MM thoughts
Food is normally king, but hammers that increase food later can be better. So I mapped out 50 turns of the game assuming 2-3WB and fast settler (1 turn revolt from BW about T40). I have considered 2-3WB then another
unit in the analysis.
Executive summary
- Maximise use of netted clam tiles
- Manage output to get each new WB ASAP given the above (i.e. do not grow first, then work nets, PFH and a non-net)
- Minimize overflows that will not be immediately useful (e.g. into box just before we build a settler) if you can get useful things instead (like overflowed onto that settler).
- Be prepared to juggle between corn, PFH and GF if that will speed a growth or a build leading to a growth. A turn not working a net that leads to finishing a WB a turn faster to work a compensating extra turn on the new net can even be useful.
Gory details
T0-10 we work 3 corn to grow to size 2. This has put 1
on the WB. Then four plans come to mind:
A) T11 we work PFH+GF to get WB in 9 turns, switch GF to nets to get WB in 11 turns (no overflow). If we then go for a fast settler, we'll start a third WB, and switch to settler at size 4. This can be whipped T47.
B) T11 we work PFH+(1GF 10corn) to get WB in 11 turns, switch corn to nets to get WB in 11 turns (no overflow). If we then go for a fast settler, we'll start a third WB, and switch to settler at size 4. This can be whipped T48.
C) As for A, but if we then go for a third WB, we'll work two nets at size 2, then two nets+PFH, then one turn of nets+PFH+2GF, then start settler with one turn of 2 nets+PFH+GF, then three nets+something (gives up a turn on a net to get WB a turn earlier and thus another net turn back, plus earlier start on settler). Settler whippable T51 (on the nose).
D) As for A, but if we then go for a third WB, we'll work two nets at size 2, then two nets+corn, then two nets+PFH+GF. Assuming settler next, we mostly work three nets+something, whippable T54.
I spreadsheeted these to see total production (have done a fair amount of comparison with the game for verification of the spreadsheet). Numbers are total
(i.e. including that which we ate, but not
building a settler) / total
(including
building a settler, but ignoring whether we've whipped or not) / total raw
(including palace). These metrics are invariant over whether we work PFH or unimproved corn when building a settler (for example).
Code:
T22 T33 T48 Settler whippable?
A 103/ 64/202 177/102/327 316/180/509 T47
B 109/ 56/198 175/100/319 319/170/501 T48
C 103/ 64/202 177/102/327 311/174/511 T51
D 103/ 64/202 177/102/327 348/148/509 T54 (third WB in place T48 but not planted)
A and B have the expected early behaviour (A has
, B has
) but by T33 A has made up the
deficit and moved ahead in
too. That's from earlier working of both the first and second net. That converts into earlier access to a settler with less food stored in the box, still with the same
lead. So that makes A look better than B, to me. If the third build is a
unit then A will still be ahead.
The third net doesn't come online until T46/49 for C/D, respectively. The third net does 5 turns work for each settler. Then, they're not good for anything post-whip until city-3E is up (i.e. T54/T57 respectively). Then they're great by comparison with A, so long as city-3E can still get (or does not need to get) the Oracle.
One consideration is that the very fast settler in A is in some danger of arriving before we get Pottery. In fact, during the settler, you need to switch off some other tile and work the unimproved clams to get enough
. Four turns of that mid-settler is enough to be sure of Pottery and no delay on the whip. Call this plan A'. Its T48 stats are 316/175/519, since we're shuffling 5
into 10
. This makes sure the first
in Fish Hills go onto a granary, which might be critically important for an Oracle.
The comparison gets very messy once we whip, of course.
Did I miss any options back in the T10-30 window?