RNG with regard to reshuffling locations

[PinkRose]

I feel that this should be looked at.

I can give example numbers on my next run through, but my anecdotal evidence is to high to ignore.

My experience is that when a card gets reshuffled into a deck, usually location, it ends up back on top.

This goes for barriers, henchmen and villains.

I know others have the same experience which is why it can't be random.

Please look into this.

If you do, my next post will be creative and witty, I promise.

[Srvannor]

Ya, I farm quite a bit with Meri. My experience with the general store has been the same. When I evade a monster the deck gets shuffled, I get to redraw (because of the general store trait), and VERY often that same monster is right back on top of the deck. Fortunately, I can just keep evading forever. But literally out of a deck with 10 cards left, I'll see that same monster on top 4-5x in a row. This is of course somewhat anecdotal - I haven't kept exact count.

[Hannibal_PJV]

Well I have been more lucky in that aspect. I have seen the same card in the top of the deck very seldom. Sometimes it happens, but I have seen it happening only few times.

[Srvannor]

Ya I dunno what the actual data is, and it doesn't seem to happen 'every' time. It does seem to happen enough to make me raise an eyebrow and wonder if it's really random or not.

[Bampop]

It doesn't happen when I want it to. I used Augury to find the villian at the Prison early on. I moved Meri there to be healed by whomever was already there and then explored like a damn fool. (I still had locations to close and Meri had no way to fight at the time) Evaded and never found the villian again.

Edited June 5, 2016 by Bampop

[Borissimo]

I have a feeling this one might be confirmation bias. When you find the same monster again, it sticks in your memory; when you don't, you take it for granted. If you've already explored 4-5 cards out of the deck, the odds of finding the monster (in the general store as Meri, for instance) 4 times in a row are 1/125 or 1/216, which is small but will definitely happen to everybody at some point. In one of my videos, I failed a roll that I had only something like a 1/700+ chance of failing -- that stuff will happen!

 

Edit: I just realized my post may look as though I'm dismissing everyone's claims. I promise I'm not! It's just that human brains are so bad at estimating true probability that game developers are unlikely to ever do anything in matters such as these based on anecdotal evidence alone. There may well be a bug with the shuffler in some instances, but to convince the devs, you'll need hard numbers.

 

I remember back when Hearthstone first came out, someone proved that a card with a 50% chance effect was only triggering 25% of the time by tracking several hundred instances and doing a chi-squared analysis. You'd think that if something was happening half as often as it should be, it would be uncovered as a bug right away, but despite everyone's anecdotal complaints about the card, the devs didn't do anything until the hard numbers came in. And unfortunately, the devs were absolutely right in how they handled the issue -- there are so many false complaints about shufflers and randomness in online games that as a dev you kind of just have to ignore them all until you're presented with something more convincing than human intuition.

Edited June 5, 2016 by Borissimo

[Hannibal_PJV]

And things happens. We way playing tabletop version of this and one player did throw three times double 1:ns one after another she was sure that she was cursed in that day ;-)

[PinkRose]

I totally agree that there could be confirmation bias, but I was tracking a Sneak with Meriseal, and the odds were found on top on stack of 7, top on stack of 6, Medusa Mask, Top on stack of 4.

That's 1 in 24, I think. 

And that happens more than 1in 24 times.

[hfm]

I recall in the tabletop version we kept hitting the same monster after shuffling, turned out there was more than one of them in the location deck.

[codepoetz]

It is not uncommon for software-based random number generators (RNG) to exhibit some kind of bias. A few weeks ago, I ran a battery of statistical tests on our random number generation code looking for any kind of bias that might contribute to strange die rolls. After analyzing a million cycles, I found no evidence of statistical bias at all. In fact, the base RNG that we are using behaved extremely well under pressure. However, my investigation was in the context of die rolls. Card shuffling is a bit different, so I promise to run some more focused tests on that soon. I appreciate the feedback because I'm very curious about how game players perceive randomness and it's one of those things that I can't evaluate independently.

[Jehren]

Regarding dice rolls: Wouldn't a bias in the player's favor be a good thing, as opposed to no bias?  Actual random results will usually feel unfair.  I seem to remember reading a gamasutra article about players needing something like a 60-70% win rate or else they felt the computer was cheating.  Though I also remember reading that you guys are tweaking die rolls, so I guess you're keeping the generation pure and just creating a bias after the fact?

[zeroth_hour]

I actually saw no issues with location randomness when I was using Augury (and I use Augury a lot).

 

I did have a few "3 henchmen in a row on top" stuff but those streaks are part of randomness.

 

Tweaking die rolls only happens during the tutorial and nowhere else.

[Hastur]

I've had the same views regarding barriers, but I always wonder if that's merely bias in my perception. I'm usually careful with that, though, and I would be curious how the numbers came out for evading barriers because I seem to often have the outcome of seeing the barrier as the next pull after evasion. Typically, I just plow through stuff, but even if I use an ability rarely, I'd still like to have it work the way I'd expect. Like most things, though, I'm pretty much for giving the developers the benefit of the doubt.

[elcoderdude]

I haven't kept track, but I definitely notice when the evaded card ends back on top, and pay no attention when it doesn't.

[ikarinokami]

I notice this all the time with regards to monsters or barriers being reshuffled into the deck either because they were undefeated or evaded. I always assume the evaded monster is going to come up next or undefeated barrier is going to come up next. I don't think it's confirmation bias either. The dice rolling is pretty good no issues there that I have seen.