth it and would only be workable with non-intuitive turn-based counters and complex conditional logic (with hex coding?)


The front hole in the wall will be radical/difficult to change. My ideal scenario is melee being able to hit through walls anywhere with 5% chance of destroying a wall in the process, but closing that front hole in the wall.
Problems:
*making more spots walkable within that 4X4 (including corners and center) and teaching AI to move around to engage any part of wall and move archers out of the borders.

[jsb]

You're just treating me badly because I'm a contrarian asshole. But have you read the yom kippur war pdf? Contrarian assholes are useful.

*Sigh* You're seriously so butt-hurt that Seravy doesn't like YOUR proposal for HIS mod that you start multiple threads about it, whining and throwing superlatives and insults - then you try to lecture him and draw some kind of asinine comparison about leading a game-development to the effing Yom Kippur war?? OK, buddy - maybe time to work on your communication skills instead of blaming everyone else for not realizing what a useful genius you are?

This has entertained me

Read the PDF, it's a good read. It's not about the Yom Kippur war, that's merely its localisation. It's about groupthink instead. Can you see the relation now, oh faithful whiteknight?

Oh - THANK YOU - I didn't REALIZE that must be because I'm a monkey right?

Do you have any more "good reads" lying about your hard-drive perhaps? I think we all need to be more enlightened... 

Look, I'm not overly fond of the running-away mechanics myself - but I certainly won't get dragged into a "discussion" about it if part of it is going to be having to read smug posts by a self-proclaimed contrarian asshole. You might think it's thought provoking or edgy - or whatever - but it distracts from any real ideas you might have had about the mechanics - which was MY POINT! It's certainly not wrong to have a different opinion on game mechanics. YES, even contrary to the work derived from people who spent thousands of hours actually working on them - it's fine. But here's the lesson: show a little common courtesy and don't expect anyone to respect or listen to you until you do. 

TL;DR - Nobody will read your pdf-files, any threads where you post this way runs great risk of degrading to name-calling and shitposting - just like THIS!

[Seravy]

Sure, you implement the trick there, but it matters on the entire realm. So, basic psych - use it because you don't like to lose something - is not relevant here. You use it to avoid risking the capital.

Except, you know, there is such a thing as AI priorities. They won't go after your small settlements if they have a strong enough army to attack the capital and can reach it. The AI might be dumb but not that dumb fortunately.

...I looked at your save. You barricaded the bottom half of the continent where blue is coming from, making ALL your cities in the upper half unreachable for the blue AI. That's what's causing you to win - you aren't letting the AI attack cities where they could be doing real damage. Given the state of the map, they can't possibly attack anything except Cantebury unless they have an intercontinental stack. The presence of the city there is not even relevant - you can have the exact same result by using the running tactic on a noncity tile which the AI is forced to attack instead - and they will do so if they can't reach a city. So this save only proves one point - that changing city mechanics would make no difference at all.

On fleeing: yes, I've read that. Could you either,
- point me to the relevant thread
- make a short summary here

Well, if you really want to know, download the ida files I posted yesterday, open up wizards.idb, find the fleeing procedure, and try to understand how it works, and figure out how it can be changed. If you are willing to accept a shorter answer, I don't want to deal with that procedure. It's complicated, has bad redundancy and is confusing.