(August 6th, 2024, 19:04)Erasmas Wrote: Any idea why no one is reporting? Do you want more lurker feed-back? I prefer to follow one thread but was spoiled before I decided whom to follow so my natural inclanatation to shut up unless I have something to tell is even more active. (Coldrain I chose you so please report ).
I would expect FFH2 lures more lurkers being non-standard.
I've been meaning to report, but didn't have good opportunities to sit down and write for a longer session. I'll update soon, thanks for the interest. That being said, I would also appreciate other players keeping up with reporting so I can read the game from their point of view when it is finished.
I'd suggest us sharing our civ's passwords after the game, that way we can all log in to different turns and civilizations to check the game from other player perspective. That's something I would like to do after the game is finished.
By no means I meant to imply it is some sort of obligation to report. I just find the FFH games so fascinating that I had expected a much bigger audience. It is true that if you do not know what the tech tree looks like you cannot give much advice. On the other hand, there is much more reason to report the mechanisms of the game, as opposed to the "Noob explanations" I see in some of the regular threads.
Anyway, sorry if I was not clear.
I tried to explain a few mechanics in my german thread at civforum.de. I expected the audience here to have greater FFH experience.
So if you can read german (or want to use a translator), there are a few explanations.
And I really need to download the savegames again, or ?
I "lost" a bunch because I reinstalled civ a few days ago.
(August 6th, 2024, 19:04)Erasmas Wrote: Any idea why no one is reporting? Do you want more lurker feed-back? I prefer to follow one thread but was spoiled before I decided whom to follow so my natural inclanatation to shut up unless I have something to tell is even more active. (Coldrain I chose you so please report ).
I would expect FFH2 lures more lurkers being non-standard.
I'm in 2 civ4 games, 3 each for aow1 and aow2, and 7 for dom6
So it's a challenge for me to even think about my turns let alone talk about them - sorry.
My progress so far has just been peaceful development that has turned out decently (as elohim)
unfortunately I had brain failure when I saw svartalf scout exploring dungeon in my lands, I couldn't cancel open borders to stop it (recent agreement) and was too retarded to remember to declar war instead
so there is now 10 strength zombie killing my galleys lol
(August 6th, 2024, 19:04)Erasmas Wrote: Any idea why no one is reporting? Do you want more lurker feed-back? I prefer to follow one thread but was spoiled before I decided whom to follow so my natural inclanatation to shut up unless I have something to tell is even more active. (Coldrain I chose you so please report ).
I would expect FFH2 lures more lurkers being non-standard.
I'm in 2 civ4 games, 3 each for aow1 and aow2, and 7 for dom6
So it's a challenge for me to even think about my turns let alone talk about them - sorry.
My progress so far has just been peaceful development that has turned out decently (as elohim)
unfortunately I had brain failure when I saw svartalf scout exploring dungeon in my lands, I couldn't cancel open borders to stop it (recent agreement) and was too retarded to remember to declar war instead
so there is now 10 strength zombie killing my galleys lol
So ... though it may be forlorn, I'm still hoping for this game to somehow continue, and I have a slightly crazy proposal for how to maybe do it. First, some background on what I've tried so far, not a spoiler in spite of the tags; if you don't care about the background, feel free to skip it:
Since the problem stems from circumstances that led Aurorarcher to a convinction that mackoti cheated, my first hope was that I'd be able to learn what actually happened, to find (if there was no foul play) that Auro's conviction was incorrect because of incomplete information, and communicate that fact to Auro without giving away in-game secrets he doesn't already know. Unfortunately, that didn't work: Though mack tried to cooperate, even sending me a copy of the latest save he received, I have as yet not even been able to reproduce the events of the critical turn as shown in the log. Since mack's reasons for taking the actions he did are not verifiable (i.e. they are based not just on the game state but on his personal impressions, which of course only he can know) there's not much hope of reconciliation along those lines.
(There is one other possibility that I'm trying to explore, that should either enable me to figure out why I can't reproduce the turn's events or obviate the need to do so; whether it helps or not presumably depends on what I find that way, but I don't know if it would save the game either way. I still want to try, and will do my best, but....)
So what to do? Aurorarcher has said that the best hope is probably to find someone to replace him, and I don't know what efforts have been made to do that so far. The game has clearly stalled out though, so either no one has tried (I haven't yet; I've been hoping we could go on with the players we've got) or the attempts have thus far been in vain. So, either while looking for another solution or as an attempt at a maybe-solution in itself, I have a completely crazy proposal which might not be attractive to anyone, but might be worth ... something?
This is predicated on the assumption that Auro will be willing to play if he can be sure mack plays fair going forward, that mack is willing to play on under far-from-ideal conditions instead of letting the game die in recriminations against him like his last game, and that all of the players would be willing to accept a likely much-reduced turn pace as better than what is looking increasingly like no game at all. It's possible one or several or all of those assumptions are false, but if they're not, this is my (admittedly crazy and extreme) idea:
Instead of being sent to mackoti, the saves for the Malakim could be sent to me (or another trusted lurker or a group of them, even thoroughly spoiled) so that I-or-the-lurkers could send screenshots to mack, answer questions about what's visible in the save, and so forth, and he could instruct me/us on what to do with each turn. Then I/we would send the To-The-Next-Player save on to the next person in line, and create a post-turn save (that is, save after ending turn, when the "game paused" message is up, with no changes possible anymore) to send on to mack for purposes of planning and reference for his next turn. This would certainly slow the game down - maybe not too much during peace-time since there may not be a lot of mid-turn information to acquire, but war-time turns would take a long time except on days when mack and I-or-whoever could communicate e.g. by chat in real time. (i.e. "Send the new zepellin on a recon mission to the tile 1N of SevenSpirits's Aluminum." Okay, here's a screenshot. "Use my new Tank to attack the city of MadeUpExample!" It died to the C5 Ambush spear, but the spear's down to 0.3 strength. "Argh, stupid pRNG! Clean up with my supermedic Jaguar.")
So ... I dunno. I'm just trying to brainstorm here, in the hope that the game can continue in a way that will still be fun for everybody. Hopefully?
I'm not sure if any attempt has actually been made to get a replacement player, or if it's only been discussed. If the former, it's been a long time with no takers. If the latter, and people still want to continue, maybe that should be done? I didn't see any posts about it on CivFanatics, for instance. But either way....
...
I hate this, but I did the work, and I'm going to post the results. It's going to be a wall of text, so I'll spoiler it, but there are no important spoilers for the game itself:
I have now tested the situation from mack's latest turn fairly extensively. I still can't duplicate the results of mack's attacks on that turn, but at least I now know why. It took a long time, partly because the time I can spend finding out what went wrong on something like this is limited anyway, and partly because I wanted to do a comprehensive job. I asked coldrain for savefiles and the password to access them, and found that by comparing what I saw in his save from any turn N+1 with the one from turn N, I could infer what must have been done on turn N, duplicate it exactly, and get exactly the same results that occurred in-game. I could then end turn, pass the resulting save-for-mackoti to myself, and get the same event and multi-turn lair results that mack got in the game - and sometimes get the same combat results mack got on turns when combat occurred, but not consistently. (As with the coldrain control group saves, it was always possible to discover the order in which units attacked, what promotions they took when, the tiles from which they attacked, and where they went afterward, by observing their state in the save for the following turn. There could have been ambiguous cases in theory, but there weren't in practice for any of the saves.) I also asked Auro for the very latest save so I could roll it forward and see what everything looked like at the start of mack's following turn, to be sure I was trying to duplicate the right things. And duplicating the moves was no problem. The combat results are what wouldn't line up, no matter what I did.
Since I couldn't duplicate them, I couldn't test (and thus could neither support nor disprove) Auro's hypothesis that mack save-scummed to try different attack orders to settle on the one with the best results for him. But I did eventually discover what might have caused the discrepency ... and also that it didn't matter. For the latter, I'm not going to post the combat log in detail (here at least; I can do so in mack's thread if desired) in case the game continues somehow after all and it provides spoiler information. I'll just provide the overall picture:
Prior to turn 89 - the one which led Auro to accuse mack of cheating - throughout the course of the entire game, mackoti attacked almost 30 units (I'm excluding defensive combats which mack didn't initiate directly, and which in any case were very few) including several individual attacks with odds below 30% apiece. And (including the defensive fights, though I'm not counting them toward the total odds) he won every single fight, without exception, all game long, before the turn when Auro became suspicious of him. The chance of this happening given his odds for the fights involved is less than one tenth of one percent. Including the outcome of the disputed turn, the odds that the attacks mack made this game would go at least as well as they actually did - by chance alone - are less than one in fourteen thousand. [EDIT: Removed an erronious and unnecessary restatement.] There have been collections of events less likely than this that actually happened in the past (see e.g. Fintourist vs the Impossibly Lucky Spear in PB49) but those are limited sequences with artificial end-points, selection bias, and cases cherry-picked to highlight the emotions they caused. That's not what we have in this case: We have an unbroken series of ~30 combats - every single one mack took, throughout the entire game - of which the only outcomes that weren't strictly wins were a <35% withdrawal on T89 itself when the victory odds were just 1.3%, and the one token loss, also T89, when the odds of victory were zero, to more decimel places than Civ bothers to track.
Then there's the question of duplicating results. I tested this on two different computers with two different operating systems, and moving units, initiating combat, selecting and executing worker actions, selecting builds, adjusting tile micro, checking various advisor screens (checking the F9 statistics screen and sortint by number of units lost confirms that I didn't miss anything: Zero units lost for mack prior to T89) and no matter what order I did them in, I got the same combat results on both machines, as long as the combats themselves were taken in the same order, no matter what else I did first or in between. And the results I got, when I attacked with the same units as mack, in the same order, always the same for me on each machine, were not the results that mack got. So I tried some other things. I'd rather not describe what worked and what didn't unless people insist, but I discovered there is a way to cheat in a pbem that advances or modifies the fixed random seed used for combat in a different way on each of my two machines. That is to say: When I did this, I not only got new results that I hadn't gotten without it, but got different new results on my new Linux computer than on my ancient one that runs Windows XP. Of course what this suggests is that mack did exactly this, resulting in a different change to the random seed than on either of my machines, and chose his attack order (and potentially further pRNG manipulation) to work with (and/or further modify) the new random seed that resulted on his particular computer. Once it became obvious that mack had cheated, I would have liked to exactly duplicate the way he cheated, and if asked, show (at least in his thread) what it must have been, but it appears that isn't possible, since the type of pRNG manipulation he must have used has a different effect on different machines. (It might be e.g. OS-based, in which case someone else might be able to copy mack's moves, including the pRNG manipulation, and get the same results he did, if their computer is set up the same way mack's is; that I can't tell - I do know my setup is definitely not the same as his, with or without reference to these data!) I would love to find out that I'm wrong about mack cheating, but ... well, the probability of that happening is looking infinitesimally small to me.
The upshot of all that text is this: I've concluded that Mackoti manipulated the pRNG in such a way as to get favorable results that would not normally occur for any player just playing the save. This did not happen just on turn 89, but every time he could do so to avoid a combat loss throughout the game. As a result, he lost zero total combats prior to turn 89 (and no winnable combats then) in spite of taking low-odds attacks, such that the combined chances of getting results as good as mack's by chance in all the battles he took, all game long, rounds to zero.
So I don't know what to do with this. If we do look for a replacement player, I would say the player to replace should be mackoti rather than Auro. If we can't replace him ... well ... maybe having this information will help prevent this situation cropping up in the future -- I hope. It's extremely frustrating, but hopefully better to know.
Thanks for all of the work you put into investigating this, RefSteel. I read your description of findings, but there's a lot of information for me to digest there. I'll read again and post my thoughts on the matter after some more thought.
).
!
