Improvements to worker automation

[HardCoder]

My favorite bugaboos with the stupid automated workers in BTS.

I don't know whether these have been implemented; I haven't followed Better AI in a while.

* Stop workers building forts on resources. Not only is there no point to this but they provide safety for invaders

* Workers should always improve and connect all resources within cultural borders

* Stop unescorted workers from moving beyond safe visibility during wartime, except to reconnect resources ... idiot workers getting horked by unseen mounted units is infuriating

* Stop workers from converting resources with alternative improvements (e.g. river spice with town) to the conventional improvement if player already has a source of the resource. If a player needs a resource, and it is already improved inside fat crosses, improve the resource outside fat crosses in preference to converting one inside a fat cross

* If at war, workers should assign high priority to chopping jungle (and forest if permitted and it is part of the improvement plan for a city) on tiles adjacent to cities, particularly on hills; this should also be a higher than average priority during peacetime

* Workers should assign high priority to building roads on hills within fat crosses, starting with the hills adjacent to the city (vastly better for defense)

* Workers should build roads with minimal river crossings between cities (why does the stupid AI unnecessarily cross rivers?)

* Workers should strongly bias for hammers for small cities that have important needed improvements but are seriously lacking in hammers, then convert improvements to more appropriate types later (say, when a city goes from radius one w/ a hammer shortage to radius two with more hammers available)

* Workers automated to a city should improve according to city governor bias (add an option for this), and change improvements slowly if bias changes

* Add "automated workers chop only jungle" option

* Workers should build mines on bare green hills to start, then if city isn't automated to hammers, convert to windmills

* If city is automated to gold, workers should consider cottaging hills if sufficient hammers are available elsewhere

* Workers should put a high priority on chopping jungle next to polluted cities

* Workers should remove/convert workshops that are causing pollution problems and seem to be unnecessary

 

[HardCoder]

Also:

* Workers should build watermills on bare river tiles if nothing else is needed (the AI hardly ever builds watermills if at all)

* Something is drastically wrong with the balance between building lumbermills and forest preserves

 

[Yakk]

My favorite bugaboos with the stupid automated workers in BTS.

I don't know whether these have been implemented; I haven't followed Better AI in a while.

* Stop workers building forts on resources. Not only is there no point to this but they provide safety for invaders

The fort connects up the resource. It also means you can defend it if you choose.

I'll admit the AI uses this better, with their surplus of units and poor ability to allocate them effectively.

* Stop workers from converting resources with alternative improvements (e.g. river spice with town) to the conventional improvement if player already has a source of the resource. If a player needs a resource, and it is already improved inside fat crosses, improve the resource outside fat crosses in preference to converting one inside a fat cross

Questionable. A town vs a river spice that is. You can usually trade a way a spice for at least 3 to 5 coin per turn -- so it isn't that common that a spice, even your 2nd, is worse than a town.

* Workers should build roads with minimal river crossings between cities (why does the stupid AI unnecessarily cross rivers?)

I expect the AI ignores rivers when building roads? But ya, without bridge building, that does suck.

* Workers should strongly bias for hammers for small cities that have important needed improvements but are seriously lacking in hammers, then convert improvements to more appropriate types later (say, when a city goes from radius one w/ a hammer shortage to radius two with more hammers available)

That is an interesting position. With slavery, pushing population growth, whipping your improvements, is often a good strategy if you have a good food resource.

Even without slavery, going for hyper-food, and after you have a decent size going hyper-improvement with heavy production, often seems to work well for me.

Ie, grow to a sufficient size with lots of food improvements, then strip down to 1 food improvement and as many production improvements as you can to pump out some infrastructure, then do another growth spurt.

* Workers automated to a city should improve according to city governor bias (add an option for this), and change improvements slowly if bias changes

*nod*. I was thinking about this problem in general for the AI.

Work out the potential of each city as a growth/commerce/production city using dynamic programming. Pick which one seems most attractive. Using that bias, pick what improvements are ideal for the city specialization, probably with some bias towards leaving things as they are (for the town problem).

Done right, the decision to do things like cottage hills falls out naturally, instead of being hard coded in.

 

[Minor Annoyance]

The fort connects up the resource. It also means you can defend it if you choose.

Also the fort is harder for aircraft to destroy. I haven't gotten around to seeing if it's harder for spies to destroy. Probably not.