Cross's Hit % is a liar.

[kamrom]

Ive found that if you select a 2-stamina attack and it misses, it almost always misses again if you select the 2-stamina attack again. This happens far far more often than could possibly be coincidence. Often times, even the *third* will miss. Has anyone else noticed this? Because it really seems like a mistake. 

[Tactless]

an attack with 67% hit rate just happens to hit a lot less than 67% of the time. i'm sure it's not completely accurate

[utunnels]

Didn't notice that. But people tends to notice failure than successful hit, I just assume. Unless you prove that by trying 100 times, but even that is not enough, because random factor on computer is different than those in real life. In Chrono Cross, the random factor is a pre-generated array(and it never changes in game), the game picks a value from the array every time and move pointer to next slot, so if you are using a save state or something like that, the result is doubtful.

It is possible you miss 4 times in a row and land next 6 htis successfully. 

[Kodokami]

Yeah, sometimes it seems like a total liar. Like when I had just used Eagle Eye, and still missed a 99% hit. Of course, it was still a possibility, and we do often notice failure more than success.

Just yesterday, I fought the Wingapede (much farther in the game than usual, mind you). I started with two fierce attacks from Norris. Both hit, both crit; battle over. What are the chances of that?

[alfadorredux]

More accurately, you will never see true random numbers in a video game, only what programmers call pseudorandom numbers, created by a generator whose output is predictable if you know how the guts work. In applications like cryptography, where true randomness is important, it isn't generated programatically, but taken from a source not controlled by the computer: anything from atomic decay for the highest-end applications to the user's mouse movements across the screen for low-end ones. For games, that would be overkill (and cause problems if the pool of true random numbers were exhausted, which can happen periodically).

Judging from utunnels' description, Cross's random number generator is more than usually crappy, though. If we could plug in a nice Mersenne Twister or similar to replace it, the results would be better distributed.

[utunnels]

Heh, I just assume. And found I was not right on that.

Someone on gamefaqs has figured out how exactly it works: http://www.gamefaqs.com/boards/196917-chrono-cross/55191996
I think I'll have a headache if I read them over. 


Sometimes you see a random roguelike player complain his character misses a zombie or a legendary dodger killed by a nomad from a single critical hit. Then you know how random number tricks you.

[alfadorredux]

@utunnels: Judging from that post, it looks like you were half right—it's just that the table is supplemented by the standard C RNG, which is better but not great and can be gamed. (I remember hearing of a guy who was able to time the start of his Nethack games to ensure the generation of a Wand of Wishing—normally a rare item—on the first level of the dungeon, just by gaming the RNG. He must have been a pathological case, though.)

Fact is, though, randomness, like statistics, is counterintuitive, and the average player will see fair random number generation, accurately reported, as slanted against him. I wouldn't be surprised if that isn't why Squeenix put that table in there in the first place: to allow them to slant the results to make them look "more fair" (although they obviously didn't do a perfect job).

[Synchronization]

1