RNG test idea

[Prizzle]

In some recent thread I read someone talking about successful combines and getting a skillup at same time, yielding to having more chance of skillup per succesfull combine being due to the not-so-random RNG.

This person suggested as proof that one should check groups rolling for an item and see that random numbers generated closely would be in short range of one another (ie /rand 1 100 done less than a second apart would yield numbers tightly bound together rather than complete randomness).

With this in mind, has anyone used the /rand command to try figure out an eventual periodicity and, say, start combining only when /rand yields high numbers (or low numbers, depending on what would yield success, less than or greater than a given difficulty) ?

[Pojodan]

I've pondered the idea of sitting down in a lagless, empty zone at a slow time of the day and punching a /rand macro for 10 minutes or so then making some sort of VB program that parses the log to get the time and rand value then input it into a graph.

After tradeskilling as much as I have I'm absolutely certain that there is about a 5 minute 'cycle' during which
1) there will be a sudden surge of skill ups followed by
2) a period of occational skillups and moderate failures then
3) a period of a ton of failures and very few skills ups then
4) a period of tons of successes and no skillups.

Then the cycle repeats. Hitting #3 is a tradeskiller's nightmare as you can go through 100 combines with about 30-40 successes and only 1 or 2 skill-ups, if any.

I've considered the idea of taking a cheep, easy skill (like baking) making combines untill I've been in #4 for awhile then switching to an expensive skill (like tailoring) and, hopefully, hitting #1 and getting 3 or 4 skill ups in one stack or less.


There seems to be more to it, however.
Anyone notice that over a period of fighting in melee there will be these periods of time when only ONE skill will go up?

Recently I was on a PoA raid, my offence is maxed, but my defence, peircing, 1HB, and Duel Weild where several points below max. I use a Journeyman's Walking stick in secondary and a Slime Covered Harpoon for Primary. Beastlord pets do not work in PoA So I became a gimped monk for the duration of the raid. Anyway, for most of it I had no skill ups untill we hit Aviak Island. over the course of taking out the various birds so that we could farm the Lord a few times I got FOUR skill ups in Duel Weild and none in anything else.
I've also noticed that defence never seems to want to raise, but once or twice a week it will go up a point and I can practically count on it going up another point or two within a few minutes, yet none of my other skills budge.

/boggle

[Prizzle]

I have also witnessed your 4 steps process. I was more under the impression that the "cycle" was around 10 or 15 minutes. Then again, mindlessly combining may make it seem like it's much longer than it really is hehe.

On another non related topic (I dont feel like making a new thread for this really) it would be nice to have the "number of mouse clicks" per recipe, giving us a kind of estimate of how long a final combine is going to take (ie. celestial essence: 5 clicks or 6 if u consider putting it back in an inventory slot)

[Twistagain]

BST Pets in PoA

Sorry to hijack the thread a bit, but I am curious as to why you say BST pets don't work in PoA. Is this a throwback to the days of the Summon Warder skill and level buff? (You couldn't summon one in those days, because the summoned pet was too low to "exist" in PoA.)

I haven't personally tried to summon a warder in the PoA since then, but I was under the impression that it would no longer be a problem.

Or maybe I'm missing something else.

Tigarri, 56 Vah Shir Animist, Innoruuk

[kiztent]

I too have noticed that the RNG is very periodic too. When grinding JC, I'd notice that I'd get no skill ups for 40 combines, then a pair of combines in a stack or two of combines - fairly consistantly from 190ish to 220.

Once I hit 230, I got bored with doing the same combine over and over and got 4 different gems and rotated amongst them (plat + pearl, plat + star rose quartz, plat + jade, plat + emerald, repeat - or whatever), I went from 230 to 243 in about 200 combines, anyway way lower than the 40 combines/skill I experienced on single recipes up to that point). No idea if that's a post 230 thing or rotating helped.

Shrug.

[Marteeny]

I too have noticed that the RNG is very periodic too. When grinding JC, I'd notice that I'd get no skill ups for 40 combines, then a pair of combines in a stack or two of combines - fairly consistantly from 190ish to 220.

Ummmm, sounds like it is working perfectly then. It is supposed to be random. If you roll a 20 sided die, are you going to roll a 1 if you roll it 20 times? Odds say yes. But you have to understand that it is over an incredibly huge amount of rolls that this would be the average. I guarantee that if you give 10 people a 20 sided die and they all roll it 20 times, some people will not roll a single 1, while others will roll a 1 three or four times.

When is everyone going to realize that it is called a random number generator for a reason? There are NO patterns to a RNG!

I have plenty of experience in tradeskills. I have experienced the huge droughts as well as the 3 skillups in 5 attempts. It all evens out in the end.

Marteeny
35 Enchanter
Vazaelle

[Yalum]

A standard algorithmic RNG will have a period on the order of a billion values, a "perfect" RNG with a 32-bit seed has a period of 2^32. Say one billion for simplicity. It's certainly possible that a busy zone might run through the whole cycle in a relatively short time, but to do it in a day requires 11,574 random numbers per second. A fifteen minutes period means 1100 random numbers are being used every millisecond.

Even if it does cycle every 15 minutes, the chance of your current hundred combines overlapping some previous set of a hundred combines is 1 in 100,000. And to get even that high you have to be able to do all one hundred combines in less than a microsecond or the noise of other people hitting the RNG will drown you out.

[kiztent]

When is everyone going to realize that it is called a random number generator for a reason? There are NO patterns to a RNG!

Actually it's called a random number generator because people are too lazy to use the correct term. It's really a pseudo-random number generation algorithm, which is why people are so interested in establishing periodicity. If it's a really bad algorithm, it might be fairly easy to establish periodicity and skill up quicker. If it's a good one, you just get a lot of superstition.

The success formula has been cracked, so why assume that there is no way to establish patterns in the RNG?

[Pojodan]

Re: BST Pets in PoA

Sorry to hijack the thread a bit, but I am curious as to why you say BST pets don't work in PoA. Is this a throwback to the days of the Summon Warder skill and level buff? (You couldn't summon one in those days, because the summoned pet was too low to "exist" in PoA.)

I haven't personally tried to summon a warder in the PoA since then, but I was under the impression that it would no longer be a problem.

Or maybe I'm missing something else.

Tigarri, 56 Vah Shir Animist, Innoruuk

Try it sometime, it says 'This spell cannot be cast in this plane' or something to that effect. T'was kinda disheartening when I first got there, especially since all the necros and Mages could cast pets. It was either a shamn or a druid that was also unable, but they didn't need their pets to be effective. I was stuck meleeing everything and passing out Spiritual Radiance.
<Edit: It may be that my warder is still to low level since my current warder is lvl 39 once cast. Perhaps the Necros and mages could have their pets since they where high enough level to have pets that where 46 or higher themselves.)

As far as EQ's RNG being perfect it's definatly far from it. Certainly after about 5000 combines your average success rate will be about constant, but the fact that you can go for 200 combines without a skill-up and then get 6 or more in a period of 20 combines reaks of poor randomness.

In any random number system (Even true ones, such as taking samples from the background noise of the universe) there will be streaks where only higher or low numbers will come out, but not 200 times in a row!

EQ's RNG is like almost all computer generated RNGs, it has a seed and an algorythem that produces it. Stick the same seed number into it and you get the same set of 'random' numbers. Eventually a pattern ensues that can be noticed, especially in EQ where thousands of people are grabing random numbers several times a second each.

There is no such thing as randomness is a system as logical as a computer, only a perception of randomness to the user.