RB Pitboss #3 [SPOILERS]- Suryavarman II of England

[Cyneheard]

Agreed that the stone/furs might not be the best option, but it's certainly viable. Roading to it isn't that big of a deal, we have the worker turns available before Pottery comes in. I think Pottery will come in around T56-57. One disadvantage of the furs/stone city is that neither food-resource is first ring, so we have 5t before the workers can improve it. That's why the roading isn't a big deal: the worker doesn't need to be at the wheat (5/1/1 > 5/0/1 sheep) until T53 or so.

Ideally, within a turn or 2 of Pottery coming in, I think we should shoot for:
sufficient warriors for defense & exploration (2 out of London, 1 out of Animal Farm).
1 chariot for zone defense (slow-built out of London?). A future chariot can head off to the unexplored direction, but I'll feel a lot better with even just 1 good unit.
2 extra workers (1 each in Animal Farm and London is quite doable). Worker MM might let us chop quickly in AF upon the completion of Pottery. London can quickly build its Granary without chops or whips (4t, more if we switch to cottages during production). With 2 workers in each spot, and 1 worker improving our third city, we'll be up and running fairly quickly.
This includes 0 whips pre-Pottery; it's not worth whipping GHMs away without a granary, and Animal Farm should never go below 3 pop in non-emergency situations: resource tiles >> whipped pop. I'll try to work out this MM plan at some point, but it might be a while before I get the time to fine-tune it. The worker labor depends on where City #3 is going.

[antisocialmunky]

Yeah, London can fuel quite a few cottages and still have decent production potential. But I agree with Hap, we will probably need a commerce heavy city to the SE to support our expansion. Beaver Rock(?) doesn't seem too ideal for that :-\

That'll put the Gold/Corn uber hill off until city 5.

Also 41 turns without meeting anyone yet. That's just... meh. Atleast the barbs will start keeping us company.

[haphazard1]

Cyneheard,

I am trying to work my way through the distance maintenance formula from the article you linked (thanks!), and am not getting a reasonable value for DistanceScale. I am sure I am messing something up, but can not see what.

We have:

[[[[[25 * Distance * (Pop + 7)/10] * Building_m] * WorldSize_m] * Handicap_m] / DistanceScale]

With each [...] indicating to round down (floor function).

Distance = 5 (5.5, rounded down)
Pop = 1
Building_m = 1 (no courthouse)
WorldSize_m = 1 (Huge map, stated in the Settings screen)
Handicap_m = 0.95 (Emperor level, known)

This would reduce in steps as:

[[[[[1000/10] * 1] * 1] * 0.95] / DistanceScale] = 1.97

Or [95 / DistanceScale] = 1.97

This gives a DistanceScale of about 48 or 49. Assuming toroidal wrap, that gives us Height + Width = about 96...which seems WAY too low for a map with over 4000 land tiles.

The prospective 84 x 50 you mention would give DistanceScale of 67 with toroidal wrapping (and even larger with cylindrical or no wrapping), significantly larger than what seems to be emerging from the formula.

Where am I going wrong?

[Cyneheard]

I've been toying in my head with the idea of making a run for The Great Wall. This would have to be after settling Beaver Rock, of course. On Emperor barbs, it's well worth it. With stone, it's only 75h to build, and we'd probably need 8 worker turns for the quarry, 2 chops, and a whip overflow, to get it done. It could be done with 2-3 workers in something like 6t or so from completing Masonry (3 workers could: pre-road the stone, in 3t build the quarry and have 1 worker move away to start chopping, and if both forests are pre-roaded, we could be done 5t after finishing Masonry. 4, if properly pre-chopped forests + full whip overflow timed with Stone Quarrying). 2 workers would have a different work schedule, and probably get it done in 5-6.

Either we could try to get Masonry after Pottery (ETA of Masonry: T70-75), or after Writing (ETA: T80-90). Or not bother at all, but it was just a thought.

GSpy @ 2GPP/t takes 50t to pop. We can pop a GSci in 17t after finishing a library. I don't think it would be difficult to get a GSci in time to bump the Spy back to at least the 200GPP mark, if not farther.

A quarry is 8 worker turns, right? I can never remember. If it's no more than 9, the 3 workers could finish 2 chops 4t post-Masonry at the minimum total cost of:
Quarry turns + 2 for road
2 (road) + 1 (move) + 3 (chop) per forest x 2 forests = 22 workers turns @ 8 turns for the quarry, including prep work. I think that's more than manageable.

Edit: @Hap: I did the same thing you did, then what I did was I pulled up the WB game, which had a 77x92 toroidal map costing us -1.58/turn. Scaling relative to that map, our costs of -1.97, has 84x50 be within 2% or so. 70x60 or 65x65 would also work.

[haphazard1]

I am skeptical about the Great Wall, but we will see what Gulliver reveals over the next couple turns. That southeastern site could become high priority if there is something nice in the fog.

Great Wall pros: safety from barbs, gpp, culture
Great Wall cons: could fail to get it, lots of worker turns, forests chopped, could fail to get it, requires settling Beaver Rock next, diverts our tech path from what we would otherwise choose next, did I mention we could fail to get it?

With industrious civs and quite probably other civs with stone (since we had some close to our capital), the risk is very large. I also do not expect barbs to be much of a problem -- how much free space will there be to spawn in, with so many teams on the map? Maybe we are next to former Korea and this will be an issue, but I doubt it.

On the maintenance formula, I do not like the results not matching the expected formula. Number of cities maintenance matches perfectly, so there is something wrong with the distance calc. I will try to dig into this further later today, but it will have to be after lunch. Too hungry for brain to work properly right now.

[haphazard1]

OK, dug into the code a bit. (Knowldege hunger once again beats out food hunger. ) The article Cyneheard linked to is not quite correct about how DistanceScale is calculated.

In CvMap.cpp the function maxPlotDistance calculates the value by calling plotDistance (found in CvGameCoreUtils.h as an inline) passing the coordinates to be compared as (0,0) -- one corner of the map -- and the width and height of the map as modified by wrap; i.e. if that dimension wraps it is passed as 1/2 the max value.

The key is that plotDistance treats diagonals as 1.5 tiles. So the result is that DistanceScale for a toroidal map is actually:

(larger dimension max/2) + (smaller dimension max/4)

So our obvserved DistanceScale value of 48-49 would still be too small with Cyneheard's estimate of an 84x50 map:

84/2 + 50/4 = 42 + 12.5 = 54.5

So it is a bit smaller, maybe 74 x 48? That would give 49, which is about right. Might be a little squarer, maybe 72 x 50? That would give 48.5 and is almost dead on. But 72 x 50 only gives 3600 tiles total, and we know the land value is over 4000.

Hmmm, needs the smaller dimension to be bigger to get more tiles without boosting DistanceScale too much, since it is 1/4 in the formula. So let's consider a square map of 64 x 64, gives 48 DistanceScale (almost right) and 4096 total tiles, also about right.

So it looks like the map is "squarer" than most scripts, and about 64 x 64 or something very close to that. Water will be very scarce, just a bit over 1%. (Sucks to be Portugal or Carthage.) And there is no question that toroidal wrap is on, the DistanceScale and total land tiles numbers would not even come close for cylindrical or none.

[haphazard1]

One more thought, just to spam the thread a bit more :

64 x 64 map could be divided pretty neatly into 16 sections, each 16 x 16 in size. Nice regular spacing and everything with the toroidal wrap.

Of course, we started with 17 teams...hmmm. Maybe not so useful a thought after all.

Still, the overall values support our earlier working assumption that distances between starting sites (allowing for movement before settling the capital) should be about 15-16 tiles.

It is quite likely that we have explored into the edges of "other people's territory" already. The edges of what we can see will be disputed land. This fits pretty neatly with the gold being where it is -- that site is probably just about exactly between us and some other civ to our north. We may find equally rare and valuable resources at roughly the same distance in other directions, as a meta-game guess.

[Cyneheard]

Hm. I think we're overthinking Sullla's willingness to control every single tile to have the centerpoints between 2 sides be all the rich stuff. He's just not going to put that much time into it. Our starting locations will be somewhat similar, and strategic resources as well. He'll put his one trick into the map (no copper? I doubt there will be any ivory on this map at all, and don't consider that a trick). No iron as well would be unspeakably cruel. But the gold/corn city isn't so far out of reach for us, it's about 9 tiles north of our true starting location, and 10 north of London: rapid expansion can claim it for us, and with horses we can defend it. And it's not clear that our neighbor is exactly straight north of us.

Still, a square 64x64 makes sense. That means our 4 lake tiles are 4/49s of the water on this entire map! Marginally high (expected value for our exploration would be about 2.5-3), but not enough to change that 64x64 map. And we know of 3 lake tiles in our opponents' starting locations, again about right for 49 water tiles.

Re: TGW, the cost of building it's very cheap. The cost of getting Masonry's the sticking point. Just thought it was worth bringing up. I'm not worried about having 2 chops and 1 whip of overflow converted into cash (and, frankly, there's a very short window where that would be a problem: We'd be only putting hammers into TGW over 2 turns).

[haphazard1]

I could see pretty careful control over critical resources to make sure they are fairly distributed. And silver, gold, and gems are all critical -- early happiness which never expires, doubled by forges, and providing big commerce yields. I think it is reasonable to assume these were checked for fairness and moved/created as required.

On starting placement, yes, things will not be exactly lined up everywhere. Especially with a 17th team being squeezed in somehow. But the average distances between starting points should be about right. This means that gold is just beyond halfway on the "expected average" distance to the next civ. We can get it, but we are likely to have to make an effort. That site is very attractive and will be targetted by anyone who sees it and is semi-close.

The Great Wall is worth considering -- your points are all good ones. I am more concerned about the indirect effects such as delaying Writing to tech Masonry, or settling Beaver Rock next rather than a better site elsewhere (assuming such exists, of course). Not to mention spending a bunch of worker turns which might be better spent on cottages, roads, etc. Also, the risk of being beaten to the wonder is large -- only 100 hammers for an industrious civ, and no need to find stone, settle, hook it up, etc. I actually expect the Great Wall will be built quite soon, and this discussion will be moot.

[Cyneheard]

How about we agree to reevaluate TGW when Pottery's about to come in? If either Ind civ seems to have teched Masonry (very hard to tell if Huayna gets it: it'll cost them the same as Pottery, and they have the ability to research either. Right now, Bismarck doesn't have the Wheel. Bismarck has both Masonry pre-reqs, and no Wheel), then we'd be crazy to go for it. Heck, if it seems like anyone has Masonry, then there's absolutely no reason to go for it. GLight is useless, Pyramids are crazy expensive (and not what our civ needs anyway): forge + GE is probably the fastest (and cheapest) way to get that one, unless Bismarck or Huayna found stone. Especially since we'll be looking at 40-60% slider at that point (even just 3 cities...), we can run 100% gold for a while, and give ourselves a few extra turns to decide.