Why no one should waste their time helping with this mod.

[billiard]

While I agree with IDLF's observations about game play, I don't feel that a deterministic approach is the answer. In essence, there are two components to this issue:
1. Countries often don't end up in the way we expect.
2. The AI can't handle the historical events as well as the human player, and therefore suffers a major handicap. This translates into (for example) a one province Portugal that is a patsy for Castile, and a free ride for the human player after the first 100 or so years when your vastly superior human intellect has created a superpower that the AI controlled powers run from in fear.

Number 1 I don't see as a problem; yes, I can understand that others may disagree, but I play the game as a "what if" proposition. If I wanted the countries to develop precisely as they did historically, I'd skip the game and read a history book, but this is game play. It's those variations from history that make the game fun. Yes, if you are playing as Austria and are waiting for that confrontation with Prussia, it might be dissapointing to discover that Frederick the Great never appears because Prussia has been absorbed into Poland and Brandenburg was swallowed by another German state. But play another game, and the situation will almost certainly be different.

That being said, I don't have a problem with minor AI cheats - hell, the game already allows the AI to have about twice as many troops as the human player, and attrition is nil even when the AI has three or four times the province limit for years in situ. The AI is certainly a hell of a lot smarter than I when dealing with revolts - a 2 to 3K army will defeat 12K or more of rebels, and it is a rarity when the AI fails to quell a revolt in a province contiguous to some portion of the country on the first try (as long as it's not fighting a war at the same time, anyway).

Yes, after seeing Portugal get clobbered in game after game, I agree that this is not optimal for game play. Sweden never turns in to the Northern Superpower it should be in the 17th century. Of late, Austria never gains Bohemia and Hungary. Again, I'm not being inconsistent - I like the variations that game play entails. However, I'd like to see some methodology by which the true big powers have a better than 50/50 chance of developing along close to their historic lines. So some AI "cheat" that gives these powers a leg up is fine with me - I just wouldn't like to see it become totally deterministic; a high level of variability is good. It's the lack of variability, when these states never achieve even close to their historical power, that has me frustrated.

Number 2 is problematic. For me, the first 100 years is a battle, but by the 16th century I'm well on my way to superpower status. By the 17th century I'm usually a hegemon in my region. The 18th century gets a bit dull, and am always vainly hoping (even if playing China or Mogul empire) that France has the revolution just so I can watch it unfold. Now part of the issue is the countries I choose to play - I haven't picked minor Asian nations or one province European countries, so I have given myself a big head start. But nonetheless, I rarely get attacked after about 1700 or so in AGCEEP (unless I border France, in which case they'll suicidally attack me every ten years like clockwork), whereas in vanilla I'd have countries ganging up on me constantly.

I play the game for the "what if's". I loved the KOI events that made for a new strategy whilst playing as Milan. I've proposed a series of events in a separate thread for a unified HRE and/or a unified Kingdom of Germany, while stating that these events should be very rare, but the known presence of which would alter (towards more realistic diplomacy) how Austria and France interact with the minor German states. I know we tend to view history as unchangeable facts, and rightly so. But in game play, we get to try to change history, if just for a few hours and just on our personal computers. I love the events, and wouldn't want to see these diminished in an effort to help out the AI. However, I wouldn't have a problem if these historical events were more positive and/or less onerous for the AI than for the human player. That would alleviate some of IDLF's concerns, while at the same time not diminishing the number of events for the human player.

 

[Tomas the Great]

That may well be the case, but is, at the very least unproven. I would speculate that it's a small number of events (less than 100) that actually make a difference, and that the effect of the other 95% or whatever it might be is so small as to be meaningless. That's just a guess on my part, but so is your theory.

Well, independently on this debate I tried to make a list of influential (scripted) events, which usually makes troubles, i.e. there is high percentage that they have dramatically ahistorical outcome killing the game. So far my list is about 25 events long. Never happened to me that all great powers were great, say, in 1648. (I learned how to help the game, rewrite the war section of the save files after roughly every half an hour of playing.)
But may be there is better solution: let us make the events without choices! No A, no B; simply this year this happened. Like the event: Czar captured by Tartars. Any choices offered? No, it simply happened. Charles the Bold died, so let him die and make Burgundy unplayable, so what?
This way events will not push things out of the road but the contrary it will help to restore the historical path.
Since HP is weaker in utilizing money (many things are twice that expensive for us) the events should hit harder with money.
But still I have been successfully fighting whole world as a "dishonourable scum" too many times to hope that anything than more cheats for AI will make the game challenging.

 

[unmerged(8894)]

Hey I just thought that it would be good to say here that i personally love the AGCEEP and have since the start and also personally find it to be much more enjoyable than the Vanella EUII.

 

[billiard]

In thinking a bit more about this topic, and reading Tomas the Great's synopsis of his research and proposed solution, I have a further refinement of my "AI cheats" statement.

I certainly understand Tomas The Great's desire (and IDLF, may he rest in peace, never ever to return, etc.) to make AGCEEP more of a competitive struggle in the SP game rather than a shoo in for the human player. I appreciate his efforts at looking into the "events" that may cause difficulty for the AI. However, I strenuously disagree with the effort to either decrease the number of events or to remake all events to be informational only.

I play AGCEEP because of the additional events, and the choices they allow the player to make that alter (in hopefully historically plausible ways) history by making different choices that actually were available at the time. Yes, when playing as Brandenburg it was a bit too easy to keep Wurzburg or to claim the throne of Bohemia, but that doesn't make these less historically plausible, it only means that the penalties in the game for making that choice were not sufficiently onerous to dissuade the human player on at least most occasions from taking the alternate choice.

That being said, and acknowledging that the AI doesn't handle these multiple choice events as well as the human player, how do we keep the human from beating up the AI even more severely when we add in more and more multiple choice events? IDLF's proposal seemed to be to cut the number of events substantially. Tomas the Great's is to do away with the multiple choice events altogether. But aren't the events the very reason most of us play AGCEEP? And even vanilla has multiple choice events, and I don't think anyone is proposing doing away with these (the AGC-noeventproject?).

My proposed solution is in two parts; the first, outlined previously, is to have these event choices either less onerous or more positive for the AI than for the human player. Hypothetical example - Historic event gives a choice of gaining a core on a territory. Taking the core might mean a -4 stab hit for the player and a revolt, but only a -1 stab hit for the AI and without the revolt. Not taking the core gives the human player a plus 1 stability and plus 1 centralization, but the AI gets plus 2 stability and plus 1 centralization (this part is debatable - sometimes I wonder if the AI can get hurt in RR by these types of slider adjustments). The human player has to think a bit harder now about whether he/she can afford this choice, but the AI isn't too bad off either way.

My second, and newer, piece of the proposed solution is to have the human player handicapped in diplomacy compared to the AI. Human player needs higher relations to gain RM's, and ESPECIALLY higher relations for diplo annexations and alliances, than the AI. I don't know if it can be accomplished, but I'd even like to see a long term averaging of relations, so that those last minute State Gifts just before annexation or RM proposals are no longer the killer tools they are now. This would force two things that would make the human player less aggressive towards the AI. First, you'd have fewer allies than the typical AI country - you are handicapped right there. Second, were you to use the typical aggressive "I don't care what my BB is" approach, you'd find you'd have even less allies and diplo annexing would become extremely hard without lots of cash for State Gifts.

To flesh out the "averaging" of relations from above, instead of the three state gifts approach right before a diplo annex that get you to +200 relations, you'd need an AVERAGE of +190 over the course of the past year (or pick another number if you prefer, this is just for discussion purposes) in order for the proposal to have a chance of acceptance. For alliances, maybe you'd need an AVERAGE of +125 over the past 5 years (or whatever game time has went on for years up to 1424) for the alliance proposal to have a chance of acceptance. This would prevent the human from winning a war, taking a core province of another country and force vassalizing them, then turn around and buy short term good relations to get into an alliance with your erswhile foe, forget about them for > 10 years, then again buy short term good relations (as you own one of their cores, their relations with you have steadily deteriorated in the decade or more since the war) to attempt a diplo annex. It would be akin to the newly unified Germany getting into an alliance with France after the Franco-Prussian war by giving them money, then for 10 years allow the hate in France to fester due to the loss of Alsace-Lorraine, then give them more money in the 1880's to try to convince them to become part of Germany. OK, so France wasn't a vassal nor is the example within the EUII time frame, but I'm more familiar with latter 19th into 20th century history and am attempting to give a real world example of the ridiculousness of the approaches human players use in the game. This may be a big reason why the AI has difficulty - we humans are much better at exploiting just these kinds of anomalies in the game. This proposal not only helps the AI, but cuts the ability of the human to "game" the game.

I have no way of knowing whether or not either of these proposed solutions to the weak AI dilemna is possible from a programming perspective or feasible from a computer processing power perspective. My reason for responding is to voice my disagreement with the positions that would either cut the number of, or choices inherent in, the events (with all due respect to the well-reasoned and logically argued points presented by Tomas and IDLF), and I would not be responsible were I to attempt to argue against other's (again, well-reasoned, logically argued, and additionally well intentioned) proposed solutions without at least offering some ideas of my own.

 

[unmerged(16323)]

In thinking a bit more about this topic, and reading Tomas the Great's synopsis of his research and proposed solution, I have a further refinement of my "AI cheats" statement.

I certainly understand Tomas The Great's desire (and IDLF, may he rest in peace, never ever to return, etc.) to make AGCEEP more of a competitive struggle in the SP game rather than a shoo in for the human player. I appreciate his efforts at looking into the "events" that may cause difficulty for the AI. However, I strenuously disagree with the effort to either decrease the number of events or to remake all events to be informational only.

I thought about a solution to that problem lately. And a simple idea came to my mind. The problem seams to be that the propability of AI choosing the B choice is too high and they too often choose them causing Ai countries to be weaker and ahistorical.

As far as I can see from teh event files mos events have only 2 choices: 1st is the historical one and the 2nd is just the opposite. The B choices oftern cause revolts and stability hits, because they're made for human player to allow them changing history. That's ok for the player, but bad for AI.

My idea is to supply tehse kinds of events with a mid-choice. A compromise between historical outcome and the ahistorical one. The best example are the szlachta's privilages events for poland, for example:

event = {

	id = 3468
	trigger = {
			domestic = {
				type = ARISTOCRACY
				value = 6
				}
		NOT = {
			domestic = {
				type = CENTRALIZATION
				value = 6
				}
			}
		}
	random = no
	country = POL
	name = "EVENTNAME3468"
	desc = "EVENTHIST3468"
	style = 1

	date = { day = 1 month = january year = 1495 }
	offset = 30
	deathdate = { day = 1 month = january year = 1498 }

	action_a ={				#Enact#
		name = "ACTIONNAME3468A"
		command = { type = domestic which = SERFDOM value = 3 }
		command = { type = domestic which = CENTRALIZATION value = -2 }
		command = { type = domestic which = INNOVATIVE value = -2 }
		command = { type = domestic which = MERCANTILISM value = 3 }
		command = { type = stability value = 2 }
	}

	action_b ={				#Enact a mild variant#
		name = "ACTIONNAME3468B"
		command = { type = domestic which = SERFDOM value = 2 }
		command = { type = domestic which = CENTRALIZATION value = -2 }
		command = { type = domestic which = INNOVATIVE value = -1 }
		command = { type = domestic which = MERCANTILISM value = 1 }
		command = { type = stability value = 1 }
	}

	action_c ={				#Refuse to enact#
		name = "ACTIONNAME3468C"
		command = { type = domestic which = SERFDOM value = -1 }
		command = { type = domestic which = CENTRALIZATION value = -1 }
		command = { type = domestic which = INNOVATIVE value = 1 }
		command = { type = domestic which = MERCANTILISM value = -2 }
		command = { type = stability value = -1 }
		command = { type = revoltrisk which = 24 value = 2 }
	}

}

Even if the AI doesn't choose A it will most likely go for B. It still makes Polang go in the historical direction and there is additionaly choice C provided for the human player. I never actually saw a not historical Poland (mayby except for when I play it ) so this approach gives good results. Any thoughts about this?

 

[Norrefeldt]

You dont get a much better choice by the AI. With two choices the ahistorical b is taken 5.6% of the time. With three options the c is taken in 5.3% of the time.

 

[unmerged(16323)]

You dont get a much better choice by the AI. With two choices the ahistorical b is taken 5.6% of the time. With three options the c is taken in 5.3% of the time.

Oh.. seems like it wasn't a great idea after all

 

[Tomas the Great]

Oh.. seems like it wasn't a great idea after all

Unfortunately, even the idea of so called sequence, which in later events is trying to balance 'a wrong choice' from the first event to follow the historical path eventually, does not work. I have to say it despite my admiration of people who made these sequences and managed to base them on historical plausibility.
There are simply far too many events for things not to go wrong.

 

[Old Joe]

topic starter is just funny guy, his speech like we'r in corporative board session, advocating his options through cheesy pR, really meaning his work just not been properly noticed

strange for me, but in general i'm really satisfied how this mod is acting by now