RNG weighted in favor of fumbles? 08/11/2016 09:16 PM CDT
Has anyone ever done some RNG research on various RNG applications?

Because, this happens to me a LOT:

You channel at a cold guardian.
[Spell Hindrance for some elven fighting leathers is 2% with current Armor Use skill, d100= 1]
Your armor prevents the spell from working correctly.
Cast Roundtime 3 Seconds.

You channel at a cold guardian.
d100 == 1 FUMBLE!
Cast Roundtime 3 Seconds.
Reply
Re: RNG weighted in favor of fumbles? 08/11/2016 11:03 PM CDT
Nope. And I doubt it is biased.

The chance of two 1's in a row is 1/10,000, but that's not what happened: you didn't hit armor in between. The chance of two 1's in three rolls is almost three times as likely.

Beyond that, when you roll a 1, you think, "Ugh, I'm going to roll another 1 with this darn biased RNG."

But here's the thing: you start looking for that other fumble after the first one. So in your case, you had a 3% chance to feel like the RNG was biased.

First, a 2% chance to hit your armor a second time. Then another 1% chance to fumble the actual cast. 2.98% chance of hitting one or the other. That's not so unusual.
Reply
Re: RNG weighted in favor of fumbles? 08/11/2016 11:42 PM CDT
I was more referring to the fact that I see several rolls of "1" in a 15 minute hunt. Figuring 3 second cast RT's * 15 minutes = 300 casts (high estimate since there is also looting, storing items and waiting for critters to gen). That should be on average 3 rolls of 1. I am consistently seeing at least double that, if not more. Swearing starts after the 3rd one. ;-)
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 09:04 AM CDT
My worst run ever was back in miners, shortly after we deICEd. Major Repel Undead was still the 25th level spell in the Cleric list, and it had three (3) possible results:
Roll nets to 100 or less, you blow the full 25 mana;
roll is 101-149, creature runs to another room [unless there is NO exit, in which case it dies] and you pay 0 mana;
roll is 150+, critter dies [search fast!] and you pay 1 mana.

Most of the time, you spent it with nearly-full mana because if the spell worked at all it cost you 1 at most, if it worked sort-of-a-little-bit you just had to chase the critter down and try to kill it again.

At 42nd level I had 109 mana. Between 1==Fumble and armor hindrance (I think I was at 5%?), I blew through my entire head of mana, Wracked, blew through all of that, Wracked again (dwarf recovery!), and blew through about half of that load before killing it. Probably only 2-4 of them were armor, the rest were Fumbles (and at least a few of them were back to back).
Stuff drains away fast when you're using it 25 at a time.

What's that, ten or eleven rolls?
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 09:08 AM CDT
And I finally took a page from someone else who commented on the RNG, and put colored highlights on
+96
+97
+98
+99
+100

Sure, every and then an AS or DS number will work out to be exactly that, but for the most part I just see the top 5% of the die-rolls.

They really do happen a lot more frequently--even back-to-back--than I had given them credit for. You just don't remember the good luck, when stuff dies easily. :)
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 09:19 AM CDT
>And I finally took a page from someone else who commented on the RNG, and put colored highlights on
>+96
>+97
>+98
?+99
>+100

I've had 85 to 100 highlighted since the old autoroller days.

So, is there a d100 to determine if you hit your armor, then another to see if you fumble, followed by the actual attack roll? If that's the case, if I don't fumble, the rolls come out 100, 100, 1!

::sigh::
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 09:50 AM CDT
You're still ignoring the fact that hindrance is a separate roll, so there's a minimum of two rolls per cast, not 1. More if you're casting AOEs, since each CS roll of the AOE could fumble.

Statistics can be weird. With 2% armor hindrance, casting only single-target spells, you have a 2.98% chance to fail a cast. But you have over a 50% chance of failing at least 1 spell after only 23 casts.

Beyond all that, you're seeing the clustering illusion and experiencing confirmation bias. The RNG, being random, will have times where it has clusters of low rolls. You see those times, but not the times it has clusters of rolls in the 65-66 range, or the 99-100 range, because the game draws more attention to them, because they have a more dramatic effect, and because you're looking for them.

To counter anecdote with anecdote, my warrior uses feint probably once per target he kills, and therefore uses open rolls frequently. I see dramatic results on rolls on both sides: in the 1-5 range and the 96-100 range. If fumbles were more likely, I'd see far more negative open rolls than >100 open rolls, but that isn't the case.
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 09:58 AM CDT
I found a solution...

99 Sided Dice
No more 1's!

Fumbles got you down? Critters teasing your clumsiness? Rub this die twice and show them who's boss with renewed confidence! These magical dice will eliminate the number one evil from your hunts for 9 minutes at a time! 9 to a bag! Buy as many as you like!

Appearance: "A 99 Sided Golden Die"
(Only works on attacking rolls against critters)

Discounted for a limited time!

999 Simucoins!
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 10:04 AM CDT
I should add that while I know my probability, I'm not an expert. I studied a few dozen hours for Exam 1/P[1] 15 years ago but chose not to take it, because it's just not my thing. If anyone has more of a background in it, feel free to correct whatever mistakes I made.

[1] The first test along the way to become an actuary. 3 hours long, 40% pass rate, and the rule of thumb is study 100 hours per hour of test. https://www.soa.org/education/exam-req/edu-exam-p-detail.aspx https://www.soa.org/Files/Edu/edu-exam-p-sample-quest.pdf
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 10:13 AM CDT
>Appearance: "A 99 Sided Golden Die"
(Only works on attacking rolls against critters)

The login rewards luck boost beats that. It gives you a dice with only the best 90 sides.
Reply
Re: RNG weighted in favor of fumbles? 08/12/2016 10:23 AM CDT
>The login rewards luck boost beats that. It gives you a dice with only the best 90 sides.

Oi! If they sold those boosters in the Simucoin store, I'd be broke.
Reply