Seems unfair, Gauntlet
![[NDQ]Azzi](https://us.v-cdn.net/6030345/uploads/defaultavatar/nL5515UJH9DYJ.jpg)
[NDQ]Azzi
✭✭
in The Bridge
Yes yes, RNG is a fickle mistress, I know.
12
This discussion has been closed.
Comments
Jim
She loves me.
She loves me not.
Fleet mate ranks 5th, Banjo man.
Happy for him.
Annoyed at system
Don't be sad! You would have lost even with 0 crits from opponent.
See it as 6 crits, that won't come in other matchups, where they would taken your win.
Just look at these almost twin rolls: 172, 170 or 223, 209 or 184, 189 or 298, 297. This RNG is a disaster. It's completely unbelievable.
Again, improbable twin rolls: 302, 303 or 263, 272. DB uses very poor quality RNG.
https://betterexplained.com/articles/understanding-the-birthday-paradox/
Right? Why is the RNG so unpredictable?
One? Maybe. But four in one duel?
Is that what your point was? Seemed to me you were saying the RNG is bad because the rolls shouldn't be close. In other words, an unexpected result - close rolls - indicates poor randomness.
So, where do you draw the line at "similar"? You're claiming a 209 and 223 roll are "similar. If we're saying +/- 20 on any given roll, then, for someone like Valeris who has a 250-500 span, there is a 41/250 -> 16.4% chance that any two given rolls are "similar". That is NOT a small probability. If you want to actually prove this, you will need more than a handful of screenshots and tinfoil hat eye squinting.
NO, I said too many closed rolls is what defines bad RNG. You got 4 twin rolls in one duel. And this is not an exception. You can observe such rolls all the time.
See: Birthday paradox description I linked earlier. With a loose enough definition, you can find any pattern you want. You just look like you're desperately grasping for something to be wrong you're just making up patterns that aren't really there.
I had this one teacher for three different programming courses in college who ran that birthday paradox survey in each class, each quarter. Not one single class ever had all unique birthdays. There was always at least 2 pairs of people who had the same birthday. And this was a class of 20-25 people each time.
Nice try but there are still three other pairs there with delta which equals 1, 2 and 5...
You on the other hand desperately try to justify/dismiss this pretty obvious problem. DB clearly has some problem with RNG seed.
Did I say that?
The seed would have nothing to do with the problem you are stating. The seed is just used to start the generator.
Here number of rolls was only 9, not 20-25.
And the rolls were also not duplicated.
Real randomness cares not for streaks and distributions and so forth. It just is what it is. Designed randomness has to carefully evaluate the presence of both, and more. Distribution is all fine and dandy, but many designers/users seem to forget to evaluate the streakiness.
No but the criteria selected above is much less than 1 in 365. Even if you class a twin roll as being +/-1 we’re talking approx 1:50 (depending on the character). Apply the birthday paradox logic and it becomes incredibly common. Common enough to hit human confirmation bias.
contrary to popular belief the game is not out to get certain individuals...
... except Kortar, who given I never beat in the gauntlet even with vastly superior crew must have it in for me personally.
For instance, there could be a single random generator feeding the entire game. Let's say it is spitting out an unsigned 64-bit numbers. Well that number is typically truncated down to a domain using a modulo operation (remainder of division). There are many numbers that will reduce down to that same value between 0 and 18,446,744,073,709,551,615 (assuming an unsigned 64-bit number). So
873,457,447 % 255 = 82
1,008,841,159,792 % 255 = 82
The rolls themselves are VERY far apart. But after some math, they end up the same.
We also don't know if it grabs 12 numbers all at once, or if they are grabbed individually between other requests for numbers. This could also introduce more entropy and mean the numbers are not actually as sequential as they appear. There could be thousands of random number requests between the values you see.
Exactly, too many cases of subsequent extremely similar rolls shows something smells bad. The key part is "too many". Observe your own rolls and you will see it too.