17

I was playing a Terror mission last night where after my unit would enter an area, a Cyber Disc and two Drones would come around the corner of a building. All four of my other units were in a position to shoot. My sniper with DoubleTap showed an 85% chance to hit the Cyber Disc. So I fired missing once and hitting once. My heavy missed twice, and my two assault guys hit once and missed once. I wasn't able to kill the Cyber Disc. I then reloaded from a save point and tried again, and again, and again.

Every time I got the same scenario, sniper miss, then hit. Heavy miss, miss. Assault hit, assault miss. I tried different weapon modifiers, but after five attempts I decided either the game was cheating or I was just so unlikely that given an 85% chance to hit I missed 5 / 5.

11
  • 5
    I would say it is not the game that is cheating, but the person who keeps trying to save and reload hoping for a different outcome! :)
    – Sterno
    Oct 11, 2012 at 19:29
  • 1
    It isn't cheating if you are using something that is part of the game (being able to save at any point and restore to any point). It is cheating if the game lies to me by saying "You have a 85% chance to hit, but I already decided you will never ever make this shot, ever." Oct 11, 2012 at 19:31
  • 8
    By saving and reloading repeatedly, you're trying to turn that 85% chance into a 100% chance. Anyway, I didn't mean to argue the morals of it with you. Think of it as the game rolling the dice before you saved, so even if you reload, the dice roll isn't going to magically change for you. The computer is just sitting there saying, "No, dude, we already rolled for that. It was a bad roll. Suck it up."
    – Sterno
    Oct 11, 2012 at 19:34
  • 2
    Insanity is repeating the same mistakes and expecting different results. Oct 11, 2012 at 19:47
  • 2
    @ShivanDragon: Assumptions like this are dangerous, so I don't make them. In the end, Fireaxis is being nice here with making it blatantly clear that they are essentially pre-computing the results. They could do much worse - in particular, they could record your hit results vs. the percentages, weighted by the "threat factor" of the situation, and simply silently lowered your chances in future attempts if your hit results are above average. The hit chance would remain the same ... calculated over the course of the whole game. Oct 15, 2012 at 13:01

4 Answers 4

30

It's not cheating as such; the game preserves it's random seed on re-load. Reloading a turn and repeating the same actions in order will produce the same results. This is because the game saves the seed it's using for the random number generator precisely to prevent this kind of constant reloading. Jake Solomon, the creative lead for XCOM was quoted in this article:

We use synchronous random in combat so the player can’t just reload when they miss a shot. Now, obviously there are ways around this, but this is a decent way of ensuring that the player’s choices do matter.

You can "game" this, if you like, by taking different actions and changing up your order. Assuming you are shooting at the cyber disc, try [something to miss], sniper, [something to miss], [heavy or assault to hit].

Bear in mind that the actually aim-scores will impact here. It's possible someone less skilled, or with an inaccurate weapon, would be unable to hit in actions 2 and 4. Likewise, your sniper could hit in actions 3 and 5, because he's got the most accurate weapon. And yes, if you do different things on these actions (aim at different targets), you may not get hits where you expect it. The sniper using action 2 might hit the Cyber Disk but miss the Muton in full cover you decide to aim at instead.

According to this article at PcGamesN, the X-com expansion will include an option to reset the random seed on loading!

8
  • 1
    I'd be careful assuming that you can determine the random outcome of action B based on what the outcome of action A was with the "same number". Its quite possible that they use different parts of the number, or manipulate the number in different ways, or that there are different number queues for different actions, etc.
    – bwarner
    Oct 11, 2012 at 20:03
  • Which is why I noted it was not guaranteed. If a high-aim soldier hits in Action 2, a low-aim soldier will not necessarily do so. Further a high-aim solider aiming at a different, more well-defended target, could miss. Current aim/current defense are not necessarily the only factors here but they are the most discernable. But my example assumed that in all 5 actions, he's aiming at the cyber disk. I'll edit to make that clear.
    – peacedog
    Oct 11, 2012 at 20:09
  • 1
    In support of this: I could not quickly find reference to an option for xcom, but Civ5 (another firaxis game) has an advanced option called "New Random Seed" which "Allows for different outcomes to random events (such as combat) when reloading a game.
    – horatio
    Oct 11, 2012 at 20:33
  • @horatio - right, some games do have a setting for that but X-com does not AFAIK. Not every game goes that route.
    – peacedog
    Oct 11, 2012 at 20:50
  • 1
    I agree. While the setting may not be exposed to users, it certainly points to it being a Firaxis default design philosophy (and probably a very useful debugging tool)
    – horatio
    Oct 11, 2012 at 20:55
3

At the begining of each game, a random kernel is generated. To put this simple, the game generates a more or less random line of numbers, ranging from 0-99. So lets say the line is 46, 58, 99, 0 and you are about to fire using your soldiers. The first shot will have to have better chance than 46% to hit. The second shot will need to be better than 58 and the third one will have to be a 100% hit. Fourth shot will be a hit even if it's a 1% chance to hit. The AI doesn't base its behavior on this. It will always take the most likely shot. The kernel will reload only if you restart the whole level. This method isn't any different than generating random numbers on the spot but players cannot cheat by reloading their saves. Well, technically, they can by performing different actions and getting different numbers, but they can't keep reloading the game until they hit all their shots.

-2

I had a similar problem where I didn't want to lose my best medic, she was the only medic I brought and got shot down while retreating.

After picking him off with my sniper I reloaded the save, and had the sniper try to shoot the muton first...and repeated that like 8-10 times missing every time.

Then I decided to try and see if I could get the game to seed me a new number by first shooting at the muton with my heavy, the heavy missed him too but right after that my sniper was able to take him down just like that snaps fingers.

But I still think the game is cheating, I miss 80% of the time for some reason, I can be standing on an elevated surface, right next to the guy, shooting and get all misses, CONSTANTLY. I had a Rocket fly off a full explosion-radius to the left and kill two of my own guys when the heavy had a 90% chance to hit during one mission. 3 missions ago I spent 15 turns trying to kill 2 Mutons when 6 guys focused their fire on them, nobody hit them for 14 rounds, I was lucky as he** nobody died (+my medic helped).

-2

Frankly, I think their percentage engine is meaningless, and they simply set some Boolean flag before each turn which decides if you hit or miss. I just found this question, because I wondered about the same issue on reloading. It is merely an 1/0 issue and not an 0 to 100 percentage question as they display it.

The most annoying thing is, even if the seed remains the same when you reload, improving your position (or trying to attack someone else) should somehow change it, but not in this game it wont. I spent 20 minutes now on trying different combos with reloading each time, and the game showing different hit possibility percentage, ranging from 40% to 95% and i missed at all the retakes I took. This leads me to the unavoidable conclusion their "seed" simply decides before the turn if the soldier is going to hit or miss in that turn, no matter what. For me, this ruins the game completely, but that's just me.

6
  • Your answer is not bad, but, could you put it in a way that doesn't makes it looks so much as an opinion ?
    – Michel
    Dec 9, 2012 at 22:14
  • i dont have a way to prove what im saying is true, except for maybe mailing Jake Solomon. But im not sure he will admit this is how it works, because then its really a pre-determined game and has nothing to do with our decisions as players
    – YavgenyP
    Dec 10, 2012 at 6:21
  • 1
    This answer is false. Game does not use boolean values for hitting. If you move, it only affects if your chance to hit changes. For example, if your chance to hit was 45% in one position and 45% in another, the same result will always happen. However, if you move to another place with higher than 45% chance to hit, say, 55% chance to hit, then there is a chance that the outcome might chance.
    – Mandemon
    Feb 18, 2013 at 22:55
  • @Mandemon, what you say is based on the numbers theyu show. what I said was based on my experience with the game. I think the percentage they show is a pure lie, but this goes back to the mathemtical question of what this number even means. In pure mathematical terms, a 90% chance to hit means that you will hit 90 out of 100 times. This is NOT what happens in the game, so basicly this number is meaningless. Even if to set the flag before you fire they used some advanced math to give you a "better chance", its meaningless once they've set it, which is what happens, according to the interviews.
    – YavgenyP
    Feb 19, 2013 at 12:38
  • 1
    Incorrect. 90% does not mean that 90 out of 100 always hits. You are using too small sample. Numerous studies have been made and they all conclude: RNG works as it should. Another thing is, that when you take the shot, you do not know what the roll will be. If it shows 90%, there is 90% chance of hitting. This is simple as it can be. "Seed" only matters if you keep reloading and repeating the same actions: This is done to prevent cheating. I have collected nearly 2000 shots and numbers do match up: docs.google.com/spreadsheet/…
    – Mandemon
    Feb 19, 2013 at 16:18

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .