[SPOILERS] JR4 tries his luck in PB 37

[JR4]

There are lots of talking points now! Let`s start with the situation between CML (Tectomoc) and Couerva. That naming scheme of Tectomoc`s suggests a rather unfriendly attitude toward his NE neighbor. The name of Couerva`s border city doesn`t fit with the rest of his cities so it`s probably a message to Tectomoc. Ouch. Yuri has a mountain to climb to get into the game. Well, he won`t be able to as there isn`t enough land for him and he`s got too few cities, even if he can somehow find a way to settle the border with Couerva.
   

I must say that the border situation between Gavagai and Dark Savant looks really unstable. Gavagai probably has the best cards on his hands as he`s got a hill city and an option of hitting the border city on the first turn of the war. Gavagai and Krill seem to have roaded toward each other so that`s not exactly a friendly border either.

PB 37 looks very, very competitive right now. I still think Joey is the slight favorite to win but all the players with 7+ cities have a decent chance as well. In long term planning I think Krill`s going to be our best friend. He needs us to keep Joey (and dtay) from becoming a runaway and we`re not in competition for land. I think having a strong Incan Empire is good for us as well. If he starts snowballing into a big lead we`ll reconsider that, of course.

[JR4]

Graphs and demos: Someone is gaining a lot of food right now, we can`t seem to entirely close the gap. That`s not necessarily a big problem as we`re pretty close to the top dogs.

Btw, as I checked the map for info, Joey moved a phalanx to join his worker on the forest currently being chopped by his worker 3E of Seven Mysteries. Is he planning to build S of that spot or is he just anxious about us attempting an amphibious attack on his worker?

[RefSteel]

I generally agree on all your points, although I'm not inclined to read too much into the naming schemes.

Krill definitely does seem like a longish-term ally; the way he moved his starting Settler effectively makes him a third player (with Russia and China) with whom we aren't realistically in direct competition for territory. To the extent we can tie up GJ's resources in the north and he can tie them up in the south, we'll both be better off here.

I think the Phalanx is there to protect the Worker and/or prevent us from moving our own Worker to the jungle copper without an escort, but the question is, why is the WORKER there? My guess is that, yes, he plans to put a city at the end of that little peninsula. It'll be a junk city until it pops borders, but with TGLH, that's not the end of the world for him, and it helps to shelter Seven Mysteries from invasion while giving him another cultural beach-head toward (and visibility into) Nambui. This of course means that we'll want to get Buddhism into Nambui soon, though we obviously have a lot of priorities for that already; it's a good thing we've got a monument there already, before he's even founded his city!

[JR4]

Krill definitely does seem like a longish-term ally;

I think the Phalanx is there to protect the Worker and/or prevent us from moving our own Worker to the jungle copper without an escort, but the question is, why is the WORKER there?  My guess is that, yes, he plans to put a city at the end of that little peninsula.

It shouldn`t be too long now before we meet Krill. It`s really interesting how the actions of one civ leads to another civ playing aggressively against a third civ. The idea of putting a super city site directly in the direction of the NE neighbor is a very good one I have to admit. Lots of action and early tension can only be good (except for those who got settled up on hard).

Hmmm, if Joey wants to put a city down there we can`t stop him. Maybe he wants to challenge for control of the copper tile with two cities being able to put culture on that tile. Another interpretation could be that Seven Mysteries is about to expand borders and the chop comes in right on time with that.

Couerva has Alphabet now, I assume other players are researching it soon as well. As long as we have a trade route I`m inclined to offer open borders with at least two civs that is not Joey and dtay to fill up our own routes. Krill is a good option. Maybe Gavagai or Dark Savant as well?

Also, I`ve checked the game with the 2600 Grandmaster with a chess program (it`s strong enough to beat the GM!). It seems that he played a vast number of first choices, systematically enlarging his advantage. At least I didn`t blunder horribly in that game.

[RefSteel]

That forest is fourth ring for Seven Mysteries. It'll provide a few hammers for State of Zijin if he chops it even from outside his borders, but I doubt if that's worth the worker turns and repositioning the Phallanx - especially as he has so little military in general.

Couerva has Alphabet now, I assume other players are researching it soon as well. As long as we have a trade route I`m inclined to offer open borders with at least to civs that is not Joey and dtay to fill up our own routes. Krill is a good option. Maybe Gavagai or Dark Savant as well?

Our domestic routes to our island cities are worth as much as international routes to cities on the same landmass. So only our island cities would benefit from international trade routes unless our trade partners have island cities of their own. (I think in BtS, we would still have to fill up our trade routes to get any benefit even in the island cities, and this might still be true in this mod; I'm not sure. I think the issue is/was that the game assigns trade routes in city number order - so in our case, it's Borte first, then Muqa, etc - and after it runs out of foreign island city routes, it starts "using up" mainland foreign trade routes too, even though Borte and Muqa would get the same yield from domestic routes to our islands. Then by the time the game gets to Kokejin, all our foreign routes to the mainland might be used up, so it just trades with Borte instead. I don't know if the mod fixes this so that domestic routes are favored over foreign routes of equal yield though.)

Also, I`ve checked the game with the 2600 Grandmaster with a chess program (it`s strong enough to beat the GM!). It seems that he played a vast number of first choices, systematically enlarging his advantage. At least I didn`t blunder horribly in that game.

Neat! My understanding (from afar) is that there are chess computers that even Magnus Carlsen wouldn't beat - a testament not only to modern computer power, but to what's possible when using well-programmed computers to unite the expertise of a large number of people "in one place." At the same time, a human (like the GM you faced) seeing and understanding so many of the "first choices" a powerful computer can calculate is really impressive. The fact that such a player had to steadily overcome you with consistently slightly-better play rather than just steamrolling you speaks well of your ability too. I played a little chess a long time ago, never competitively; I'm pretty sure if you saw the games I played, you'd recognize many of my moves as blunders that I just didn't see.

[JR4]

Trade route mechanics are hard to get right, that`s for sure. I read somewhere that the size of the cities matter as well, though I can`t remember how big they have to be in order to receive extra commerce. If we can get a trade connection with e.g. Couerva, who`s built at least one island city we`d still get some extra commerce, even if it`s in Borte, right?

The fact that the strongest chess engines easily beat the best human in a match comes down to the brute force of calculation. A game such as chess where all information is available at all times is especially well suited for the computer. In a turn based game with lots of hidden information (like civ) it`ll never beat the best human player.

[RefSteel]

Trade route mechanics are hard to get right, that`s for sure. I read somewhere that the size of the cities matter as well, though I can`t remember how big they have to be in order to receive extra commerce. If we can get a trade connection with e.g. Couerva, who`s built at least one island city we`d still get some extra commerce, even if it`s in Borte, right?

That is correct as far as I know - although I don't know exactly how big a city has to be. Trade route mechanics are fairly arcane.

The fact that the strongest chess engines easily beat the best human in a match comes down to   the brute force of calculation. A game such as chess where all information is available at all times is especially well suited for the computer. In a turn based game with lots of hidden information (like civ) it`ll never beat the best human player.

Don't underestimate the cleverness of computer programmers (and the power of Moore's law). Chess is complex enough that it isn't a "solved problem" even for the most powerful computers, but computers have been taught to "understand" the chess board well enough to bring their "brute force" calculations to bear to unbeatable effect. An even more extreme case is Go, where the number of possible moves on each turn is so large that "brute force" approaches on their own are useless - but this past May, a Go computer just went 3-0 in a three-game set against the Go player who at the time was rated #1 in the world, because the best AIs no longer rely on "brute force" calculation alone. If civilization doesn't collapse first, I don't think there's anything at which a computer will never be able to "beat" a human being, because it's possible to combine so much human expertise - along with extraordinary processing power - into the program of a single machine.

[JR4]

Yes, I remember reading about the computer winning against the go champion. It was seen as a big surprise. I don`t know much about the rules of that game, unfortunately. I have to admit that it`s a very impressive thing to do.

I still think that the AI winning over a determined team of mackoti and SevenSpirits in civ 4 is highly unlikely even in the future (maybe I shouldn`t have used the word never). How are you going to teach the computer to defend against complex naval attacks and forking? Maybe the AI can stay reasonably close in the early stages but it still needs to place it`s cities perfectly, as well as microing everything, in addition to making the right tech choices at all stages of the game.

[shallow_thought]

Yes, I remember reading about the computer winning against the go champion. It was seen as a big surprise. I don`t know much about  the rules of that game, unfortunately. I have to admit that it`s a very impressive thing to do.

I still think that the AI winning over a determined team of mackoti and SevenSpirits in civ 4 is highly unlikely even in the future (maybe I shouldn`t have used the word never). How are you going to teach the computer to defend against complex naval attacks and forking? Maybe the AI can stay reasonably close in the early stages but it still needs to place it`s cities perfectly, as well as microing everything, in addition to  making the right tech choices at all stages of the game.

I think the mechanics that you mention are within the bounds of possibility for the current range of computers - think about how little effort has been put into improving Civ AI compared to chess or go (and no-one AFAIK has tried to run a Civ AI on separate, high-performance hardware). Micro, dot maps - these are just in-game facts that could plausibly be understood. The problem is "grand strategy"* and - even more arcane - the psychology of the other players. I can easily believe that we're decades or more away from computers being able to cope with that.

*I've repeatedly failed to draft a definition of this I'm happy with while putting this email together  - may try again later.

I've not played chess at as high a level as JR4, but when I was following it closely I was struck by how Kasparov dominated other players not only by his skill but also by his physical presence (I watched an afternoon of him playing Short at the Savoy) and emotional readiness to take risks, to enter tactical whirlwinds and assume that he would come out ahead more often than not even when he acknowledged he couldn't be certain. I find the latter highly reminiscent of top tier play here at RB. Of course, he didn't have this edge against a computer. While the day of the machine was always coming, I remember the commentary saying that - good as he was - there were other players who were perhaps more likely to beat Deep Blue at the time, as they were less reliant on tactical brilliance and forcing their opponents off-balance. Computers don't yet know fear, or fatigue.

[JR4]

I think the mechanics that you mention are within the bounds of possibility for the current range of computers - think about how little effort has been put into improving Civ AI compared to chess or go.


I've not played chess at as high a level as JR4, but when I was following it closely I was struck by how Kasparov dominated other players not only by his skill but also by his physical presence. Computers don't yet know fear, or fatigue.

1) I might be underestimating how well the AI can be taught to play. My point is that it`s far easier to make it play chess than a game where a lot of the information is hidden. We`re not likely to get the definite answer to this, unfortunately.

2) Oh, that`s a fair point. Psychology is a big part of the picture, not only the mechanics. Poor Nigel Short has a remarkably bad score against Kasparov. He`s actually described the ex champion as "The monster with a thousand eyes", an expression that speaks for itself. In some ways, having a reputation for being a top player will help you, in civ like in chess. Kasparov, of course, knew how to take advantage of this to the greatest extent.