Anyone else finding the randomness to not be so random?

[Maebon]

I realize that with true randomness you can see apparent patterns and I may have just been extremely unlucky, so I want to see if other people have noticed anything similar.

 

I usually play one or two adventures on quest mode a night with a 6 person party. Typically I see the same locations, and I mean the exact same 6 locations, game after game. Sometimes I'll get 1 or 2 new ones, but often enough it's the same 6 from the game before. I frequently get streaks of the same mastermind as well.

 

I've also noticed that if I fail against a monster or a trap it tends to get shuffled back into the top 2 cards of the deck. I just had a very unfortunate game where I failed against the Zombie Horde card 4 times (with 6 characters someone will probably get unlucky with their check) and each time it was shuffled back as either the top card or right under the top card. For this to happen 4 times in a deck of 6+ cards is very unlikely.

Then the mastermind was the guy where after the check you roll a d6 and on a 1 or 2 start the check over. Apparently that roll happens even if you would lose the check, because he rolled a 1 five times in a row. Not a 1 or a 2, he kept rolling '1' and I thought the game was stuck (was long out of cards to add to the check). This is of course after rolling a 1 the previous round as well and failing the second check.

 

So as I said, either I'm just extremely unlucky and hit the 1 in 7776 chance of rolling a 1 five times in a row, or something is off in the randomizer. (It was also a stage where he got stronger for each henchman defeated. I didn't stand a chance.)

 

Anyone else experiencing anything similar?

[Pidesco]

Everyone that has ever played a game with a RNG has experienced this. It's usually all in your head.

[RedPred]

Personally I have the same problem. The auto-shuffle of the location deck really love to reset on the top the same monster or trap.

But if we are just you and me among thousands of players it could be.

Otherwise is an algorithm lack and other players doesn't take care of it (or think it's just in their heads).

 

I don't play a lot Quest lately, but if I understood well how it works, that game mode unlock specific tier at specific level and it could be also exclusive. In other words, after the level 10 I didn't meet anymore the previous villains but only the ones in the tier 1 group. So it could be for the locations for induction (every released adventure has its own locations with something new).

 

So I preferred to play Story mode to have higher variety in villains and location and loot collection. Also I tried easily different builds for some characters, cause in 18 scenarios you can fully rebuild any character you have. In Quest mode with 18 scenarios you get at most a couple of levels and 1 feat unlock (after level 15 and easy difficulty, cause I don't have good equipment) ... and cards from the same tier (the improvement in my decks there is quite slower than the amazing collection among all the adventures and their tiers of Story mode).

[Ethics Gradient]

So as I said, either I'm just extremely unlucky and hit the 1 in 7776 chance of rolling a 1 five times in a row, or something is off in the randomizer. (It was also a stage where he got stronger for each henchman defeated. I didn't stand a chance.)

Bad luck, sorry.  

 

Rolls appear to be handled by some pretty simple and standard Unity random functions.  It may not be be "cryptographic" randomness, but we're just throwing dice here, not securing financial data.  Considering how heavily this game relies on random numbers, if the "randomizer" were truly broken, much weirder problems would manifest than the occasional reports of five-in-a-row rolls.

 

I also sometimes feel like more stuff gets shuffled to the top of a stack than seems normal, and it is one of the things I've been trying to analyze how digital Pathfinder handles.  It appears to work, but I haven't quite mapped out how or why.  Too many references referencing references referencing references _{​referencing references...}, and I've got a limit on how much assembly language I will load into my brain at any given time .

 

Anyway, despite all the reported bugs over the last couple months, I still give a lot of credit to the devs for making it work as well as it does.  The game mechanics appear straightforward if you've played it tabletop, but from a programming standpoint, every other card seems to require special handling or a rules exemption.  Compared to other digital card games like Hearthstone, which was designed for digital from the ground-up, the Pathfinder rulebook constantly assumes there are a couple humans playing the game that can parse the plain-language text on the cards and find a way to balance that against whatever's going on in the scenario.

[RedPred]

I have an admission to do. There are a couple of situations where I use some issues at my advantage (exploit). I see the limitation, I understand the back-end (without reverse engineering) and I use specific sequence of iterations to avoid some bugs.. That to say: there are mechanics where different and equivalent sequences produce a different output and they wouldn't. I still see a beta version.

 

Then it's only matter of how many programmers you have on the job, how many hours they works on a project, and how many projects they work on.

[HPL_fan]

For checks I usually find the mean of the dice I have laid out and roll if the mean is equal to or greater than the check.  For important checks (like closing a location) I'll add another gratuitous die just to stack the odds in my favor.  The odd thing is, working from the mean means that I should get that number or higher only 50% of the time.  My anecdotal experience says that works more like 80% of the time.  That defies an even distribution of die rolls. 

 

Anecdotal, I know, but it wins scenarios.  Also - please don't 'fix' this if it isn't intentional  

[saxmaam]

If the expected value of a roll is 4.5, then there is a 50% change of rolling five or more. But if the expected value is 5.0, the odds of rolling a 5 are better than 50% because the result is inclusive of the "exactly five" condition.

 

5.0 is the expected value (or mean) of rolling 2d4. There are just 16 possible rolls, so it is not hard to prove this to yourself.

Edited July 25, 2016 by saxmaam

[Ethics Gradient]

My anecdotal experience says that works more like 80% of the time.  That defies an even distribution of die rolls. 

 

Anecdotal, I know, but it wins scenarios.  Also - please don't 'fix' this if it isn't intentional  

I know it sounds obvious to say, but there's no indication that the game fudges dice rolls in your favor.  Everything is pretty clearly defined.  For instance, a "d12" is boring old random number from 1 to 12 (inclusive).  

 

It would be cool if the game gave you the odds for each check you're attempting.  That way, when you feel like you're on a hot streak you'd actually find out the 3d6+4 you're throwing at a Bunyip really has a 95% success rate.

[Bajie]

List of random things seemingly not random or rolls designed to screw me (my play experience on questmodes)

 

1. That Boss that re-rolls on 1 and 2 on a d6?! Yeah 6 in a row..and I havent even trapped him..needless to say, all blessings were used.

2. Just about almost every closing check where I can only fail at a 1..at higher level, i double up on dice when this happen!

3. Yeah...locations..specially on legendary..I get random early on..but once Grind on..it seems it stops on certain combination of location and get stuck there..

4. The Spectre that always get shuffle on top of the deck because I don't have any magic/divine atm!

 

But I agree..most of these are just in my head..just exaggerated by frustrations specially when things don't go my way. Something like worse comes in the worst possible time or anything that could go wrong, will go wrong ^^

[RedPred]

If what you say is true at point 1:

[1,2] on d6 is 1/3 chance,

6 in a row is (1/3)^6 = 1/729 chance = 0.137 % chance

 

With all 'you have in your head' you could win the lottery!

 

Anyway I m taking statistics now: shuffled location deck with changed top card vs unchanged ones. Also with number of cards to have a parametric function, so I flushed old data. But I m not paranoid, just don't spy my data!

 

For the rest honestly it's seams normal casual extraction. I also got [1,1,1,1] rolling 4d6 (1/1296 chance) once, but I have thousands of rolls.

[Hannibal_PJV]

Yes, has done that with real dice too... They just hate you sometimes )

[jcalton]

The good thing is with a game like this, you don't need to wonder if there is intent behind certain RNG events.

Obviously this is already a board game with the same rolls, and it's not really pay to play. It's not free but it's also not one of "those" games.

[RedPred]

Personally I thought also to buy the game, because it's exactly the type of game I was looking for since years.

What stopped me:

- really high prize for an Android game (there are old full pc/console games at 10€, or recent pc games at 30€)

- the fact I can't avoid to observe the game still as a beta release (too many bugs and instabilities) with slow fix rate.

 

Anyway the RNG is a thing, the re-mapping of the numerical sequence on every procedure that uses it is a different thing. In other words, the generation problem is different from the re-ordering problem.

[Ethics Gradient]

Anyway the RNG is a thing, the re-mapping of the numerical sequence on every procedure that uses it is a different thing. In other words, the generation problem is different from the re-ordering problem.

I'm not sure what you're looking for.  The game seems to use the Unity Engine's Random() and Random.Range() functions.  It is quick, automatically seeded, and likely how most every other Unity game handles randomization.  Yes, there are more complicated methods using C# System.Random or Crypto libraries, but its debatable if the effort is worth it to simulate dice throws unless you are running an online casino.

 

Now that I'm looking a little harder, shuffling/reordering appears to be handed by a standard algorithm.  Other than arguing that Unity's RNG core is flawed, perceived bias is likely a human problem, not a technical issue.

 

As far as I'm concerned, the randomization is still far better than sloppy shuffles and leaning dice you get when playing the physical Pathfinder ACG.  Its actually pretty difficult to get a "good" shuffle on all those tiny stacks of cards when you're playing with a group.  (especially when the scenario is running late into the night...)

Edited November 15, 2016 by Ethics Gradient

[RedPred]

I'm not sure about your model for location deck and shuffle function really exists at all

 

When they say 'deck is shuffled' they mean really that the adt of object cards is reordered.... what do you think? A game of words just to write down something? Your modelization is inconsistent and your programming style is quite old school... really old... 

[Hannibal_PJV]

From dev post there Are card tables and Draws Are based on those.

Not sure, but I think that "cards" in the deck Are pointers to those card tables. I am not professinal programmer, only studied it lightyears ago... So I supose that random function make pointers to those card tables and other random funtion change those pointers locations in location tables. Someone with real experience with Unity engine can make much better estimations how this work in practice.

But what is known is that cards Are arranget in somekind of loot/boon tables and you can add, remove, move cards in those tables. So that above cards in quest mode Are in different pool of cards than in story mode.

Old fachion random Number vere based on seed Number. If you give the same seed, you get same result. So you have to had a prosedure that changed that seed Number to each random check. I don't have faintest idea how random generation works in unity, but post above seems to say that it is basically the same even today.

[Ethics Gradient]

Old fachion random Number vere based on seed Number. If you give the same seed, you get same result. So you have to had a prosedure that changed that seed Number to each random check. I don't have faintest idea how random generation works in unity, but post above seems to say that it is basically the same even today.

In Unity, the UnityEngine.Random functions seed themselves and quickly pop out a handy no-mess random number that is good enough for most purposes.  You can seed it if you want, but it takes care of that behind the scenes.  Though, since all UnityEngine.Random numbers are part of the same seeded stream, there is an extraordinarily unlikely theoretical possibility that multiple threads could end up with the same result if they were to pull a random number simultaneously.  Mostly an academic debate; Pathfinder and other games seem to use the Unity functions due to simplicity.

 

System.Random is a more "classic" RNG.  Needs a seed, takes a few more lines to implement, but allows you to have a one or multiple continuous streams of random numbers if that is of importance.  Otherwise works the same, and is technically part of the .NET/Mono family, not Unity.

 

RNGCryptoServiceProvider is the bonkers option.  Found in Unity via .NET/Mono as part of the System.Security.Cryptography library, it might output the "best" random numbers, but it is the slow option with more overhead.  When you consider that the physical game is played with mass-produced plastic dice or whatever cool D&D dice you found from college, maybe there's a point where a cryptographically secure RNG may end up provide a slightly different gaming experience from tabletop.  The app is trying to emulate an actual game that actual people play with their actual hands (I apologize to any self-aware robots that may happen to be playing Pathfinder Adventures).

 

Anyway, I suppose figuring out how the game works has been its own mini-game while I'm waiting for AD4.  I've mostly been trying to learn the hows and whys of where the Pathfinder app diverges from tabletop play.

[Ethics Gradient]

So I supose that random function make pointers to those card tables and other random funtion change those pointers locations in location tables. Someone with real experience with Unity engine can make much better estimations how this work in practice.

You're pretty much on track.  Card tables (the box) are the source of various lists (location decks, etc...) which then have actions performed upon them (draw, shuffle, banish, etc..).

 

Most of the mechanics happen in a simple, abstract form that mirror the tabletop rules pretty closely.  The hard stuff is getting that all represented in the interface.  

 

But what is known is that cards Are arranget in somekind of loot/boon tables and you can add, remove, move cards in those tables. So that above cards in quest mode Are in different pool of cards than in story mode.

 

Quite true.  Quest mode and story mode essentially have different card pools to build a game, but the "tables" themselves do not change unless the game updates.  They're more of an index of all the cards that should be available during play.  

 

That's the root of all the Treasure card shenanigans last month when the game had certain unobtainable cards left in the card table "index" and it needed a fix via update.

[Hastur]

The only thing I've thought wasn't properly random was evaded cards, but the sample size was too small to really get a feel for.  That, and I can't arse myself to do the math.