Smelting

[Theo_GL]

I'm a GM Miner.

I'm diging ore.

I dig 20 Valorite Ore.

Smelt, fail, smelt, fail, smelt, fail....

From one pile of 20 Ore I got ZERO ingots.

Thats right, I failed like 7 times in a row.

What are the success odds at GM mining? Like 45% success on Valorite?

How do I fail 7 TIMES IN A ROW?? That is a statistically annomaly about as rare as winning the lottery.

Devs - the RNG is junk. Why not get rid of smelting failure altogether? If you want to 'punish us' just always smelt Val ore down by a factor of 40%. Don't give us some lame 'random number' generator that is nowhere near random.

That is like flipping a coin and having it come up heads 7 times in a row. Is it possible? Yes. Is it probable - hell no.

 

[Mishra_]

Congrats you won the lottery

 

[Nestorius]

I guess that's the reason why 'they' tell you to smelt each piece of ore one at a time.

 

[Kyrie_Elaison]

I'm a GM Miner.

I'm diging ore.

I dig 20 Valorite Ore.

Smelt, fail, smelt, fail, smelt, fail....

From one pile of 20 Ore I got ZERO ingots.

Thats right, I failed like 7 times in a row.

What are the success odds at GM mining? Like 45% success on Valorite?

How do I fail 7 TIMES IN A ROW?? That is a statistically annomaly about as rare as winning the lottery.

Devs - the RNG is junk. Why not get rid of smelting failure altogether? If you want to 'punish us' just always smelt Val ore down by a factor of 40%. Don't give us some lame 'random number' generator that is nowhere near random.

That is like flipping a coin and having it come up heads 7 times in a row. Is it possible? Yes. Is it probable - hell no.

I agree! There is no random chance of losing lumber, so why should there be one on ingots? Especially when I have better luck mining my new ore cart than in the hills.

 

[Kyrie_Elaison]

I guess that's the reason why 'they' tell you to smelt each piece of ore one at a time.

For ages I have taken my high end ore and minimized it to the little Z shaped ones, and then smelted them 2 at a time. It's time for a change.

At least they could give us dishing stumps for eleven year rewards that would give us a bonus to smelting.

 

[Jhym]

That's... because... the... chances... are... INDEPENDENT.

Just because one roll gives you a failure doesn't mean the next roll will be any more or less likely to do so. It's the same exact chance each time. Even using the 45% failure chance, that's a fairly likely chance of failure, so your results don't surprise me. I've smelted a stack of 450 valorite smalls and managed to get 8 ingots out of it. Did I cry? Yes. Do I think the RNG needs changing? No.

The independence of individual rolls is what confuses people and makes them delude themselves into seeing patterns where there are none.

"Oh, I ate food just before I smelted and I got 100% smelt off the pile"

"Oh, I killed three pigs and suddenly I hit every single time on the next three monsters"

"I kissed the elf when I got his quest and he gave me a runic!!"

It... is... all... RANDOM.


Everytime this comes up I want to strangle people. Does NO ONE take statistics anymore?

 

[Theo_GL]

I guess that's the reason why 'they' tell you to smelt each piece of ore one at a time.

Yes, with a larger number of trials you are likely to get closer to expected results - however, who has the time to separate and smelt hundreds of ore individually? Thats right - only scripters.

They shoud just take out the ore failure rate. You already randomized it to make Val impossible to find why punish us on the smelt too?

Either that or give me a 120 mining PS so I can get to 100% smelt rate.

 

[Traveller]

How do I fail 7 TIMES IN A ROW?? That is a statistically annomaly about as rare as winning the lottery.

Incorrect. By at least four orders of magnitude, probably five. The likelyhood to fail 7 in a row is about 1.5%. Which is obviously low, but FAR from impossible. Or even from winning the lottery (which usually is 1 in millions).

How many times you have smelt 7 times in a row? I bet more than 100, so that's fine. As much as it sucks, it balances out with those other times when you smelt the whole pile at first try.

 

[Kyrie_Elaison]

We need dishing stumps.

 

[UOKaiser]

I'm a GM Miner.

I'm diging ore.

I dig 20 Valorite Ore.

Smelt, fail, smelt, fail, smelt, fail....

From one pile of 20 Ore I got ZERO ingots.

Thats right, I failed like 7 times in a row.

What are the success odds at GM mining? Like 45% success on Valorite?

How do I fail 7 TIMES IN A ROW?? That is a statistically annomaly about as rare as winning the lottery.

Devs - the RNG is junk. Why not get rid of smelting failure altogether? If you want to 'punish us' just always smelt Val ore down by a factor of 40%. Don't give us some lame 'random number' generator that is nowhere near random.

That is like flipping a coin and having it come up heads 7 times in a row. Is it possible? Yes. Is it probable - hell no.

Yep the great system of smelting
But thats why gold and above are stored untill you have time to smelt 1 at a time or reach 10k of each or so and smelt 1k at a time.

 

[Gwendar-SP]

I guess that's the reason why 'they' tell you to smelt each piece of ore one at a time.

I never go that far but do smelt in even numbers. When you fail you lose half rounded UP except for the small half ingot ore. So his smelting was 20-10-5-2-1- then two smalls.

 

[Maplestone]

Things are nicely scaled to allow the possibility of mining powerscrolls (I'd gladly sacrifice another 20 skill points to be able to smelt 100% of the time)

 

[ZippyTwitch]

I agree! There is no random chance of losing lumber, so why should there be one on ingots? Especially when I have better luck mining my new ore cart than in the hills.

Also because valorite ore is a hell of alot easier to find than say frostwood. Seriously the smelting issue for this has been around since day one. No point in complaining now.

 

[T_Amon_from_work]

Now, ya see? You take all the fun out of things!

My GM Miner has a worse return of ingotsre with Verite than Valorite ... wearing +5 gloves even.

 

[Setnaffa]

I've been on many long streaks of failure and success. One time (before the change to Val BOD frequency) I received two Val BODs in a row. What are the odds of that happening? In UO, the odds are alot higher than reality.

They need to seriously look at the RND code and re-work it.

Maybe instead of losing all your ore, you'd get a percentage of the ingots. So if you had GM Mining, you'd always get at least 50% of your ingots from Valorite ore.

 

[J0KING]

Smelting ore has been something that has always troubled me. I understand the need to increase the difficulty with smelting higher end ore but... since the ore is now more difficult to aquire couldnt we ease up on the chance for failure?

Back when we could hit all the Valorite spots in our rune books and collect what we needed there was a need... but this is no longer the case. Hence the reason that I take the time consuming method of smelting them two piles at a time.

 

[RoycroftLS]

That's... because... the... chances... are... INDEPENDENT.

The independence of individual rolls is what confuses people and makes them delude themselves into seeing patterns where there are none.

Everytime this comes up I want to strangle people. Does NO ONE take statistics anymore?

Does "NO ONE" actually read forum posts anymore? Nothing the OP said gave the impression that he misunderstood basic statistics. He overstated the overall odds of such an occurrence, yes. But he did correctly compare the scenario to the flipping of a coin.

Granted, if you play UO frequently there will be enough statistical trials so that it is likely to have something improbable happen to you. And it's also given that people generally underestimate how "streaky" an ideal RNG should be.

But even taking those two factors into account, there is plenty of empirical evidence from people over the years that supports the theory that Ultima Online's RNG is not statistically random on a local level. That is, the success chances of consecutive events are likely not independent. If this is indeed the case, it's possible that the game is designed in such a manner, but the more likely culprit would be an inadequate RNG.

My personal anecdote is from a time when I was short on gear, and thought a 99% lrc suit would suffice on my mage while hunting to get the last piece I needed. I didn't bring spare regs because I figured I could survive a single "need more regs" inconvenience every so often while hunting. The chance of it happening even twice in a row was rather insignificant, assuming independent events.

However, in less than one weekend of normal gameplay I had three consecutive "need more regs" on two separate occasions. Needless to say, that was enough to convince my mage to never again leave home without 100% lrc.

 

[Pink Dragoness]

The only time that is beneficial, is in gaining mining skill. I hate serperating ore down to one at time. I want to smelt high end ore like I did back in 2002. I got a lot more back then for my time and swings then I do today. :[

 

[Basara]

Everytime this comes up I want to strangle people. Does NO ONE take statistics anymore?

Start with yourself - it's a PROBABILITY problem, not a statistics one - any decent Probability & Statistics teacher would reprimand you for confusing the two.

As someone else pointed out, there are local inconsistancies that indicate a problem with the psuedo-random number generator.

IF you do multiple-roll experiments with the numbers (and people HAVE, regardless of what you might think) -

The MEAN is what is expected, but the distribution curve of the data resulting from the experiments is anything BUT a normal curve.

 

[Warrior of Time]

Also because valorite ore is a hell of alot easier to find than say frostwood. Seriously the smelting issue for this has been around since day one. No point in complaining now.

Random ore has not been around since day one so why chang that one and not the other.
I say remove random ore, and put it back as it once was or add 100% smelt.
Hard enough to find the ore much less lose what you do find.

Why bother giveing a better chance to get the bods for a val hammer, if you can't get enough ingots to fill them.

You lose a lot of ingots as it is making exc. val plate armor. Then it also comes down to should I enhance with val or save for the bods that I hope to get to finish a set.

 

[Traveller]

As someone else pointed out, there are local inconsistancies that indicate a problem with the psuedo-random number generator.

That someone also does NOT tell us what some inconsistencies are, except "there are". Actually, in case people do not remember, the RNG USED to have inconsistencies some time ago, and was fixed (by mrTact if memory serves). I have a strong hunch that people that say RNG is not working are referring to urban legends born before the fix.

IF you do multiple-roll experiments with the numbers (and people HAVE, regardless of what you might think) -

Actually I and some friends did. _After_ the RNG changes of some time ago, a friend of mine went on test center and started rolling dices for thousands of trials, putting all the data into a spreadsheet. Then he passed the sheet to me (I think I should still have it somewhere) and I evaluated all the main statistical indicators (variance, higher level moments, etc) of the stuff. They were the same of a normal distribution. I dedicated specific attention to correlation coefficients as even in a normal distribution attempts close in time or sequence may suffer from correlation. I didn't find anything significant (i.e. correlations from 1 to 5 were almost 0).

The MEAN is what is expected, but the distribution curve of the data resulting from the experiments is anything BUT a normal curve.

Actually it's not even enough for the curve to be a normal. The overall form of the curve could still be normal, but close attempts might still possess high correlation. HOWEVER, I suppose you have some proof of the form of the curve or of its correlation, right? What I can see in this thread is just someone saying that after playing for years hit a 1 in 100 chance he doesn't like.

If you have opposite results, please let me know, I would be glad to look them over. Just note that the results obtained before the RNG chance have to be considered as null and void. Only the info after the RNG changes are meaningful.

 

[Basara]

All testing was done long after Mr. Tact's changes were ancient history.

Traveller, all your insistance won't change the reality that somethign is going wrong either at the RNG level, or what the game does AFTER it gets hold of that number. Testing the in-game DICE has little bearing on the RNG issues behind everything else. After all, everything else has OTHER variables influencing the results - like the missing ")" that messed up the Smith BOD results from Publish 19 through Publish 51 or so. What may be a "fair" roll from the RNG may not be so, once it goes through all the decision points - each a RNG check that may NOT be fair due to a programming error.

Try doing tests on, say, 2000 valorite ore, smelted 1 at a time, at 99 mining (should be 50/50 success/failure) - and tabulate the length and requency of success/failure runs. When you do, get back to us.

Note that with 6 different characters, all high enough level to get colored BODs, it is ROUTINE out of 6 characters, making BOD runs back to back, for 2-4 of them to get a weapon BOD for the exact same weapon type (quality & count may vary, but often do not). The RNG should not be producing streaks like that nearly EVERY DAY (and I typically only make 1 or 2 BOD runs per day), when there is only a 2% chance (50% for weapon chance, 1 of 25 small weapon BOD types) that any two consecutive BODs (note, I'm talking CONSECUTIVE BODs - would would expect there to be a reasonable chance of duplication if it was 2 of ANY 6 BODs) should be similar in that manner. It's occurring much closer to 20-25% of the time. While the overall distribution of BODs (in terms of quality, color, category, etc.) all came back within 0.02% of expected numbers, after 2500 BODs (all collected since February), the streakiness was apparent by looking at a day's "samples" - 4 of 6 smiths shouldn't all get normal Hammer Pick BODs, on the same trip (3 of the smith's runs consecutive, in a 90 second period) - and perhaps 1 day a week would go by where two consecutive smiths didn't get something near-identical on their runs.

Had I known that the obvious would be so vehemently questioned, perhaps I would have kept more specific data - but at the time, I was only sampling for normal vs. exceptional, and colored Armor vs. Iron weapon vs. Iron Armor.

 

[It Lives]

I have come to accept Ill miss on valorite.

What really heats me up is a 3 in a row on gold with a gm miner and +5 gloves on( gold color +5's) if that even works any more.

Twice this week this has happend to me.

 

[lham52]

I agree with those who say the smelting chances for higher end ore should be re-evaluated (improved chance) since they randomized ore spawning.

 

[MalagAste]

I know even with mining gloves on bringing my mining up to 105... I still fail an awful lot with Valorite... but I suppose that's the way things are ..... Look at how often one fails at enhancing... I mean I'm a 120 smith and a 120 tailor... and I fail more at both of those even with a talisman of over 20 bonus's to exceptional and to skill on ..... with a +60 ASH hammer..... I fail more often trying to enhance metals or leathers than I do with my GM fletching.... go figure...

 

[Radun]

I bet you wish you had bought a lottery ticket instead of sitting down to play UO that day

 

[Traveller]

All testing was done long after Mr. Tact's changes were ancient history.

Good.

Traveller, all your insistance

I am merely insisting on having hard data instead of being fed the same urban legends that have been told since before AoS. Especially after somebody shouted at "winning the lottery" just because he bumped into a 1/100 chance.

the reality that somethign is going wrong either at the RNG level, or what the game does AFTER it gets hold of that number.

Which are two separate issues, as you can understand. Shouting "something is wrong with the RNG" instead of "something is wrong with smelting" is only increasing the confusion. As far as I am concerned I have no interest in smelting, so I never checked that out. Instead I gave you some information about the RNG, which was the (incorrect) point of the OP.

Try doing tests on, say, 2000 valorite ore, smelted 1 at a time, at 99 mining (should be 50/50 success/failure) - and tabulate the length and requency of success/failure runs. When you do, get back to us.

?? Why should I? I have no interest in smelting. It is you who are interested, so it's your interest to attracting devs and other people's attention, isn't it? If so, use SERIOUS data (like those you have given for BODs), not "lottery" sensationalism, devs won't be listening to that (at least not for RNG-related stuff). Why in this day and age it is still so difficult for people to be scientific?

The RNG should not be producing streaks like that nearly EVERY DAY

I agree, it shouldn't.

4 of 6 smiths shouldn't all get normal Hammer Pick BODs, on the same trip (3 of the smith's runs consecutive, in a 90 second period) - and perhaps 1 day a week would go by where two consecutive smiths didn't get something near-identical on their runs.

Mhhh... I wonder if the problem is nothing more that the server is using the time of the day in MINUTES to feed the RNG for BODs, or directly to decide the weapon. That would be really stupid, but wouldn't be the first time I see something like that being done. It would be interesting to make bod runs more interspersed during the day and see if the streakiness persists, but I assume that people have also a life besides BOD running...

Had I known that the obvious would be so vehemently questioned,

Maybe you should consider that not everybody plays a bod runner? I was not aware of it, as I am sure about the 100% of those who don't run bods. I merely questioned the "OMG I hit a 1/100 chance!" which still remains a silly stance.

perhaps I would have kept more specific data

Yes, you should have. If you want the problem fixed, you should first convince the devs that the problem exist. Send them a bug report saying "bods don't work!" and you know what will happen to that report. Send them a report with a proper statistical analysis and at least you will make them aware of the problem, though I cannot say about the priority given.

 

[Theo_GL]

A few random follow-ups:

My main rant was around the fact that due to a HIGHLY unlikely streak I started with a stack of 20 ore and ended up with ZERO. I'm not a newbie miner. I am GM. I am the highest skill level the game allows (minus gloves). I don't' believe it is fair that you can have cases where a stack of 20 ore (that should be 40 ingots) ends up with a big fat zero.

I also understand that 7 fails (or actually 9 b/c I failed 2 times on the next stack right after) is statistically possible in any case. I minored in statistics back in college. I'm fully aware of distribution curves, probability theory and the like. The problem I have with UO is that results do not seem random.

For example - I'll have an 80% chance of EX making Plate Helms. I go to make 20 for a bod. I make 13 straight EX and then i'll make 4 straight NM. This is not a 'freak occurance' it is something I regularly come across. I may end up averaging 4 NM and 20 EX over the run but to have 4 in a row is very odd and falls outside 1 or 2 deviations. It is not a probable random distribution. Its almost like the RNG is tied somehow to clock cycles or something that seed it to not be random at times.

At any rate - my main point was that smelting is a poor system. They should change it to one of :
* Smelt a portion of it correct and lose the rest
* Change it so there are no fails
* Add PowerScrolls so that you can get to 100% smelt for all ores

The statistical bickering is amusing though. Carry on....

 

[Olahorand]

Hm, I am sure I replied yesterday, but no?
I wonder anyway - where is the metal going from the fails? Come on, 50 percent loss at one try?
While things, you could destroy in "reality", always stay intact?
For smelting ore you need to have Mining skill - and loose.
For cutting barbed leather you need 0 skill and never destroy anything (even if you could snip it too small). And on top it must not even be cut to use it.
The second goes for wood.

I also wonder, if the +5 over GM with gloves do still count for the difficulty equation.
While we never failed smelting Gold with them years ago, this somewhen changed and there is a chance of failure with 105 mining.

*Salute*
Olahorand

 

[Uvtha]

Smelting is LONG overdue for a change. The valorite smelt rates are perposterous. At worst smelting failure should be like 75% for the high ores.
I mean even with 100% you still only get like 40 ingots if you are REALLY lucky. Generally you will get like 10 colored ore from a spot.

 

[Traveller]

Smelting is LONG overdue for a change. The valorite smelt rates are perposterous. At worst smelting failure should be like 75% for the high ores.
I mean even with 100% you still only get like 40 ingots if you are REALLY lucky. Generally you will get like 10 colored ore from a spot.

I wonder if people realize the consequences of what they want. One thing is to ask for smelting to be fixed as it was designed (assuming, and proving, that it was actually broken) another, completely different thing, is to change the design.

If you do that, so that you smelt much more from a single batch, just be prepared to see a steep drop in val prices. You might like it, or not, but what I see it is that people ask for changes without even considering the consequences. I am sure that smiths would like that, but miners? I doubt it.

 

[Eslake]

That's... because... the... chances... are... INDEPENDENT.

Just because one roll gives you a failure doesn't mean the next roll will be any more or less likely to do so. It's the same exact chance each time.
..
It... is... all... RANDOM.


Everytime this comes up I want to strangle people. Does NO ONE take statistics anymore?

And every time I see someone try to make this argument I get a chuckle.
You are applying 2 Different mathematical principles together in a way that does not work, in order to prove someone (who is correct) is making the mistake.

A roll of a die has a 1 in 6 chance, true.
Each roll has that same chance, regardless of previous or future chances, true.
But you are also indicating that the chance of rolling a 6 7 times in a row is 1 in 6, which is Not True.

Successive same results in chance are exponentially diminishing. That is why in cards/yahtzee/etc 5/6 of a kind are more valuable than a full house. And why you don't get a flush in 25% of your poker hands.
The chance of any 1 card being hearts is the same, but the chance of getting them in succession decreases with each card you take.


As to the OT, I could not agree more that smelting is a joke.
You don't lose lumber when you make boards.
You don't lose leather cutting hides.
Why should we lose metal when we smelt?

Even ignoring the failures, mining is slower than lumberjacking or leather farming (unless maybe you're farming dragons for barbed).

--
FYI, it has nothing to do with STATISTICs.
Statisics are solution-to-problem mathematics.
In statistics, I learned how to have 20 people chose between pepsi and coca cola, and if 18 chose pepsi, use Statistics to prove that coca cola was more popular.

I guess that is why the federal government employs so many statisticians.

 

[Basara]

FYI, it has nothing to do with STATISTICs.
Statisics are solution-to-problem mathematics.
In statistics, I learned how to have 20 people chose between pepsi and coca cola, and if 18 chose pepsi, use Statistics to prove that coca cola was more popular.

I guess that is why the federal government employs so many statisticians.

I remember a famous quote - most Americans attribute it to Mark Twain, but he was actually quoting Benjamin Disraeli, twice Prime Minister of the United Kingdom during the Victorian era.

"There are three kinds of lies: Lies, Damned Lies, and Statistics."

 

[Ozymandies]

Yes, it's probability, not statistics.

No, the streakiness is NOT abnormal, but very normal.

It is this streakiness of random events that has given rise to the concept of luck.

We wouldn't even talk about luck if it were not so.

You were unlucky, the streak did not work in your favor.

OZ

 

[Basara]

Except we've repeately proven, over the years, the streakiness we are seeing IS abnormal well above and beyond the normal streakiness one would see from a fair RNG....

Either something is wrong with the seed generation for the RNG, or several applications of the RNG results have flaws that skew results from being fair on the small sample scale, that balance out with similarly skewed results being on both sides of the mean.

If one takes a sample of 1000 coin flips, and the first 500 are heads, and the last 500 are Tails, the overall derived means of looking at the experiment will look totally NORMAL - but one could not deny on looking at the RAW DATA, something was seriously wrong with the data. While not as extreme as that one (one taken to the absurd extreme), THIS is the category of abnormalities we are reporting - and it is especially apparent when dealing with numbers that are significantly different from equal odds of either result.

 

[Plum Cake]

use +5 mining glove

 

[Ozymandies]

The OP said he failed 20 times, not 500. Not in the same category.

I have never failed 20 times in a row, its not that common.

Also, you can always just move to another tile, as that resets the random seed.

OZ

 

[Eslake]

The OP said he failed 20 times, not 500. Not in the same category.

I have never failed 20 times in a row, its not that common.

Also, you can always just move to another tile, as that resets the random seed.

OZ

They took location out of the equation back when they removed 8x8ing.

As to 20 vs 500, yes the difference is huge.
But the pseudorandomization UO uses is even less "random" than those that come as part of the various high level languages.

When assembler was still a required course for computer science majors, the creation of a random number generator was always part of it. UO's would be lucky to recieve a D.

If we could find a way to get UO to spew out numeric random values, I guarantee it would fail the arc-test 100% of the time with every set ranging from 1,000 to 100,000 values.

 

[Basara]

Finally, someone gets it... Thanks Eslake.

And note that when I used the "500" number I EVEN POINTED OUT I was taking it to an absurd, extreme, example. but even 20 failures (or successes) in a row, at what should be 52% (Grandmaster) or 57% (105) Mining should be throwing up major red flags for anyone familiar with computer programming. Probability and streaks are well and fine, but the people who keep dismissing the issue miss the point that there is no such thing as a TRUE random number generator with a computer. In other words, there is no "fair Die" for their probability scenarios - just an approximation of one, and one can judge the closeness a given psuedo-RNG is to APPROXIMATING fair, from looking at its output individually, and over stretches of use. And UO's Fails horribly.

I took Assembly Language back in 1986, then had to leave college for a while, and it was 1994 before I resumed computer courses in college (and of course, by then, most schools were transitioning to C++ from Pascal, and my 80s courses actually started with FORTRAN before moving to assembler). It's been so long I've forgotten everything I knew from those old classes. I remember basics of the syllabus, but not their implementation - including the RNG assignment.

 

[Silly Seadog]

Arrrr, methinks me still gots me Fortran and Pascal books somewhere!

 

[Traveller]

In other words, there is no "fair Die" for their probability scenarios - just an approximation of one, and one can judge the closeness a given psuedo-RNG is to APPROXIMATING fair, from looking at its output individually, and over stretches of use.

This is true. However I do not understand why people who advocate the following:

And UO's Fails horribly.

do not actually take their time to evaluate the statistical indicators specifically designed to evaluate the streakiness, i.e. the correlation coefficients. Oh well, not my problem...

 

[Basara]

You do not understand, because you continue to stick your head in the sand.

We keep providing you with statements of observed behavior - you expect us to provide physical evidence of something you'd have to be WATCHING OUR MONITORS to see. The only test you've CLAIMED to have done (the dice rolling) is one that is the least likely to show any problems, because it is such a simple test (and, again, we only have your word, just as you only have ours). There are a lot more complicated probabilities in nearly every aspect of the game - and that's before you start adding modifiers to the results of the "rolls" like luck bonuses for arty drops, the decision tree for BODs (which invokes at least a half-dozen different random rolls, PLUS is modified by the skill level of the requesting crafter), etc.

Quit talking technobabble, as you can't even TEST the things you refer to (was it you or another who stated they have never had a miner), unless you're an EA employee in the UO department of the Mythic division of the company, with access to the code. You are as much in the dark as we are, moreso in that you've never tested the very thing we are asking BE tested, to see if there's really something wrong. You're just trying to bury the real observations with big words, acting as if you have the slightest clue about what we're talking about, when you know you don't.

Note that the Hubble Telescope Mirror was "Perfect", during installation & launch - it wasn't until they saw the errors in it, that they went back and found out all the calibration & testing equipment had been mis-calibrated. The point: don't get so involved with testing one element, to the exclusion of others, when the others' failure can render the tests you did moot.

 

[Ozymandies]

Well, I don't think Eslake can prove anything about the arctest, either.

And what does the Hubble telescope have to do with anything?

Yes, failing 20 times in a row at 50% is over a million one. But if it NEVER happened, it wouldn't be very random. The question is not "if" it happens, but "how many times" does it happen.

If it NEVER happened, I would be worried. If it happens multiple times per day, I would be worried. Once every 3 months, no problem. How many times does UO roll the dice per day?

A good combined method will satisfy the chi-squared test for a million rolls or so (don't have a reference, sorry). I have programmed LCG method and done this test in Fortran language about well . . . sometime in the early 90s doing Monte Carlo sims of vibrating structures.

Hopefully, they have put some thought and effort into a good RNG.

Having said all that, I, too, have found myself screaming at the computer when smelting.

OZ

PS Just re-read OP who said he failed only 7 times in a row. That happens all the time. What was I talking about?

 

[Tycolby]

Success Rates in smelting are a joke. To make the success rate into 100% for GM miners would kill the already damaged market. It would only benefit the scriptors and push any Legit miner out of business. Yes its tough to find Val these days but there is ways to combat this.

First of all the random changing of resource made it seem at first that the scriptors would have a tough time in mining but soon that was proven wrong.
The more you dig in one spot the better it seems to find the ore you want.

Second of all with the removal of Brock from Mining, Scriptors have been unrestricted in they mining production for the last 7 months. Ever since the begining of Feb when the Devs turned them off. Any attending miner could out run Brock with ease. It was only the scriptors that complained that they were getting killed.

Ever since these changes have been done I have found that the old Miner Temp was unsuited for this new world. To improve you success rate at getting High end Ingots, one must rebuild their miners into Combat miners. Ones that can defeat any ore ellie. Using Prospector tolls and Gargoyle picks a miner can now bump decsent ore site up two levels and also have a chance of popping up an ore ellie. The ore alone on Ellies is worth the risk. 25 large piles of colored ore. After killing 163 Valorite Ellies, my smelting rate off the corpse is at 72.39% success. Thats 118 successful smelts. Remember this is directly from the corpse unto my Fire bug. I never once removed the ore from the corpse.

The Temp I use is as follows

100 Mining
120 Blacksmith
73.6 Chiv
99.6 swords
100 tactics
97.5 Parry
100 Anat
Human Male
using a Blackthorn's ellie slayer
I dont use any +5 mining gloves either

I have found that my success rate drops down nearly 47% success rate smelting from my pack. Thats with any size pile of high end ore.
Too many times I have seen a nice size pile of Bronze reduced to nothing and I have to say WTF! But you know what, anytime you smelt, its a gamble. To win Big you have to bet Big.

To change the success rate to 100% would kill the mining market to where only scriptors win. The everyday Joe would never be able to compete in that type of market.

Just My two cents.

 

[Restroom Cowboy]

Congrats you won the lottery

Heh...sometimes its interesting watching that RNG work.

 

[Restroom Cowboy]

Arrrr, methinks me still gots me Fortran and Pascal books somewhere!

What the hell are you doing with those on a pirate ship???

 

[Nyses]

That's... because... the... chances... are... INDEPENDENT.

Just because one roll gives you a failure doesn't mean the next roll will be any more or less likely to do so. It's the same exact chance each time. Even using the 45% failure chance, that's a fairly likely chance of failure, so your results don't surprise me. I've smelted a stack of 450 valorite smalls and managed to get 8 ingots out of it. Did I cry? Yes. Do I think the RNG needs changing? No.

The independence of individual rolls is what confuses people and makes them delude themselves into seeing patterns where there are none.

"Oh, I ate food just before I smelted and I got 100% smelt off the pile"

"Oh, I killed three pigs and suddenly I hit every single time on the next three monsters"

"I kissed the elf when I got his quest and he gave me a runic!!"

It... is... all... RANDOM.


Everytime this comes up I want to strangle people. Does NO ONE take statistics anymore?

Ya, talk about, ignorant , notice I did not say stupid. Go read a book on statistics some time, you will notice that if you flip a quarter 7 times, the chance of getting heads all 7 times is .5 x .5 x.5 x.5 x.5 x.5 x.5= .007813, which is almost 8 times in 1000, while it is possible, I have seen it happen FAR to ofter to be "chance".

 

[pavel.vesely]

to anyone who thinks to know anything about random number generators: You should test result on appropriate numbers. Testing smelting on twenty tries is joke. Go mine 100k ingots and compare results to what you should get. I dont see anything wrong in my own.

 

[Silly Seadog]

Arrrr, methinks me still gots me Fortran and Pascal books somewhere!

What the hell are you doing with those on a pirate ship???

Arrrr, an alien done gave 'em to me in the Bermuda Triangle!

 

[Warrior of Time]

Had they not of changed the chance to go back to the same spot and get 100% chance to get the same ore no one would have said a thing. Befor you had a chance to get more or less at the spot was bad enough. Then to have less chance to smelt it makes the chance harder yet. I think we need to have at the very least a chance to get the same ore at that spot or get 100% ingots for the amount you was able to get.

As it is we first have a chance of it being there. compounded by the amount you get If you get any. On top that we have the chance to smelt what we do get. There is too much chance to end up with nothing for the time spent looking and smelting what we need for bods.

Saying all that I think we need a brake someplace.