Announcement

Collapse
No announcement yet.

The RNG.. What's behind it?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Brell-

    hear hear!!

    Humans are very bad at observing, detecting and predicting random events.

    Ask someone to pick a number between 1 and 10. Fewer than 10 percent will say 9. But I would guess quite a few would guess 4 to 7. 8)

    Anyone remember "dragon dice" ... a nifty tool they came out with decades ago to replace real dice at the pen-and-paper Role-Playing games table? You just toggle a little switch and press a button and the "dragon dice" generates a "random" number and displays it.

    Most games I played in banned them within 6 months.

    1) they weren't nearly as random

    2) they could be tampered with

    3) you couldn't roll 12 skeleton's attacks all at once

    4) they were actually harder to read than real dice

    (but very minor changes to the 1-6 range so that it generated 2-7 and rounded down 7's to 6's.... well... *sigh* those were the weeks) :twisted:

    For those bad at math ...

    1 2 3 4 5 6 = average roll of 3.5
    2 3 4 5 6 6 = average roll of 4.33 (not much, about .83 more damage an arrow)

    3d6 = average of 10.5
    3d(broken) = average of 13 (still not much, I just roll well)

    6d3d6 = average of 63
    6d3d(broken) = average of 78 (ok, finally a character WORTH playing)

    Too bad it didn't take a genius to figure out "Hey, Mr. Slightly Above Average over there NEVER rolls a critical miss, or a one damage, or a 3 stat" ...

    No, I never took a DD apart and messed with it. But I did once play with a guy that NEVER rolled a one. On a 10 sided dice game he NEVER rolled a one. In literally hundreds of rolls.

    Was he

    a) very lucky
    b) cheating

    yeah, he's the world's most obvious cheat

    Not to worry, the GameMaster caught on quickly (read: the second game night) to "Mr. I never critically fail" and bullets aimed at his characters did remarkably well. He never caught on. :roll:
    In My (Not Always) Humble Opinion, except where I quote someone. If I don't know I say so.
    I suck at this game, your mileage WILL vary. My path is probably NON-optimal.
    Private Messages attended to promptly.

    Comment


    • #17
      Wow. Three posts in a row (edit: whoops - maybe four) that support the idea that there may well be nothing wrong with the eq RNG at all.

      What are the chances of that?

      Comment


      • #18
        Originally posted by Xenephex
        Wow. Three posts in a row (edit: whoops - maybe four) that support the idea that there may well be nothing wrong with the eq RNG at all.

        What are the chances of that?
        TRUELY random!

        <ducks and runs for cover while cackling>
        Morani
        Wanderer of Tunare,
        Protector of The Mother's children.

        Comment


        • #19
          Hmm.

          I know that the standard random functions are pretty streaky. Computer Science Major here.

          There are ways around this--there are special hardware accelerators you can get (usually radioactive decay, and very expensive) or you can use another algorithm.

          I know OpenBSD has a cryptographically strong RNG but I don't know the algorithm exactly. It is a much 'better' RNG from what I've seen of it. The reason they have it is because they use it for a *lot* of security reasons. http://openbsd.org if you care.

          I wonder if the EQ RNG is server-, client- (oh gods), or zone-specific? Wouldn't you expect that many many people doing tradeskills in the same area would get rid of the streakiness?

          That'd be a fun thing to coordinate someday on say, test.

          [/url]
          --"Captain" Antoin Nolouj
          59th Dark Knight, Absolved
          Firiona Vie

          ((DISCLAIMER: If I say something's tradeable, please don't tell me I'm wrong.))

          Comment


          • #20
            This reminds me of a collection of essays I once read called "The Dreams of Dragons", by Lyall Watson. It had some very interesting essays, but one in particular had to do with human perception of random sequences. To summarize, the basic conclusion was that humans will try to perceive a pattern in any sequence of numbers regardless of whether or not that pattern exists. It has something to do with how we rationalize things.

            Comment


            • #21
              Maesembra-

              Yes, humans are VERY BAD at math and detection of randoms.

              Antoin-

              Hi, CS major as well. The "streakyness" of the default RNG lies in it's reliance on 'time seeds' rather than on the algorithim. Change the seed generation method and the "streakyness" goes away. Instantly. Given that ALL of the programmers for VI/SOE -have- a BSCS or better, one would presume they know this.

              Again, the /rand rolls USED to be client generated. They fixed that. It's well established the seeds and algorithim are all on the server now. (Um, you do realize that several zones are all actually on one server machine, so saying "zone specific or server specific" has no distinction. Server specific and zone specific are the same thing.)

              Ever met Michael Graff? He's one of the team that broke the 40-bit RSA public key encryption. He and I were both at Iowa State University at the time, and we were semi-close friends. I learned a lot about public key encryption from him. The difference between "cryptographicly strong" algorithims and "basic pRNGs" is simply the range over which they have a smooth bell curve distribution. Since few people need a 1028-bit random number in EQ there is little reason to add to the server load generating one and then "washing" it down to a /rand 100 roll. (Not to mention that doing so would actually BREAK the smooth distribution curve.)

              The pRNG of EQ is an artifact. It exists and won't visibly change. Having no choice but to accept it's results... we are left accepting it's results.

              Group: Ok, let's roll for etheral parchment.
              (oh goodie, I got two in a row last week)
              Itek: /rand 100
              **Itek has rolled a 9**
              /em sigh

              So they may or may not use an "RSA quality" pRNG with an analog input. Does that matter? I lost the roll. Maybe a "perfect" RNG would have generated a 10 instead of a 9. Whee.

              Humans are bad at math, at logic and at detecting patterns.

              Don't believe me?

              Walk into ANY store in the USA. Find ALL the items that are over $.75 that DON'T end in $.99 ....

              Do you honestly think to yourself "Wow, this is a penny cheaper than if they charged me a whole dollar..."

              Then drive past any commercial district and look at the number of places with "SALE!!" signs in the window.

              Wouldn't one presume that things are ALWAYS on sale in a store?

              "With our huge discounts the more you buy the more you save!!"

              Um, no, the more I buy the more I spend. I only SAVE money by buying things I had ALREADY planned on buying, but since I hadn't been PLANNING on buying 3 pairs of shoes this month...
              In My (Not Always) Humble Opinion, except where I quote someone. If I don't know I say so.
              I suck at this game, your mileage WILL vary. My path is probably NON-optimal.
              Private Messages attended to promptly.

              Comment


              • #22
                Historicla note:

                Again, the /rand rolls USED to be client generated. They fixed that. It's well established the seeds and algorithim are all on the server now
                However, the tradeskill algorithms were moved to the server much earlier.

                During beta 2, people were fishing up platemail and acquiring fortunes by manipulating tradeskill success. Right then, 989 moved tradeskills to the server.

                Comment

                Working...
                X