Don't Waste Worker Turns

[Um the Muse]

Build your road network out of X's and not +'s. Moving in the secondary directions (1379) is faster than moving in the cardinal directions (2468), so build a road network that exploits this. Don't build a road that goes 8888 when you can build one that goes 7979. It gives better coverage of adjacent tiles, and it will be faster to build new roads to other destinations. This isn't as rigorous as 100 turn worker plans a la SGOTM, but leads to fewer facepalm moments in the early game when you don't know what's around you.

It's probably obvious, but probably worth saying out loud.

I can honestly say I never thought of this, even though this is how I like to do my initial exploring for exactly this reason.

 

[krikav]

Build your road network out of X's and not +'s. Moving in the secondary directions (1379) is faster than moving in the cardinal directions (2468).
It's probably obvious, but probably worth saying out loud.

Absolutely.

I tend to screw up roading on a regular basis.
Really, building a road network requires as much planning as dotmapping does.
And I seldom give the roads the attention they deserve...

 

[Archon_Wing]

Amazing stuff.

Reminds me of stutter step micro in Starcraft

 

[Carboniferous]

Another one I noticed is that the worker AI likes to build a road before starting the improvement. This delays the tile yield improvement which can be significant early on if it's improving a bonus. Sometimes, it also roads to the bonus plot, rather than from it, further delaying the yield benefits. This is more arguable, as the road can help with retreat. Of course, this is not a problem if you don't automate your workers, but it does handicap the AIs.

 

[Nick Carpathia]

Build your road network out of X's and not +'s. Moving in the secondary directions (1379) is faster than moving in the cardinal directions (2468), so build a road network that exploits this. Don't build a road that goes 8888 when you can build one that goes 7979. It gives better coverage of adjacent tiles, and it will be faster to build new roads to other destinations. This isn't as rigorous as 100 turn worker plans a la SGOTM, but leads to fewer facepalm moments in the early game when you don't know what's around you.

It's probably obvious, but probably worth saying out loud.

It's a weird quirk of the engine, where diagonals are cheap on things like movement and expensive on things like culture. This leads to moving on the diagonals objectively better than the others. The best way to normalise that is to restrict movement to the cardinals (as Advance Wars and Fire Emblem), but it's probably not worth fixing.

 

[Manco Capac]

Another one I noticed is that the worker AI likes to build a road before starting the improvement. This delays the tile yield improvement which can be significant early on if it's improving a bonus. Sometimes, it also roads to the bonus plot, rather than from it, further delaying the yield benefits. This is more arguable, as the road can help with retreat. Of course, this is not a problem if you don't automate your workers, but it does handicap the AIs.

I think this has to do with a funtion "road to" the human often uses if micro isn't their strong side.
It doesn't happen at the time with resources, but between cities it occurs more often. You sure point out one problem in regards to get profit asap on special resources yields, but at war, this way of roading may also be a problem when it is time to protect their threatened workers.

 

[Um the Muse]

It's a weird quirk of the engine, where diagonals are cheap on things like movement and expensive on things like culture. This leads to moving on the diagonals objectively better than the others. The best way to normalise that is to restrict movement to the cardinals (as Advance Wars and Fire Emblem), but it's probably not worth fixing.

What about changing the shape of tiles? Hexagons and octagons are common tile shapes, triangles might work. The original Avalon Hill Civilization had irregular shapes, though I don't know if that would be possible for a computer or even desirable.

OT, tachywaxon's post reminded me of something. On MP, workers don't stop improving stuff when they are threatened but have a queue. You have to be careful when building road networks, have auto workers, etc.

 

[TheMeInTeam]

AI workers function differently from auto-workers on a lot of levels, which is very annoying because human auto workers will try very hard to kill themselves while you are at war. While one might occasionally see an AI worker get itself out of position, you can't simply declare on an AI and watch it repeatedly move workers on a tile adjacent to your border to be captured (or in AI's case, killed) over and over and over again. What a "great" way to go from 14 workers to 2 workers.

And yes, I count this as one of many UI/control strikes against failaxis. Why give us an option in-game if it is so utterly terrible as to be unusable, when there are obviously better algorithms ALREADY IN THE GAME...they literally gave us an inferior-programmed worker set just to not use it. Ludicrous.

Incidentally, tile-yield output with currently existing workers should be one of the things the AI is VERY GOOD at; it should be able to determine pathing that offers the highest yields of each kind of output, not to mention totals. The AI even has a rating system for , , and existing already IIRC...so why don't workers (auto human OR AI) calculate out actions that are better-optimized? I'd have taken that in a patch in a heartbeat over the chain disgraces we got from 3.13 on or so.

OT, tachywaxon's post reminded me of something. On MP, workers don't stop improving stuff when they are threatened but have a queue

Yes, this is a great way to equalize skill, make it so that you can't give orders at the start of a turn AND so that the units perform an action despite attempts to stop it from doing so. I can almost picture the failaxis design meeting!

"You know what would be more fun?"
"What"
"Let's put some situations in the game where the controls don't do anything!"
"Haha! Wait, you seriously want this in the game?!"
"...do it or I'll report your "islandtarget" code."

Hexes actually do make movement a bit more "fair" in that sense, because you have the same cost regardless of the direction you chose. Other games use tiles but incur fractional extra movement cost along the diagonals, but those games usually have higher total movement points where it matters (HOMM series for example). Ironically while hexes weren't well received in civ V initially by most it was one of the things about the game I actually thought was an improvement and it's a shame there were so many other issues that were not.

 

[shakabrade]

Incidentally, tile-yield output with currently existing workers should be one of the things the AI is VERY GOOD at; it should be able to determine pathing that offers the highest yields of each kind of output, not to mention totals. The AI even has a rating system for , , and existing already IIRC...so why don't workers (auto human OR AI) calculate out actions that are better-optimized? I'd have taken that in a patch in a heartbeat over the chain disgraces we got from 3.13 on or so.

This reminds me of a travelling salesman problem, a role-model of optimization. Civilization is a complex game with lots of variables, and tiles to optimize.
I remember I was given the task to determine how much time would a 1GHz computer need to find the best possible route for travelling salesman to visit 28 cities, and IIRC it was more than million years since 28! (!=factorial) is a huge number. And it was a symmetrical problem, something Civilization map isn't.
So you're proposing even longer loading turns.

http://en.wikipedia.org/wiki/Travelling_salesman_problem

But I agree they could have made a much better algorithm, but that's something a HUMAN and not a computer would do better.

 

[dalamb]

If you're willing to take a "good" answer rather than a "best" one there are much faster Travelling Salesman algorithms (that is, of much lower order) -- eminently suited to use in a game program.

 

[TheMeInTeam]

If you're willing to take a "good" answer rather than a "best" one there are much faster Travelling Salesman algorithms (that is, of much lower order) -- eminently suited to use in a game program.

Indeed. Right now, even an algorithm like "move to highest post-improvement yield tile and improve it, else improve near cities working unimproved tiles" would largely improve AI worker functionality too. It can't be THAT hard to come up with something better than that :/.

 

[Carboniferous]

Indeed. Right now, even an algorithm like "move to highest post-improvement yield tile and improve it, else improve near cities working unimproved tiles" would largely improve AI worker functionality too. It can't be THAT hard to come up with something better than that :/.

This is pretty much how the worker AI works in my mod (see my sig). Tile improvements falls into two categories; ones which are immediately beneficial, and those which will be useful later. Tracking improvements at a city level also removes some (not all though) of the constant switching of improvements. Although I am not sure that my workers do better overall than the original ones, I'd be interested to hear any feedback from those of you who are always complaining how bad the worker AI is

 

[Nick Carpathia]

What about changing the shape of tiles? Hexagons and octagons are common tile shapes, triangles might work. The original Avalon Hill Civilization had irregular shapes, though I don't know if that would be possible for a computer or even desirable.

OT, tachywaxon's post reminded me of something. On MP, workers don't stop improving stuff when they are threatened but have a queue. You have to be careful when building road networks, have auto workers, etc.

Agreed about Civ 5's hexes being a nice idea. Apparently, the reason Sid Meier chose squares in the first place was to differentiate his game from other hex-based wargames.

Something else the AI has to fix: Stop paving over old improvements! It's incredibly common to see AI workers switching grassland river cottages to farms to cottages and back again. Fixing that could make the AI so much more competitive. It's even more painful watching AI workers demolish villages and towns just before your cannon stack captures the city (though they're really just unintentionally trolling you).

 

[Dirk1302]

The only good idea out of CivV? Actually its fine as it is, agree with Sid here, hexes are the last thing we need to worry about here.

 

[vranasm]

I can easily see skipping this micro on youtube as planning workers correctly in this way takes some time. Building/cancelling a road while underway is probably doable but sometimes you need to plan a bit further to maximize worker potential which is hard on youtube.

most important it would be very boring for viewers and most probably won't provide enough benefit for provider of such videos (the benefit here is the number of "hits" on your videos, you do this for viewers after all).

Ironic enough if I check my yt videos I got around ~1k views on the only video of me playing fantasy wars scenario (and very badly at it, since I actually lost the scenario) compared to ~200 views on civ video... I am just boring ;-).

 

[ben-jammin]

This is pretty obvious but some people might not do it...prep new cities as much as possible before settling them. Getting the trade route roaded to cities far from the capital is huge on immortal and deity. Often tiles that are outside a current city's BFC but still within your culture can be worked by a new city and those can be improved so that they're workable as soon as you settle.

Also dont forget that when a worker or two run out of more pressing things to do you should road into AI land for better trade routes.

 

[vicawoo]

most important it would be very boring for viewers and most probably won't provide enough benefit for provider of such videos (the benefit here is the number of "hits" on your videos, you do this for viewers after all).

Ironic enough if I check my yt videos I got around ~1k views on the only video of me playing fantasy wars scenario (and very badly at it, since I actually lost the scenario) compared to ~200 views on civ video... I am just boring ;-).

It's pretty fast if you hotkey all your workers.
1 right click r 1 backspace
2 alt-c 2 backspace

 

[Carboniferous]

This is pretty obvious but some people might not do it...prep new cities as much as possible before settling them. Getting the trade route roaded to cities far from the capital is huge on immortal and deity. Often tiles that are outside a current city's BFC but still within your culture can be worked by a new city and those can be improved so that they're workable as soon as you settle.

Also dont forget that when a worker or two run out of more pressing things to do you should road into AI land for better trade routes.

This is why the AI's love of forts for out of city radius bonuses is not so good, better to improve bonuses with their real improvements so they provide a better yield straightaway. The only case where forts are optimal is where you know the improvement is there, but can't improve it yet (only oil, afaik - you can mine uranium with physics, you just don't get the resource).

 

[Jab4849]

Here's a simple one: If you're trying to chop something as fast as possible don't move more than one worker to each forest, send a worker to each forest you want to chop so you don't waste extra turns stepping onto them (Plus you keep the production if you're working the tile)

 

[Um the Muse]

Does anybody have an opinion on how valuable a settler turn is? I'm thinking specifically about cases where I road to my target city spot, including the spot itself if it'll save a turn. My main reason is that, if barbs come, I'd rather put the worker in danger. Kinda dumb though; if I save time, I'll put less stuff into danger altogether, so it's better to chop.