Why the AI is so bad at colonization?

[unmerged(29041)]

The test should involve emptying the exploration list. If you have a large enough list the AI will find places in the list to go.

There is the possibility that this is a Mac-related freakness. I doubt it (because the port is extremely compatible and well done), but it is important to know.

 

[sturmvogel]

And I'm a bit puzzled why de Gonneville made a rush for the Gulf of Suez first then to the Persian Gulf every single time. Unlike Fodoron's tests.

 

[Garbon]

And I'm a bit puzzled why de Gonneville made a rush for the Gulf of Suez first then to the Persian Gulf every single time. Unlike Fodoron's tests.

*nods*

This is what is strange to me, as I know I've never seen anything like that. I've never actually seen anyone rush to the Persian Gulf.

 

[Garbon]

While it is too soon to say anything definite, perhaps there is something wrong with the GulfSea on the mac version.

 

[zdlugasz]

Once or twice I have seen explorer (IIRC one English and one Spanish) in the Gulf of Suez. Additionally, sometimes I load one of colonial power POV and I have seen that they often know Gulf of Suez, however that knowledge might have been obtaned in other way.

 

[sturmvogel]

Fodoron and I are both Mac users, but I don't know about zdlugasz. That's why I was wondering why de Gonneville went to the Gulf of Suez first and Fodoron's explorers went to the Persian Gulf first.

I suppose I should really see exactly what Verrazzano does during his time on this earth.

 

[Norrefeldt]

I haven't seen the French explorers drawn to that woman in Isfahan while testing the current AI with PC. I can test in again, and if you like to compare, load up France 1520 in AGCEEP 1.42 and see what it is up to.
sturmvogel, the settings in post 54 look correct, as you removed
region = { South America West Africa }, which was both incorrect. Well, Africa and America was correct, but there are no regions called South or West.

 

[zdlugasz]

Two fascinating tales: Voyages of Gil Eanes

Setup: I am playing Castille and pumping all the money to Portugal, even become their vassal, Eanes appears in 1419, in August 1419 Portugal receives Cao's AI (expansion = 6 IIRC) without areas (other AIs are slept), modified Portugese event gives them 1000 ducats and -4 land and -4 mercantlism in 1419, war with Morocco removed

PC with AGCEEP beta 1.50

First try:
Eanes is so tired that stays in Tago till December 1421. Goes around Africa and TPs in Senegal, Casamanca are established, shortcut to avoid Kongo and Angola, TP in Table, Cape of Good Hope in November 1423, TP in Zambezia (later destroyed), May 1425 near Somalia, January 1426 Isfahan! Pal is so tired that had to recover his stamina and stayed 4 months, goes south and visit Table (colony with population 150) in January 1427, turns back and goes to Gulf of Suez (february 28) where stays in known places till April, in the meantime Portugal establishes TP in Nampuia, Good Hope in December 1428, July 1429 near Mauretanian Coast and using equatorial current goes to Winward Isles (december 1429) , TP in Zambezia, Eanes turns to Europe and sights of Portugese coasts in Gulf of Cadiz in May 1430 is enough for him! he turns south and traverses Atlantic once again discovering Lesser antilles that time in November 1430, turns back and goes to Azores, Canarian Islands, back to the Gulf of Cadiz and south around Africa and in March 1433 is back in Isfahan. Turns to Table where arrives in December and waits 3 months, goes to Casamance where stays 9 months. Portugal starts TP in Tassaret, Eanes discovers see around Cape Verde in January 1436 and sails back to Casamance, waits 4 months and sails back to the Gulf of Suez where arrives in October 1437. back to western Africa, central Atlantic, in March 1439 swims along Azores and Portugal offers to annex me . In April 1439 Eanes discovers north atlantic drift and in August 1439 discovers White See. In February 1440 finally goes back to Tago. I felt so deeply offended by Portugese offer of annexation that I have restarted game

Second try:
This time Gil moves his ass as soon as November 1419! One see square west of Portugal and known route along Africa (avoiding dangers of Kongo and Angola), TPs in Nouadibuch, Gambia and Leone are established. in Januray 1422 Cape Good Hope, first attempt to colonize Ciskei, August 1423 Somalia and in March 1424 Isfahan, that time climate or women were bad and he turns back immediately, TP in Mtawa is established, Gil swims to Mtawa and turns back, it is time to check what they offern in Gulf of Suez (June 1425). Third attempt to establish colony fails, first one in Transkei is undertaken, Gil swims to the coast of Mauretania (Mtawa gets destroyed and Transkei attempt fails) and tries his luck on west, TP in Niasa is established and finally colony in Ciskei. Eanes discovers Winward Islands in March 1427, swims back to the Mauretania Coast and back to the America discovering Lesser Antilles. Portugal establishes TP in Nouakshot. Gil Swims to the Gulf of Cadiz, Azores, little west and back to Mauretania, south and finally discovers gulf of Nigeria. Climate is too hot so Eanes turns back to the cental Atlantic, Azores and swims NW! to the New Foundland and St Lawrence in March 1430. And I notice that stupid Portugese AI is using my money to build Fine Arts manufactury! They offer me vassalization in april 1430. Gil turns back, to Azores and west to discover straits of Puerto Rico, back to the coast of Mauretania , I notice that Portugal started new TP near Guinea, Gil swims south, in Ciskei there is already city so I am sure that he wants to visit Isfahan or Gulf of Suez, around Good Hope and north, Gil avoids pirates in Delta of Rufiji discovering coast of Madagascar and goes north all the time, coast of Somalia and I thaink that Isfahan was his choice but game crashed

That is story of great voyages of Gil Eanes.


summary
- AI sends troops only when natives takes over province, if TP is wiped out AI does nothing
- Eanes did not like Gulf of Nigeria and kongo/Angola coast, he never visited Kongo and Angola and only very late Gulf of Nigeria (both travels), he always took "shortcut" south of Gulf of Guinea
- even on autopilot Eanes was able to go around pirate, or possibly this pirate was staying there year or longer
- Eanes did not need to go inside port to replenish strength, proximity of owned province did woners for him
- however proximity of TP did only small wonder and eventually he had to go back to the western African coast or Europe in his second travel
- lack of colonies in RPA drasticaly reduces number of visits in Arabian brothels


Edit: and I have increased Gil's movement to 5

 

[zdlugasz]

Third voyage of Tuf, ehm.. Gil

Eanes moves his ass in December 1419, one Atlantic sqare and south around Africa, TPs in Senegal, and Casamance are established, shortcut south of Gulf of Guinea, Good Hope in February 1422, TP in Table and later Zambezia are established, in April 1424 reaches Isfahan, after 2 months decides to leave this beach resort and goes south, near TP in Zambezia he regained his strength and turns back, In November 1425 visits Gulf of Suez, turns back, August 1426 Good Hope, (Zambezia was colony overrun by rebels at that time), in March 1427 reache coasts of Mauretania and enters Atlantic current, August 1427 Winward islands, turns back, in the meantime TPs in Mtawa, Lounga and Nouakchott were established, Lounga got destroyed later, Transkei attempt failed and TP in Nouadibuch failed or was destoyed (I have seen natives). Gil goes to the Guld of Cadiz and fully replenished visits a few Atlantic squares, In March 1429 is near Azores but turns back and in January 1430 discovers Gulf of Nigeria. Azores in June 1430, discovers Gulf Stream and Cheasapeak Bayin November , back to Azores, Atlantic and in May 1431 visit port in Tago.
After short stay swims west-north and in October 1431 discovers Newfoundland, later St. Lawrence, Bermudy in April 1432, in August 1432 Straits of Puerto Rico, swims south and in January 1433 discovers see around Cape Verde, swims to rest in Senegal. Rest was rather long, so there is rather strong suspition that black market connected with Arab slavers was already developed in Senegal. In June 1434 Gil moves west and discovers Coast of Brazil in December 1434, some Atlantic squares in August 1435 in back in Senegal. In October 1435 Eanes goes south, Gulf of Guinea shortcut but Gil feels stong urge and turns east discovering Gulf of Congo in MArch 1436, back to full Atlantic and south.
At this moment Portugal proposed to annex me, so I saved game and accepted proposal. And wow, I GOT ANNEXED!!!

According to save file Gil was heading stright into the Gulf of Suez, which forces me to change my suspicion regarding target of second Gil's voyage (when game crashed)

Summary: finally Congo was discovered and this time Gil discovered really many Atlantic provinces but when I have thought that he is cured from "Isfahan", call of Gulf of Suez appeared to be overwhelming
It seems that Isfahan is visited first, but maybe due to the lower quality of service in Isfahan Gil prefered Gulf of Suez

And yes, human player may be inherited (at least in normal diplomatic action)

 

[unmerged(29041)]

This are the type of events that I want to test to see if I can increase the speed of colonization by Spain. The system has four types of events:

1. Early events, before conquest of Aztecs. Spain needs lots of money and colonists.

2. Mid events, after conquest of Aztecs but before conquest of Inca. Spain still needs some money, and lots of colonists

3. Late events, after conquest of Inca. Spain no longer needs money, but still needs colonists.

4. Historically conquered provinces jump their population on correct dates to speed colonization without giving money or colonists.

These are examples of the events to be commented before I spend too much time on testing them. They can easily be made AI specific if we think is the best way to proceed.

##########################################
# Part I Semiannual generic events without Zacatecas, Cuzco, or Lesser Antillas

#(1495-1573) Signing of Capitulaciones for colonization
# by Fodoron
event = {
	id =
	trigger = {
		atwar = no
		NOT = {
			owned = { province = 27 data = SPA } #Zacatecas
			owned = { province = 169 data = SPA } #Cuzco
			owned = { province = 145 data = SPA } #St. Thomas
			owned = { province = 146 data = SPA } #St. Martin
			owned = { province = 147 data = SPA } #Antigua
			owned = { province = 148 data = SPA } #Dominica
			owned = { province = 149 data = SPA } #Guadeloupe
			owned = { province = 150 data = SPA } #Martinique
			owned = { province = 152 data = SPA } #Barbados
			owned = { province = 153 data = SPA } #Curacao
		}
	}
	random = no
	country = SPA
	name = "EVENTNAME" # Signing of Capitulaciones for colonization
	desc = "EVENTHIST"

	date = { day = 1 month = january year = 1495 }
	offset = 90
	deathdate = { day = 30 month = july year = 1495 }

	action_a = {
		name = "ACTIONNAMEA" # Name a new Adelantado
		command = { type = treasury value = 80 }
		command = { type = colonists value = 3 }
	}
	action_b = {
		name = "ACTIONNAMEB" # We are expanding too fast
		command = { type = colonists value = -1 }
	}
}

EVENTNAME;Signing of Capitulaciones for colonization;;;;;;;;;;
EVENTHIST;Under the unique colonization model adopted by Spain, the Capitulación, the conquest and colonization was not funded by the state, neither left to the private initiative, nor entrusted to monopolistic trading companies. The Capitulación was a contract by which the State retained full control of the conquest and colonization, that was to be carried out by private individuals under a credit system based on the hypothetical riches to be found. The Adelantado, who had to advance the cost, and was in charge also of colonizing the assigned territory, if succesful received land grants and titles, and retained 4/5 of the bounty to pay the cost and as profit. The crown provided nothing but paperwork. The system proved wildly succesful until it started to create problems to control such huge territory, and was carefully regulated and toned down in the Regulations of 1573. After that date, Capitulaciones became unfrequent. Use this money to help you colonize historically, or reject it.;;;;;;;;;;
ACTIONNAMEA;Name a new Adelantado;;;;;;;;;;
ACTIONNAMEB;We are expanding too fast;;;;;;;;;;

##########################################
# Part II Semiannual generic events without Cuzco, or Lesser Antillas

#(1515-1573) Signing of Capitulaciones for colonization
# by Fodoron
event = {
	id =
	trigger = {
		atwar = no
		owned = { province = 27 data = SPA } #Zacatecas
		NOT = {
			owned = { province = 169 data = SPA } #Cuzco
			owned = { province = 145 data = SPA } #St. Thomas
			owned = { province = 146 data = SPA } #St. Martin
			owned = { province = 147 data = SPA } #Antigua
			owned = { province = 148 data = SPA } #Dominica
			owned = { province = 149 data = SPA } #Guadeloupe
			owned = { province = 150 data = SPA } #Martinique
			owned = { province = 152 data = SPA } #Barbados
			owned = { province = 153 data = SPA } #Curacao
		}
	}
	random = no
	country = SPA
	name = "EVENTNAME" # Signing of Capitulaciones for colonization
	desc = "EVENTHIST"

	date = { day = 1 month = january year = 1515 }
	offset = 90
	deathdate = { day = 30 month = july year = 1515 }

	action_a = {
		name = "ACTIONNAMEA" # Name a new Adelantado
		command = { type = treasury value = 40 }
		command = { type = colonists value = 3 }
	}
	action_b = {
		name = "ACTIONNAMEB" # We are expanding too fast
		command = { type = colonists value = -1 }
	}
}

##########################################
# Part III Annual generic events without Lesser Antillas

#(1532-1573) Signing of Capitulaciones for colonization
# by Fodoron
event = {
	id =
	trigger = {
		atwar = no
		owned = { province = 27 data = SPA } #Zacatecas
		owned = { province = 169 data = SPA } #Cuzco
		NOT = {
			owned = { province = 145 data = SPA } #St. Thomas
			owned = { province = 146 data = SPA } #St. Martin
			owned = { province = 147 data = SPA } #Antigua
			owned = { province = 148 data = SPA } #Dominica
			owned = { province = 149 data = SPA } #Guadeloupe
			owned = { province = 150 data = SPA } #Martinique
			owned = { province = 152 data = SPA } #Barbados
			owned = { province = 153 data = SPA } #Curacao
		}
	}
	random = no
	country = SPA
	name = "EVENTNAME" # Signing of Capitulaciones for colonization
	desc = "EVENTHIST"

	date = { day = 1 month = january year = 1532 }
	offset = 150
	deathdate = { day = 30 month = december year = 1532 }

	action_a = {
		name = "ACTIONNAMEA" # Name a new Adelantado
		command = { type = colonists value = 3 }
	}
	action_b = {
		name = "ACTIONNAMEB" # We are expanding too fast
		command = { type = colonists value = -1 }
	}
}

 

[sturmvogel]

For the three Caribbean gold mine sequences that I added in 1.5 in Global_ProvinceSpec I added small amounts of population when the gold was found and removed it when the mines began to fail. Rather than add more events we could possibly add more population to them and not remove them later. They also add several hundred gold each to the treasury. They cover Tortuga, Moron, and Puerto Rico.

 

[unmerged(29041)]

Tests from colonization boosting events:
A) Set-up: A saved game for Spain in Dec 1491 with historic provinces and choices. DPs were INNO=1, LAND=1. Number of colonists per year was 4.80. Healthy economy but almost no trade, and over 200 ducats in cash. Columbus is at port in Andalucia.
The game was loaded as Magindanao and run until Dec 1500.

B) Variables: 12 boosting events for SPA set to trigger twice a year between Jan 1495 and Dec 1500 giving 80 ducats and 3 colonists as option A, and -1 colonist as option B. The trigger conditions are met only if at peace, but being at peace was not a test condition. The games were tested without or with the events. Number of provinces owned, number and level of TPs, number and level of colonies, number of city-level provinces, number of colonists under way and minimal number of colonists necessary to give that result were recorded. The number of tests (games) for each condition was four.

C) Results:

                         Provinces owned          City-level          min. # colonists
-----------------------------------------------------------------------------------------
Without events               3.25                    0                     3.75
with events                  5.75                    1.5                  21.0

The effect on colonization was very noticeable, a 560% increase in the minimum number of colonists sent. Basically the AI was colonizing 5 times faster with the events. In most games with events, the AI had already two fully developped provinces.

D) Discussion:
The effect is surprisingly strong considering that the amount of money given is not that big, 160 ducats/year. It seems that in fact the AI is using the money essentially for colonization, but still the AI is adding quite a lot of money from its own pocket. I would consider also the possibility that temporary lack of colonists and/or money might stall colonization more than it should, so the events might have a synergistic (bigger than additive) effect. In any case this tests prove that the AI does not colonize faster because is chocking in lack of money and/or colonists.

IMHO these events are needed, and we should discuss how to implement them.

 

[unmerged(40707)]

testing is good but it doesn't explain to me why especially France is so bad at colonizing despite big money and many colonists. There must be something in AI files too but I have no idea. In many games, I saw France building many manufactories instead and I suspect she was just saving money for that.

We could say the same for England but lack of money is maybe just the cause here and it is not as bad as for France in my tests.

 

[unmerged(29041)]

testing is good but it doesn't explain to me why especially France is so bad at colonizing despite big money and many colonists. There must be something in AI files too but I have no idea. In many games, I saw France building many manufactories instead and I suspect she was just saving money for that.

We could say the same for England but lack of money is maybe just the cause here and it is not as bad as for France in my tests.

We have to study each case individually. As I have done just with the first AI for Spain, every AI for any explorer nation should be tested and fine-tuned to produce the desired result. If you apply these events to France in the game period you want to test, you will see soon if lack of money and/or colonists is an issue. But my recommendation is to test the AI and fix it first, so the events can be functional.

My plan is to finish with Spain before beginning with other nations, but we will proceed much faster if more people check and try to fix other nations AIs, as sturmvogel and zdlugasz are doing.

The need for extra money for colonization by Spain, even when it is in perfect condition and under (ejem) expert hands, has been demonstrated by me in the first post of this thread. I don't think we should give money unless a need for it can be demonstrated. I am however more inclined to give free colonists to big colonizers without proof for need if it shows that it helps.

EDIT: I re-read this and I want to make clear that I am not suggesting that you test France yourself, Yoda. I will get to France eventually, but if somebody interested (French forum?) picks it up, it will be faster.

 

[Norrefeldt]

I'm inclined to give those events only to the AI. With this thread it's been proved that we cannot ever hope for getting the AI to colonise properly. Our alternatives as I see it are:

• Help the AI perform properly (which is cheating).
• Help the AI and player alike, reducing the challenge of playing Spain.
• Accept that the AI wont be succesful in colonising, reducing the late game challenge for all nations.

I have always been against adding AI cheating event, but in this case I think the alternatives are worse. This kind of events would apply to just a handful states that have massive colonising to do, and not each and every minor coloniser.

 

[unmerged(29041)]

I'm inclined to give those events only to the AI. With this thread it's been proved that we cannot ever hope for getting the AI to colonise properly. Our alternatives as I see it are:

• Help the AI perform properly (which is cheating).
• Help the AI and player alike, reducing the challenge of playing Spain.
• Accept that the AI wont be succesful in colonising, reducing the late game challenge for all nations.

I have always been against adding AI cheating event, but in this case I think the alternatives are worse. This kind of events would apply to just a handful states that have massive colonising to do, and not each and every minor coloniser.

AI-only or everybody is the same for me. The historic cost of colonization was widly variable, so a short-lived Scottish colony in Darien bankrupted the Scottish Government, while Spain colonized half a continent almost gratis. Game concerns have to be taken into consideration. While few people use AGCEEP for MP, a significant help to Spain would change the current situation, which I don't know what it is, since I don't play MP. Regarding SP, all big nations are extremely easy. I would argue that significantly easier than vanilla, since our events try to push the situation towards its historic result, and that helps historic winners at the expense of historic losers.
The way I have designed the events, if the human player touches the lesser Antilles, he loses the help. He will also lose it when he conquers the Incas and Aztecs, at least the money, that is no longer needed. So they should provide only moderate help, and a mild incentive to colonize historically, at least until 1534 for the money or until 1573 for the colonists. To put things in perspective, and according to the data that I presented in post 1, these events provide only 1/5 of the necessary money to somebody colonizing and conquering historically. You could increase that amount slightly by delaying the conquest of Aztecs and Incas, but why would you do that? The gold provides a lot more money.

 

[Norrefeldt]

The limitations of the events' design does make it possible to have them both for players and AI. The AI wont colonise ahistorically. I don't know whether the AI colonise while at war, at least it doesn't explore.
As they are right now, the option B are only handicaps for the AI, as a player wont take them. I'm fine with making it one-option (to maximise the AI gain) or give some 10s of infrastructure.
Perhaps not going totally against some chosen historical DP settings can be made a requisite too? That would further decrease the attractiveness for at least some players, while not affecting the AI at all.

 

[unmerged(29041)]

The limitations of the events' design does make it possible to have them both for players and AI. The AI wont colonise ahistorically. I don't know whether the AI colonise while at war, at least it doesn't explore.

Wars are very disruptive for colonization, even for us, humans

As they are right now, the option B are only handicaps for the AI, as a player wont take them. I'm fine with making it one-option (to maximise the AI gain) or give some 10s of infrastructure.

If the events are AI-only, then option B has no reason to exist. As I designed the events pending our decission if for AI or humans also, call me old-fashioned if you want, but I put option B under a honor system that if the player is not going to use the money to colonize he should not take it:

EVENTHIST;Under the unique colonization model adopted by Spain, the Capitulación, the conquest and colonization was not funded by the state, neither left to the private initiative, nor entrusted to monopolistic trading companies. The Capitulación was a contract by which the State retained full control of the conquest and colonization, that was to be carried out by private individuals under a credit system based on the hypothetical riches to be found. The Adelantado, who had to advance the cost, and was in charge also of colonizing the assigned territory, if succesful received land grants and titles, and retained 4/5 of the bounty to pay the cost and as profit. The crown provided nothing but paperwork. The system proved wildly succesful until it started to create problems to control such huge territory, and was carefully regulated and toned down in the Regulations of 1573. After that date, Capitulaciones became unfrequent. Use this money to help you colonize historically, or reject it.;;;;;;;;;;

I figured that a 5% discount in the help provided to the AI is hardly significative, as there will be about 80 of these events with money, we can predict that 4 will fail for the AI.

Perhaps not going totally against some chosen historical DP settings can be made a requisite too? That would further decrease the attractiveness for at least some players, while not affecting the AI at all.

Yes, this is a very good idea. Expected policies for Spain if choice A is always taken are:

          1479 1500 1550 1600
Aristo      6    7    8    8
Central     7    7    8    9
Innovat     4    3    2    0
Mercant     9   10   10   10
Offensive   5    6    8    8
Land        5    4    4    4
Quality     4    5    8    8
Serfdom     7    7    7    7

Which ones would you like to limit? I can think Inno=5 could be a máximum. Perhaps Aristo=5 could be a minimum.

 

[Norrefeldt]

If the events are AI-only, then option B has no reason to exist. As I designed the events pending our decission if for AI or humans also, call me old-fashion if you want, but I put option B under a honor system that if the player is not going to use the money to colonize he should not take it:

I figured that a 5% discount in the help provided to the AI is hardly significative, as there will be about 80 of these events with money, we can predict that 4 will fail for the AI.

I can understand that, but it as the AI has no sense for honor, and many players lack it too I don't think we should have it. We frequently tell new modders to fix their unbalanced events too.
As the idea is to help the AI, let's maximize its chances. Option A could cost a diplomat or reduce ADM for a few months, while option B could add/take something useless, and not reducing colonists.

Yes, this is a very good idea. Expected policies for Spain if choice A is always taken are:

          1479 1500 1550 1600
Aristo      6    7    8    8
Central     7    7    8    9
Innovat     4    3    2    0
Mercant     9   10   10   10
Offensive   5    6    8    8
Land        5    4    4    4
Quality     4    5    8    8
Serfdom     7    7    7    7

Which ones would you like to limit? I can think Inno=5 could be a máximum. Perhaps Aristo=5 could be a minimum.

Those are good. It sounds like a very centralised system too, but most players will go centralised anyhow.

 

[unmerged(29041)]

Did I tell you how much the AI of EU2 sucks? Well, let me tell you again:

The AI sucks!

More examples:

-The AI doesn't know how to count to 10. It will raise a colony to level 8, and then it might send 5 colonists together to that colony, raising its population to over 1300 and wasting 3 perfectly good expensive colonists. Moreover, that colony may remain at >1300 for a while and then the AI might decide to send yet another colonist to it. I've seen that several times already.
It seems as if the AI does not check the pop of a colony before sending a colonist. It certainly does not check the number of colonists left to reach city-level before sending a batch of them.
I believe the mechanism to be in place is that the AI sends one or more colonists to the only province in its list for colonization (only one receives colonists*), and right after sending them checks if the province is full city or not. If not, the province remains in the list regardless of what happens with the colonists sent.

-Land exploration seems to be done with the same list of naval exploration. But unlike naval exploration, land exploration uses all available conquistadores. However, what the AI does is trace a route from all conquistadores positions to a province in the area that occupies the top position in the list of areas to explore. Two explorers in the same province will therefore follow the exact same route, exploring the same provinces and changing to new goal ones at the same time. Additionally, land exploration has a strong tendency to stall when naval exploration is active. This is probably due to an area or region that is not land accesible occupying the top position in area or regions to explore. At this point, conquistadores will just sit iddle until the top slot in the list is again accesible to them.

Several strategies can be drawn from this hypothesis and facts:
-The order in the list of areas to explore is not irrelevant. Mostly land and mostly sea areas should be intercalated.
-It would be useful to periodically create short windows with only conquistadores, so land exploration is activated.
-Events that raise population during colonization are less useful to the AI because it does not count properly.

*A second province might receive one colonist and wait like that until the first one is finished.