Random Numbers Generator

"

1) the drop and enchant system in Path Of Exile isn't purely random. it can't be.


It's quite possible it's not distinguishable from random. And technically it could be purely random. I think they could draw from an external source that uses atomic decay as a source, although I'll personally go uncle a monkey if they do. Ooh, ooh, ahh, ahh. LOL.

"
2) the developers are likely using a ready-made pseudo-RNG, since making one from scratch is needless, messy, time-consuming work.


Probably just the stock rand() function from the OS, almost certainly, and small chance indeed that they are pulling a distribution, although those toolkits are readily available also.

"
3) that RNG accepts parameters from the other game code, and it's those parameters that need to be re-balanced, to avoid making the whole game into one giant lottery, where progress is based on luck and not player skill or amount of effort. especially with orbs that create and link sockets in items.


While I generally agree with the sentiment of some other posters that randomness ought play less of a role in POE than it does now under certain select circumstances, I frankly admit that I do not know what you are trying to state with point 3. Could you rephrase? I don't see how changing the "parameters" to the random generator would affect any outcome you desire, and suspect you may be overly specific with your technical request here.


Last edited by Courageous#0687 on Mar 16, 2013, 12:02:15 PM
"
Courageous wrote:

While I generally agree with the sentiment of some other posters that randomness ought play less of a role in POE than it does now under certain select circumstances, I frankly admit that I do not know what you are trying to state with point 3. Could you rephrase? I don't see how changing the "parameters" to the random generator would affect any outcome you desire, and suspect you may be overly specific with your technical request here.


I'll try to be as clear as possible about this.

parameters are what you "ask" of the RNG to do, and what you do with the result it returns.

suppose the RNG is some weird machine in a giant black box, that gets a parameter n and tosses n completely fair coins, then returns how many fell on "heads".
assume any result between 0 and n (inclusive) is equally likely (it isn't).

asking it "throw 100 coins" and if it returns exactly 12 - give the player a unique from the monster he just slain, is fair.

however, asking it the same question and if it returned 51 - create an extra socket in the item the player just used a Jeweler's Orb on, and remove at least one socket otherwise - is by no means fair. especially if you do this each time independently.
I believe that's what you were trying to say in your "RNGitis" thread as well.

the coin-tossing machine cannot be changed.
parameters and result-handling ( = the way you ask it to do something, and what you do with the result) can and should, as far as currency is concerned at least.
Alva: I'm sweating like a hog in heat
Shadow: That was fun
Last edited by johnKeys#6083 on Mar 16, 2013, 12:23:24 PM
Well, I always entertained the notion that hack'n slays are the slot machines among video games.
You sit at them and push buttons in time (or not), hoping for randomness but only receiving chance loaded with odds to your disadvantage.
Being repetive, tiresome and dull the game only fascinates as long as it has you believe that you got 'a fair chance'.
PoE somewhat lacks in giving me that believe.


"
Courageous wrote:

I was not being a smart ass. I was super specific, quite intentionally, on describing the event, after the fact. I even stated that I was looking at the quarter, after it was flipped. The keywords I was looking for was "randomness" versus "uncertainty." Depending what field you come from, you can regard these distinctions as relevant, or irrelevant, depending. But they are important. Because, I, holding privileged information, have considerably different uncertainty than do you. This applies in a huge variety of real world circumstances.

A practical example would be casinos in the united states, which are regulated by law. If you see in the control center for major casino, you can see who's going to get an (electronic) payout before they do, and from your perspective, the gambler is engaging in something that is not particularly random at all.

I think a quantum physicist (I always want to call them "quantum mechanics," ha ha) would argue that there is not practicable difference between randonmness and uncertainty as far as they are concerned. Someone else might say otherwise. Particularly if where you are playing involves people who hold privileged information.

Do you see what I am getting at?

Computer RNG's and cryptography both have something in common. They are not random, but are, when well-implemented, seemingly random, to a party without privileged information. So. Are these two distinctions any different, as a matter of reduction to practice? I think not, in this case.

Just a bit of idle philosophy for you.

Rather pointless, but it will make you (perhaps not very) entertaining at parties.


"Randomness" exists only in the model you make them exist.

"Randomness" is based on a mathematical model: Random variables, domain space, probability functions, etc.

You choose entirely what model to use on a specific reality you have with the knowledge you have.
For a random dude in the casino, his model of the slot machine will be different than the guy that programmed it.

However, it's safe to create models for the "ignorant uninformed" parties in a specific reality. This way uninformed parties gain more information on reality, while the informed parties still get info on their model of reality as well


Also for specific sequences of events, you can test whether they are "random" or not (basically test if the events are independent which is what matters).
If RNG passes that test, then yes, you can basically just call it "random" and use it as any other uniform random event.
I would say it is all simple math, as I have said before,
ergo : x_n = (a_1 x_{n-1} + a_5 x_{n-5}) mod m
simplified three pages into three lines


(edit) until your account gets FLAGGED for being a math nerd
Last edited by HpDarkwaters#3565 on Jun 27, 2016, 8:21:31 AM

Report Forum Post

Report Account:

Report Type

Additional Info