*Author

Tobriand

  • Guest
Random Generator https://elementscommunity.org/forum/index.php?topic=43.msg1568#msg1568
« Reply #24 on: December 15, 2009, 10:09:33 pm »

Well... I don't know about the RNG being quite as bad as Oki has found it. But I must admit I've noticed two things...
1) Whenever I manage to get one of the miracles in my deck, I nearly always get the next within 3 cards.
2) It is MUCH easier to spin cards at 3am (2am GMT) than it is the rest of the day. Usually at this time of night, I'll get 1-2 cards from every fight (False Gods are VERY profitable to fight), where the rest of the day, I'll often go 6 fights or so before I get a single card...

oki

  • Guest
Random Generator https://elementscommunity.org/forum/index.php?topic=43.msg2025#msg2025
« Reply #25 on: December 15, 2009, 10:09:35 pm »

It's clearly a flawed implementation. I need at least one Ruttler in the first 10 cards, which far too often doesn't happen. Sometimes not even until the 15th card. I have 6 Ruttlers in the deck and a deck of only 30 cards, so I cannot optimize it further.


Tobriand

  • Guest
Random Generator https://elementscommunity.org/forum/index.php?topic=43.msg2026#msg2026
« Reply #26 on: December 15, 2009, 10:09:35 pm »

The other thing in terms of randomness... I recall that oddly enough, clumping is a sign of randomness rather than a sign against it. That is, in a truly random sample, it is far more likely that multiple similar units will clump together than it is that they'll be spread proportionally through the deck. If you think about this, it makes sense - statistics works only when applied over a massive number of iterations, after all. I'm not mathematically minded enough to try and give a full example, but it's probably not as surprising as it seems at first glance...

the sage

  • Guest
Random Generator https://elementscommunity.org/forum/index.php?topic=43.msg2865#msg2865
« Reply #27 on: December 15, 2009, 10:09:48 pm »

I agree that the random number generator appears to be flawed. The number of games in which I start with 0 pillars or novas is just way too high Since they make up 1/3 of my deck, the chance of the first 8 cards not being pillar or nova shoud be (2/3)^8th (ignoring the fact that it should get more likely with each next card)= 4%. The chance of getting only one is about (2/3)^7=6%. The number of games where I start with 0 or 1 nova or pillar is more like 40% than 10%.
A fix would be most welcome, as the initial number of pillars becomes a far too major predictor of who will win (if the game manages to finish without other gamechanging bugs, that is).

 

blarg: