Appreciate the effort to wait on me. My backlog will keep me tied up for at least another six hours. I'll try to offer some input before I go to bed tonight.
SGOTM 08 - XTeam
- Thread starter AlanH
- Start date
Hi CP,
Nice to see you again. Hope all is well.
Take your time as we're not in too big of a rush...
Much to work out still.
Nice to see you again. Hope all is well.
Take your time as we're not in too big of a rush...
Much to work out still.
Mad Professor
Deity
Mad Professor - keeping us honest, from wherever he may be, because he's missing
I'm coming and going a bit Leif. Caught up again.
Don't be in a hurry to start - the discussion is good, and the scout on the hill adds lots of info. Talk more first rather than rush into it.
The discussion on where to settle is interesting. I think SCT sums it up nicely though. The cow site is wonderful early on, but if the game is going to go long, settling in place might be better. It just depends what other sites are available nearby though. We might finish up better off long term settling on the cows if there's a spot just to the west or south west which compensates for the longer term disadvantages of the cow site? Problem is we need to make the decision before we find that out!
For what it's worth, my personal default is if in doubt, settle on the spot.
Having looked at the start position and leif's abbreviated thread (certainly appreciate the effort to save my time), I've not much to add but questions to ask about some considerations not discussed fully (at least in the abbreviated thread). . .
The discussion implies that, even if we hurry to quarry the marble tile, it would not be connected and would therefore not halve our speed in building the Oracle. Is this correct and does it hold even if the marble is in the fat cross and we have sailing and a road on it? If settling two east actually delays utilizing the marble, then I'm inclined to settle in place and anticipate establishing a city to utilize the marble.
We will certainly want the marble to speed building several early wonders. What is the best way to gain use of it? If we settle in place, on what turn would our capital expand to include the marble tile? Do I understand that in Bts we could build a fort south of the marble and somehow connect the marble (even if the fort and the marble are not within our culture? or does a fort have culture? and on approximately what turn will we be able to build forts?)?
I didn't read anything about building both Stonehenge and the Oracle in the same city to rather quickly generate an early Great Prophet(s). Is that worth considering -- either in the capital or, if we establish our capital to the east, in a nearby western city that can utilize the abundant forests we'd lose by establishing east?
The discussion implies that, even if we hurry to quarry the marble tile, it would not be connected and would therefore not halve our speed in building the Oracle. Is this correct and does it hold even if the marble is in the fat cross and we have sailing and a road on it? If settling two east actually delays utilizing the marble, then I'm inclined to settle in place and anticipate establishing a city to utilize the marble.
We will certainly want the marble to speed building several early wonders. What is the best way to gain use of it? If we settle in place, on what turn would our capital expand to include the marble tile? Do I understand that in Bts we could build a fort south of the marble and somehow connect the marble (even if the fort and the marble are not within our culture? or does a fort have culture? and on approximately what turn will we be able to build forts?)?
I didn't read anything about building both Stonehenge and the Oracle in the same city to rather quickly generate an early Great Prophet(s). Is that worth considering -- either in the capital or, if we establish our capital to the east, in a nearby western city that can utilize the abundant forests we'd lose by establishing east?
ShannonCT
Deity
To use the marble, we must have a city or fort on that island. If we settle 2 east, we could build a quarry on the marble for the hammers, and a fort south of it and 2 roads to hook it.
If we settle in place, it will be around turn 80 when our culture expands to the marble. Only then can we build a fort (enabled by Math, which will come sooner). If we settle on the cow, our borders reach the marble on turn 25. The techs required to build the fort and hook the marble (Math, Wheel, and Sailing) should come between turns 50-60.
I'm not having trouble getting 3-4 prophets by turn 200 without Stonehenge. Oracle, Shwedagon Paya, Chichen, Apostolic Palace, Ankor Wat, and Maoi all give prophet points, and preist specialists can be used as well. I don't see much use for an early prophet unless trying for an AP diplo win.
Frederiksberg
Deity
Haven't had time to look at the save yet, but I did spend some time during the weekend to make a small analysis tool for predicting E(t)/t for different tech paths.
Input to the tool is a number of "mandatory" techs and a "beeline tech". Using this input the tool computes the number of techs needed including prerequisite techs and the cost of these techs in beakers. It also computes the number of Wonders enabled by the techs, E(t). Finally, using an estimate of the beakers produced as a function of the turn number, it produces an estimate of the number of turns required to discover all the techs. Thus a rough estimate of E(t)/t can be calculated for any beeline tech.
I tried to run the program with Theology, Civil Service, Machinery, Construction, Alphabet, Currency and Literature as mandatory techs and got these result for the best beeline techs:
These are very preliminary results and it might be better to run a search for a combination of beeline techs. It's also important to try and vary the beaker estimate to see if there is a big difference between slow and fast research. Other scenarios with other mandatory techs could also be interesting - in the case above I have assumed that Astronomy is not needed.
Input to the tool is a number of "mandatory" techs and a "beeline tech". Using this input the tool computes the number of techs needed including prerequisite techs and the cost of these techs in beakers. It also computes the number of Wonders enabled by the techs, E(t). Finally, using an estimate of the beakers produced as a function of the turn number, it produces an estimate of the number of turns required to discover all the techs. Thus a rough estimate of E(t)/t can be calculated for any beeline tech.
I tried to run the program with Theology, Civil Service, Machinery, Construction, Alphabet, Currency and Literature as mandatory techs and got these result for the best beeline techs:
Spoiler :
These are very preliminary results and it might be better to run a search for a combination of beeline techs. It's also important to try and vary the beaker estimate to see if there is a big difference between slow and fast research. Other scenarios with other mandatory techs could also be interesting - in the case above I have assumed that Astronomy is not needed.
ShannonCT
Deity
No. I'm completing CS-slingshot around turn 70-75 in my practice games. From there, I'm going for Sailing and Masonry for Great Lighthouse, and Pottery if still lacking. We could be getting Aesthetics around turn 80 if we want it then.
ShannonCT
Deity
Just revisiting this estimate. I played a practice game last night with Leif's new "Practice Base" game. I got to Democracy, Communism, Mass Media, and Medicine around turn 225. That opens up 52 wonders, or one wonder every 4.33 turns. Now the real trick is building all of the expensive wonders without many engineers. We might be able to grow one or two engineers, but that still leaves a lot of expensive wonders.
It seems like the best way to build them quickly is to focus on raising a lot of cash at the end of the game. The best way to raise a lot of cash is to create a specialized gold city, containing Sushi, Jewelers, Wall Street, shrines, and other gold multipliers. And we want to think about turning research to 0% at the right time, before we research our last tech, because with Representation, our specialists can finish the final techs. With Kremlin reducing cash rushing cost by 33%, wonders can be rushed at the cost of 4 gold per hammer. If we can generate 1000 GPT or more, we should be able to finish the late wonders before turn 250.
This seems to confirm KC's assertion that a long game is better.
ShannonCT
Deity
A little graph gazing:
I see Smurkz and OSS finished their first turnset with 4 culture less than the max, meaning they settled on the second turn. Murky and CRC look to have settled on the first turn.
CRC has skipped bronze working but most have gotten it by turn 40.
I see Smurkz and OSS finished their first turnset with 4 culture less than the max, meaning they settled on the second turn. Murky and CRC look to have settled on the first turn.
CRC has skipped bronze working but most have gotten it by turn 40.
Frederiksberg
Deity
ShannonCT said:
Can you post some values of the beaker output as a function of the turn number. Then I could use this for my E(t)/t simulation.
When you look at my preliminary numbers it doesn't point towards Mass Media as the best beeline tech since it has a significantly lower predicted E(t)/t compared to Military Tradition. A higher tech rate may change the conclusion.
Frederiksberg
Deity
Played a bit more with the E(t)/t simulator. By multiplying my current beakers per turn function with a constant I try to simulate fast and slow research (would be better with some test game data). The simulator quite consistently reports an E(t)/t difference of 0.26 between MilTrad and MM when using the following mandatory techs: Theology, Civil Service, Machinery, Engineering, Alphabet, Currency, Literature, Education, Philosophy, Astronomy, Divine Right. The "fast tech" simulation predicted E(t)/t=1.17 for MM and 1.44 for MilTrad.
Would be interesting to see if a test game can confirm these numbers. Question is, of course, if the build gap and the capture gap can be closed efficiently with this tech path in mind.
Would be interesting to see if a test game can confirm these numbers. Question is, of course, if the build gap and the capture gap can be closed efficiently with this tech path in mind.
Frederiksberg
Deity
ShannonCT said:
I guess that means that Smurkz and OSS went for the eastern forest that grabs marble and fish in the FC.
Then I'm for settling in place (visceral rejection of settling on cows, though that's certainly a clever idea).
The argument for roughly a 250-turn game is compelling, and the idea of taking over one civ early if possible to increase our production base certainly is a desirable goal.
ShannonCT
Deity
I ran some regression analysis of some data points, with beakers from great people averaged in (all bulbing came in the second half of the game), and found the best estimator was a cubic function B=.000102t3+9
As for stopping at Military Tradition, if we're mostly using curassiers, airships, galleons, and frigates to capture all of the wonders we missed, stopping at MT is not really an option. The capture gap wil be too big at that point, and we wont be ready to claim victory anyway. If we're declaring on most AI to capture their wonders, we're going to need a domination/conquest victory.
Frederiksberg
Deity
ShannonCT said:
I would rather have some data points since this is what my program can readily use.
ShannonCT said:
Religious Victory is an option that requires a minimum of hammers to achieve as JT remarked earlier. War can be fought with many types of units. If we shoot for an earlier victory we must obviously use the units available at that time.
Domination/conquest is probably not an economical VC since we need to spend time and hammers capturing cities without any Wonders. If most Wonders are indeed built in a few cities it may be better to capture them one by one using a smaller army. A difference in E(t)/t of 0.25 is the equivalent of about 7 wonders so even if the capture gap is a little more difficult to close we can afford this.
killercane
Deity
Stopping at MT I dont think is an option either. We cant cash rush or even enjoy Kremlin's power. Also we cant use Redentor, etc. Lots of posts today!
I am against moving for double fish. If we stay in place we can put a city on the marble and use it for Oracle. Using only 2 chops for Oracle is pretty excellent, and leaves more for CS-ToA-Lighthouse. By the time those things are done lit should be upon us. Lets hope we have an abundance of AIs to make use of Lighthouse-ToA! If we do we can make 300 bpt at 250 AD, and another hundred once Oxford is built, probably ~450-500 with unis in place.
Im still thinking about cow though; staying or cow seems to be the best two options of the 3.
I am against moving for double fish. If we stay in place we can put a city on the marble and use it for Oracle. Using only 2 chops for Oracle is pretty excellent, and leaves more for CS-ToA-Lighthouse. By the time those things are done lit should be upon us. Lets hope we have an abundance of AIs to make use of Lighthouse-ToA! If we do we can make 300 bpt at 250 AD, and another hundred once Oxford is built, probably ~450-500 with unis in place.
Im still thinking about cow though; staying or cow seems to be the best two options of the 3.
Frederiksberg
Deity
Maybe I should reiterate the chain of thoughts behind the strategy simulations:
We can, of course, choose to forgo the extra 0.26 in E(t)/t because it seems easier to close the build and capture gaps if we tech further on. Or we can choose not to lower the bar and take up the challenge of meeting the - admittedly - more difficult task of closing the gaps with fewer techs at our disposal.
In some of our earlier SGOTM's our problem has been that we didn't act early enough and didn't push aggressively enough. As an example we waited far too long with the attack in SGOTM7 because we wanted Artillery in our stacks. It turned out that other teams beat us by starting war much earlier and without using a single piece of artillery in their wars. I don't want us to make this mistake again if it is indeed feasible to go for a more ambitious plan.
- As we all know the goal of the game is to maximize W(t)/t = 5*#Wonders/#turns for the turn where the game is won.
- As shown earlier W(t)/t is upper bounded by E(t)/t where E(t) is 5 times the number of Wonders that can be built or are already built at turn t.
- In order to maximize W(t)/t we should try to maximize E(t)/t and at the same time close the build gap which is (5x) the number of Wonders that can be built but are not yet built and the capture gap which is (5x) the number of Wonders owned by the AI.
- So far the simulations I have done indicate that E(t)/t has a maximum when researching up to Theology, Civil Service, Machinery, Engineering, Alphabet, Currency, Literature, Education, Philosophy, Astronomy, Divine Right and Military Tradition. Compared to a beeline for Mass Media there seems to be a difference of 0.26 in E(t)/t which is considerable since a winning score is probably somewhere around 1.0
We can, of course, choose to forgo the extra 0.26 in E(t)/t because it seems easier to close the build and capture gaps if we tech further on. Or we can choose not to lower the bar and take up the challenge of meeting the - admittedly - more difficult task of closing the gaps with fewer techs at our disposal.
In some of our earlier SGOTM's our problem has been that we didn't act early enough and didn't push aggressively enough. As an example we waited far too long with the attack in SGOTM7 because we wanted Artillery in our stacks. It turned out that other teams beat us by starting war much earlier and without using a single piece of artillery in their wars. I don't want us to make this mistake again if it is indeed feasible to go for a more ambitious plan.
Have you tried the simulation when Astronomy is required?
How would that change the W(t)?
Capture Gap would be a bit more difficult?
EDIT - Missed Astronomy, you already considered it...
Sorry.
How would that change the W(t)?
Capture Gap would be a bit more difficult?
EDIT - Missed Astronomy, you already considered it...
Sorry.
It seems counter-intuitive that giving up the cow (and forests) for slightly earlier development would be optimal in a 250-turn game during which we want to build maximum wonders?
BTW, kc, do you think your Vols are missing David Cuttcliffe?
Similar threads
