General Turn Discussion

[Sullla]

sooooo, your analogy with a lottery draw is a poor one. Yes, the odds of any one person winning the lottery are miniscule, but the odds of someone winning, out of millions of iterations (sold tickets) are nothing special. By making the analogy to "a lottery winner in the dock", you're already admitting that SANTCA has "won the lottery" with regards to combat luck. That makes sense if there are millions of teams playing the game at once, not so much in this singular case. If (in order to prove your case), you already have to assume one-in-a-million odds, that doesn't exactly dispel suspicion!

What we have here is more akin to someone purchasing several dozen lottery scratch-offs and winning on every single one of them. I think anyone in their right mind would find that pretty fishy. That's not going to happen in a truly random situation.

Whomp, I have no idea how one would cheat; I don't know anything about that. Nor do I have a quick and easy solution to this situation. I really, really didn't want to raise this issue at all. I simply don't believe that it's possible for all of the random battles to swing in one team's favor, over and over again, across such a lengthy period of time. I'm a little surprised that no one on SANTCA seems to find any of this even the slightest bit suspicious.

Since Memphus seems to control 99% of what goes on with SANCTA, one possible solution would be to have someone else assume the reigns of the team and have Memphus recede into the background for a while, at least until the current war ends. Of course, that's no easy fix either, since Memphus is virtually the sole person running the team right now. I think that might be the best option, however.

 

[azzaman333]

I can't see how Memphus could be cheating.

I also think that we've been incredibly lucky, and that surely the RNG will turn against us at some point, but no matter how unlikely it is we could've had this luck, it's still possible. 1 in a million odds is still a possibility.

 

[Memphus]

Well Sulla i go back to my other post then...

If Someone else plays and we get good luck again then what?
What if it ls neutral? Or really bad? All point to build on the past case.

What if I chat to that player on man and direct what unit to attack with when and why? Or is that not remove enough as my 'vodoo' could still be at work.

FYI the only way I can think to cheat is to hack into the server take the game down, (without breaking the link to civstats somehow) then with the game down test the attack. If you like the results upload and repeate. If you don't like the results then you don't attack. But I don't think you can actually change the rng.

Now I am good with a comp. But hacking into a server? Well let's see that is also illegal... So risk going to jail over a mtdg? That and the ability to sync with civ stats is way beyond my capability.

But with all of that said the easiest way to disprove that would be for someone else to be in the game at the same time....to witness the attacks as they happen.

 

[IamJohn]

First of all Sullla, I really appreciate how you are dealing with the issue. You've made sure that the other team is getting their right to make sure everything is on the up and up, but you're also making sure things aren't getting out of hand, and that Memphus, and by extension, our team, is getting a fair chance as well. Witch hunts are generally a bad idea, and I'm glad to see we're dealing with the issue without them.

My position right now at least, such that it is, is that yeah, it's certainly suspicious, but a) We've never had a reason to distrust Memphus in the past, and although sure he (and Krill) have been willing to push the rules as far as they're allowed (ie preventing the other teams from seeing our tech research by manipulating what we're "researching", but nothing so hugely drastic as manipulating battle results throughout the course of an entire, multi-year multiplayer game; and b) There's utterly no actual proof. True the battle results are crazy. I'm not denying that, but accusing someone of cheating on such a large scale is very serious, particularly in a community like this one that is built on trust. Memphus has been extremely dutiful in recording every battle that's gone on over the course of the game, and in any case Daveshack should have by this point a wide variety of saves from just before our combat. If I remember right we set preserve random seed, so battle results shouldn't change (right?) if everything is played the same way. Have the admins replay a couple saves starting a couple turns before some of the battles and see if it's different.

In any case it may be worth it to talk to some of the HOF mods/admins (Denniz is probably the best choice right now), they're probably the people who've put the most work into discovering how to detect cheating.

I support Memphus. He's put an enormous amount of work into our team, and has continued to plow on ahead even when a lot/most people haven't really been able to be here. That doesn't mean it might not be best to have Memphus play indirectly for the time being simply so the involved teams can have confidence in what's going on, but unless someone can bring real proof it feels wrong to toss aside a friend who's dedicated so much for the team. I totally agree the number of victories we've won is beyond suspicious, and if someone is found to be cheating I want heads to roll, but until then we need to recognize we can't just burn someone with anything more then circumstantial proof, even if it is something that looks extremely strong.

 

[sooooo]

sooooo, your analogy with a lottery draw is a poor one. Yes, the odds of any one person winning the lottery are miniscule, but the odds of someone winning, out of millions of iterations (sold tickets) are nothing special.

Exactly.

By making the analogy to "a lottery winner in the dock", you're already admitting that SANTCA has "won the lottery" with regards to combat luck.

Well, as I said before, the odds should not be as high as cav propose but it should be the odds of Memphus getting results good enough to be accused of cheating. But yes they are still low, though not close to 1 in a million.

That makes sense if there are millions of teams playing the game at once, not so much in this singular case.

No! And this there is the prosecutor's fallacy. You should not restrict your data set just to people playing this game. You need to consider every game that has ever been played by a person that would inform the civ community if they got such impressive results. And that's a lot of games of civ.

Maybe this analogy is better: There's a report in a newspaper that someone has the "perfect" hand in bridge. Should you accuse the player of cheating? Of course not, without further evidence. You cannot say "But, even if you consider all of the hands in that game, the chance of the player getting such a hand is so miniscule that he must have cheated". Because clearly you can't just restrict it to hands in that game. A very large number of hands of bridge have been played around the world since the game was invented, and if it occurs then it's fairly likely that someone will tell the newspapers.

What we have here is more akin to someone purchasing several dozen lottery scratch-offs and winning on every single one of them.

Well, the time frame doesn't matter - it's the sequential odds that matters.

If (in order to prove your case)

Again - I don't have a case. I'm not trying to prove Memphus' innocence. I just am a fan of statistics and hate to see them misused.

 

[mostly-harmless]

First of all, thanks Sullla, for bringing it to our attention, before things get nasty in the public forums.

Consider this:
The odds on rolling a certain number with a dice is allegedly 1/6, right?

So given a large enough number of samples all numbers will appear equally.

Now the crux is, what is a large enough number of samples?

I start rolling the dice and get 5 sixes in the first 10 throws and not a single two. Bad dice!?
I keep on rolling. After 100 throws, the sixes are still way ahead at 35 occurrences. Bad bad dice.

I keep on rolling. After 1000 throws I get a slight pain in my wrist but the numbers are slowly levelling out.
And now consider a 1000 people doing the same 1000throw experiment. What happens if one of them get 700 sixes??? Unlikely? Yes, highly. Cheating? Not at all.
And now consider, that guy would have been the first and only one to do the experiment. He throws 700 sixes in 1000 throws. Would it be any closer to cheating?

So it really comes down to the number of samples. And stating only 30-40 battles is not a sufficient number of samples for battle results.

Even if Memphus keeps up those battle results for 10,000 battles there is no evidence at all he is cheating. He just got lucky! We are talking odds and RandomNG here.

Sullla, remember the streak of bad luck in our Apolyton battles, where we lost what, 3 battles at 95+% odds in a row? What would have happened if we would have lost the next 2 battles at 80+% odds as well? and the next two at favourable odds as well?
We didn't, because it was unlikely, but we could have, since there is a small chance the RNG throws up "the same" numbers in a row.

I remember how this forum was ripe with people screaming "AI cheat" after loosing battles at 95+% odds repeatedly. (I think it probably is still going on?)

And on top of that, I would like to chime in with:
How can cheating wrt to battle outcome even been realized?

mh

 

[dutchfire]

lurker's comment:
Couple of questions/notes:
We were attacking most of the time. Does this affect things?
Memphus, you say you made simulations of combats, how did you do that? Could you give us your simulator? Did it use the civ4 combat engine itself? (IE. a python mod of civ4 or something like that)
Is there a way to get the random seed out of a pitboss game and use it for tests? That's the only way of cheating I can imagine right now.

 

[azzaman333]

I would suspect that Memphus' simulations are WBing the situation, and running through it 100 or so times.

Or at least, something like that.

 

[Krill]

Whomp, I have no idea how one would cheat; I don't know anything about that. Nor do I have a quick and easy solution to this situation. I really, really didn't want to raise this issue at all. I simply don't believe that it's possible for all of the random battles to swing in one team's favor, over and over again, across such a lengthy period of time. I'm a little surprised that no one on SANTCA seems to find any of this even the slightest bit suspicious.

I'm pretty sure it's partly down to us being involved in every single war.

Since Memphus seems to control 99% of what goes on with SANCTA, one possible solution would be to have someone else assume the reigns of the team and have Memphus recede into the background for a while, at least until the current war ends. Of course, that's no easy fix either, since Memphus is virtually the sole person running the team right now. I think that might be the best option, however.

I'd rather disband the entire civ than have other teams demand that someone stop playing because they are losing.

 

[Krill]

OK, I don't really need this *%&^ at this time, I'm kinda busy as it is, but whatever:

1. Geronimo, after killing the first big Kaz stack, was because we needed to consider how to position troops for the second wave, and what to do with the GG. We were playing it safe. How's that for irony, huh?
   
   
2. Trireme. Kaz had 2 triremes and could easily have blocked us at PM, or killed our trireme and then blockaded us. So it was basically 100% odds of getting screwed over, or 30% odds of winning, getting the promo (key!) and then being able to stand off on the ocean tile while keeping 50% odds on Kaz trireme. And AT was the player who said attack there, Memphus through that one out to the team.
   
   
3. Killing the first Kaz stack: we won the 10% odds battle at first, when we expected to damage the mace. Did we get lucky? Yeah, but considering that the entire battle plan was going to work so long as we just dented the mace enough I don't see what the big problem is.
   
   
4. In the above battle, we had to kill the HA so that they couldn't push forward and fork cities that we couldn't then defend, or keep their stack together and push for TKY while hte HA cut the road leaving us stranded in Zero. That was why we continued to push so hard to kill that stack (we also new a second wave was coming and needed to heal/stop them getting fort boni).
   
   
5. Next battle, in the second wave at 1AD, I see only a 26.7% battle that seems odd (LB v. Mace). That battle was going to happen anyway, too many damaged goodies to kill underneath. Now the *%&^&ing just seems to be whining.
   
   
6. Knights v. Muskets, IIRC they also had retreat odds due to flanking. Please include those odds, because the likelyhood of our units winning or retreating compared to winning is slightly different when you consider we got hte free flanking 1 event.
   
   
7. The last battle does seem nuts though. I was never around for that (I'm busy ffs!) and the first thing I knew of this was Cyneherd saying that some *%&^ had just blown up over here. Someone tell me how we supposedly cheated, or they can STFU and keep on playing.

Now, I also want to say that you've ignored barb attacks in the list, even though you bothered to comment on them originally. Remember Zuurstof! We lost a 99.5% battle there, right when we needed units to deal with Kaz (and lets be frank, the barbs gave me and Memphus more headaches than Kaz just because of where they were attacking from). Doesn;t matter that it was to the barbs, we still needed him.

Yes, we've gotten very lucky. I don't think anyone is denying that. Every single turn was planned out so we knew how badly it had to go to stop the attack and turtle for the turn. We knew how to change the worker orders to chop out max units to repenish our supplies. Kaz just got outplayed completely in that first war.

 

[Krill]

And something constructive. Prebuild knights. I want those cannons flanked out of the game and the rifles can finish off the remainder of the weakened red coats. THe second Cav stack will be delayed enough for us to just flank the crap out of and then stay behind TKYs' culture. Hopefully anyway.

So long as we can hold TKY this is counts as a succesful defense IMO

 

[AutomatedTeller]

I havent' been around all that much recently.

Couple of things:

1) To figure out the actual odds of the results shown, we really need to look at the round results, since those are the actual random events.

2) Do we know that a turn played out in exactly the same way will turn out the same way on different computers? I mean, assuming that you have the exact same software. If so, the solution is easy - have someone else play each attack in exactly the same way memphus did and see if we get the same results - that person would probably have to be Sulla.

3) We really need to take into account encounters with barbarians when doing this analysis.

 

[shadyforce]

Sullla,

Clearly I find the situation very suspicious and I certainly agree with you bringing it up and asking questions. The RNG outcomes are crazy. We don't have our head in the ground pretending nothing weird is going on. I am just a lurker on this team. I don't even know Memphus so I don't really have any bias or loyalty or anything. But from my perspective:

1. Soooo's lottery analogy is perfectly accurate and his understanding of the use and misuse of probability is the same as mine. I also like MH's dice analogy. If millions of dice have been rolled throghout time, someone somewhere could well end up rolling twenty-five 6s in a block of twenty-eight rolls, without cheating whatsoever.
2. I don't understand what you want him to say. How is he meant to defend himself? How is he meant to prove he wasn't cheating? And why has he got the burden of proof anyway?
3. Nobody seems to be aware of a viable way to cheat.

Clearly Memphus lucked into a highly favourable circumstance... crazy luck can supplement proof of cheating were it to be presented but in the absense of any evidence whatsoever, luck is just luck.

 

[Mojoqmeyvam]

3. Nobody seems to be aware of a viable way to cheat.

Right. Either there's a previously-unknown way to remotely hack the RNG to give the answers you want directly, or or a previously-unknown way to hack the RNG to tell you what the next number's going to be, and then either using it for the battle, or skipping to the next one. I don't see how either of those are plausible.

As I said before, disproving that we got two lucky streaks from the RNG in 50BC and 940AD, and were generally lucky (about 1% chance) of the single combats going the way they did, losing no units, isn't happening. 1AD was a "normal" good round of combat. So we play on, and pray that the next battle isn't a ridiculously streaky one, when these questions will come at us again.

 

[Krill]

I'd just wait for a plausible method of cheating to be published. If anyone can say "This is how you fix the game" then I 100% want it looked into by everyone. Until that happens everyone can suck it up IMO.

 

[Memphus]

On more interesting note....No other team has attacked another human team, so Sulla for you to not include the barbs, but only attacks on other teams isn't justifiable.

We are the only ones on offence. (which goes Back to Krill's point that the finger can be pointed at us since we have been in every battle in this game)

@At #2: the RNG is on the server, not the player playing, therefore anyone who replayed the turn from their machine would get the same results, given the same attack order.
Sulla ask DaveS for the saves prior to all the battles. You have the sequences so run through them. When you win in the same fashion then can we go back to playing?

 

[Krill]

Hmmm...Not sure, but I think the order of moving units may affect the RNG (i.e. worker moves). If so, and because we didn't log the exact order everything was moved, replaying all of the relevant saves isn't going to prove anything.

 

[Memphus]

yeah good point.... :sigh: I have detailed notes but not that good.

But back to planning. It will take 2 more turns for MT. We have a road being built from OO over to the Gems city so east coast can reinforce (OO is buiding troops)

We can save 350 gold a turn. saturn gifts us ~160. Knights cost 110 to upgrade.

Do we gift TKY to Kaz? or do we try to make a stand? We can have 8 Rifles in TKY to try and hold for 1-2 turns before we have cavalry ready.

Worst Case is if Kaz splits thier stack and head for DT too. We have no units there as everything is headed to TKY.

Also it is 200 gold to upgrade the CGIII LB to a Rifle. worth it?

 

[Krill]

Every single Cavalry we have has odds on attack, and flanks. No cannons means we should be able to hold out. So the main question is, can we flank every single cannon Kaz has, and then Cavs', if they beeline TKY? That's why I think we need to upgrade every knight and HA we have and prebuild the Cavalry to wreck Kaz first, now we bought ourselves another turn against Cav.

That lone upgrade might be enough to stop Kaz for a turn to bombard tbh, but I doubt it...kinda hope they just wait for a turn and then match Cavs time table to walk up to TKY on the same turn.

 

[mostly-harmless]

Hmmm...Not sure, but I think the order of moving units may affect the RNG (i.e. worker moves). If so, and because we didn't log the exact order everything was moved, replaying all of the relevant saves isn't going to prove anything.

I was under the impression that the RNG only changes when it is called by the program.
ie. unit movements do not affect the RNG.
Say single play:
You attack at 75% odds but loose. You reload and try to spread a religion first (RNG is called) and then attack, there is a chance that you now win at 75% odds in the same battle as the RNG has to come up with a new number.

mh