rng solved?

[coloeb]

Since the release of the tradeskill interface, I have noticed that my success rates have been based on time between success's . this is mearly an observation on my part. It seems that waiting 80 seconds between success give or take a few combines, gives me anouther success. This was done between skill 200 and 240 and for all tradeskills. Your results will prolly be different as I am sure things like Int. Wis etc. would also come into play.
The message I have is for you to Pay attention and see if you can find that nitch that is right for you. Also keep in mind that the time may be different for different recipies. This method has increased my tailoring skill from 220 to 240 on 2 stacks of combines. thats right GM'ed tailoring. Same thing with fletching and baking I am know GM of all tradeskills. Hope this helps at least one other person, good luck!
BTW if for some reason you cant make this work it may have very well been a fluke or a bug, I have no idea. I just know it worked for me.

[Tinile]

There is a reason it is called the Random number generator.

[coloeb]

edit

in my original post skill ups are what I call a success sorry if that was not clear.

[Carmaris]

I'm going to have to say it was a fluke. A random number generator shouldn't have any relation to time. Also, if it were just a matter of time, people across the board would see fairly consistent numbers when skilling up, ie if my time (could/would vary according to wis/int etc) is 100 seconds and I can do ~25 combines in 100 seconds, then I would expect to be able to predict a skillup every 25 combines or so. Just doesn't happen that way, unfortunately.

[Hibashira]

Still, no RNG that a computer is responsible for is true random. If the condition is absolutely identical, the numbers will be the same. Its consider good enough simply because truely identicial condition is almost impossible to achieve. Time is almost always an element in RNG routine simply because it is NEVER the same. But if by some fluke .... well .... its still a stretch .... hmmm ...

[Freya]

It would make sense if there was some sort of pause time between sucessful combines..it would punish those who use macro's to tradeskill, since with a macro the time between combines with be virtually identical.

[Tradeskiller]

Well it may be a just a fluke here to but I remember not to long ago with the new UI I was doing some combines for skill ups, forget which tradeskill it was. But I did approx. 45 or so combines and no skill up, left the tradeskill UI open while I had to go afk and the first combine I did when I came back was a skill up. I didnt think too much about it but was happy none the less then maybe a few days later the same thing happend after coming back from afk. I never thought anything about it until now.

[Gardeltron]

Most random number generators are known to have temporal autocorrelation in them. That means that numbers at one time are not totally independent of the numbers drawn previously. Some random number generator algorithms are better than others. If EQ uses a simple one, it is likely that there is a significant degree of autocorrelation in their random number generator

There is also the intriguing possibility that EQ purposely builds in temporal autocorrelation into their random numbers.

I almost always find that my skill ups, whether in trade sklls or ther skills, come in "batches".

[splunge]

/tosses another post in the RNG bin

[Hibashira]

"temporal autocorrelation"

I like that term.

[Tradeskiller]

Since there is a RNG then whats the deal with needing to have max wis/int or whatnot. Or does the RNG look at your wis/int and say ohh hes got soandso wis/int so lets give him a skill up at X combine while this other guy has 75 less wis or int so lets make him do more combines for skill ups. What gets me is some people with 355 wis or int can do 150+ combines and get no skill ups.

coloeb why not post what you used to skill up on?

[Horanaf]

Now I utterly suck at math, but here is how I imagine the system works, in a round about way.

Let's use the old /ran 1 101 that lots and lots of people have used to roll on loot since forever (on prexus anyways).

System does 2 checks, one for skillups and 1 for success/failure. First check is like this.

1-80 = no skill up (with base int/wis, pick a number for this)
81-101 = skillup

Lets say you reach a point in wis.int that it is beneficial for skillups. For the sake of argument, we will go in 50 point steps fromm 100.

100 int/wis would be 1-101
150 would be 5-101
200 would be 10-101
250 would be 15-101
300 would be 20-101

However, this doesn't account for the increasing difficulty in getting skill ups in higher skill levels, so lets change the base roll a bit,adding the wis/int bonus to the front number of the range.

at skill 1, it would be 70-101
at skill 50, it would be 50-101
at skill 100 it would be 45-101
at skill 150 it would be 20-101
at skill 200+, it would be 1-201

This is just an example, and likely highly inaccurate, but would account for the benefit of high wis/int while still being random.

The success/fail check sould be similar I imagine, but using skill level/trivial to determine the range for success.

Of course, I may be (and probably am) completely wrong. But that's how I always pictured the system.

[coloeb]

combines I used

The combines I used for tailoring were othmir fur caps , for fletching I did bows and followed the trivial curve, for baking I did the crab/ tuna recipies along the curve. Barbequed ones till 242 . hope that helps.

[Naxy]

I almost always find that my skill ups, whether in trade sklls or ther skills, come in "batches".

I have the same experience. And I find that it goes for failures in combines as well. My theory is that the "seed" for the random number generator is refreshed every tick. And some ticks you have a seed which is "friendlier" than at other times.

Now, I know this doesn't match what I learned about random number generators, but I am an experimental kind of guy. If theory doesn't match experience, the theory goes in the trash.

So I make quick combines just after I get a skillup and more often than not do I get another skillup just following it. /shrug

Whatever voodoo floats my boat.

[scarbrowed]

I recently read a story about a professor teaching probablility. (I wish I could remember where I read it). He would have students write down heads or tails 50 times, then had them flip a coin 50 times. The real coin always had more streaks of the same flip than the lists the students had.

The point is that random numbers like coin flipping are far more likely to have these clumps than people expect. So in fact the streakiness people see seems to me validation that EQ is using a good random number. If streaks were rare, we would know it was rigged.