Mission percentages bugged

[Werner]

Ive been noticing more problems with this as time goes on. It seems that at certain points the game won't let you perform certain things no matter how many times you try. I noticed this awhile ago when I was trying to get my princess to marry a general. She had 8/10 charm but no matter how many times I reloaded the game it never went right and she always "ran off with her lover".

The other day I tried to kill off an enemy faction leader, some lunnybin that had undoubtably been sniffing paint and drinking lead for most of his life, with my top notice assassin (12/10). The mission percentage was 37% and my assassin died. I tried reloading about 10 times in a row and every single time my assassin was killed. On a 1 star faction leader ?! WTF

Now I know what your going to say, the percentages just played against you and you were very unlucky. Well today I was trying to sneek my spy into an enemy town. It said that he had 100% chance of success and he got killed.

[vonsch]

I've seen the 100% success rated death with spies too.

[JCoyote]

All mission percentages are rounded to the nearest 5%. IIRC from other TW games, no missions has a true success rate of much over 99%. Hence, on M2TW, if your percentage is 96%, it'll say 100% even though it isn't.

[KHPike]

So far the highest percentage I've ever seen is 95%. I've encountered this bug before, which happens commonly with priests, assassins and merchants. On the other hand, I've managed to acquire merchants much more powerful after a few tries. On odds 15% and below don't bother trying anything though. The odds will be so bad that you'll blunder.

As for the spy, I can't explain that. Bug I guess?

[Foz]

As for the reloading thing, people speculate that the random number generator's current seed is saved along with your save game. Possibly the seed is somehow otherwise based on the state of your game though. Either way, reloading any number of times should always give you the same result, provided the next action you take is the same one you've always been taking. The generator may advance once you take some different action, but it's also possible that whatever the current random roll is, it sticks around until you use it for something. So the short of it is that doing something else, or possibly more specifically doing something else that requires use of a random number, may be the only way to achieve a different result than the one you're seeing.

[TevashSzat]

Yeah, people always remember to do something after you reload to reset the seed or you will always get the same thing over and over again

[Nebuchadnezzar]

Definately bugged!!

Well I've seen spies in a settlement with a 113% chance of opening gates and fail!!

and a 95% chance of success means the mission will almost always be successful. This is certainly not the case as all my assassins which take on 95% missions end up in failure and death after a few attempts.

Reloading does not reset probability values unless you also include an additional move, besides reloading is a cheat but possibly the only way to make assassins work.

[TevashSzat]

Umm....I am pretty sure you cant have 113% chance. You can only have max 95% chance for anything i think

[Hollerbach]

If you reload you will have the same 'roll' as the seed is preserved. If you do some other action that uses a random roll you get a different roll, but it will be the same different one if you reload again and do I different thing before trying again. I hope the made some sense

As for having a 113% chance of completing a mission?? I ain't never seen anything like it. I've never had a spy fail a 100% infiltrate mission and other odds seem to go alright. Why reload though? IMO the game is easy enough without reloading whenever something doesn't go your way. If you find assassins underpowered then mod them, but reloading all the time would be a real pain!

[Nebuchadnezzar]

Umm....I am pretty sure you cant have 113% chance. You can only have max 95% chance for anything i think

Honestly it was several spies in the one settlement nevertheless unless my eyes deceive me that was the case. I might still have a screeny at home of it.

I never checked to see if I could reproduce this. I just no longer paid much attention to the values again.

^^Regarding assassins. No I never said I reload. What I said is to get uber assassins you need to be reloading. Their are a couple of posts in which players are assassinating the entire Mongol 10 star generals which usually have a 10-15% chance of success with a 10 eye assassin. How is this at all possible without reloading even with guild HQ?

[Carl]

i've seen the over 100% chance to open the gates BTW. You need about 5 3 skill spies though.

[Nebuchadnezzar]

i've seen the over 100% chance to open the gates BTW. You need about 5 3 skill spies though.

Accumulated rounding error then?

[Carl]

I don't think so. i've had it hit nearlly 200%. I think each spy has his own chance then they all get added together and thats the chance.

however i suspect a hardcode ovveride ensures that theirs allways a 5% failure chance.

[HoreTore]

Regarding reloading failed attempts...

It seems that the game only lets you reload one time with a new calculation, that calculation is then remembered in the save, so that any subsequent reloads are given the same calculation.

If you've played Civ, you'd be familiar with the system. Honestly, I think it's good. Stops some of that reload cheating ;)

EDIT: just noticed I'm the third one to post this. Note to self: read entire threads before replying...

[Whacker]

Yeah, people always remember to do something after you reload to reset the seed or you will always get the same thing over and over again

I can confirm that this is absolutely not true, at least the last bit. I've had both successes (rarely) and failures (mostly) on just simple reloading and not doing anything else but the mission execution.

[Werner]

I used to think that the "random seed' was saved also, but its not. I've used a 10/10 assassin to kill the pope before and around the 3rd or 4th reload it'll usually work. I think its more of a game mechanic, they dont think you should be able to kill so-in-so because set amount of faction heirs are left or set amount of factions are against this faction. For example, most of the time this problem has occured against smaller factions who are financially poor, dont have big armies or are getting pushed off the map. Maybe its a balance tool so you can't just spam your enemy out of existence with assassins? Either way it sucks and should be fixed

BTW I have also seen the 113% chance of success, it happens when there are a bunch of spies inside an enemy city.

NOTE: To the guy who was talking about all percentages being rounded up or down by 5%. Huh? I could show you a million pictures of spies with 100% ratings for getting into a city/spying on someone etc..etc..

[Whacker]

NOTE: To the guy who was talking about all percentages being rounded up or down by 5%. Huh? I could show you a million pictures of spies with 100% ratings for getting into a city/spying on someone etc..etc..

My thought as well, but I wasn't going to get into that. The game has always shown me percentages down to the ones digit not rounded, for example 38%, 53%, etc etc.

[holycow]

from my experience, the percentages do not correlate to %success at all. I've had 17% and succeed and 95% and fail. Obviously save before you committ. And after reload, go move something else or what I do is try an inconsequential spy action on diplomat/princess/etc to get diff seed.

[HoreTore]

from my experience, the percentages do not correlate to %success at all. I've had 17% and succeed and 95% and fail.

Why should that be wrong? 95% doesn't mean a guaranteed success, and 17% is almost a one in five chance...

[Foz]

from my experience, the percentages do not correlate to %success at all. I've had 17% and succeed and 95% and fail. Obviously save before you committ. And after reload, go move something else or what I do is try an inconsequential spy action on diplomat/princess/etc to get diff seed.

Holy cow! Seriously, this is about the worst understanding of statistics and probability that I have ever seen. 95% success means 5% failure, which is still 1 in 20. It's not ridiculous at all to fail sometimes when you have a 95% success rate. Just because a result is likely to happen does not mean the random system is broken if it does not. The converse also holds true: just because something is very unlikely does not mean you should expect to never see it. In poker, for instance, the odds of hitting a natural royal flush are 1 in 649,740. The odds when extra cards are involved (draws, or community cards for instance) are a bit better, but still nothing remotely likely. Odds like this should make it impossible to see a royal flush, right? Nope. I've seen 3 already and am only 26 years old - one of them was actually dealt to me. I can sit there and expect not to see that royal flush all day long, and most times I'm right - but sometimes, it does happen. When it does, contrary to feeling like the deck isn't sufficiently being randomized to prevent the royal flush from being dealt, I take heart: a seemingly impossible event should still happen sometimes. It's the occasional unlikely event that really lets you know a random system is working correctly - any simple system can produce the favored result. Those unlikely events shouldn't happen often, but anything with a chance to happen will generally happen sometime if you wait long enough.

The point is that the long odds in M2TW are nothing even close to those of the royal flush, so you can't go around claiming the random system in the game is broken just because you see some things happen that only have 1 in 20 odds of happening. They should be happening a noticeable amount of the time, and in my experience do occur at approximately the 1 in 20 times that they should.

[Werner]

I agree but something else is going on in this game. The percentages dont mean anything...

[holycow]

cripes! who cares about understanding stats and probability, this is only a game! I play this game to dull my higher brain functions anyways. if it was designed to be unpredictable for balance or playability, then let's make it so and do away w/ percentages. All I WAS SAYING IS "FROM MY EXPERIENCE" NOTE AGAIN "FROM MY EXPERIENCE!!!!!" NOT WRITING A DISSERTATION. So get a grip people, sheeesh.

[Whacker]

cripes! who cares about understanding stats and probability, this is only a game!

I do. In terms of the game itself, that is. As werner pointed out, there are probabilities given for certain tasks, but anecdotal evidence points to game mechanics that aren't taken into account (namely spies in proximity). Thus his point, the prob. displayed is most likely misleading or invalid. This is uncertainty that I definitely want fixed/removed. Either don't display a probability at all (no way), or tell us exactly what our chances are with all things accounted for (definitely).

[gardibolt]

And I suspect that even when the spy thinks he has a 100% chance to infiltrate there's still a possibility of something going badly wrong---I hearken back to the old D&D days where a lot of people played that a 1 on a d20 was always a miss, even if by the charts your guy should automatically hit. That is to say, there's never a totally risk-free infiltration.

[Whacker]

And I suspect that even when the spy thinks he has a 100% chance to infiltrate there's still a possibility of something going badly wrong---I hearken back to the old D&D days where a lot of people played that a 1 on a d20 was always a miss, even if by the charts your guy should automatically hit. That is to say, there's never a totally risk-free infiltration.

Indeed, they could handle this like that, or something like the character resistances in say like Diablo II. You could actually have gear and skills that raised your various elemental resistances way above 100%, but the max you could ever get to was 80% because the game limited you to that. Likewise I think perhaps 95% as a max chance would be good, it still leaves a very small margin of error, so nothing is actually guaranteed.

[dismal]

Accumulated rounding error then?

You can't ever have a probability over 100%.

Sounds like someone at CA missed the lecture on conditional probability.

The probability of two spies with a 75% chance of opening the gate woould be:

p(spy one succeeds) + p(spy two succeeds: given spy one failed)

this would be (75%) + (75%)*(25%) = 93.75%

Add a third spy with a 75% chance and it's:

75% + (75%*25%)+(75%*6.25) = 98.43%

You can never get all the way to 100%.

[Werner]

This is uncertainty that I definitely want fixed/removed. Either don't display a probability at all (no way), or tell us exactly what our chances are with all things accounted for (definitely).

Exactly, the least CA can do is fix the problem. Whats the point in using percentages if their not actually true?

[Foz]

You can't ever have a probability over 100%.

Sounds like someone at CA missed the lecture on conditional probability.

The probability of two spies with a 75% chance of opening the gate woould be:

p(spy one succeeds) + p(spy two succeeds: given spy one failed)

this would be (75%) + (75%)*(25%) = 93.75%

Add a third spy with a 75% chance and it's:

75% + (75%*25%)+(75%*6.25) = 98.43%

You can never get all the way to 100%.

For the record it's far easier to determine the chance that all spies fail and then realize that the balance of the difference to 1.00 accounts for all possible success cases (since only 1 combination can possibly fail entirely to open the gate, all the rest must open it). In the last case, 3 spies fail is (.25)^3 = .015625. 1 - .015625 = .984375, the answer we're looking for. 5 spies for instance give (.25)^5 = 0.0009765625, with 1 - 0.0009765625 = 0.9990234375.

Two points then. One, CA may have decided to simply use one roll to determine the result, instead of multiple ones. In that case, they would simply sum the chances for the agents to succeed, and if the roll is less than the sum, success. In that case you still can only have a 100% chance of success at best, but numbers higher than 100 could be the "number to beat" as displayed and used by the game, lending some amount of sense to what the game displays.

Two, while no number of spies could theoretically make the success rate reach 1 (if they each have a roll), there may be a point where the game can no longer distinguish the success rate from 1, in which case failure would become impossible. That is, depending on the decimal accuracy of the random number generator and the way the devs implement the random tests, failure may not be an option even if it should be statistically. In the case where success = 0.9990234375 above, we've already passed 99.9% success. The game displays integer success percentages, so anything at or above 0.995 could be simply rounded up, and treated as 100% by the game, i.e. will never fail.

[dismal]

For the record it's far easier to determine the chance that all spies fail

In this case yes, but then you don't get to learn about conditional probability!

Two points then. One, CA may have decided to simply use one roll o determine the result, instead of multiple ones.

It's easy enough to calculate the probabilty correctly and then do one roll at the correct probability

We just did it.

[TevashSzat]

Some modder here should try to take a look at the files regarding mission success percentage and try to figure out if they are working properly or needs to be fixed

I have heard way too much anectdotal envidence to believe that people are just simply getting a bit lucky or has no luck