SGOTM 16 - Plastic Ducks

We had 4 cities when PD had 2 and Kauk had 3.

sometimes slower REX while securing critical wonders can be better, especially in such long game as the Ft is.

GLH with nearby island/s is always killer wonder and this time there was pretty big island west from main continent.

I would argue that on map where you can get island cities and a lot of coastal on main continent GLH is much better then Mids. but I am usually alone on the "bigger rex is better" in such discussions :).

Was very impressed by PD's quick rex after critical wonders.
 
We had 4 cities when PD had 2 and Kauk had 3.

It'll be nice to compare the hammers investment of the first 100 turns or so. As I said a bit earlier, we invested some early hammers into granary to support the Hereditary Rule-whip-REX which allowed for faster cities after 3. As long as a city is a net benefit (especially in a 200+ turns game), you don't care how early one of them is settled, but instead the total amount of city-turns (I guess you could say instead population-turns) you get in the end.

As LTC said a few times in the past, it's total population that's most important (paraphrasing).
 
The one voting for Mids may argue that this is a trade-off between 300 hammers and 360 beakers(monachy) in the short term. However, the return rate of Classical Techs on average is much lower than granaries and settlers, expecially after TGL is on hand.

[Edit: If with stone, the deal is changed to 180 hammers for monachy. It's close to the effect of oracle, so why not?]

With stone, the cost of mids is 500/2.5=200 base hammers, that's double as Oracle even without marble (150/1.5=100H), moreover, Oracle will be done ~T60, while mids can only come ~T90, this period is the critical moment for Orleans to overcome the happy cap to whip.

The Colossus does not fit to our style very well as we whipped a lot, but this is not enough to convince me. We have many water tiles and a long period between MC and Astronomy to have it work. What else is against this cheap wonder with only 160 normalized hammers?

First, Astro can not be delayed in this game, it provides

1. ~40C from trade routes
2. UU of free artist and 25% science bonus
3. some happiness resources
4. Galleons for swift war with Ram and barely enough to stop Hammy from going for liberalism.

Therefore, if we went MC before CS and built it ~T125, that's ~800C gain until T166 at the cost of

1. 250/1.5 = 167 base hammers
2. delayed bureo + OR and therefore delayed 3 cities which are waiting to be settled after civic switch
3. delayed the settlers from Orleans due to building forge+TC

Therefore TC is a weaker wonder in this game. there's not much difference of T100 or T125 TC since the effect of TC depends on how many water tiles being worked, earlier TC won't give much advantage.

I don't against wonders, but I always try to evaluate the gain from wonders to see if they are better than settlers. That's why I suggested building a weak wonder like MoM in Orleans since the peaceful expansion was done at that time.

If you have time, I'd suggest you to read dingding's posts in our SGOM11 thread for the evaluation of wonders (in the last few pages). In term of hammer vs commerce ratio, GLH is like 1:50 in this game while TC is like 1:2 and mids is like 1:1.
 
I partly agree on your view of Pyramids, Duckweed.
They been mediocre in this game like all sgotms cos GLH exists. Otherwise they have been what kept us in this race beakers wise. So you are right, we focused too much on them and forgot about GLH.
Peoples arguing it's been okay building them..we do that cos we still liked how we played overall, place 3 would be goody for us. So you should not call that amusing, not everybody can be the top team ;)

You don't need to agree with me.;) I respect other players choices. The only case I will try to convince others is our team members.

Kaku IS a top team in my eyes, they are like previous OSS (SGOTM11-13) with quite good strategical thinking and most important -- top teamwork. Honestly, I feel only our SGOM11 teamwork was comparable. Their tech pace in this game is only 5~10 turns behind us, that says, if they have thought about the spy trick and realized of going UN, they will likely finish ~T240. That's a very competitive date with us.
 
@DW

Ah, mids costs 500h, not 450h. I forgot it.

TGL is definitely better than TC, but the gap of 1:50 and 1:2 is overstated. It does not take time into consideration. 1C at T125 is just 0.5C at T100. In the mean time, the cost of Techs and other gains have to be taken into consideration.

I calculated the return rate of MC path and CS path now. CS is about 40% higher than that of MC(TC), so overall CS first is the right choice in our game.


Spoiler :
The investment on CS path is about 1600 beakers (CoL + CS), 1 turn of anarchy(150~200 h/c) and a few missionaries(150~200h). The gain is 25% buiding bonus(25h/turn) + Beurau (30~40 h/c per turn at T123)

The investment on MC path is about 700 beakers + 167 hammers(TC) and a few forges(400 h). The gain is 25% hammer bonus(20h/turn) + 10c/turn at T123


However, if we get the gem city earlier, CS path will be quite close to MC path.
 
@DW

Ah, mids costs 500h, not 450h. I forgot it.

I calculated the return rate of MC path and CS path now. CS is about 2~3 times of MC(TC), so overall CS first is the right choice in this game.

How can you calculate that?

What do your models look like? Willing to share?:)
 
How can you calculate that?

What do your models look like? Willing to share?:)

just edit it. Some errors in the first shot. The difference of the two paths is not that big actually.

There's some simplication in the model about the missionary cost and forge cost, considering the factors below
1. some of the religion is spreaded by Brennus
2. missionaries can also bring 1 happiness as a warrior does
3. the city number is different in the two paths
4. 25% hammer bonus can be applied in all cases but there's no much advantage in this case vs 25% buidling bonus

The data could be more accurate if I download the save and count one by one. However, it's normally not required in the real game since our target is just to compare and know which way is better.
 
Well we got mids T75 so that is 15 turns ahead of Duckweeds target. Maths chops and forge really helped.

T180 what did we do? We naval attacked Hammy with Mace and trebs. We took a beating but slowed him down tech wise. He was getting close to Economics and lib. Hammy retook Babylon with a 7-8 strong stack 1-2 turns later. He was a tough nut to crack. So we came back with cannons a few turns later.
 
You didn't account earlier Sankore into CS path.
It is almost impossible to make a long term model in civ. You can't involve all the parameters and then optimize them.

For example, if you had Colossus, you'd work coast tiles in Tours instead of riverside plains cottages, and would optimize your game to max the effect of said wonder. And it is like that with all the other things. If we could make anything near a good model, we'd also have good AIs in games. And those never happened. Anywhere ever.

Feeling of a good player >>> simplified models.

Because civ is much more complicated and totally non-linear problem.

If difference is huge, like 10:1 simplified models using return rates can be used. But civ also has build Wealth/Research options which can totally blow your mind cause they can but don't have to snowball into something huge. Wars add to complexity the most. For example. CS path is 3 turns faster than MC over period of 100 turns. Almost negligible. But, 3 turns faster war can, for example again, kill Hammy just before he researched Liberalism. 3 turns later, we either lose Lib, or lose Libbing Medicine. And that delays sushi and that is really bad since you lose a lot of turns on Sushi, and after that, on Mining Inc. since you can't whip without Sushi. So, you see, it is almost a deterministic chaos.

Having said that, I think/feel Colossus at the time we could have built it was so cheap compared to the potential output of the empire, that it would pay for itself. We saw the research OF in the end. Pure waste. That is couple of renaissance techs (10 turns of research at least). TC lasting for 50 turns would probably speed our research up in mid game at little expense of research in late game (less maintenance - higher slider, many coasts contributing +1 commerce). I think that'd be more optimal. But don't have any numbers to prove it, nor will I have the ability to do so. I felt that way before we started Sankore (when DW asked what did I think next wonder was) but didn't feel like debating when I can't bring what people only look at, and that includes me - numbers.
 
For a wonder such as TC, with copper easily accessible, and having a city that has nothing to build (other than wealth), TC SHOULD be a wonder to go for. With copper accessible, it's effective cost is 90 hammers or so. At around turn 90 for US, 6 beakers were worth 7 gold empire wide. So 90 hammers equates to 110 gold. 110 gold equates to 95 beakers. So for TC to pay back, it needs to result in 95 beaker for us via the extra one commerce per water tile between now and Astronomy. Assuming we only work the minimal tiles (cities that have fish food), that's AT LEAST 6 extra commerce per turn. US kept the slider at roughly 60% for most of this period iirc. So 6 commerce converts to 4.5 beakers (assuming library in every city, which should be almost a necessity considering we are CRE) and 2.4 gold. Going by 6 beakers are equal to 7 gold (not necessarily true, but it works fairly okay), 6 commerce gets us 6.6 beakers per turn. Pay off time for TC is therefore 14.5 turns. It's not a bad wonder and definitely worth building if there's nothing else. Obviously, opportunity costs are something else entirely because it's an entirely different situation if having to choose between building a settler or TC.
 
@shaka

If we count the return rate of "paper + sankre", it's lower. (800 beakers+ 275h for 20 beakers per turn: all normalized and 10 religious building assumed at that time)This is why we go MC after CS, not paper.

Building wealth is just to invest on techs. The return rate depends on what the tech is and how many beakers are left to research the tech.

I don't say that the calculation is all. It's just reference. However, I don't trust feeling, even it's my own. This game is eventually a game of operational research. When my model fails, I need to figure out what is the reason behind and polish it.

Definitely a lot of events and rating will change the result, but that's beyond our control. It is something like statistics and risk control, but it can still be calculated IMO.

Going back to talk about TC, I don't think it's a weak wonder---from my instinct with so many watertiles and 50 turns before Astro. The gain from TC and the pop is linear correlated, and the return rate of pop increase itself is usually high. However, the calculation shows that it still cannot win over the CS approach. You cannot look too far because both paths will be finished in 20 turns. What we need to decide is just to choose which path first and the marginal increase is the one we need to measure. If there was just one gold/silver/gem, I think I would go MC as the TC is unique.

PR's session is better for TC. They have less cities but more pop. They oracle MC if my memory is correct. However, they have no interest on islands, water and Hereditary rules. They invest too many on GPs(a kind of Tech), but the return rate of a lot of Techs are lower than island cities.
 
Well we got mids T75 so that is 15 turns ahead of Duckweeds target. Maths chops and forge really helped.

T180 what did we do? We naval attacked Hammy with Mace and trebs. We took a beating but slowed him down tech wise. He was getting close to Economics and lib. Hammy retook Babylon with a 7-8 strong stack 1-2 turns later. He was a tough nut to crack. So we came back with cannons a few turns later.

Slowly working on thread summary all the way back in post #2.

[...]
Having said that, I think/feel Colossus at the time we could have built it was so cheap compared to the potential output of the empire, that it would pay for itself. We saw the research OF in the end. Pure waste. That is couple of renaissance techs (10 turns of research at least).
[...]

The excess OF is my fault as I didn't re-create the test game well enough. Had I done so, T234 would have been 100% certain (queue Rifling earlier instead of taking 2 turns on a tech) and T233 probable as well (AP win).
There are more examples which slowed us down, ex. Kremlin was delayed 1 turn unnecessarily which cost ~2000 beakers.

As you say, completing those renaissance techs earlier would snowball big time which is why I tried finding ways to accelerate our research (banks :lol:).

Anyway, Civil Service was also very important for irrigation, our food situation was pretty bad in a few cities. I recall that we wanted to settle the first GP if it was a merchant.

@sossos
Sankore had a much longer viable period than Colossus, up to turn 220 and, of course, more and more religious buildings with time.
EDIT: obviously, grabbing both would have been best :D
 
Slowly working on thread summary all the way back in post #2.

I got to turn 174 on ours. Then we pretty much did 1-2 turns for rest of game. hard to say much in 1T. With 10-15 turn turnsets it was easier.
 
Well we got mids T75 so that is 15 turns ahead of Duckweeds target. Maths chops and forge really helped.

T180 what did we do? We naval attacked Hammy with Mace and trebs. We took a beating but slowed him down tech wise. He was getting close to Economics and lib. Hammy retook Babylon with a 7-8 strong stack 1-2 turns later. He was a tough nut to crack. So we came back with cannons a few turns later.

You also then need to take into account the beakers 'lost' because you bulbed a cheap tech such as maths because 45n probably doesn't offset the lost beakers. (n is the number of scientists you run). Although I guess you could at the same time argue, your cities grow larger faster, and thus snowball effect. But that depends on whether your cities grew :)
 
@sossos
Sankore had a much longer viable period than Colossus, up to turn 220 and, of course, more and more religious buildings with time.
EDIT: obviously, grabbing both would have been best :D

Sankore is like a creature with longevity but lower reproducing power: 2 beakers each city after Scientific method is on line.

TC is like a creature dying soon but with high reproducing power: 1C per pop on water. With island cities, it's easily more than 5C per city.

Smiles bring more smiles, and assets bring more assets. As I mentioned, 2 beakers of T200 is close to 1 beaker of T175 and 0.5 beaker of T150 on average. The waste of your last turn and the waste on EP points may delay our victory a little bit, but the difference is not that big if we take the compound interest into consideration.

Yes, we lust for everything but we have to make choice.
 
@shaka

If we count the return rate of "paper + sankre", it's lower. (800 beakers+ 275h for 20 beakers per turn: all normalized and 10 religious building assumed at that time)This is why we go MC after CS, not paper.

We agreed CS first was better for research. Thus we had Paper earlier no matter how many techs we squeezed between CS and Paper. That gives more time to UoS to pay for itself in any case. Qualitative assessment is just enough.

We also could have gotten MC in trade if we had some luck with AIs and also one more reason to go CS before MC. One more unmeasurable parameter (since you can't replay the same map and take samples to do statistics mumbo-jumbo). But it is obvious MC is cheaper if gotten for free and that's why we, humans, will always try that with no need of 1000000 iterations. Delaying MC had high opportunity costs plus we didn't have any more buildings to produce in main land cities.



Building wealth is just to invest on techs. The return rate depends on what the tech is and how many beakers are left to research the tech.

Building wealth to get military tech earlier can be dozens of times better than building, let's say CH. Building CH can be considered a local optimum in a way while earlier war with better units can lead to global optimum although your cities will bleed money at the start of conquering. Also, CH before wealth is better for later war and can generate more research later. Knowing exactly when to do which is what makes a good player. Maths can't give you exact answers/boundaries, especially in changing environment. That's why we lean towards short-term benefits. They are usually war oriented. And we are good at war and can actually assume control over the game then, making it predictable. PD always win via war, more or less limited.

I don't say that the calculation is all. It's just reference. However, I don't trust feeling, even it's my own. This game is eventually a game of operational research. When my model fails, I need to figure out what is the reason behind and polish it.

If you try to have a reason for everything you do in a game, you'll soon realize your faults and way to go. That's why SGOTM games are so much better. No move is just to hit end turn.

Definitely a lot of events and rating will change the result, but that's beyond our control. It is something like statistics and risk control, but it can still be calculated IMO.

Going back to talk about TC, I don't think it's a weak wonder---from my instinct with so many watertiles and 50 turns before Astro. The gain from TC and the pop is linear correlated, and the return rate of pop increase itself is usually high. However, the calculation shows that it still cannot win over the CS approach. You cannot look too far because both paths will be finished in 20 turns. What we need to decide is just to choose which path first and the marginal increase is the one we need to measure.

We actually agree on this. What I wanted to say we could have had profit from later Colossus too (after CS). I think we even had forests to chop. MoM was nice, but it is usually easy to capture.

PR's session is better for TC. They have less cities but more pop. They oracle MC if my memory is correct. However, they have no interest on islands, water and Hereditary rules. They invest too many on GPs(a kind of Tech), but the return rate of a lot of Techs are lower than island cities.

I agree. That is typical example of not playing the map and staying in your comfort zone (Pyramids). However, it could be they didn't know how many islands and seafood was there in the first place. In any case, island cities available - GLH no brainer.

The excess OF is my fault as I didn't re-create the test game well enough. Had I done so, T234 would have been 100% certain (queue Rifling earlier instead of taking 2 turns on a tech) and T233 probable as well (AP win).
There are more examples which slowed us down, ex. Kremlin was delayed 1 turn unnecessarily which cost ~2000 beakers.

Yes, kossin and jata are the only ones responsible. Prepare the stake and a torch.:D
C'mon. Mistakes are counted in. Recreating whole game, as much as minor effort as you claim it is can't be done better without food bar options and many other.


As you say, completing those renaissance techs earlier would snowball big time which is why I tried finding ways to accelerate our research (banks :lol:).

We could have built some banks earlier for the cost of unnecessary religious buildings we have built. Some of them by most humble me. It'd be a better game definitely.

Anyway, Civil Service was also very important for irrigation, our food situation was pretty bad in a few cities. I recall that we wanted to settle the first GP if it was a merchant.

Than, idea of bulbing CS arose too. Combined with post above, we really had good reasons for CS first. Glad PD recognized that.

Anyway, congratulations team!

Atmosphere was great, without pressure with great degree of freedom while still meeting the deadline.
I'd like to thank you all for opportunity to learn and I have learnt a lot. Especially around micro and whipping. So, many thanks.:D
 
@shakabrade
I really enjoyed this one as well, it feels like every SGOTM I learn a lot more than I thought possible.
Thank you everyone!

~~~
Thread summary (a bit short) updated, post #2.

Let me know if anything is wrong or if I should include anything else.
 
^Before I joined SGOTM, the only things in my memory was DanF articles and snaaty's guild.

Afterward, what I have learned most are from some xGOTMs and SGOTMs in these years.:)

BTW, this game and SGOTM12 is the best platform for competition since SSV like victory covers almost all the skills of CIV game and the effect of luck factors has the minimal effect.
 
Yes, SGOTM is a very good place to test and compare which strategy is optimal. In the meantime each player has to share his/her knowledge to help the team to win.

Very good mechanism to attract more civ fans to have fun, I should say.
 
Top Bottom