I will look over the plan in more detail tonight, but I'm just wondering right now if we want to look a bit ahead to the 5th workboat. Since we probably want to put nets on 2nd city's fish asap.
Too risky to delay Oracle a turn or 2 and max overflow from Oracle whip?
Indeed, we do want a fast WB. After the Oracle, I was expecting Fish Hills to keep working the two mines to get out that WB in at most six turns (45/8 rounded up) while losing 1
/turn. The capital is building a settler at size 4, so all our good tiles are being worked, and no alternative plan can do this and get a faster WB without extra whipping.
My previous Oracle plan got that Fish Hills WB in 5 turns. I improved my Oracle tile-working plan today to get 5
overflow after the Oracle, to get that WB in 4 turns.
Since the rate at which we can put hammers on a workboat is the same as we can do on the Oracle, there's no advantage to delaying the Oracle to transfer those hammers to the workboat. Neither is it worthwhile to get the WB early (unless CC is bored! impossible!) since we can't work the fish until 2 turns after the Oracle builds.
I'm thinking about the 3rd workboat... the
on it are beginning to expire.
They look like they're expiring, but if you play it out, they don't yet. That is true of the hammers on the barracks, too.
I presume there is no MM strategy that builds WB before the worker?
The max overflow off the settler is currently going to build nearly half the worker, so that we can whip the worker the next turn once the normal
go on also. We *could* put those hammers onto the workboat instead, which will build straight away, putting some irritating food into the box, then whip out the worker ASAP. That delays the worker by 5 turns (once you consider the overflow effects). If we were to keep the worker plan the same, that's 5 turns slower on each of the three improvements for Fish Hills, so naively 10
+15
. Call that 3-4 Oracle turns. However we'd probably skip the three turns on the corn road and not have to temporize one turn on a FPH road, so that probably picks up a turn on the Oracle at cost of some food in both cities (no corn =>
at size 5/4 respectively). So it is an option worth considering.
That delay does jeopardize our 1000
from Oracle->MC. Izzy without marble (IIRC) has built Oracles in the T90-99 range in some of my test games. Later times are more common, though.
Or maybe get it out before the granary, and send it as far away as possible to hunt for AI?
In the abstract, the granary is our best building. The earlier map knowledge we gain also has a value (albeit unknown and wide-ranging). This is tough. We'd like to be 2-whipping the capital about T80, when the double-whip
expires and we're at size 4 again. Meanwhile we've only put about 28
on the build item, and haven't researched Sailing yet. So the only T80 2-whip available
is the granary, and that gets the fourth WB out T92. One-whipping the WB at T80 means we regrow fast to 4, and then the granary is 2-whippable with max overflow about T89, but we haven't teched Sailing yet and we don't necessarily want to snowball the whip-
yet.
We can reasonably gain about 12 turns exploration by compromising the granary by at least 9 turns, and compromising CC to some other degree (doing several undesirable 1-whips, adding extra whip-
, or something) and maybe delaying a forge or the third city, which would be delaying a 3-wonders run or a Duckweed-Pyramids.
What would we do differently if we found Ragnar's capital close by with our WB on T90 compared with T102? Joao's capital? Met someone's WB? Found a big ocean 6 tiles away?
Both alternatives are risky. I think we're best losing the 12 turns of exploration. That does come with the reward of faster growth to offset the potential costs of slower research. In my test games I did legitimately have time to find city sites with the combination of the exploring WB and short galley expeditions (e.g. we already know we can settle the stone usefully even without a food supply, the clam on its own would be acceptable, maybe even marble without food is good enough if we plan GLib, NEpic and HEpic). Our explorers are never likely to definitively prove we're isolated in time for us to react (and the first one might go the wrong way entirely when we
aren't isolated).
FYI: Research formula is floor(floor(((1+cities)*ai_known/total_ai*0.3) * prereq_bonus)
Depending on the prereq bonus, each AI we meet that has knowledge of the tech we're going for is worth an extra 5-6%
.
Worth knowing, thanks. There's still a big random component on how many turns we might know AIs while doing relevant tech, however. There's another big random component on how much jeopardy we'd be accepting on the Oracle or expansion, too.