(September 12th, 2013, 11:06)WilliamLP Wrote: The first question about a simulation is always "how well does it track reality", isn't it? Honestly I'm not sure how well the actual growth and production of a city would actually match, and there's a danger of picking something that optimizes an idealized model (and not necessarily value in game), because of how we chose that model.
It won't track to reality at all, of course. More of a model for 'potential' rather than actual outcomes.
(September 12th, 2013, 11:06)WilliamLP Wrote: There are lots of complications, and I'm not convinced they don't matter, even a lot:
- Which tiles to work is not a static decision. To stick with one tile set isn't just sub-optimal, it is really poor play that makes a big difference.
- Related, whether to share the grass cow from the capital is not a yes / no decision. In a turn when the capital is building a setter or a worker and can't grow, it doesn't need any food surplus at all! +0 is fine and hammers are more valuable than food for workers. And so the cow is absolutely more valuable to use for growth elsewhere in these times. This is an advantage for a city sharing it, and not a small one.
True, but tile sharing would be brutal to model so won't be touching that with a 10-foot pole. I did build in a modifier while testing against Master Krill's PB5 screenshots. But that's just a manual 'adjustments' input.
(September 12th, 2013, 11:06)WilliamLP Wrote: - How many worker turns are needed at any given time to make a plan work matters, and would be very hard to model.
Actually that would be pretty easy once you've defined tiles. The spreadsheet knows when a city grows and could easily figure out worker improvement requirements. Would have to assume that workers travel 100% efficiently except for the move to the improvement target should it be forest/hill. That actually might be useful for ballparking worker requirements and timing anyway.
(September 12th, 2013, 11:06)WilliamLP Wrote: - It's highly likely that any city is going to spend some turns on workers or settlers, and not grow those turns.
As an input that would be ok since the only effect to the sheet is that it delays growth.
(September 12th, 2013, 11:06)WilliamLP Wrote: - Whether to whip is never going to be as cut and dried as doing it every 10 turns.
Wouldn't even think about touching that one.
(September 12th, 2013, 11:06)WilliamLP Wrote: - Hammer output vs time is not the only objective function.
Hmmm,...this is why I said that it was only useful for a production city,...but,...it wouldn't be impossible to calculate cumulative turns put into a tile defined as a cottage. Might be doable in spreadsheet without going full 'app'. Not sure that I would want to get into that level of complexity.
(September 12th, 2013, 11:06)WilliamLP Wrote: So I don't know if in practice it can be done better than just going through each turn and considering each separately! I love the idea though as a tool in the kit, but I am skeptical about leaning on such a thing for winning Civ decisions.
If 'BetterAI' can't beat human decision making there's no way my 2 hour spreadsheet is going to. All conclusion to be taken with a pound of salt.
You know, I fully support your movement to bring Sabermetrics to Civ 4. With the state of the current art though I suspect that the gut, plus fuzzier reasoning, do a better job. But if you want to play a game in another PB that is very heavy on number crunching and analytics I'll happily lurk you!
If I were going to try and model this, I might start just by looking at "how good is a city at size 1, size 2, size 3, etc." and also quantifying how quickly it can grow based on available food. Trying to multiply that together into one master number is a little tenuous in my thoughts. And again, a wildcard supporting the cow-sharing sites, that 2-city micro can have 1 city growing while the other one is building a worker or settler, and vice versa. The benefits of 2 cities mutually alternating between growth and production if they share tiles isn't something that I've been conscious of before, and and I'm definitely learning things here.
At any rate, the site 3W1N is a good one by any measure.
For the current turn, there is no power change visible on F9 (great!) and retep's units have all disappeared out of sight.
For our unit placement, what you see there is what we have. The repositioning is going on in the NW, and 1 axe is returning to Barbiere, leaving an axe and an archer over there. There's 2 axes and an archer in Barbiere now, with another archer that can return immediately. It's kind of light at the moment which is why I want them fortified! Retep can't see in the city, which is a really good thing.
He roaded the plain hill 1S of the staging tile, and moved his worker away without mining it. I'm not sure exactly what to make of that; he wants tactical options I guess. If he wanted trade routes from us, connecting that river would be the way to do it, though they'd be useless anyway even at peace.
In the east we have an archer, a couple of quechuas, and the forking archer can get there immediately. Given that we can see Retep coming from a mile away, and Ichabod has almost literally no military yet, I don't think we need more than that at this moment! We have basically the minimum viable defense everywhere, but I think that's what the goal is.
My worker management in this game has been pretty bad, sorry. The "3T farm" tile should be finished.
Actually I figured out a big flaw in my understanding of worker micro! Starting from the "3T Cottage" tile, I thought a worker could go NW-NE (crossing the river ends the semi-turn), then N-Farm in the same move. It can't! Crossing the river uses the two half-moves. So a worker's move isn't quite as good as 2 turns of a 1-move unit, like I thought. This explains a lot of mysterious "lost" worker and scout moves I've seen in other games.
Oh, writing finishes very soon. What would you think about opening borders with Ichabod? There is a river that is a natural trade connection. The commerce boost would help him more than us, but it would help us. Too much of a deal with the devil? I'm inclined not to care and take it anyway, just to improve our standing relative to retep and bacchus with the goal of trying to be dominant (eventually) in this west part of the world.
Retep could be roading that tile for trade routes to a southern city that he plans to settle or has settled already?
He could also be roading to plant a city along that southern corn. That would be an uncomfortable border.
Open Borders: Yeah, why not. Alternatively, we could OB after we plant the gold-corn city. Not be too friendly until we've sealed him off from our Western area? Other than that consideration, our primary problem right now is Retep over Ichabod. Although Ichabod's probably much closer to crashing his economy.
It's probably obvious, but I assume you have cottages growing (to hamlet, village, town) in your model for computing total commerce? In practice, I don't know if anyone would actually start a whip cycle as soon as city finished a granary - they'd probably wait for it to grow to the happy cap first, but who knows, with a city that had a massive food surplus (like a farmed flood plain site) maybe it's even a good idea, I don't know.
Anyway, today in the news:
Retep and Bacchus have peace, so I suppose we need to look out for him now that he's not as tied down. Also the age of the barb Zerg rush seems to be here. I killed 2 warriors this turn, which will level up an axe and the archer in that shot.
Retep's power is steady. I'll post the full suite of graphs as soon as we get Slowcheetah's, either next turn or the turn after.
Unit moves:
I had the thought that I think we can now tell when retep roads the staging tile, because the graphic for the road on the plains hill south of it should change! Indirect scouting for the win, then.
The N axe was diverted yet again on a heroic quest, but she killed another warrior and will get a promotion out of it, and will eventually get back to Barbiere.
The settler finishes this turn.
Questions:
1. I'm not saying it's a good idea, but what do you think about rushing a city next that claims wines somehow? Either tweaking the location of marble city (bleh) or trying to get a site like 3E1S of Carmen? (That would be "only" 8 tiles from our capital, fwiw.)
2. What would you research next? The two main options are Math (for forest chops, and cornerstone tech for most of everything) or Poly->Mono (for free switch to Organized Religion and building multiplier, build missionaries). I'm kind of torn. Trying to blitz out Retep's capital with a catapult total war is an all or nothing play, obviously. And the way things are going now, it doesn't seem like cats for defense are going to be needed immediately.
Bacchus showed up with 2 chariots at the right time, when Carmen is underdefended.
We have a slightly wounded CG1 archer and the W2 quechua in there. The chance of the archer holding is about 73%, so it's down to luck now.
If he wins, the city will auto-raze and he'll lose both chariots for a bit of gold.
A warning: if we lose this city I might need to check out of this game into Azza-mode and just play the turns without a lot of update in this thread. With 2 PB games going on this level of reporting is a lot of investment. Having two warmonger neighbours rushing at once before 1000BC is a little much for me.
I totally wasn't aware of this:
Writing enables map trading in RBMod! Accepting is a clear win / win.
Odd resource distribution, and a wasteland between us and Cheetah.
I offered peace to retep, hoping that he won't just reject it and interpret it as the right time to attack. It could be a really bad play, but I'm a little on tilt right now and not sorry for it!
Update: it looks like the CG1 archer was enough to make Bacchus pack up and leave, and he was just probing to see if we could defend and get free gold if we could not.
Still since he has chariots running around we need a couple of spears now. But the game goes on!
Retep wanted no part of peace. We got another set of graphs and world maps from Ichabod so I have massive amount of info to dump here, but it will wait until tomorrow morning.
Bacchus's chariot threat and me saying "oh shit" and panicking threw my worker management way off - I moved a few back to make sure he wouldn't steal them if he took the city, which will slow some things down, unfortunately.
The next game I play (and hopefully more than 7 tiles from someone) I'm going to try and plan worker movements exactly more than one turn in advance. It's been a huge area that needs improvement for me, this game shows.
With the state of the current art though I suspect that the gut, plus fuzzier reasoning, do a better job. But if you want to play a game in another PB that is very heavy on number crunching and analytics I'll happily lurk you!
