Poll About RNG (Random Number Generator)

[King_Momo]

I've been playing this game since it was released last April and although it has provided me a lot of fun, the rolls have always seemed not quite right to me. I've played the actual card game (finished RotR and SnS) and in my experience, I've never rolled so many 1's in both playthroughs combined as I have in a single playthrough on the digital game.

 

Now, I'm tempted to actually record the numbers of minimum rolls and maximum rolls across several games but the dice disappear so quickly that it's not always easy to see what you've rolled exactly. If anyone is interested and has time on their hands, perhaps they'd like to try this and  share their findings?

 

What I'm trying to say is that  I enjoy this game and I've tried to get my tabletop buddies and gamer friends to like it too. They've all tried it but those that  don't like it say it feels like there are 2 layers they are fighting against - the actual game and the system itself. They feel that the mechanics of the game are designed to make them fail more regularly than in an organic game, that there's another bug in the RNG. Some of them even suggest Obsidian has deliberately programmed it this way to generate a need for players to buy more chests to make their decks better.

 

Regardless if the conspiracy holds any truth, the most common gripe I've heard about the game outside of bugs and release timings is that  the RNG in this game has left a very bad taste in their mouths and the rolls just don't feel right. What do you guys think?

[Hannibal_PJV]

I generally get much better rolls in the app than in the real tabletop version... Maybe Paizo has a conspirasy with dice makers so that They can sell better dice to me

Nah... I think that They Are the same, but I remember the bad rolls from the tabletop version better, because They Are more important to the team. When I play the app, the rolls has less meaning, because there is not that sosial aspect. But I definitely feel that I have newer rolled so badly in the app than I have in the real game. Maybe because I have played the tabletop version so much more than the app, so there is Also more those bad rolls Also.

Edited January 3, 2017 by Hannibal_PJV

[StormbringerGT]

This came up months ago so we had our programmer run simulations for a while. Long story short they are about as fair as fair can be.

In my experiences I have horrible luck in the app, but I have horrible luck outside the app as well (People delight in playing dice based games against me).
 

[Chris_R]

'Fession time- playing the app helped me to confront how often I 'fudge' dice rolls in the solo tabletop game. My Conclusion? We need a 'nudge the table' button on the app.

 

That said, I do get long runs of bad results on the app and I do feel close rolls fail more often than I feel they should. I have often considered doing a run and cataloguing the dice rolls. But on the other hand, most rolls in the game come under 3 categories

 

- no chance of passing, just take the lumps (don't even bother looking at the dice here )

- essentially a dead cert (don't look at the dice much here)

- should make it but might not.

 

Because of risk/reward in the app, there's little incentive to take a punt on longshots (say, where you need 7 on a d8) - either you don't care if it comes up or you get some help from a blessing or a spell. So most of the rolls we actually pay much attention to are ones where we really might fail. Unsurprisingly some fail and often, because of the nature of these rolls we have committed resources or they genuinely matter. So those failures sting.

Edited January 3, 2017 by Chris_R

[Ripe]

Rolls are mostly fine but card shuffler have some issues... I can't remember how many times I evaded something with Merisiel only to have it be a top card of the deck. And having to face Enchantress five times in a row on a same turn at General Store (with 8 cards in a location deck!) was rather fun....

[Chris_R]

Rolls are mostly fine but card shuffler have some issues... I can't remember how many times I evaded something with Merisiel only to have it be a top card of the deck. And having to face Enchantress five times in a row on a same turn at General Store (with 8 cards in a location deck!) was rather fun....

Yes, I'm usually confident that I can shuffle chests back into the deck and then move Meri back to the location secure in the knowledge it'll show up next card up.

[Dunesparrow]

Hey there

 

I am on my third play through with six characters and I have failed rolls with a high expected success rate and succeeded at rolls with a low expected success rate.

 

Some banes and boons have shuffled back to the top and others have been spread throughout the deck

 

Sometimes I have had all villains and henchmen in the top 3 cards of the 8 location decks, other times spread out and some all toward the bottom of the decks.

 

I expected that the dice and shuffling was fair because it was random and this has been borne out in my experience.

[Ethics Gradient]

I'm with @StormbringerGT on this one.  The part of the game that does random things randomly appears pretty solid.  

Random numbers are pulled out of the air by a Unity function that is a xorshift128 variant and shouldn't see noticeable loops until 2¹²⁸ rolls.  The shuffler is Fisher-Yates Algorithm, which is practically a Classic Car of computer programing.

 

If there is a debate to be had, it is better framed as: "under what circumstances could the game go *burp* and screw up all these high-quality random numbers it made?" not simply "is the RNG bugged?"

 

I was curious about stuff like this for a while, and tried a few practical tests.  For example, I loaded out Ezren with 8 Augries and Augured the heck out of a location deck (once he gets in the zone, he can keep going almost indefinitely).  Things came up different enough each time that I wasn't terribly concerned that there was a systematic problem.

 

I haven't yet run into anything that is both suspicious and reproducible.  Anyone found any potential RNG quirks that can easily be reproduced?

 

#TeamRandom

[Thyraxus]

Fisher-Yates may be "classic" algorithm, but that doesn't mean you don't have to be careful when implementing it. For example, a little, easy to make (and easy to miss when looking over the code) off-by-one error when picking the index for the card to be swapped in each iteration can completely skew the results (reference for the interested here).

Edited January 4, 2017 by Thyraxus

[Amentep]

'Fession time- playing the app helped me to confront how often I 'fudge' dice rolls in the solo tabletop game. My Conclusion? We need a 'nudge the table' button on the app.

I don't fudge dice results playing solo tabletop so much as I play on an uneven surface and sometimes it takes a re-roll to get a legitimate roll not interfered with by outside physical influencers or to resolve situations where there is no clear "face up" on the die.

 

Mind you trying to playtest a self-created character to see if they're balanced and comparable to the existing ones is a good cure for fudging (most) rolls.

 

EDIT:

 

As to the app, I find that I seem to hit hot/cold streaks comparable to real life with rolls so it seems fine to me.  Sometimes you can't buy a success and other times you can't mess up a success not matter how you try.

[mrmarcus]

I wonder of late. I'm having huge fits getting past Jorgenfist Heroic due to the boss always seeming to roll 1s after I beat him at least twice per attempt.

[Ethics Gradient]

Fisher-Yates may be "classic" algorithm, but that doesn't mean you don't have to be careful when implementing it. 

Exactly.  That is one of the reasons I had Ezren Augur a deck over and over again.  I wanted to rule out something like a derangement shuffle (aka Sattolo).  Implementation otherwise feels right, and PRNG space is more than enough to handle shuffling all the tiny decks.  It is hard to rule out modulo biases, because checking for that would be a bear from the user-end.  You'd have to pry that bit of code out of the app and check the results after a few million runs.  In the end, I still feel the game shuffles/rolls at least as well as humans would do so playing tabletop.

 

Though, going back to my other post, what I find interesting is where the game has a random number or shuffled array fresh from the RNG, then something silly just happens to it.  

 

A somewhat recent example is this bug.  There was a period of time where you drag some cards from your deck and play them as if it were your hand (fun way to cheat).  But the down-side was that touching your cards would de-shuffle your deck and permanently sort it in order by type (Ally, Armor, Blessing, etc...).  

 

Maybe the reason I never thought that the RNG is rigged against me is that I just happen to play the game in such a way that it doesn't consistently happen.  I'm not saying that anyone is doing things "wrong," but how we tap, drag, and poke the UI may have more effect than we think.  

 

Some people are convinced that Evaded monsters stay on the top of a location, others see otherwise.  It doesn't seem too outlandish if it somehow came down how you interact with the UI.

[Sines314]

This came up months ago so we had our programmer run simulations for a while. Long story short they are about as fair as fair can be.

 

In my experiences I have horrible luck in the app, but I have horrible luck outside the app as well (People delight in playing dice based games against me).

 

 

There was a period in my life where I had such terrible dice rolls, I asked the DM if I could treat all my d20 rolls as being equal to 21 - the face value, so that a roll of 6 was a 15. This did, quite strangely, work. I had much better rolls for a while. I've since dropped it and things seem back to normal. I'm not a superstitious man, and I'm still using the same dice, so I have no idea what happened there. But hey, I figured that there was no harm in doing that, as that method would result in exactly the same spread on a fair die anyway :D

 

And as far as this game goes, I seem to be having about the right amount of luck. Rolling 1s on two or more dice at once is painful and memorable, but I don't think it happens more often than it should.

 

Still, it's also possible the OP isn't imagining it. For every player with an amazing streak of luck, there's going to be some poor guy who has constant bad rolls. You could legitimately be rolling regularly low simply because the dice gods have decreed it. The good news is that the dice gods are by nature capricious, and that this could turn around at any moment. Either that, or program a mod to flip the dice rolls, in order to emulate my solution ^_^

[aznxknight]

I rolled 5x dice for disintegrate and four 1s and one 2 came up I shoulda screenshotted it . 

I would like to say yes but come on... it doesn't mean it's not likely to come up. Just because you failed doesn't mean the game is against you. Think about all the times you did succeed though. Now if you put your role as the evil dungeonmaster, you'd have the same reaction.

[Effusion]

This topic comes up regularly in any RNG based game. Unfortunately, humans have evolved to be bad at estimating statistics thanks to a negative bias which causes negative outcomes stand out more to us than positive ones. This means that most of the time when a system is fair it will feel unfair. I think games tend to amplify this when the player has the agency to avoid situations where they're unlikely to succeed (eg, using blessings instead of hoping for a 15 on 3d6) because this causes them to experience more unlikely failures than unlikely successes. It's quite difficult to create a system which feels fair without actually tipping the scales heavily in the player's favor.

 

In this game I deal with it by ignoring average rolls and just focusing on minimum rolls for anything important. Aside from rolling the occasional 0 on a d10+2, this works pretty well for me.

Edited January 5, 2017 by Effusion

[Sines314]

I think this game is prone to making the player think they've got bad luck more than other games. There is pretty much no reason to care about rolling high. With a few exceptions, meeting the DC, and exceeding it by 20 have exactly the same results. As such, we never really 'log' the really high rolls, because they don't matter. Likewise, if a roll needs something like an 8 on a d8, we're not likely to remember that roll either, as if it were important enough for us to care a lot about succeeding, we'd be spending some blessings or other buffs on it.

 

Contrast with, for example, XCOM. For every important 90% shot you miss, you're going to score that 10% critical on a key target, or make a 50/50 shot and roll max damage. In Pathfinder Adventures, there are simply far fewer situations where the game swings on a very lucky roll, than on a very unlucky roll. So we're even more inclined than usual to count the losses more than the hits.

At the very least, that's my experience playing a larger party. Perhaps smaller parties with fewer blessings and skills to go around are more likely to gamble on low probability actions. But even then, there's still the fact that a high damage roll will almost always be forgotten, since it doesn't matter.

Edited January 5, 2017 by Sines314

[bigremy]

I was looking for a discussion such as this.  For the vast majority of the game, I have had no complaints over the dice roller and RNG.

 

Until I got to 6-3: Scaling Mhar Massif.  I've played through that quest about 15 plus times now, and still have not beat it.  I will grant my team is likely not optimized, but going purely on dice rolls and type of dice (d4 vs d6 vs d8 etc etc) I have found that a larger than average percentage of time 1/3 to 1/2 of my dice rolls a 1 or 2.  Maybe I just notice it more because of the condition that makes those zeroes, but when I consistently am rolling 4-5 dice because of blessings and other effects, primarily d8 or above, and consistently two to three of those dice return a 1 or 2 every single roll it begins to feel like more than just bad luck.