A small reason not to move starting settler

[DaviddesJ]

Short version - the AI knew about the increased tile yield that undiscovered resources would eventually provide (though not the resource itself per se), and placed cities with that knowledge. Probably the blue-circle algorithm did the same.

What evidence do you have for "probably"? I don't know if this particular bug did or did not affect the suggestions as well as the AI behavior (and Soren doesn't say one way or the other so I don't see what evidence you could have for your opinion), but even if it did, as Soren says, this factor would be so minor that it would almost never make a difference in practice. Calling this a "theoretical" issue seems pretty accurate.

Soren said that this would be fixed, so maybe this is no longer the case in 1.52. Or it may have been too late for 1.52 and will be in the next patch. But it is an established fact that the AI did (approximately) know about unrevealed resources before that.

He said it would be in the upcoming patch, so I assume it was in the 1.52 patch, that came out more than a week after that posting. And the 1.52 documentation does have the comment "Fixed AI city targeting bug", which seems likely to be this problem. In any case, I think your last remark is very misleading: the effect on targeting of just one extra shield from a hidden resource is going to be much, much less than would be the effect if the algorithm knew about the resource itself, so it's misleading to suggest that it would ever have behaved as if it knew where the resource was.

 

[Wodan]

Very good point about towns versus coast. By "etc.", I meant lighthouse and the temptation to build Great Lighthouse and Collosus.

Lighthouse just adds 1 food. Same as farming. So, trade Hammers to build Lighthouse for Worker time. Plus, the farm will get better with Biology.

Great lighthouse and Colossus are good to have, definitely. Worth it? Debatable. Consider the Capitol w/Towns vs the Capitol w/Coast & wonders. Cottages start with 1 commerce, while coast has 2. However, you're not working the Coast, you're working the Grassland and Hills. Only as your city gets above 5-6 Population are you working the coast. By then, the cottages are 3-4 commerce. If you get the Wonders you change 2 commerce to 3. Still not as good as the Towns. What's the other do... I don't have my book here. Extra trade route? Well, that's good for 2-3 gold per coastal city. A couple of Towns will beat that.

So, at best, it's a break even proposition. And, you would have to spend tons of Hammers on those wonders, and possibly the AI beat you to them anyway.

All that said, it's not like I don't often go for those wonders myself, especially on archipelago.

Seriously, without a Civ3 Platform giving end-game sea production boost, I think the sea has gotten the short end of the stick.

And those coastal squares that I don't get by not moving my starting settler are never going to be worked because no other city is going to be able to reach them.

You mean because you moved your capitol 2 tiles in (so it reaches barely to the beaches)? You could consider moving it 4-5 tiles in, to leave room for a city or two. Depends on the map.

Still overall I think you're right. I need to seriously re-evaluate my tendency to want capitals on the coast. Thanks.

you're welcome. Again YMMV.

Actually, I enjoy talking about this stuff... I find it sparks my own creative thinking and gets me pondering different strategies. Inevitably what I end up doing is starting my next game specifically to explore a technique.

Wodan

 

[Wodan]

By the time you actually need to build a palace, it isn't expensive - and it is well worth the minor investment.

Actually, despite everything I've said, I tend to agree somewhat with you Jantis.

Honestly, I think there's value in both our viewpoints. Depends a lot on the situation.

Above all, I guess my two cents is "don't just hit Build City when you start the game... spend 20-30 seconds looking and deciding if maybe you should spend a turn or two to optimize your start spot."

And really, I haven't had to build a palace very often. But I am more willing to risk needing to do so versus potentially wasting several starting turns looking for a central location and being wrong.

I'm very willing to "waste" those turns. 2 turns is NOTHING compared to what you could gain/lose.

Also, even if you plant your starting Settler right there with the eventual plan to move your capitol, there's a good chance it's a crappy city site, period. e.g., it's just a bit from the corner of the continent, leaving 4-5 tiles either totally unworked or else a city that's half size.

Unless you know the map ahead of time (or play on an obvious coast-is-evil map like Pangaea), you have no way of knowing where the center of your empire will be. Assuming that moving a couple of turns (or worse... several turns) will actually put you near the center of your eventual empire is just taking a risk.

I don't think anyone assumed anything. The premise is this: coast is more often than not less preferred for your capitol. That's it.

Oh, a minor point but I won't quibble: I personally think it's worth a turn or two to look around. If you end up coming back to the same spot I think it's still worth it.

My latest game put me on the coast to start. It is a continent map, and I was sincerely considering moving inland, but the starting spot was just too appealing. So I took it. Based on the map layout and a few hundred years of expansion, I found that my starting spot could not have been more perfect, putting me smack dab in the center of my empire.

Blind dumb luck will beat planning any day.

Now, tell me how to get my start spot in the same place as I would put it if I had the whole map revealed.

YMM certainly V! I agree with ya there.

Yep!

Wodan

 

[Mikey_rex]

... move it to a river! When I start off one to two spaces to the coast, I'll move there since it is shorter than moving 3 or more spaces away to leave that area free for the growth of a future city. But my #1 rule is move to a river. River and coast is good, but river is almost a must (can you tell I don't like the inland lake scenarios )

When you put a city next to a river you get the health bonus, instant trade route to other cities along the river, and a defensive barrier. Yes, there is a movement cost, but it's worth it to me.

The health bonus is great because it allows you to grow bigger before having to deal with aquaducts, pastures, etc.

The trade route is great because it generates more income that you can spend on techs (I'm including saving for when you want to run into the red) and the incredibly expensive cost to upgrade warriors to spearmen/axemen. BTW - the colossus and great lighthouse make the coast trade route even more beneficial... and you are so much more likely to build either of them first if you have a coastal capital.

The defensive barrier is great since the AI doesn't seem to care enough to work around it.... in fact, has anyone notice that when moving more than one space to attack a city, the AI generator always seems to want to have you attack from accross a river? I frequently have to move my units in stages (within the same turn) to get them to where they can attack (again, same turn, think mounted units here or other units within your borders) without having the river get in the way? This isn't my favorite gripe about C4 though... what I hate most about C4 is the diplomacy... as if a puny weak country would actually rather go to war with you than open it's borders or trade a tech, especially when their main complaint is that your borders are too close.

 

[Sim-Liver]

Generally I don't want my capitol to be on the coast. City maintenance encourages your capital to be in the middle of an encircling ring of cities. Over the entire span of the game, that added maintenance (by having your capitol on the -edge- of your empire) will be huge.

Wodan

I have noticed this also but I tried setting it up on the coast then when I had the outer boundries to my empire developed I simply build a palace in the middle of the entire empire (only viable on single continents/Pangaea maps) Not on islands expecially if I have several isls. Will be to rough to center empire appropriately, and does not work out to save maintenance costs.

 

[Renata]

Re: starting positions and resources --

Starting positions are balanced in the map script to all be roughly the same strength. (Don't ask me by what algorithm 'strength' is calculated, but it definitely includes *all* resources that are in the area, even those you can't see.) So moving significantly away from the starting spot may very well lose you critical later-appearing resources. Also, every starting spot will have at least one food resource, I think. If you can't see it in 4000BC and guess wrong, you could very well move away from it.

For those reasons, I very rarely move my settler, and virtually never if I don't know where the food is.

The blue circles are a separate matter, and the bug with them was mentioned in the links on the last page. It's supposed to be fixed in 1.52.

 

[Beamup]

What evidence do you have for "probably"? I don't know if this particular bug did or did not affect the suggestions as well as the AI behavior (and Soren doesn't say one way or the other so I don't see what evidence you could have for your opinion), but even if it did, as Soren says, this factor would be so minor that it would almost never make a difference in practice. Calling this a "theoretical" issue seems pretty accurate.

I say it probably affects the blue-circle algorithm because, from a programming standpoint, the logical way to set it up would be to have them share the same code. It would be frankly ridiculous to do otherwise.

Beyond that, whether it's a theoretical or practical issue is immaterial. The issue exist(s/ed) despite your kneejerk rejection of the possiblity.

He said it would be in the upcoming patch, so I assume it was in the 1.52 patch, that came out more than a week after that posting. And the 1.52 documentation does have the comment "Fixed AI city targeting bug", which seems likely to be this problem.

Most likely yes. However, I have no direct evidence that it is the problem, and it's entirely credible that a week before patch release could be too late to include something (due to QA issues and the like). Hence I choose not to make a categorical statement without foundation.

In any case, I think your last remark is very misleading: the effect on targeting of just one extra shield from a hidden resource is going to be much, much less than would be the effect if the algorithm knew about the resource itself, so it's misleading to suggest that it would ever have behaved as if it knew where the resource was.

Don't be ridiculous. My "last remark" is clearly a summary of the post and could not possibly be mistaken for a statement of fact in its own right (particularly given the inclusion of qualification). The only way to distort it into a claim that the AI behaved as if it entirely knew where the resource is would be to completely ignore the entire post.

 

[Smirk]

Isn't it odd that finally after so many years your people decide its time to settle down from the nomadic life, but they have no map to speak of and barely have a grasp of their surroundings?

You'd think nomadic people would have a better knowledge of the area they've previously nomaded around.


Anyway, coasts suck. Which is unfortunate because this is so very unlike reality.

 

[AnotherJon]

coasts suck.

Yet the only strategy I have seen posted for beating deity at standard settings relies on the potential advantages of a coastal start .

http://forums.civfanatics.com/showthread.php?t=148242

 

[DaviddesJ]

I say it probably affects the blue-circle algorithm because, from a programming standpoint, the logical way to set it up would be to have them share the same code. It would be frankly ridiculous to do otherwise.

I don't think it's likely to use the same code because the two functions don't have much in common. E.g., the suggestion algorithm always seems to offer two choices, while of course the AI is always going to choose a single place to put its settler. You could imagine that there's a code that ranks locations, and the AI chooses the best location, while the suggestion code suggests the best two. But the suggestions are also always within a certain distance, while the best place for the AI to settle might be quite far away (e.g., in the middle of the game when all of the good nearby locations are taken). The suggestion algorithm also seems to have diversity built into it; it doesn't tend to suggest two locations right next to each other. While if it were just ranking results according to a fixed set of criteria, it would pretty often be the case that the two best choices would be right next to each other. Etc., etc. They just don't seem very similar at all, or likely to share code.

I recognize your opinion that they might, but it seems entirely hypothetical or suppositious to me.

My "last remark" is clearly a summary of the post and could not possibly be mistaken for a statement of fact in its own right (particularly given the inclusion of qualification). The only way to distort it into a claim that the AI behaved as if it entirely knew where the resource is would be to completely ignore the entire post.

The original post argued that one should accept the starting positions suggested by the computer because "the little blue circle placement algorithm knows about not-yet-revealed resources." This would only make sense if the presence of those resources has a strong effect, not if it has a very minor effect that almost always wouldn't even matter. If it's only a very minor factor, then using the superior judgment of human players would make a lot more sense than being guided by a dumb algorithm with a very, very weak "hint" built into it.

Furthermore, the thread you point to ends with the conclusion by those who did the tests that the behavior has changed and no one can find the effect any more in v1.52. Confirming the comment that seems to relate to fixing it, and contradicting the assertions of others in this thread.

 

[DynamicSpirit]

Re: starting positions and resources --

Starting positions are balanced in the map script to all be roughly the same strength. (Don't ask me by what algorithm 'strength' is calculated, but it definitely includes *all* resources that are in the area, even those you can't see.) So moving significantly away from the starting spot may very well lose you critical later-appearing resources. Also, every starting spot will have at least one food resource, I think. If you can't see it in 4000BC and guess wrong, you could very well move away from it.

For those reasons, I very rarely move my settler, and virtually never if I don't know where the food is.

I tend to take the opposite approach. Unless there is I think the starting square really is ideal, I'll almost always take a turn or two to move around.

Couple of factors in my thinking:

1. Some of the resources that the AI knows about might not be available until much too late to help my initial civ development. One of the best examples is when you get a spot surrounded by things that need a plantation. In that situation I'd much rather move and build my capital somewhere that has something I can use *now*-ish and then put a city near the initial spot once I'm near discovering calendar. A similar example that occurs less often is where the resource is on jungle. Even if it's something like rice that only needs a farm, that is no help until you have iron working and can chop down the jungle first. By the time you have iron working, the fact that you weren't able to use the resource will probably already have significantly hurt your development.

2. I tend to put a lot of mental effort into planning a way that I can distribute cities so they are just far enough apart not to inhibit growth, but I can eventually put as many cities as possible in an area in a way to use as many resources as possible. That's a much more complicated proposition than just finding a good site for one city. It may be that the starting position is excellent for your first city but would then make it very difficult to use some other resources 3-4 squares away.

3. If there's a barbarian hut one or two squares from my settler, then I'd rather visit it with the settler than the warrior - from experience that seems to give you the same advantage as a scout: You are guaranteed a friendly tribe. Plus my settler can move faster and might get to it faster (getting a scout or a map revealed on the first or second turn can give you an extra advantage in deciding your starting location)

 

[DynamicSpirit]

Isn't it odd that finally after so many years your people decide its time to settle down from the nomadic life, but they have no map to speak of and barely have a grasp of their surroundings?

You'd think nomadic people would have a better knowledge of the area they've previously nomaded around.

LOL! That thought has occurred to me a few times. On similar lines, it always amuses me that you can be getting income from trading with cities that you have no idea where they are (have the merchants who ply those routes sworn an oath never to reveal to you the trade routes they are using?) and that you can develop diplomatic relations, share technologies with other civs, develop literature and a civil service, yet have done all this without knowing how to sketch a map on a bit of paper and exchange it with another civ ('cos your wise men haven't discovered paper yet).

I guess in the end it comes down to the balance between realism and enjoyability. And exploring the world is one of the most exciting bits of civ, which I guess means you sacrifice a lot of realism to keep it that way

 

[KerThud]

I'm glad to learn that the blue circles are ignorant with v1.52. And now I must think more about where to place that initial settler. I continue to be impressed by the decisions CIV forces me to make. Good food for thought here, thanks to all.

 

[Thalassicus]

I often relocate my capital to a more central location, but usually to a coastal city. There are several reasons for this:

- I like Financial civs, which (among other things) receive +50% commerce on coastal tiles (3, up from 2).
- The Great Lighthouse and Colossus both significantly improve commerce for coastal cities. The Great Lighthouse is also useful for an amazingly long time (doesn't obsolete until Corporation).
- Harbors then improve the trade income by 50%. Harbors also provide up to +3 extra health, which can help your capital grow several population points larger.
- Bureaucracy improves Commerce and Production by 50%, after the other bonuses have been calculated.

I do prefer starts that are non-coastal though, it allows you to expand in all directions with a minimal increase in maintenance or upkeep. I then pick a coastal city that's relativly centrally located, with about the second-best production (best is reserved for the military city) and move my Palace there.

 

[DynamicSpirit]

I often relocate my capital to a more central location, but usually to a coastal city. There are several reasons for this:

- I like Financial civs, which (among other things) receive +50% commerce on coastal tiles (3, up from 2).
- The Great Lighthouse and Colossus both significantly improve commerce for coastal cities. The Great Lighthouse is also useful for an amazingly long time (doesn't obsolete until Corporation).
- Harbors then improve the trade income by 50%. Harbors also provide up to +3 extra health, which can help your capital grow several population points larger.
- Bureaucracy improves Commerce and Production by 50%, after the other bonuses have been calculated.

I do prefer starts that are non-coastal though, it allows you to expand in all directions with a minimal increase in maintenance or upkeep. I then pick a coastal city that's relativly centrally located, with about the second-best production (best is reserved for the military city) and move my Palace there.

I don't quite understand why you'd specifically want to move your capital to a coastal city - any chance you can clarify? I agree with what you say about benefits of a coastal city, but surely those benefits - other than the bureaucracy one - apply equally whether or not the coastal city happens to be your capital? For the bureaucracy benefit - obviously if you're going to use it then you want your capital to be a high production city to maximize the 50% gain, but I don't see any reason to expect that to be a coastal city - coastal tends to maximize gold at the expense of hammers (in most of my higher level games, I tend to be short of hammers more than gold so getting them up is my priority).

Also, why don't you want your main military-producing city to be your capital? Is there some disadvantage to that that I've missed?

 

[sandman_civ]

Is that really the case? Maybe I should pay more attention to it's recommendations!

This appears to be the case as all of these tech-related resources are pre-defined anyway. Still the blue circle might not be optimum as it might take into account uranium which is only of use late game, and it might be better to lose a turn scouting for a better starting position, or especially one closer to more hills. Annoying though when iron is discovered on a grassland

 

[Beamup]

I don't think it's likely to use the same code because the two functions don't have much in common. E.g., the suggestion algorithm always seems to offer two choices, while of course the AI is always going to choose a single place to put its settler. You could imagine that there's a code that ranks locations, and the AI chooses the best location, while the suggestion code suggests the best two. But the suggestions are also always within a certain distance, while the best place for the AI to settle might be quite far away (e.g., in the middle of the game when all of the good nearby locations are taken). The suggestion algorithm also seems to have diversity built into it; it doesn't tend to suggest two locations right next to each other. While if it were just ranking results according to a fixed set of criteria, it would pretty often be the case that the two best choices would be right next to each other. Etc., etc. They just don't seem very similar at all, or likely to share code.

The two functions are, in fact, virtually identical. They are both evaluations of city sites. What they do with those evaluations, and which sites they evaluate, are indeed different. But the core of both algorithms, and the part that would consume (conservatively) 95% of the programming time, is identical.

I recognize your opinion that they might, but it seems entirely hypothetical or suppositious to me.

Gee, do ya think that maybe that's why I explicitly put a "probably" on it?

The original post argued that one should accept the starting positions suggested by the computer because "the little blue circle placement algorithm knows about not-yet-revealed resources." This would only make sense if the presence of those resources has a strong effect, not if it has a very minor effect that almost always wouldn't even matter. If it's only a very minor factor, then using the superior judgment of human players would make a lot more sense than being guided by a dumb algorithm with a very, very weak "hint" built into it.

Furthermore, the thread you point to ends with the conclusion by those who did the tests that the behavior has changed and no one can find the effect any more in v1.52. Confirming the comment that seems to relate to fixing it, and contradicting the assertions of others in this thread.

I wasn't replying to the original post. I wasn't discussing whether or not it has been fixed. I was stating that your (apparent) original assumption that any such knowledge was ridiculous was, in fact, completely incorrect.

If I meant to say that you should place more credence in the blue circles because of that, I would have said that. You're reading FAR too much into things.

 

[Thalassicus]

I don't quite understand why you'd specifically want to move your capital to a coastal city - any chance you can clarify? I agree with what you say about benefits of a coastal city, but surely those benefits - other than the bureaucracy one - apply equally whether or not the coastal city happens to be your capital? For the bureaucracy benefit - obviously if you're going to use it then you want your capital to be a high production city to maximize the 50% gain, but I don't see any reason to expect that to be a coastal city - coastal tends to maximize gold at the expense of hammers (in most of my higher level games, I tend to be short of hammers more than gold so getting them up is my priority).

Also, why don't you want your main military-producing city to be your capital? Is there some disadvantage to that that I've missed?

It's a difference in focus. I gather you are primarily a warmonger, so the +50% production boost would be the strongest benefit of Bureaucracy in that case, going in the most productive city. Playing as Financial civs I focus my capital primarily on commerce with Oxford University and (occasionally, depending on where religions appeared) Wall Street.

The advantage here is that the +50% commerce boost from Bureaucracy has a much larger effect in a coastal city than a non-coastal one, since it has the added benefits described above. The city can grow larger, has more trade and income from trade, and has more base commerce early in the game for a Financial civ. The +50% production boost then helps counteract the lower production coastal cities typically have.

My city specialization typically is like this:
Most productive city - Wonders, and mid to late-game Military City. Ironworks + West Point.
Second most productive city - Early military city, with Heroic Epic and later the Red Cross
Another coastal city with good production, preferrably next to a river (for extra commerce) and centrally located - Commerce city with Palace and Oxford
City with most religions founded + commerce - Wall Street

It usually doesn't end up anywhere like that (depending on the starting location), but that's roughly how I make the decision of which cities to specialize.

 

[Wodan]

I'd like somebody to do the math... a capitol with 100% grasslands & Towns vs a Capitol with 100% Coast (on a 1 tile island) with Harbor, Great Lighthouse, & Colossus.

I'd be curious to see the results.

If nobody else does it when I get home I'll figure it out.

I get the feeling people are thinking the wonders will win hands down, but I'm not so sure. And, of course, this is totally ignoring production... it would be impossible to build those wonders in the first place. Still, as a purely commerce argument, it would be interesting to see.

Wodan

 

[Abegweit]

On similar lines, it always amuses me that you can be getting income from trading with cities that you have no idea where they are (have the merchants who ply those routes sworn an oath never to reveal to you the trade routes they are using?).

This part is historically accurate. Think Spice Islands.