Fire starting odds Fail

[ernestww]

Hi Hinterland,

Just tried to start a fire with an 80% chance of success three time and failed each time. I guess there could be mitigating factors that I don't know about, like encumbered, or fatigue. Or just very bad luck. But I have noticed Fails in other random rolls like crafting that seemed like unusually bad luck as well, but I would run out of metal scraps before concluding a bug. It's a bit like the roll value is made 1 - 100 but is not reset per action attempt. Or the algorithm is not truly random. Perhaps making the random roll the 1/100 of a second game time the actual mouse click happens on?

http://pastebin.com/w2y0pL2A

[joki]

Just tried to start a fire with an 80% chance of success three time and failed each time.

And you find this surprising? The first time you roll the dice, and expect to be successfull 4 out of 5 trials. You lose.

Then you roll the dice again, your chances are now still 4 out of 5. They aren't higher, because you think "i have just lost already, so this time it should work!". These are independant events. So you roll and you lose again. You are surprised, but roll the dice a third time. For this third roll you have a (no surprise) a chance of 80%. It is not higher now, because you already lost two times and expece the universe to "make up for it". The universe doesn't care, that you have just lost twice. If you however try it 10 million times than you should be - more or less - 8 million times successfull.

Or just very bad luck.

You could call it like that, if you like.

Or the algorithm is not truly random. Perhaps making the random roll the 1/100 of a second game time the actual mouse click happens on?

As with all non-security critical random number generators, a PRNG (pseudo random name generator) is used. How exactly it works is an implementation detail (usually the numbers are generated of a timestamp, since how long ago you started your system, the last access time of files or of user input data, since mouse coordinates over a period of time).

If you want "true" random numbers (like banks, online gambling sites, and research companies use) you would need to buy special hardware, that generates numbers by measuring environmental data, such as picking up noises from a microphone or measuring temperatures, light reflection of surfaces and such.

[ernestww]