Game Four is finished! We had another result that was unexpected in a long, long game. There was another amazing turnout for the Livestream today, averaging 132 viewers on a 7 hour stream. Thank you so much to everyone who was able to catch at least some of the broadcast! Here's a batch of links for the next game:
Congratulations to Alternaroll and Kuro as the winners of Game Four's picking contest with 16 points apiece. Scoring was well below normal in this game with an average score of 5.34 points. I won't spoil things but everyone was way off on one of the victory conditions. We are continuing with our normal Friday date for next week's game. This time we a wide open field with Charlemagne and Shaka as the seeded leaders - not exactly a murderer's row of competitors. Who do you have this week?
"Pericles has a lot of potential land; I'll hope Hannibal doesn't attack him early before he can get metal" was not, in practice, a good bet.
Does the AI actually somehow take into account that a neighbor doesn't have a metal source? We also saw a brutal and successful early attack on metal-less Pacal last year.
No metal will result in lower power, but what about more directly?
"Pericles has a lot of potential land; I'll hope Hannibal doesn't attack him early before he can get metal" was not, in practice, a good bet.
Does the AI actually somehow take into account that a neighbor doesn't have a metal source? We also saw a brutal and successful early attack on metal-less Pacal last year.
No metal will result in lower power, but what about more directly?
I think it's more just that we remember these attacks much better than the metal-y ones, because they're successful so much more often.
Maybe the power bar is lower for metal-less AIs which makes them more tempting, though?
(June 19th, 2020, 15:39)GeneralKilCavalry Wrote: Could the EP difference between observer and AI be prompting AI espionage spending? Or is the AI unaffected?
No it definitely has a huge effect and has been brought up basically every season. I think it really skews things since once a civ hits a breakpoint against sullla in courthouses or whatever they can turn off their espionage slider and launch their research way above their competitors still wasting a fifth of their entire GDP on the observer civ.
Can't remember which poster has posted the text edit solution to this for several years and sullla keeps saying he doesn't want to mod the game.
But it is the debug mode that enables zooming into cities, correct? So wouldn't it be better to dump espionage buildings and specialists in the observer capital, allowing for demo visibility, but not skewing results?
"I know that Kilpatrick is a hell of a damned fool, but I want just that sort of man to command my cavalry on this expedition."
- William Tecumseh Sherman
(June 19th, 2020, 19:53)Fluffball Wrote: Can't remember which poster has posted the text edit solution to this for several years and sullla keeps saying he doesn't want to mod the game.
That's me, and this issue is why I don't follow this series, it's so skewed from the real AI behavior (also being on Deity difficulty, which skews things too, by the cost of build items being cheaper relative to that of techs.)
Re the AI win condition discussions - there is a specific code path that they turn on for a culture win, where they pick 3 cities and max all the culture multipliers in them and eventually run the culture slider. Whether a particular AI does this is random and depends on the leader's flavors, so yes Gandhi does do that a lot. I'm not sure if any such targeting exists for the other win conditions. I do know the "we'd rather win the game" message comes up sometimes on refusing to trade a spaceship tech.
Except for Ragnar, all fielded AIs were competent AIs. That is, none you would expect to win, but none you'd be shocked to see win.
Ragnar had the most available land : any other AI in his starting position would have been a favourite... but that's Ragnar.
Starting spots were rather poor, as illustrated by the fact they were heavily wooded, which is the start equalizer function's way of saying "sorry".
Ramesses : no horses, no war chariots.
Saladin : his religious emphasis usually works against him. Contrary to other zealots, he won't actively try and spread his religious. As a consequence, he's usually on the receiving end of a coalition, instead of the opposite.
In the end, I went with three predictions :
1) Pericles would be metal-less, and thus quickly eliminated. Most likely by Hannibal, who'd grow into the dominant power.
2) The West would be a clusterfuck. While Kublai and Ragnar piling on Victoria seemed a logical scenario, I didn't believe in it. I rather imagined something like Victoria declares on Kublai, Ragnar piles in, Saladin attacks Ragnar.
3) It would be a late finish.
Well, that turned out to be mostly correct, "mostly" being the operative word here.
1) Pericles was indeed the FTD, owing to his initial lack of metals.
But while he didn't get the spot I'd shown, neither did Hannibal, which came as a surprise (discounting the Copper which he couldn't see initially, it was still his only 4-ressource cluster in close range).
That delayed his war declaration enough that Pericles was able to get Iron during the war. Pericles still died, but much later than I'd expected.
2) The West was indeed complete chaos, but instead of a "less beaten-up" AI emerging, we got a "game-dominant" AI with Saladin...
3) It was indeed late. But a tad later thant anticipated.
Got lucky once again with the number of wars prediction. Getting within two is good game-reading, nailing the exact number is dumb luck.
Second game in a row where the victory condition is down to a coin flip. I'm wondering whether this is owing to Diety-level. Culture or Domination wins are on the clock, and the clock might be moving too fast... (I know, weird observation to make after one of our slowest games).
(June 19th, 2020, 19:53)Fluffball Wrote: Can't remember which poster has posted the text edit solution to this for several years and sullla keeps saying he doesn't want to mod the game.
I did. Posted a reminder this year too, but not gonna insist.
Somehow, I get it : Sullla is extremely familiar with the World Builder and the debug tools, but not at all with XML editing.
When you've got a solution with tools you know, you tend to be reluctant to try another solution, even though it's a better solution, if it requires unknown tools.
Since he did mod the game this year to remove the AP, I thought he might give it a shot this time. Oh well.
(June 19th, 2020, 19:53)Fluffball Wrote: Can't remember which poster has posted the text edit solution to this for several years and sullla keeps saying he doesn't want to mod the game.
That's me, and this issue is why I don't follow this series, it's so skewed from the real AI behavior (also being on Deity difficulty, which skews things too, by the cost of build items being cheaper relative to that of techs.)
Re the AI win condition discussions - there is a specific code path that they turn on for a culture win, where they pick 3 cities and max all the culture multipliers in them and eventually run the culture slider. Whether a particular AI does this is random and depends on the leader's flavors, so yes Gandhi does do that a lot. I'm not sure if any such targeting exists for the other win conditions. I do know the "we'd rather win the game" message comes up sometimes on refusing to trade a spaceship tech.
To add to that, he doesn't have to do it on his own. Multiple people volunteered to do it for him.
By the way, it had been mentionned during one of the streams last year, but you didn't notice : pressing ALT while hovering a tile controlled by a player's cultures gives you the number of cities for that player. No need to count them every time.
I’ve found the general AI performance of AI survivor to correspond well with actual performance of AI’s both in my games and in civfanatics sample games. The only difference is that AI survivor discourages wonderbuilding more than usual (Deity level + hard to be isolated from other AI’s), punishing characters like Ramesses, Louis or Bismarck.
I'm curious how the AI determines when it's ready to start settling icy fishing villages and filler cities which claim a handful of tiles between widely spaced core cities. At the Deity level they ought to have enough spare commerce to afford them at almost any time unless they've already severely over-expanded, but they mostly seem to plonk them down once they've entered into "time to win the game" mode.
The cities, particularly the fishing villages, also seem to confer the AIs significant advantages- in addition to speeding along domination and building defensive depth, with all the production discounts they receive on Deity any city able to work a mine or watermill or two will likely be able to churn out several units and build enough infra to pay for itself. Are some leaders more willing to carpet the map than others?
Scoring was well below normal in this game with an average score of 5.34 points. I won't spoil things but everyone was way off on one of the victory conditions. We are continuing with our normal Friday date for next week's game. This time we a wide open field with Charlemagne and Shaka as the seeded leaders - not exactly a murderer's row of competitors. Who do you have this week?
