Thrugadin? Cursed or RNG Jacking Me Around?

[GrimwoodCT]

After getting an impressive number of gems for the Artisan Seal over the past months, I jaunted down to Thurgadin and bought a few sacks of Velium bars. To my shock and horror I fizzled on the unfired artisan's seal pn 8 of 20 when they were trivial.

Ok, I'm calm now. 40% fail on a trivial item...maybe its just not my day.

Yesterday I head back to Thurgadin to do some smithing, making trivial subcombines. Wow, another 30-40% fail rate.

Has the ghost of RNG's past come back to haunt me or are their others that can attest to the Thurgadin Tradekill Curse/Zon Modifier?

Grim

[Genella]

8 out of 20 fails on Artisan's Seals...ouch!
Seems like the RNG has given you a swift kick in the shin, never managed to fail one yet. Only other reason would be if your pottery was only just above the trivial.

[Verdandi]

It's the RNG. I've done a considerable amount of tradeskilling in Thurgadin over the years, and never noticed any unprecedented failure rate. Genella's got a point, what's your pottery skill?

[Silound]

Funny, thurg was always extremely kind to me when I did baking and tailoring on my alt.

[Kieroth_whiteleaf]

I did most of my baking and my last 30 or so smithing skillups there and never noticed anything - if anything I think I was getting a little better than the statistical 1/20. Had a couple bad runs (~40 or so without a skillup) but overall it was MUCH better than a lot of the horror stories I've heard about the 190-220 range. At 227 now and hoping I can get to 250 with the same rate.

SFG

[Xanafeldier]

I did most of my baking and my last 30 or so smithing skillups there and never noticed anything - if anything I think I was getting a little better than the statistical 1/20. Had a couple bad runs (~40 or so without a skillup) but overall it was MUCH better than a lot of the horror stories I've heard about the 190-220 range. At 227 now and hoping I can get to 250 with the same rate.

SFG

I noticed an unusual number of fails and runs without skill ups in various skills on more than one occaision. If you take the time out to do 100 /ran 100 and watch you can see it's the rng.

Many times I've had an average /ran 100 score in the 40's and sometimes 30's with barely any even breaking the low 80's. When I notice this it's time to stop tradeskilling.

I've noticed it when I get just one too many resists on pulling also. Stopping to do a batch of /ran 100 shows there's something screwy with the average being far enough below 50 and the highs maybe 3 in the 90's, 2 in the 80's... stinky.

The only solution would be sampling the rng from time to time to see when it's at least normal. Unless there's a bug, exploit, cheat, whatever going on and someone's cursed you or is sucking your luck away. /evil grin

[Sanadila]

Honestly, I think there's a roll for luck done each time you log in. Take my husband for example. When he charms in Halls of Honor, he has days when charm breaks in less than a minute every single darn time. I have tradeskill runs where I'll fail constantly on things that have been long trivial. I think somewhere in the code there's a luck modifier that is rolled that effect all rng calculations. I can't figure out any other way to explain it.

Okay, that's a joke *laughs*. Seriously, you can get really long strings of bad numbers. If you don't believe me, talk to professional gamblers. There's a reason they stay away from games requiring luck. The best system in the world will get hammered by a string of bad luck. RNG will always give the same average over time, but the individual numbers can and will vary widely in their frequency. It's why statistical research requires such large numbers to verify if something is statistically significant. If you calculate with two few numbers, you don't know if any results are significant or just the run of the luck.

[Ixiola]

Grimwood, what's your pottery skill? Trivial does not automatically = 5% or less failure rate. Trivial only means you'll no longer get skillups on that combine. If you refer to the stickies in these forums regarding how success/failure is calculated, it will probably help.

But in general, while at high skill levels, trivial means close to 5% chance of success, at very low levels, despite an item being trivial, it's still a pretty hefty failure rate. Since I think the combines you are referring to are around 16 trivial, to be getting that high of a failure, my guess is your skill isn't all that high AND you're having a bad run of luck.

Roughly, for every 50 skill you are above the trivial level, your failure rate should be reduced by about 25%. Generally, if you're 200+ above the trivial, the combine should be essentially no fail, although I'm not sure that's ever been proven for sure. However, at 250 skill, I haven't failed an artisan seal combine for as long as I can remember.

Given the cost of the supplies to make these, I would recommend increasing your pottery skill before you make more.

[Xanafeldier]

Random... isn't. There are times when, for some unexplainable reason, things that have a low probability of happening DO happen and more than once.

Unfortunately superstition, gut feelings, psychic ability and praying for miracles all have about the same effect on reality. Argue that it's zero or higher, I don't care, they're all equal no matter the value.

Someone planted random number generators in computers all over the place and got updates from them periodically (part of a program on the discovery science channel, I hadn't planned on quoting it so I'm afraid someone with a better memory will have to step in if anyone wants more info). There were significant positive and negative variations in each at around the same period of time. I'm not clear on exactly how a computer manages to generate it's numbers but I'd be willing to bet it's influenced in some way. Electromagnetic flux in the planetary field, sunspots, a comet knocking a dust particle out of the asteroid belt millions of miles away causing an electron to spontaneously jump locations, whatever.

Rumors of the RNG being streaky crop up for a good reason. Luck of any sort seems to go that way, usually bad and that's what most people are vocal about. Usually when I've gotten 2 skill ups in something I stop and wait at least a day to do more. Chasing a third point has always used up more stuff than the first two combined. Also if I don't get a point in the first 40 tries I think about quitting, sometimes do maybe 10 more, if there were a lot of fails and no skill I quit after 40.

My mother used to tell me that bad things come in threes. If you notice the first two early enough you can even sometimes avoid the third, or at least minimize it. The universe has to be pretty quick or sneaky with that last one now.

Oh... and don't forget Kenny Rogers. "Know when to hold 'em. Know when to fold 'em. Know when to walk away. Know when to click your jboots" or something like that. Figuring out luck can be a bit like trying to hear the sound of one hand clapping.

Thanks for reading this practically meaningless post!

[Trellium]

Random... isn't. There are times when, for some unexplainable reason, things that have a low probability of happening DO happen and more than once.

It is absolutely random, since anything you just had happen tells you nothing at all about what will happen. You don't really know what will happen next, you just think you do, based on your own random samplings of a random sequence of events, from which you draw patterns because your brain prefers patterns to randomness.

There will always, and should always, be patterns in randomness. If there isn't then the pattern is the lack of pattern and that tells you something ... which means its not random. Random means anything you saw before is completely irrelevant to what is coming next, and there is no pattern to it in EQ. Its not like you can say "If I win twice, I will fail next" and be right in the vast majority of cases. Mostly, you will be right about half the time.

[Xanafeldier]

It is absolutely random, since anything you just had happen tells you nothing at all about what will happen. You don't really know what will happen next, you just think you do, based on your own random samplings of a random sequence of events, from which you draw patterns because your brain prefers patterns to randomness.

There will always, and should always, be patterns in randomness. If there isn't then the pattern is the lack of pattern and that tells you something ... which means its not random. Random means anything you saw before is completely irrelevant to what is coming next, and there is no pattern to it in EQ. Its not like you can say "If I win twice, I will fail next" and be right in the vast majority of cases. Mostly, you will be right about half the time.

I'd feel better about the RNG if it didn't occaisionally crop up with patterns, long lasting tendencies of high or low. Even a coin toss isn't truly a random event. The amount of force applied, location and direction of the force on the coin, length of time spinning, air density in which it's left to spin all have the possibility of being controlled. If the same conditions are applied within a certain margin then the result will be identical.

The physical skills of a juggler and a knife thrower could theoretically skew the result. A coin toss is random simply because you're not paying attention to and calculating the variables involved. If you know you have the same coin, the coin is in the same position, you are about to apply the same force in the same way, you know the coin will be spinning for the same amount of time and be landing on the same surface or caused to stop in the same way, BUT the air just doesn't FEEL right... We don't have the luxury of knowing where the "random" in everquest comes from and how it's calculated, at least I don't know.

If I get 2 or 3 skillups close together it's been my experience that there will be no more for a long stretch. If I have a string of failures on trivial items and attempt a non trivial combine it's almost always doomed to fail as well. If I've succeeded on a series of non trivial combines my nervousness grows and I'm unlikely to attempt an expensive non-trivial any time soon.

I suppose if I have any point at all it's about minimizing the impact of BAD luck. I can't think of a way to alter the individual outcome. If you know darn well you're going to be MAD if you fail a combine, then wait until you're in a better mood. If you can't afford to lose, don't play. If you've already won enough to make you satisfied for the moment don't "let it ride."

Use your head and you won't need to become a member of Gambler's Anonymous.
Thinking that it's random, there's no pattern, and you could possibly win AGAIN leads people to be in debt up to their grandchildren.

[Trellium]

That is also true for thinking there is a pattern, when there is none. You have the same chance of winning a lottery next week using last weeks winning number as any other number you select, since they don't exclude numbers that won last week. That tells me how unlikely any win is.

To me, "random" is nothing more than the inability to know what will happen next and that anything that happened before is completely irrelevant. Someone would have to demonstrate repeatability in the EQ random numbers for me to believe the randomness is not random. Patterns are irrelevant, unless they prove repeatablity.

Furthermore, I assume that the servers are using a random key that involves the system time in some manner ... so that the time that you trigger the request for a random number (ie skill up success) is in some way a function of the random timing you are giving. People often assume that random numbers are generated in a pattern built just for them, when they ignore that other people or events might either be selecting from that same random number stream, or that timing also randomizes the selection. People don't want to believe that billions of possible outcomes had already occured in the random sequence, they only saw their miniscule sampling and believe they are the only ones who see a pattern. But, how good is a pattern when a miniscule sampling of random selections made by yourself? You only see a tiny, tiny part of the entire random stream so how could you ever see the entire pattern?

If they use nothing but a single long stream of random numbers, and continuously select from this list in a linear manner ... its still random. If they add to that a random sequencing of events. Someone else did a /rand command before you tried for a skillup, they might have gotten "your" random number ... blame it on them. Thats another level of randomness, over and above the original random pattern. Then, add on to it the possibilty that the computer server time (usually counting in milliseconds) also randomizes the place within the random stream of events ... and all you are doing is an miniscule random sampling of a random sequence of numbers in order to prove a pattern. A pattern that updates every millisecond.

Tying a random number generator to the system time isn't unusual, its very normal. The timing of the request itself is a function of the randomness. Had you done your skillup a few milliseconds before you did could return a completely different result. Blame that on yourself.

You say you see a pattern, but how many thousands/millions of other random events happened on that server between the times you selected for your "pattern"?

[Sanadila]

Ohhhhh....computer geek talk....honey, you know how that turns me on....

*swoons*

[Xanafeldier]

You say you see a pattern, but how many thousands/millions of other random events happened on that server between the times you selected for your "pattern"?

I wouldn't say I see a pattern in it, just long periods of time with an uncommonly low average sometimes lasting for days. If there were a pattern to it that would be something completely different.

Flipping a coin 5,000 times in a row should produce some long streaks of heads and of tails. There shouldn't be a pattern to when those streaks occur. The lack of a pattern to the sudden lack of variation means that suddenly random... isn't. /ran 100 or click combine 5,000 times and you'll get steaks of luck here and there good and bad.

My experience in EQ is that when bad luck strikes a few times it doesn't just go away. It sticks around to kick me when I'm down. Sometimes it decides not to let me up for a while. That's when I choose simple non-random or non-critical things to do untill it seems like it's gone or swung back the other way. I don't care how badly the rng wants to treat me, a spiderling isn't going to get the upper hand and it'll drop a silk here and there. On the other hand deep blue cons would most likely resist enough times to make it a painful and unrewarding experience.

You're right about the time. Sometimes changing the timing of what I'm doing or breaking out of a pattern I've gotten into can make a difference. /dance once in a while between combines for instance. The RNG could possibly come up 1 out of /ran 100 fifty times in a row. /shrug I'd rather be /dancing for some of them.

[Sukrasisx]

I wouldn't say I see a pattern in it, just long periods of time with an uncommonly low average sometimes lasting for days.

Okay, I'm going to try an experiment to test this perception of the RNG being streaky, and in particular that you can spot it via /ran.

I'm going to set up a bot somewhere running a hotkey that does /ran 1 100 five times. First off, I'll compute the frequency of the results, though I admit I'd have to break open my old textbooks to remember how to analyse whether they're off by more than randomness would expect. But the main thing I'll be looking at is an easy one: for any batch where the first four numbers are all 80 or less, how often is the fifth number also 80 or less? If each number is independently random, the answer should be about 20%. I'll go see...

Okay, I ran the above test. Got 42155 /ran's. (And annoyed the heck out of one poor sap who wanted to use the Kaladim bank during that period; it was the most underpopulated zone where I had a character that I could quickly and safely set up. I did put an explanation in my /afk message, but he just cried STOP in /ooc so he never got my afk.) So I expected each number to come up 421.55 times on average; they actually ranged from 375 to 479, which doesn't strike me as that unusual for the sample size. As for the stated experiment, I broke them up into 8431 sets of five, each set having been done by a single hotkey (though of course there could still be lots of other things causing the server to generate random numbers inbetween my requests). I then looked at those sets where the first four numbers were all <= 80. There should be about 8431 x 0.8 x 0.8 x 0.8 x 0.8 = 3453 such sets; there were actually 3503, which again doesn't seem too far off. For those 3503 sets, I checked whether the fifth number was <= 80. If the numbers are truly independent, it should be about 80% of them. It was 2788, which is 79.59%, pretty darn close.

Though it wouldn't hurt to repeat the experiment a few times, my tentative conclusion is that the RNG is not especially "streaky"; just because it's rolling low for a while doesn't mean there's any reason to expect it will roll low on the next attempt.