4000 BC Aha!

[Posidonius]

Just a note, a thing i wondered about for a while, and the insight came to me in a flash when i was thinking about something else, isn't that always the way?

Why are there no respawns in 4,000 BC? When you kill off a rival civ's original Settler in 4,000 BC, it will not respawn a new rival of the same color immediately, you have to wait until the turn rolls over to 3,980 BC for the respawn to appear. Now, think i know why. In a regular turn, the respawn algorithm runs a maximum of 200 times to find a suitable square, and that's why sometimes a respawn does not happen. When the game starts in 4,000 BC, the same civ-placement algo runs, but it MUST run until it finds a suitable spot. If you choose 4 rivals in the prelims, then the AI must place all 5 larval civs as new Settlers, no 200-attempt limit.

Have seen the AI, later in the game, place two new respawns right next to each other, so we know that after 4,000 BC the respawn algo doesn't consider units when determining the suitability of a square, only the presence/absence of cities nearby. So the 4,000 BC version of that algorithm must see a larva Settler already placed as a different thing, as if it were ALREADY a city, not just a Settler. Otherwise, we would find some new civs starting right next to each other in 4,000 BC, but we never see that.

In 4,000 BC, Settlers exist as a dual-nature entity, both a unit and a city. Kill the unit, but it's city-ness is not destroyed, because the AI needs all Settlers to have that attribute in 4,000 BC, for proper world creation. Thus if a rival is not fully destroyed, the death can not trigger a respawn.

Only an academic exercise if you play CivDOS, but in CivWIN you can save the game in 4,000 BC. This quirk in the game proves that there can never be a 0-turn win, the fastest possible victory is 1 turn. I knew this before, but now understand why it is so. Academic theory is not impotent, for now it suggests another practical possibility: a 2-turn win with 7 civs on Emperor, no matter what map you play.

Haven't done the math on that, Emperor with 7 civs, conquest in 3,960 BC. Have a feeling it'd overflow the civscore's signed-integer capacity, and you'd end up with a rating lower than Dan Quayle.

 

[tuga2112]

not really on the topic but im curious.

what is your motivation for thinking this up and considering the experiment VS starting up a new game with a new set of self-imposed rules (like, must not conquer or bribe a single city and not allow city to be destroyed by AI ) ?

 

[Posidonius]

not really on the topic but im curious. What is your motivation for thinking this up and considering the experiment ...?

Why? Imagine the civscore, my boy! And because i like to take things apart and see how they work. Every time i try a new game, while reading the rules i'm already thinking up ways to break them.

Now know why a 0-turn win is impossible, and knew that a 1-turn win can happen but only in special circumstances, and counting on luck too, but there was another big problem: Even a failed attempt would take an hour or more, and that's with only 1 rival. The civscore bonus for a conquest win in a 2-civ world is not so good. But the 1-turn win with a full slate of 6 rivals is impractical to even try, a failed attempt might take ten hours before you know it's failed. And it's a tedious 10 hours: not doing much, just mashing keys.

So, took the recent idea as a jumping-off point and considered a 2-turn win. Once you let go of preconceptions, the possibilities blossom. Suspected that a 3,960 BC win would be easier, still tedious and hard, but easier. As it turns out, it's not just easier, but a total inversion of probabilities. A 2-turn win in 3,960 BC is not just incrementally easier, but i think it could be virtually guaranteed. I think i could make it work on Emperor, with 7 civs, with any kind of map.

Err, there are "circumstances" i suppose. Would have to abandon the dream of a 0-city win, would have to discard games until you get one starting with MapMaking, and you'd have to start on a landmass with at least 4 huts. Need 1 hut for a Phalanx. Need 3 huts for 150 gp, which buys a Trireme when you add 50 gp from selling your Palace. To maximize success, i'd customize for a young and dry planet with large land.

4,000 BC: kill all rivals on your landmass, found a coastal city, buy a Trireme.
3,980 BC: get the Trireme, kill all but one of the respawns, kill all other civs.
3,960 BC: kill all respawns, kill the last rival city, win game.

It should work every time, no matter what map the game gives you. Just a matter of mashing keys, but the realization now, is that the 1-turn win would require luck and lots of time even to fail. But a 2-turn win almost can't fail, it's just a matter of putting in the key-mash time. Start with MapMaking and 4 huts at minimum, and each extra hut just makes the game go exponentially faster. That's the real Aha! here: attempting the 1-turn win with 6 rivals could mean hundreds of hours of real time. A 2-turn win might take 10 hours, but it's virtually guaranteed, and you only lose 2 civscore points by waiting the extra turn. Before, i faced an uncertain sinkhole eating swaths of time. Now, it looks like a manageable investment of time to pull it off. Aha!