[SPOILERS] WillPlunder as Mehmed of Arabia: Noobs meet world

[Old Harry]

In worldbuilder you can save the file, then when you load it offers a choice of which civ to be. You lose a lot of data (hammers invested, food box) but I think it keeps the info about injuries to units correct...

[WilliamLP]

Kuro rejected peace, traded a couple more units to kill a couple of camels, and then offered peace again. I rejected it again to see if there were any tactics first, but simply re-offered.




With two horrific strategic misplays, it's amazing how close we are to breaking through that stack, even still. Even looking at it right now I wonder if we shouldn't just whip and cash upgrade some more units and run right in again... But in 10 turns we can hit this really decisively once and for all I think.




Oi.




We have enough EP to have visibility of a couple of Bandit's cities. I don't know if he knows we know about that knight stack there, but it's not a scary one. I think it's just a standing defensive force. Bandit has way too much to worry about in the west and the sea to be thinking about charging into pikes and castles in this era. He's basically equal to us in tech, and catching up in land.




Thanks to Plunder for this idea. I think those forts will work as a canal when we control the north one with culture after popping 40% borders at C. Vanderbilt (or Bolognese). And it's worth noting that it's the 1 tile lake that makes it work.

For tech, I've decided to do what everyone else is doing: run my best guess at what a Specialist Economy is in our golden age. This will be in 4 turns unless we get an 8% prophet, which would suck. So, forget Economics for a while since we'll be running Merc / Caste / Pacifism. Production cities will produce, hence Chem for 6 hammer grass workshops during the GA.

The aim will be to get 3 great people for yet another GA... and then bulb with surplus scientists and maybe a merchant or two.

Working this out, it's 9GPP per specialist. A specialist will then produce 108GPP in the 12 turn golden age. Our GP counter will start at 700 and one other is nearly done so producing GP will take 800-1100 or so. With the free one from Merc, that means running 7-9 specialists per city over the 12 turns, which seems totally doable. (And it won't take any extreme starvation scenarios).

[TheHumanHydra]

Open the sim in Hotseat and you can be both teams.

[Qgqqqqq]

Or open the sim normally in single player, enable debug mode (in the menu toolbar click game details) and press alt-Z to switch players.

[Gavagai]

I know that Civ 4 is a much easier game than chess (or Go!)

Sorry for a kind of threadjack but this statement is far from obvious. In chess we can build an AI able to beat a world champion. Can you imagine a civ AI able to beat, for example, Krill, without ridiculous bonuses?
(Seriously, this question always intrigued me: what exactly makes it so difficult to make a good AI for civ in contrast with a good AI for chess? Fog of war? More variety? Or just much less thought invested?)

[NobleHelium]

Uh, a lot more time has been spent building an AI for chess than an AI for Civ4. And equality of knowledge about chess is much, much higher than Civ4.

[Plunder]

I think the reason why you can build an AI for chess easier than for CIV IV is that chess has a "smaller tree". Per turn there are less choices to be made than in Civ IV. This is the same reason where there is still no AI that can beat even the weakest of professional go players. The branching of choices is just too high.

Probably it is possible to have an AI play the perfect opening because there relatively few choices to be made, but then the number of choices increases too fast.

At least that's my gut feeling =)

[WilliamLP]

Sorry for a kind of threadjack but this statement is far from obvious. In chess we can build an AI able to beat a world champion. Can you imagine a civ AI able to beat, for example, Krill, without ridiculous bonuses?
(Seriously, this question always intrigued me: what exactly makes it so difficult to make a good AI for civ in contrast with a good AI for chess? Fog of war? More variety? Or just much less thought invested?)

Yeah, I agree it's far from obvious.

I feel good about saying that it would be easier to learn to be competitive and win some games against the Civ IV elite players, vs training to win a single chess game against Magnus Carlsen or Kasparov. At least, there is a lot of variance in Civ because of the map, combat results, extremely incomplete information, and so on.

I find it quite striking how much computing power and research was actually needed to beat the best humans at chess, actually. Since it appears on the surface to be a small game where computational accuracy is the critical path, and we don't find it remarkable that we can't compete with calculators in long division.

Also, anyone who wants to jack this thread is welcome to!

[Plunder]

well, it's not that hard to imagine I think.
Let's say for a second that every move in chess has 10 possibilities, and that a game takes 10 moves. (that's a massive simplification).
If you would just brute-force it, you'd have to calculate 10^10 different outcomes. That are too many possibilities to calculate.

A game of go takes maybe about 150 to 200 moves, each move has hundreds of different possibilities.

So you would need a way of only selecting the reasonable moves to reduce the "branching", and a way of actually evaluating the intermediate result (so you don't need to calculate until the end of the game).

[WilliamLP]

A really quiet turn for us, so no picture report. Tje next Great Person and Chemistry finish in 3, Oxford finishes in 2. Kuro took peace and we'll probably make another run at him in 10 turns. This time a stack supported by muskets should have more success, and he can't build all that much more in ten turns, right?

Sutt very much got the worse end of the deal with Boldy, according to the power graph.