SGOTM 12 - One Short Straw

[babybluepants]

We definitely want peace as soon as possible. Not only to avoid having to defend, but also because it's precondition to OB > tech trading > etc. We want to start accumulating the cheap plusmods (peace, OB, resources) quickly.

Oracle CS in the 1000 BC range won't be beaten by much. It's slow for regular Emp, but seems like a good move on this test map. We obviously can't really say whether it would be in-game. I don't think we need to decide yet, though, since we should know quite a bit more about the AI by the time we get through base techs and Writing.

I had a run last night, where I settled: 1NW > 1N of Rice > clams/rice/sheep/stone > 1N of iron > horse/clams, beelined Alphabet, then GLH in the second city. I managed to get GLH (600's BC), open trade routes with 3 AI by 500 BC, and get to 4 cities + 6 workers by then, with another settler and worker finishing in a couple.

I like this settling pattern (1NW + 1N of Rice) because:

• Guarantees two decent cities to start. This point is moot if Erkon was generous with land to the west, ofc. Still, it's nice to have a good shot at GLH without committing capital resources to it.
• The second city can share tiles. I worked a lot of turns of gold mine in it, while the capital was growing / regrowing.
• It just feels slightly faster in the 2500-1000 BC era than SIP. Not sure why, given that SIP has more food. I guess it has to do with the second city contributing faster, and probably some less worker-turn wastage.

Long-term this capital sucks, though. It doesn't have a large enough food surplus, unfortunately, which may be enough to disqualify it.

Like LC suggested, I'll have to stop chasing wonders and test just REX-ing from different cap locations...

 

[Mitchum]

Forgot to gift Gandhi Alpha, and he demanded it 5t later for a semi-permanent +2. The +1 fair trade will last for twice as long as +2 (even longer if they meet further AIs in the meantime), so waiting for a demand is an option.

I was able to buy peace with Wtg from 4 AI and Fish from the two others, when getting Alpha in 1800's BC.

This is good info to have. However, is there an advantage to staying at war with at least some of the AI? Will they build more units and fewer wonders, for example?

 

[Mitchum]

I check Masonry and notice that no one has researched it yet. From now on I can monitor it to see if anyone has marble or stone.

How does this work? Do you check how many turn it will take to research Masonry and then watch to see if the number of turns goes down due to us knowing an AI that has already learned it?

In order for this to work, you would have to be sure that the number of beakers being generated doesn't change during this period, which could be tough as cities grow, libraries get built, tiles get improved, etc.

SIP

CS-slingshot 1040bc. Researched Math+CoL by hand. No GS.
2 cities, second city is building a lh for GLH.

I have to say that this opening looks very strong: CS sling pre 1000 BC, two cities, GS in 8 turns, etc. Two bad things are that it wastes the clam tile (we can still net it for health reasons though) and St. Petes isn't coastal to benefit from the GLH. On a positive note, this type of capital could become a research monster and would benefit from an early academy. Also, we should be able to piggyback off of every AI when researching BW as they should all have it by now.

I think a useful comparison for testing would be just how much can we REX if we skip the CS-slingshot. With this terrain, I'm not sure how much a difference it makes at this point.

In general, I think an early CS sling without completely sacrificing REX is better than the all-out REX option. In my test, I was still at one city, which was a major drawback to that opening. As you said, with so much fallout, early REX will have to slow somewhat unless we want to found a bunch of useless cities...

 

[babybluepants]

How does this work? Do you check how many turn it will take to research Masonry and then watch to see if the number of turns goes down due to us knowing an AI that has already learned it?

Yeah, I'm wondering about this, too. That discount is the same as the Pre-req modifier, in that it gets applied to beakers rather then tech cost, right? Can you somehow tell the exact amount of beakers being applied every turn, other then estimating from number of turns remaining?

 

[mystyfly]

@mysty: I didn't mean to upload the example form the last post. Just giving another example.

Nah, I was under the impression you wanted to post a link to a strategy article about

Monitoring AIs early tech research and civ growth

@bbp: I don't like settling NW, prefer 2N2W, all it loses is crappy silk and we don't lose another turn, plus we're coastal and might even get more resources in the fog. It's a gamble though. I like to keep rice, cows and clams for a second city, something we couldn't do when settling 2NE.

 

[LowtherCastle]

In the test I linked above, the second city also shared the river gold much of the time. Problem is, of course, we don't know that site will be any good otherwise (pigs...). Furthermore, SIP is a huge risk in terms of disabling all other sites on this peninsula--might be more than just the clams, although it would be pretty unlikely, I think, that Erkon would make SIP so underpowered. Either there aren't other especially good sites or SIP gets another resource, is my guess.

But I agree with Mitchum about SIP not benefitting from GLH. Settling on the plains is better long term, imo, because it could have Moai to good benefit. But all of these capital sites are fairly lame for a space race game, which is why I think it's better to think short- to mid-term for our capital. Long-term we just move it, if we find a really nice site soon enough. Or we could even build FP there and move our capital far away later on. This all makes us less dependent on Bureau if we don't like the maintenance later on.

@bcool: Your rationale is solid but I'm still lukewarm on Feudalism. It just doesn't do much for our research or REX. +50% for our workers is good for fallout: 6t/1.5=4t, but lame for mines/pastures: 4t/1.5=3t, but overall it's a plus, I'll grant you that. Problem is, though, I think we want to run either slavery or castes most of the game. We can put 10h into a settler and then 2pop it, for example. I'd rather build Hagia Sophia, though I've never built it in my life , even if it dies with Steam Power. In my CS-sling test, Stalin has an axeman SoD coming at me, but it's too far away to be of concern. That may be misleading because he might be much closer in the real game.

I think this map will resemble the SG10 in two regards: 1) All AIs accessible by galley, and 2) Gandhi and Stalin on one landmass, the rest on another. We'll be on one of the two landmasses, my guess on Gandhi's. Furthermore, my guess is that Gandhi will lie between us and Stalin, meaning that they would soon be WEs, after we DoP with them. PUre speculation, of course, although I do know a bit about how Erkon thinks. We'll see how much that little bit is.

@Mitchum: The F9 Demographics screen tells you exactly how much of a bonus you're getting for any tech you can currently research from other AIs knowing it. What you do is unclick the tech you're researching, and check the F9 GNP, then click the tech you're interested in and check the GNP again. The difference tell you how many beakers the AI % is giving you. After figuring that out, then you calculate how many AIs know it, based on the formula for the tech percentage. You can test this by settling in place using the 4000bc test save and then checking the techs we knw the AIs have. I figured this out (just today) studying the Demographics article, oneof the most powerful articles that exists in the War Academy, imo.

Edit: Note that this is an idiosyncracy of this particular scenario, because we know all AIs from the beginning. Otherwise, we'd only be able to figure out how many of the people we know had any give tech at any given time. This game has a unique beginning in that regard. That's why I decided to re-study the Demographics article. Try to figure what all we can know on T0.

xpost with bbp

 

[babybluepants]

The F9 Demographics screen tells you exactly how much of a bonus you're getting for any tech

Ofc, it does.

Note that this is an idiosyncracy of this particular scenario, because we know all AIs from the beginning. Otherwise, we'd only be able to figure out how many of the people we know had any give tech at any given time. This game has a unique beginning in that regard. That's why I decided to re-study the Demographics article. Try to figure what all we can know on T0.

Exactly. We can definitely learn a lot. I think we should note the scoreboard and screenshot F9 every turn.

Settling on the plains is better long term, imo, because it could have Moai to good benefit. But all of these capital sites are fairly lame for a space race game, which is why I think it's better to think short- to mid-term for our capital. Long-term we just move it, if we find a really nice site soon enough. Or we could even build FP there and move our capital far away later on. This all makes us less dependent on Bureau if we don't like the maintenance later on.

That was roughly my thinking in trying out the other two sites, despite the fact that NNE seems strongest long-term. The initial slowdown from clearing fallout multiplies this further. SIP takes away other potential sites. I'm aware of the fact that the pig in the fish city makes it significantly stronger in the test save. We should remove it for further testing maybe.

I think this map will resemble the SG10 in two regards: 1) All AIs accessible by galley, and 2) Gandhi and Stalin on one landmass, the rest on another. We'll be on one of the two landmasses, my guess on Gandhi's. Furthermore, my guess is that Gandhi will lie between us and Stalin, meaning that they would soon be WEs, after we DoP with them.

That sounds like a reasonable speculation to me. Why I put them on our landmass in the test save. Gandhi's location (i.e. how easy it is to get trade routes open with him), is critical to the early game, btw. GLH isn't super-hot, if we can't get any international TRs until the AD's, for instance.

Having an AI much closer than the test save suggests may really throw the opening off, if it means they're much more likely to attack early. Perhaps we should devise a test save for that? Esp. if we decide to go worker-first, and don't get immediate exploration.

@mysty,

I agree that 1NW isn't such a good city. It's strength relative to SIP is largely dependent on what Erkon gave us immediately to the west.

What do you mean 2NW? We can't settle there because of fallout.

 

[Mitchum]

@Mitchum: The F9 Demographics screen tells you exactly how much of a bonus you're getting for any tech you can currently research from other AIs knowing it. What you do is unclick the tech you're researching, and check the F9 GNP, then click the tech you're interested in and check the GNP again. The difference tell you how many beakers the AI % is giving you. After figuring that out, then you calculate how many AIs know it, based on the formula for the tech percentage. You can test this by settling in place using the 4000bc test save and then checking the techs we knw the AIs have. I figured this out (just today) studying the Demographics article, oneof the most powerful articles that exists in the War Academy, imo.

Cool. I used to play a lot of PBEM games and I used the info on the F9 screens all the time based on the linked article -> not much else to do when you play one turn every day or two and games last a year. I could tell exactly what tiles my opponents were working, which tiles they improved and when, what they built in their cities, how many units they had, etc all from the the F9 screen and watching their score. However, I never realized that your GNP actually changed depending on what you were researching. I still have to explore/test this a bit more to make sure I can leverage this trick during a game.

What we can do is watch the score board. Every time an AI's score goes up, we can use this trick to figure out whether they grew a pop or researched a tech (and which one). We'll need a detailed list of what to check and record every turn to make this possible.

 

[LowtherCastle]

If SIP, do we beeline alpha? My testing so far shows this to be kind of pointless, because we might be unable to take advantage of it for tech trades and the DoP doesn't get us trade routes either, because we sacrificed exploration.

Beelining alphabet just to make peace even if it doesn't give us any significant trades might be a necessary play especially if we don't uncover any horses/metal nearby.

I need to clarify my above point about beelining Alpha being pointless.

Beelining Alpha is anything but pointless, no matter what. It gives us DoP. It opens up Currency for a slingshot. It gives us early spies. It could give us tech trades. What I meant by pointless was simply one thing: In my test beeline, at SIP, I didn't send out any explorers, thus leaving me without any trade routes (no tiles defogged).

By the way, bcool, there is one very powerful reason to do your Feudalism slingshot, that I just thought of (I think you and others have mentioned it): Then we don't have to DoP all AIs. That's better for our negmods in the long run, especially if we want to try to trade with someone later or whatever.

 

[babybluepants]

RE: not taking peace with all AI

What is WW a function of? Is it tech era dependent, like it appears?

There's no reason to DoP with someone like DeGaulle, if he's on another land mass. We won't get any trading from it anyway. Gandhi, whoever shares religion with Gandhi, anyone within land-based striking distance - those are important, I think.

RE: Feudalism

Aside from the massive initial slowdown due to fallout and AggAI on steroids, this doesn't strike me as so different from a typical fast space game. That suggests a general CS > Edu beeline, followed by some medieval or renaissance warfare, followed by MoM-GA and either Comm or Corp.

Feudalism sling doesn't sound good to me. It accomplishes nothing towards those goals (no research or expansion boost), and we're not gonna be teching very fast here in the BC's. Serfdom is sucky because it denies Caste/Slavery. It would be good to have in the first 50t or at the very end maybe. I'd hope we don't build any longbows at all. Vassal states are turned off.

 

[bcool]

The only reason I went for feudalism in my game was because I was unable to get civil service with the oracle. I agree it is much weaker than a civil service slingshot.

I'm very impressed by how quickly you were able to get the civil service slingshot. And agree that if we can do it before 1000 BC it is a probably the best play. (depending on what we glean from the AI's speed in the real game).

And I'm impressed by the analysis you guys figured out with the scoreboard and demo screen.

When I said vassalage was enabled by it, I meant the civic. A timed dip into vassalage might make sense if we had an early war. But yes it is a rarely used civic as well.

And I agree serfdom isn't worth giving up the other civics, but with the feudalism route I didn't have caste system so it seemed to help a bit until I did research code of laws. I did use serfdom near the end to speed up the cleanup of the last fallout in my full length test game.
(This type of switching near the end of the game might actually make Cristo Redentor worth its cost as well despite how late it appears)

 

[mdy]

Checking in.

I favour settling 1NE 1N for reasons that have already been mentioned.

Normally I would think that a CS sling would clearly be the best strategy in these circumstances, but with the warning that this game will play somewhat harder than the normal emperor game I am sceptical whether even 1040B.C. is good enough for the Oracle.

 

[LowtherCastle]

Hey mdy! Welcome home!

Did you mean NE N?

 

[mdy]

Did you mean NE N?

Yes, that was a typo.

 

[ZPV]

What is WW a function of? Is it tech era dependent, like it appears?

It's battles on ground that isn't culturally ours. I'll check whether tech era comes into play. We won't have a problem with it until we go on the warpath.

although I do know a bit about how Erkon thinks

Kill'em. Kill'em all!
Kill'em. Kill'em all!
Kill'em. Kill'em all!

 

[babybluepants]

Hey, mdy!

W and E have to be right next to each other on the keyboard, ofc.

@ZPV
You sure? That doesn't necessarily correspond with my experience...

 

[shyuhe]

So, the big points of discussion so far sound like they are (in my perceived order of importance):

1) Where to settle our capital
2) How soon do we go for alphabet
3) What (if any) are we going to seek from Oracle.

Obviously there's a lot of interplay between the three since it affects our initial build orders etc. But I think the general permutations are some combination of SIP, settling NW, NE, and NEN? Do we want to divide up the testing and have people test one permutation a few times? I can do some testing this weekend.

 

[Mitchum]

I've been playing around with the demo screen and tech rates and I can tell you that it will be very hard to predict how many AI have a given tech using this information unless I'm missing something. The problem occurs with all of the FLOOR and ROUNDDOWN functions in the beaker formulas, which make it impossible to back into the exact # of civs that know a tech. We're more subject to these rounding errors when our beaker output is low, like what occurs early in the game. Even with both golds mined and 2 towns (thank you WB), there will still be a large enough error to make the calculations difficult.

For example, all 6 AI know Archery. However, if fewer than 6 knew Archery, here is how many total beakers we would get on T1 if we are working a 1C tile (total 10 base beakers):

6 AI know Archery - 15 total beakers
5 AI know Archery - 15 total beakers
4 AI know Archery - 14 total beakers
3 AI know Archery - 14 total beakers
2 AI know Archery - 13 total beakers
1 AI knows Archery - 13 total beakers
0 AI know Archery - 13 total beakers​

Note that some of the bonus beakers are coming from the 20% prerequisite bonus.

However, if I start with 14 total beakers and use the formulas to calculate the Known Civs w/Technology, I get 1.4, not a number close to 3 or 4. As I said above, this is due to all of the rounding and flooring.

So, the only way to really use this information is to watch for trends. Watch the scoreboard. When an AI's score goes up, check each tech to see if the GNP goes up from where it was the turn before. This requires you to know what the GNP would have been for each tech the turn before though. You'd also have to make sure that your total base beakers didn't change. This all seems like a lot of work.

Also, let's say no AI knows Archery and one AI just learned it. Our GNP would not go up based on the table above. However, if you see a corresponding jump in Power on the Demo screen (Archery = 6K Soldiers), actually, the jump would be in Average Power, then we know that an AI just learned Archery. Techs like Priesthood that don't affect power are more difficult to figure out.

Does anyone else have any ideas for a practical way to use this knowledge?

 

[babybluepants]

So, the only way to really use this information is to watch for trends. Watch the scoreboard. When an AI's score goes up, check each tech to see if the GNP goes up from where it was the turn before. This requires you to know what the GNP would have been for each tech the turn before though. You'd also have to make sure that your total base beakers didn't change. This all seems like a lot of work.

Also, let's say no AI knows Archery and one AI just learned it. Our GNP would not go up based on the table above. However, if you see a corresponding jump in Power on the Demo screen (Archery = 6K Soldiers), actually, the jump would be in Average Power, then we know that an AI just learned Archery. Techs like Priesthood that don't affect power are more difficult to figure out.

Yes, that's basically it. This has a much crazier application in Pitboss MP games, if you're willing to log in after each opponent has finished each turn. Otherwise, it's imperfect, ofc. For starters, I'm mainly interested in studying Yield/MFG/Health/etc to determine whether they likely have the same fallout-ridden starts we do, and Score/Soldiers to guesstimate advanced techs.

We need a useful EP target, as well. Stalin seems like the obvious choice.

 

[Mitchum]

For starters, I'm mainly interested in studying Yield/MFG/Health/etc to determine whether they likely have the same fallout-ridden starts we do, and Score/Soldiers to guesstimate advanced techs.

... and the land totals to predict how many AI have coastal capitals which could play into our GLH aspirations.

We need a useful EP target, as well. Stalin seems like the obvious choice.

I was thinking about this too. Espionage will be a losing battle for us initially because all AI will be directing all of their points toward us since we're the only other civ they know.

Starting with Stalin seems logical. We may want to shift to one of the AI(s) on our continent once we know who they are. Or toward our intended tech trading partner so that we can avoid researching the same things.