Arathorn's partial report/rant

[Arathorn]

almost everything else in the game moves individual units

Except city captures, which involve thousands and thousands, or pillages, which can easily add up to thousands in a relatively short number of turns, or even founding a city which will be worth thousands, or anything else. Plus, the trader is giving away thousands and thousands of units, too. Or does that not matter?

You can have unlimited tech trading

Thanks for arguing a straw man. Next time, will you please read what I wrote?

Why is the mechanism picked to limit tech trading so counterintuitive?

If tech trading is good strategy, why are you arguing the AI should stop doing it? Arguing for a dumber AI is very odd, indeed.

Perhaps you can do a similar analysis for opening moves that you've done for combat next?

No way. WAY too many longterm repercussions of various moves that are too subtle or unpredictable. For example, forming an early religion and concentrating on getting the shrine. A few gpt ain't worth it. BUT, if you can assuage larger neighbors with your religion cheaply/easily (see Sullla's Adv2 game), is that worth it?

What's the value of founding a religion vs. getting a worker out earlier? WAY too many variables. And that's just one possibility. What about starting techs or popping huts or (early) contact or .... I'd like to have a reasonably comprehensive view, but there are just too many "What if"s to complete anything, really. I'm hoping enough games will give me (or others) a good feel, but.... I just don't think it's possible.

Arathorn

[Majromax]

Except city captures, which involve thousands and thousands, or pillages, which can easily add up to thousands in a relatively short number of turns, or even founding a city which will be worth thousands, or anything else. Plus, the trader is giving away thousands and thousands of units, too. Or does that not matter?

The tech trader isn't "giving away" thousands and thousands of units -- he still has them afterwards. Sharing, is more like it.

Tech trading is very much a positive-sum game. Capturing a city is negative sum (in that improvements, culture are lost), because the other civ no longer has the city. Additionally, the gains of a city capture, while certainly large for some cities, are spread out over the remainder of the game.

City founding is certainly not a "many thousands of units" decision. Firstly, simply founding the city represents (often) a net loss, because maintenance costs go up. Secondly, the city will reach its full potential only after a long time and the investment of other "many thousands of units" in production, food, and possibly cash if anything's rushbought.

Indeed, tech trading is probably the only aspect of the game that is an absolute win-win for both parties (resource trading is partially a win-win, but there are limits to numbers of resources, and hence the gains). Two otherwise equal civs will each research (effectively) twice as fast if they trade with each other than if they don't.

This simple fact is what made tech trading such a fundamental, indeed required, part of the game in Civ3 -- the inability of the human player to keep up once out of the tech trading loop alone is testament to that. Furthermore, because of a better ability to analyze strategic concerns (combined with more erratic diplomacy in general, up to and including delcaring on "best buddies" anytime the strategic situation demands it), an experienced human player is better at the tech trading game than the simple calculus of X beakers for Y beakers allows the AI to be.

In fact, the fairly common case of "it's okay if I trade Genghis this tech, because he doesn't have the resources to build Y" is an example of a "special case" that an experienced human will intuitively see, but is hard to cover (especially completely) with AI programming.

Thus, the obvious decision for Civ4 was to consider tech trading a slight negative for the AI, and thus to impose an AI limit.

I think we can agree that the need for such a limit is relatively uncontroversial.

Why is the mechanism picked to limit tech trading so counterintuitive?

Probably because it was the least obtrusive limit that the designers tried that actually worked. Your alternative suggestions do more than just limit the number of trades; most notably, your "time since last trade" method (as Sirian points out) limits when the human player can trade tech.

I would argue that perhaps it would be better to escalate the margin required after N trades, rather than impose a hard limit, but the net result is the exact same thing. The player would then just see himself unable to afford to trade (or worse -- white techs that couldn't be purchased at any price) without the "error" message of "too advanced."

It's a trade-off between gradualism and information, and I can certainly see why the decision went towards the error message.

Another mechanism that would completely work would be to have tech trades be a negative modifier on relations with AI, but that's insane for obvious reasons.

If tech trading is good strategy, why are you arguing the AI should stop doing it? Arguing for a dumber AI is very odd, indeed.

You're better at arguing than this one.

The AI shouldn't tech trade limitlessly, because in that environment the human players do it better than the AI. Hence, it's not in the AI's interests to trade limitlessly with the human.

It's basic game theory -- the human will, on average, gain more (in a strategic sense, which the AI is intrinsically bad at quantifying) than the AI. Since the human is in competition with the AI, the increased gains for the human is a bad thing from the AI's point of view.

Overall, this is a pretty good argument, and it's one that I think the civ4 community needs to have in order to better understand the civ4 design. That's not to say that there's not some design going on, but I don't think this is one of them (to first order).

[Bezhukov]

"BUT, if you can assuage larger neighbors with your religion cheaply/easily (see Sullla's Adv2 game), is that worth it?"

Doing this is not mutually exclusive with building a worker first, as Sulla himself will tell you. I'm saying I've now read two of your CivIV reports where you were wondering why you were so far behind. In one game, you neglected to develop your seafood resource at the capital for literally 2500 years (see: http://forums.civfanatics.com/showpost.p...stcount=19). In the other, you had three camps and didn't build worker first.

Establishing a religion is a separate, and I agree, much more complex issue than whether building a worker first beats the alternatives. There are some conceivable situations where one might not be able to do so, Sirian's religious denial approach might necessitate ignoring worker techs for too long, for instance, but I'm pretty sure that in the general case it is strongly optimal.

[Sirian]

If tech trading is good strategy, why are you arguing the AI should stop doing it?

Something that ruins the game is not "good strategy".

[Sirian]

Establishing a religion is a separate, and I agree, much more complex issue than whether building a worker first beats the alternatives.

Scouts First is a commanding winner in some situations. You have to start with Hunting, though, so it's not on the table for everybody. Work Boat first is a possibility. Settler first is a possibility, too. As is Stonehenge First.

I am very fond of Worker First. I'm sure that some other openings are valid, too, though. The fact that you cannot afford to have too many cities too soon means that you can shuffle the order in which you do certain things. Racing to four cities faster means slowing your research sooner. It also means breaking out the other side of the "investment period" sooner, but for a UU like the Quechua, it may be Barracks First and a bunch of veteran Q-guys, on the Q-T, if you're going to hit somebody. (Counting on taking their workers.)

I love the fact that these things are arguable. Maybe time will show that some paths are a little bit better than others, but map differences will tilt things far enough that game after game won't always turn out the same optimal path. Surely a Galley before the first settler is a valid choice on some maps! The list goes on.


- Sirian

[Bezhukov]

"Work Boat first is a possibility"

I actually prefer work boat first when it is available (a rare situation, since you need both fishing and a water resource), as it lets you grow arithmetically and geometrically both. Usually its one or the other, and I'm just saying that geometric nearly always beats arithmetic.

I agree about not rushing to get 4 cities down before you can develop the resources in the cities you already have. But getting those resources going is so valuable its hard to argue with taking all available measures to do so.

[Sirian]

"Work Boat first is a possibility"

I actually prefer work boat first when it is available (a rare situation, since you need both fishing and a water resource), as it lets you grow arithmetically and geometrically both. Usually its one or the other, and I'm just saying that geometric nearly always beats arithmetic.

I agree about not rushing to get 4 cities down before you can develop the resources in the cities you already have. But getting those resources going is so valuable its hard to argue with taking all available measures to do so.

You do need to start with Fishing, but you don't have to have a water resource. You can use the boat for scouting, with the expectation that your second city will have a water resource and your work boat will be put to immediate use once that is settled. Valid on some maps.

You can even try to circumnavigate on some maps with a work boat. (The AI is doing it! See RB7 SG )


- Sirian

[Bezhukov]

"You can even try to circumnavigate on some maps with a work boat."

You mean there's another way? :aar:

My scouting workboats get so excited exploring that sometimes they lose their taste for fishing!

I had one end up an offshore platform.