Why the AI is so bad at colonization?

[Toio]

I assume that to colonise a state needs both a positive net flow of colonists, and an AI file with expansion value 1 or higher.

the default explorer which lives for 30 years is also bad, one of these is enough to ruin the game.
I always suspected it was due to having naval DP set at 0, 1 or 2 for a very long time.

I cannot recall the old thread mentioned on when nation get default explorers

VEN has no problems in getting colonists , so to stop them , you need to prevent them getting a "defaulty"

 

[Norrefeldt]

A default explorer is given by event, national or random. The random events only happen for certain DP setting with at least naval 11 (see random events file for exact details). If the AI file has expansion 0 the AI will not use the explorer, and not colonise at all. This is the best way to stop exploration/colonisation.

 

[Toio]

on a side note in regards to lack of colonists, i , many versions ago used a random event like this

trigger was
continent = europe
year = 1520
provincereligion = catholic
countrysize = 8

command was to get 6 colonists

this was aimed mostly for france, spain ( POR missed out, which worked fine) while below

trigger was
continent = europe
year = 1550
provincereligion = protestant
countrysize = 5

command was to get 6 colonists

was for england, denmark and sweden

england now seem to not be defined religiously now as it was previous, but these random events could help in the "push" for coloies and TP's

 

[unmerged(40707)]

This is what Fodoron started to enhance for Spain with regular events. Main problem for England and France is lack of generated colonists per year as I stated some days ago (and maybe lack of money for England too because of instability with all scripted religious changes - fix proposed by Bordic and me - but maybe not only cause). Such random event won't help in this case. Maybe after, when other problems will be solved.

 

[unmerged(40707)]

For France and broken colonization, see here in BIF thread.

 

[Bordic]

Suggested improvements for a better English and hopely Spanish and Portuguese colonization

ENGLAND
areas-regions to be added to the current ENG AI files: ENG_1635.ai and ENG_1740.ai
area = { Georgia Newfoundland "Hudson Bay" Labrador }

and in ENG_1740.ai or at least in another ai file after 1770s:

region = { BarrierReefSea BarrierreefSea OceaniaSea SWPacificSea SEPacificSea Australia }
area = { "Grand Lacs" Papua Botany Kanguru Maori Papua Tasmania } as to let Britain colonize Australia and New Zealand.

(with these simple modifications now England colonizes the Hudson bay and Australia... )

As a sidenote, shouldn't we repeat in each country specific AI file the sea- landzones the AI driven country should explore-colonize as it was coded for the Daywalkers files?


We could even add
combat = { BGL MAH MOG VIJ MYS } and more other Indian countries to the ENG AI 1740 file, as to let the British form their colonial empire. Comments?

EDIT:
Just in case I would suggest following modifications if not already tested and/or not already successfully done:

SPAIN
areas-regions to be added to the current SPA AI files: starting from SPA_1545_CarlosI.ai
region = { EPacificSea SEPacificSea WPacificSea ThaiSea }

PORTUGAL
areas-regions to be added to the current POR AI files: starting from POR_1504_Manuel.ai
region = { NEIndianSea NWIndianSea SEIndianSea SWIndianSea ThaiSea }

 

[Norrefeldt]

Have the changes for SPA and POR been tested? I don't know if ENG had it's AI files tested but SPA and POR have, and any changes to them should only been done if they actually work. Programming AI behaviour have proved itself not being straight-forward nor predictable, and changes that appear to be for the better might have the opposite effect.

As a sidenote, shouldn't we repeat in each country specific AI file the sea- landzones the AI driven country should explore-colonize as it was coded for the Daywalkers files?

IF the AI should have the same focus I guess we should have that. For some AI files it might be that it's more important that the few colonies are places in area B than in area A that were supposed to be colonised earlier.

 

[Bordic]

Have the changes for SPA and POR been tested? I don't know if ENG had it's AI files tested but SPA and POR have, and any changes to them should only been done if they actually work. Programming AI behaviour have proved itself not being straight-forward nor predictable, and changes that appear to be for the better might have the opposite effect.

Not the Spanish and Portuguese. I just wanted to know how Portugal and Spain could reach SE Asia as they never colonize there, while the Netherlands always succeed. Philippines and Indonesia are always colonized by them except New Guinea.

Adding regions/area in ENG files I have seen good colonization of Newfoundland and the southernmost East coast colonies and in late XVIIIth century Australia and Fiji. Not yet seen TPs in New Zealand.

The netherlands often colonize Hudson Bay or Labrador and Lower Louisiana, where France doesn't even send troops or explore. Maybe should we change the order of zones in region/area sequence.

As for Spain, as it is now in the last beta5 and 6, it seems to me too much advanced in tech research, the most avanced one, after XVIIth century, more than France. England cannot cope with Spain in naval tech, and even the Netherlands are a bit lower than Spain. Is this to be found in the new Spanish colonization events?

I have reedited my previous post about Spain and Portugal, my interests were more towards the ENG AI files.

 

[Norrefeldt]

Not the Spanish and Portuguese. I just wanted to know how Portugal and Spain could reach SE Asia as they never colonize there, while the Netherlands always succeed. Philippines and Indonesia are always colonized by them except New Guinea.

Portugal made it through Malacca to China in time in every game. Don't recall the exact year they should be there, but as I tested it Toio knew it, and they managed it.
I don't think they colonised much in Indonesia, but that cannot be solved by adding sea files. Adding those sea files could on the other hand cause the POR ai to spend time finding out seazones in the open ocean, sea routes that were not used much early on, and that would delay the arrival in China.

 

[unmerged(40707)]

But, in my tests, I never see Portuguese in Malacca. Reason is sometimes Malacca (the state) doesn't exist anymore but often, because Portuguese don't even know indonesia around 1550 (only Lanka as the "far east").

OTOH, there is no problem for Netherlands after 1600 for Indonesia beginning with event for Jakarta.

And I don't know why but Spain doesn't colonize Philippines at all and doesn't "want" to colonize the two provinces east to Nicaragua and west to Mosquitos (I don't remember the names...) or only very late in game.

 

[Bordic]

And I don't know why but Spain doesn't colonize Philippines at all and doesn't "want" to colonize the two provinces east to Nicaragua and west to Mosquitos (I don't remember the names...) or only very late in game.

And Baja California.
As for the Caribbeans, in my tests it is Nicaragua only but it should probably be historically colonized by England, Belize iirc.

 

[unmerged(29041)]

Suggested improvements for a better English and hopely Spanish and Portuguese colonization

ENGLAND
areas-regions to be added to the current ENG AI files: ENG_1635.ai and ENG_1740.ai
area = { Georgia Newfoundland "Hudson Bay" Labrador }

and in ENG_1740.ai or at least in another ai file after 1770s:

region = { BarrierReefSea BarrierreefSea OceaniaSea SWPacificSea SEPacificSea Australia }
area = { "Grand Lacs" Papua Botany Kanguru Maori Papua Tasmania } as to let Britain colonize Australia and New Zealand.

(with these simple modifications now England colonizes the Hudson bay and Australia... )

As a sidenote, shouldn't we repeat in each country specific AI file the sea- landzones the AI driven country should explore-colonize as it was coded for the Daywalkers files?


We could even add
combat = { BGL MAH MOG VIJ MYS } and more other Indian countries to the ENG AI 1740 file, as to let the British form their colonial empire. Comments?

EDIT:
Just in case I would suggest following modifications if not already tested and/or not already successfully done:

SPAIN
areas-regions to be added to the current SPA AI files: starting from SPA_1545_CarlosI.ai
region = { EPacificSea SEPacificSea WPacificSea ThaiSea }

PORTUGAL
areas-regions to be added to the current POR AI files: starting from POR_1504_Manuel.ai
region = { NEIndianSea NWIndianSea SEIndianSea SWIndianSea ThaiSea }

These changes should be tested to see if they produce any improvement.

 

[unmerged(29041)]

As for Spain, as it is now in the last beta5 and 6, it seems to me too much advanced in tech research, the most avanced one, after XVIIth century, more than France. England cannot cope with Spain in naval tech, and even the Netherlands are a bit lower than Spain. Is this to be found in the new Spanish colonization events?

I have also noticed this. It looks like all my work for 1.37 is already useless. We are touching a lot of things and in EU2 everything seems connected. I think that with the free conversions Spain is a lot richer. We could reduce the rescue from Atahuallpa from 10,000 ducats down to perhaps 1,500, as now the AI is getting more money. All the gold is now very difficult to evaluate, but I think that Spain is wealthier than she should, and as we know, Spanish AI, as dumb as she is, is not as wasteful with money as the Habsburgs

 

[unmerged(40707)]

Always under or up, we can't reach perfection. It is a quest...

In french: "Le mieux est l'ennemi du bien" (Better is the enemy of Good enough).

But I won't play again with 1.37! I can't...

Interesting part of the process is identifying where things went wrong with additions or modifications and corrections (yes, sometimes correcting an event has a huge effect and not only in order to avoid a CTD) and adjust accordingly.

 

[Toio]

Always under or up, we can't reach perfection. It is a quest...

In french: "Le mieux est l'ennemi du bien" (Better is the enemy of Good enough).

But I won't play again with 1.37! I can't...

Interesting part of the process is identifying where things went wrong with additions or modifications and corrections (yes, sometimes correcting an event has a huge effect and not only in order to avoid a CTD) and adjust accordingly.

its like tuning a vehicle, small, very small tweaks

 

[Bordic]

I have also noticed this. It looks like all my work for 1.37 is already useless. We are touching a lot of things and in EU2 everything seems connected. I think that with the free conversions Spain is a lot richer. We could reduce the rescue from Atahuallpa from 10,000 ducats down to perhaps 1,500, as now the AI is getting more money. All the gold is now very difficult to evaluate, but I think that Spain is wealthier than she should, and as we know, Spanish AI, as dumb as she is, is not as wasteful with money as the Habsburgs

Or just trying to worsen the DPs such as big decentralization, serfdom, aristocracy and narrowmindedness or also lowering the monarch rates as until 1648 Spain should in any case be the leading nation, then France and in mid XVIIIth century England no?

As for now I am testing the English situation adopting the changes I have previously suggested, but I think I am still no good at AI tweaking.

 

[unmerged(40707)]

Btw, Fodoron, it's time to submit Spanish AI colonization events. They're fine. We always have a problem with Philippines but Spain performs as intended for America (the two provinces in Central America I can't remember the name apart but not always apparently and other little problems).

We made our part of the work, Bordic and I, for France and England and I'm proud of the result with "only" DP/stab tweakings.

 

[Bordic]

Btw, Fodoron, it's time to submit Spanish AI colonization events. They're fine. We always have a problem with Philippines but Spain performs as intended for America (the two provinces in Central America I can't remember the name apart but not always apparently and other little problems).

We made our part of the work, Bordic and I, for France and England and I'm proud of the result with "only" DP/stab tweakings.

I would like to see the same Spanish AI colonization events for England in her impossible task to properly colonize Labrador and Hudson Bay anyway, before the Netherlands and Danmark send their settlers there...

And still no French in Biloxi-Mobile, always Dutch colonists there!

 

[unmerged(40707)]

I would like to see the same Spanish AI colonization events for England in her impossible task to properly colonize Labrador and Hudson Bay anyway, before the Netherlands and Danmark send their settlers there...

And still no French in Biloxi-Mobile, always Dutch colonists there!

I forgot these problems.... o
But results are definitely better than before for both countries and East coast.

 

[Norrefeldt]

Some version ago Denmark got an AI change in 1600 in order to make them explore the world, at least around Africa to India and China and the coasts around the Atlantic, find some COT's to send traders to and hopefully not colonise too much but just set up a few TP. That would be historical. But we don't want a lot more than that!

I have evaluated the performance with the help of ten games run by Yodamaster in 1.51, and looked at saves around 1650 and around 1750.

1650:
Denmark have survived in eight of the games. In six games it has found its way around Africa. (Numbers as the test games are numbered.)
0: TP: Bangor, Karroo. New known COTs 9, 1 merchant in India. Maps to NE America, India, China.
3: TP: 3 in South Africa + Anticosti. New known COTs 5. Maps to India, Malacca, NE America.
4: No TPs. New known COTs 2. Maps to China, India.
5: TP: Guinea Colony: Leone. New known COTs 2. Maps to India.
8: TP: Table Colony: St. Thomas. New known COTs 9. Maps to India, China, N America, S America, Antilles.
9: TP: Natal, Transkei Colony: Nouadibuh. New known COTs 5. Maps to India, China, Antilles.

1750:
Denmark have survived in five of the games. In all five cases it has some TPs/colonies. (Numbers as the test games are numbered.) In all games Denmark knew the way around Africa.
4: Colony: Guniea. New known COTs 11. Maps to India, Japan, Brazil, N America.
5: TP: 3 African. New known COTs 9. Maps to N America, India, China.
7: Colonies: Kribi Colonial cities: Senegal, Antistirana (COT). New known COTs 8, 1 merchant in Isfahan, 2 in Antistirana. Maps to India, China, N America.
8: Colony: Muni Colonial cities: Table, Senegal, Namaqua, St Thomas, Nova Scotia. New known COTs: many? Maps to India, China, Phillipines, Japan, Indonesia, Brazil, Antilles, Panama, N America.
9: Colonial cities: Seven in total (forgot to write which). New known COTs: many? Maps to India, Japan, China, Brazil, Antilles, N America.

Conclusions:

1. Resulting map discoveries work fine.
2. Denmark colonise too much. Settings are now expansion 1 and tradingpost 95. Could raise tradingpost to 100. That will still lead to some colonies eventually, but fewer. If other historically big colonisers colonise better, it would be less room for Denmark. Another idea is to set the AI to start a few decades later than 1600.
3. Trade in non-European COTs is close to non-existant! This appears to be the same for most states. Is there anything we can do or is the engine broken? Should we test to raise "traders" above 100?
4. Another AI should hit in late game, to allow exploration of more peripheral areas.