Todo list for 5.44

[robinh3123]

If you've already defeated (or effectively defeated by weakening them such that they are no challenge) 3 AI by 1411, you have successfully rushed, even if not a crazy super rush. The third is expected to be a strong challenge, and not be defeated significantly earlier than 1415.

So, if you're strong enough to have defeated 3 by 1411, then you're too strong - NOT the AI being too weak.

I don't necessarily agree with this concept, as particularly on lunatic, I think it's not reasonable, but that's the design goal.

So if you defeat an ai before 1405, then you snowball and then a second before 1410 then double snowball. These acts are putting you outside the expected - progressive wouldn't solve that, the AI would still be dead already, and things like expected research get thrown out the window.

Could you specify in which difficulty level were you referring to, for your 1411, 1405, 1410 statements?

[zitro1987]

I totally agree.   I even want to suggest more snowballing effects.   When I play to a point, that I think I can win, but too lousy to win.... I quit the game, since it is too boring.
I am too powerful, but can not end the game fast... too boring.

I suggest:  When a wizard is banished for more than one (perhaps two) times, he/she can not cast spells in the following battles until his/her word of return spell is completed.  This makes snowballing easier.

I'm also thinking of a similar concept that help quicken the game and its snowballing effect.
*Wizards can only cast spell of return 3 times. Afterwards, it is automatic defeat

[Nelphine]

If you've already defeated (or effectively defeated by weakening them such that they are no challenge) 3 AI by 1411, you have successfully rushed, even if not a crazy super rush. The third is expected to be a strong challenge, and not be defeated significantly earlier than 1415.

So, if you're strong enough to have defeated 3 by 1411, then you're too strong - NOT the AI being too weak.

I don't necessarily agree with this concept, as particularly on lunatic, I think it's not reasonable, but that's the design goal.

So if you defeat an ai before 1405, then you snowball and then a second before 1410 then double snowball. These acts are putting you outside the expected - progressive wouldn't solve that, the AI would still be dead already, and things like expected research get thrown out the window.

Could you specify in which difficulty level were you referring to, for your 1411, 1405, 1410 statements?

Last time seravy discussed this, it was for all difficulties. As mentioned, I do believe that higher difficulties inherently go faster, but so far the design goal is for all difficulties.

[Nelphine]

I totally agree.   I even want to suggest more snowballing effects.   When I play to a point, that I think I can win, but too lousy to win.... I quit the game, since it is too boring.
I am too powerful, but can not end the game fast... too boring.

I suggest:  When a wizard is banished for more than one (perhaps two) times, he/she can not cast spells in the following battles until his/her word of return spell is completed.  This makes snowballing easier.

I'm also thinking of a similar concept that help quicken the game and its snowballing effect.
*Wizards can only cast spell of return 3 times. Afterwards, it is automatic defeat

@robin: this was actively changed away from what you describe (which is the original banish mechanics) because it's too easy for a human to fortress strike the AI with the humans only offensive stack and then without combat spells, destroy a huge portion of the AI empire with random units that have no value in the war otherwise.

@zitro: particularly on larger difficulties, later in the game, that still allows the human to win via fortress strike, even though the human might be losing multiple cities per turn to the ai, and the human only has one stack capable of fighting the AI (the one being used for fortress strikes). 

@both: you'd need a way to tell that the AI was no actual threat anymore, which is why the surrender mechanic was introduced. (And I've seen 3 AI surrender, so it's not a bad for for what it's supposed to do.)

[zitro1987]

I figured that if a player gets banished more than 3 times, they deserve to lose. Same for an AI, which avoids less interesting war of attrition gameplay.

[Nelphine]

I understand that when you're winning that makes sense.

Unfortunately, the AI can only play attrition warfare. It can't make surgical fortress strikes (unless the human actually leaves their fortress unbelievably weakly defended even while the ai doomstack moves straight toward it, but my assumption is no human trying to win does this). So it can only win via attrition. 

So in a sample game, the human has one stack capable of defeating any ai city. And against any city it would take losses. However it's also strong enough to take the AI fortress albeit with even more losses. 
The ai on the other hand has several stacks capable of defeating the human cities, and is destroying/conquering at least 1 city per turn even while the AI is banished, and often multiple cities per turn particularly when not banished. And the AI takes 1-3 turns to cast spell of return.

If both the human and AI have around 40 cities, the ai will win, despite being banished ~6 times during the process.

And this is actually a fairly basic 'human vs myrran ai' scenario, assuming that the AI actually break the towers. 

Any anti-attrition changes would result in the human winning, because the human can choose to fortress strike, and the AI (effectively) can't.


What the suggested change actually does is make it so the ai can't win unless the human is so weak that they literally can't attack the AI fortress. That shouldn't be the only AI victory condition - the AI should be able to win actively by its own actions - but the only actions it can do are attrition based.

[Bahgtru]

If you've already defeated (or effectively defeated by weakening them such that they are no challenge) 3 AI by 1411, you have successfully rushed, even if not a crazy super rush. The third is expected to be a strong challenge, and not be defeated significantly earlier than 1415.

Except that this stems from an example of game by the very person that is designing the game, so all theorycrafting can be put aside - this is how it's going to be, and you can define it rush or not, it does not matter.

So, given this state of reality, you can 
- change things slightly to make the end game more challenging
- argue that it's impossible and do nothing
- remove the end game (basically... the victory conditions proposals in the last 3-4 comments, which remove the end game to different degrees)

I remain convinced that giving AIs a gradual upgrade makes things better for the game. Again: rush or no rush. I insist on this: a progressive bonus damages rush strategies relatively MORE than economy ones, as rush is by definition better early but worse late compared with an economy strat. If these strats eshew this definition in the game then there's a balance issue that needs to be addressed.

[Nelphine]

What? How does him being the designer possibly mean that every game he plays must perfectly match his design? That's why they're called test games, so he can see first hand how the design is doing.

I've been a proponent of progressive bonuses before (and in some ways still am) but your reasoning for them doesn't make sense of me.

I'd like to see more games from people, not seravy or myself, where the player uses at most 2 retorts (possibly also without using ghouls). We know there are still balance issues, let's see how bad they are. If in those games, people consistently find that the early game is fine (meaning difficult, the ai sometimes wins, but not impossible), but the late game is too easy, then an argument for a progressive bonus is much stronger.

[Bahgtru]

It's not because he's the designer, but it's because he's got a gaming philosophy that is probably the most self-restrained in the board. And despite that he still manages to do what you call rush. And of course that's the case for others too. Every game I've ever seen in videos or screenshots here reaches that point, rather sooner than later. That's because the fixed advantage is lacklustre.

How's the reasoning so difficult? It's pretty simple: rush is worse than economy in the long run so a progressive bonus damages rush more than economy. That's it.

[Bahgtru]

To elaborate:

Yes, that's the same thing with different numbers.
It's not helping, the only problem with current AI is it's too vulnerable in the early game. The game has exponential growth, if the AI doesn't lose territory early, they are usually powerful enough.

What does "powerful enough" mean? Absolutely nothing. Because it's absolute. (ha... ha...)

What matters is relative strength of different strategies compared, and a progressive bonus helps slow strategies by being comparatively more damaging to fast strategies. Currently slow strategies are weaker than fast ones, so a progressive bonus is desirable.