200+ above triv failure rate question

[Cloud]

I tried the petition a GM and claimed that i failed the last cleric epic 1.5 combine but he just laughed and said nice try. /snicker

[Soriano]

200+ and No fail

I got directed to this thread from a different forums...

The 5% chance to fail still exist...regardless of how far above trivial you get...

an allowed 5% chance to fail programmed into the formula does not mean you will fail 5% of the time...

the thing to remember here is the law of averages...and the fact that your skill and the trivial is the main components to the calculation in determining the success of a combine...As you skill increases the law of averages for you to hit the 5% window get smaller and smaller...but it does not in any way mathmatically eliminate the possibility of failure.

Therefore the combine NEVER becomes no fail...it might seem that way, but that base 5% chance to fail is still there, you have just gotten so far above the trivial that the law of averages makes it VIRTUALLY no fail...but you can still fail...I know this for a fact because I have...and again it was not some ghost in the machine...lol

when a formula uses a RNG the way this formula does it will play havoc with the law of averages...and you could write a program to simulate the exact formula known for tradeskills, run it 1 million times and do some analysis, then run it another 1 million times and get varying results...

[Taushar]

The 5% chance to fail still exist...regardless of how far above trivial you get...

Absolutely untrue. I have heard directly from developers that if the unmodified skill is above 200, you will not fail based on their formula. It's absolute. However, there are few exceptions to this rule. The DoN cultural armor has hard cap on success, I believe it's 70% max. Other than these exceptions, if your unmodified skill is 200 above the trivial, you will not fail in combines.

Taushar

[Twistagain]

The 5% chance to fail still exist...regardless of how far above trivial you get...

Not so. A 5% chance to fail exists unless one of the following is true:
1) The recipe is specifically NO FAIL.
2) Your skill level is 40 points higher than the trivial. For each 40 points, the chance of failure is reduced by 1% of the original, so 4% at +40, 3% at +80, 2% at +120, 1% at +160, and 0% (NO FAIL) at +200. Note, that I am not sure if it's actually AT 40 points or ABOVE 40 points.

an allowed 5% chance to fail programmed into the formula does not mean you will fail 5% of the time...

No, but it does mean that, on average, you will fail 5% of the time. You can, of course, fail 0% for a run...or even 100%, but you will likely come close to the 5% over time.

As you skill increases the law of averages for you to hit the 5% window get smaller and smaller...but it does not in any way mathmatically eliminate the possibility of failure.

Therefore the combine NEVER becomes no fail...it might seem that way, but that base 5% chance to fail is still there, you have just gotten so far above the trivial that the law of averages makes it VIRTUALLY no fail...but you can still fail...I know this for a fact because I have...and again it was not some ghost in the machine...lol

I have absolutely no idea what you mean by that. If the "window" gets smaller, then the 5% chance isn't still 5%.

Now, of course, I have not seen the code that does this reduction. But the devs have told us many times that this is the way it works. I have yet to see a first-hand account of an incident that disproves it. I've seen several "a friend of mine told me" or "someone ingame said" accounts. You said that you have failed...what was your raw skill and what was the combine you were attempting?

[Soriano]

Lol

Unless the formula has been modified (It might have been at the time skill caps were raised to 300) although the existing formula would have worked from 250-300. Then it is TRUE...

Back in the day the original programmer to the tradeskill engine was VERY tight lipped about the formula and left most things to total speculation...

A couple years back, the programmer who took over the tradeskill process had a Q&A at one of the fan faires...I was at it, and he went to a deep a level about the formula for both success rates and skill up rates...He also gaves us (at the time the exact formula) for both...

When it came to skill ups formula he was going strictly from memory on what drove the 1-4 (type factor) used in the skill up formula...from his memory all tradeskill used the 4 but every skill you can raise in EQ used the same base formula...The melee skills were the 1, spell skills were the 2, he gave no example for the 3 as he could not from memory give us an example and tradeskills were 4....

The formula given at that time had a mathmatically 5% chance to fail on a trivial combine...

In the last month my blacksmith with a 250 raw skill failed at a metal bits combine...

IT DOES NOT become no fail mathmatically....

[Soriano]

I have heard that silk swatches are no fail, and I personally can not ever remember failing at one and I have made 1000's of silk swatches...Celestial Essence is another no fail combine...At zero skill is is still no fail even though the trivial is 16...this just means you can raise your skill zero to 16 on celestial essence...

I have even gone from 0-16 making silk swatches without a fail...(I believe it was 16)

The smithing combine for the Xanthes Earring is also no fail, at zero skill you will not fail it, and the trivial is higher than 0 because I have actually gotten a smithing skill up from 0 to 1 doing the combine...

I am not saying in my above post that the formula has not changed, if it has so be it...but the original formula allowed a 5% chance to fail any combine the 5% a mathmatical possibility...and yes over large numbers of combines it would ration out based on the law of averages...but it would also explain how a person with over 200 skill points above a trivial would fail, just not very often...

I am not suggesting some conspiracy to LIE to the community by a dev...I have heard many things in the 6+ years I have played EQ from DEVS, some true, some not, some were true at one time and changed...

But I might suggest that since the original programmer to the tradeskill engine has been long gone from sony, that there has been alot of mis-information reported and repeated...

But when I hear a dev or anyone else try to explain the rare occurence as some glitch or bug...It just is not so...A bug is a problem in the designed logic that will only occur when some unknown circumstance hits the formula...Not randomly it worked this way this time and another way another time...it always works the same way!!!!

[Verdandi]

Soriano, we have been told in no uncertain terms by a developer who is still with SOE and just left the EQ team very recently that the way Twistagain described it is correct. The thread is Fan Faire June 2005 Write-up (Plus Tanker handout) which was based on notes from things said by Dev-Tanker, and has some further input by him. At that time Tanker had been responsible for the coding for various things, including tradeskills, for some time. He knows it well enough to be trusted.

I can't find the specific threads but I know we've been asked to /bug it when failing recipes that are 200+ past the trivial.

At this point we have every reason to trust Tanker and no reason not to. Unless you can come up with specific examples and/or data, I think we need to stay away from questioning his credibility, lest it stray into personal attacks and/or SOE bashing.

[Soriano]

my post does not challenge anyones credibility...

MY point is this in one statement you say we have been told in no uncertain term something works this way...

then it get followed by to /bug something that does not follow that premise...

if ONE TIME the formula does not hold true to be no-fail at 200 above trivial, then either the formula is incorrect...or the point that a combine becomes no fail is incorrect...

Because I make reference to a 6 year period in where I have heard true and untrue statements, for you to take the leap that my statement is about the one particular dev you mention is absurb...

My point is about 1 example that I know is true, because it happened to me, and within the last month...

My 250 Blacksmith failed a combine of metal bits...

That is well over the 200+ skill difference, and the combine as of early january the combine was not no-fail...

If it ever happens again I will screenshot it...

I even make reference above that there may have been a change to the code since I had the direct conversation with a dev (over two years ago)...

[Bobaten]

The "very trivial" combine code was added 2-3 years ago. We have known the 200 above rule since then, but the specifics (4% at 40+, 3% at 80+, etc.) weren't given to us until the June fan fair.
You are correct that the devs used to be very tight lipped about the formulas behind the scenes, but that has changed greatly over the last few years. I have seen a few isolated reports of people occaisonally failing combines that are more than 200 below trivial, and your guess is probably right, the RNG probably gets it's 1 in 10000 (or however many digits they use) roll of 0000. In any case the chances of failing a combine more than 200 below your skill is far less than the 1 in 20 that a 5% chance would give.

[Soriano]

Aye...I will certainly agree with the amount of information today compared to several years ago has greatly improved...no question about it...

What led me to this thread was from a different forums where some one states that ANY combine where your skill is greater than 200 above trivial is NO FAIL...

I had a recent combine where that did not hold true...

I finally found the thread here where they have the notes from someone who talked to the developer and shared them with the community...

I think the real problem is just that it is an interpretation from a conversation...

What makes EQ a great game with great core mechanics, is the whole unknown created by the RNG, if not for this a sword that hit a mob with a specific AC and AGI would hit every single time for the same amount of damage...and on every 20th combine you would get a skill up...The fact that the formulas allow for these variances is what makes the game so good...

My point in the other thread, is that the formula does not support the premise that once your skill is greater than 200 higher than trivial a combine becomes no fail...

And in fact that the law of averages, the fact that both your skill and the trivial are part of the calculation and there there is that RNG factor means when the dev notes claim the INTENT is for all combines where skill is greater than trivial + 200 be NO FAIL...But what is really the case that as your skill progress higher and higher above the trivial that by the point you reach trivial + 200 the possibility of failure is so minimal that the combine becomes virtually no-fail, but not mathmatically no fail...And the times that you do fail something that far below your skill is not random weirdness in the code, but as you put it above you hit that 1 in 10000...

The formulas we have been given are to illustrate how a very complex code works at its core and not the actually coded formula with every variable accounted for...

[Soriano]

I have absolutely no idea what you mean by that. If the "window" gets smaller, then the 5% chance isn't still 5%.

The 5% chance to fail is on a trivial combine...if your skill is 200 and you are attempting a combine that trivs at 200 you should only fail 5% of the time...

As your skill progresses higher and higher above the trivial, the window for failure gets smaller and smaller...

[Wyvernwill]

Actually, you reach the 5% failure rate before you hit the exact skill that it trivials at.

I have seen things that i've been 200+ skill over and to this day, I have yet to fail one, besides doing thousands of varying combines. I have not one complaint about the current system because atleast I am not almost gauranteed to fail 1 combine out of 20 every single stack I do.

Reading through your posts, I miss the point other than a rant about code.

[Ngreth Thergn]

Somewhere along the way from the formula you got changed

They definitely added something so that every 40 points above trivial the change to fail gets reduced by an additional 1%. I have looked at the actual code over Tanker's shoulder.

So at 250 skill a trivial 50 combine is no-fail. You have a 0% chance to fail **unless** that combine is specifically given a minimum fail chance (though I know of nothing trivial 100 or lower that has a minimum fail chance set)

Now that said.

There seems to be a bug that no coder has been able to track down.

Even if there is a "0%" chance, even if the combine is set to no fail, there is a very rare chance that it will still fail. My personal guess is it has something to do with a bad lag pulse, but I am not sure.

For example. Silk Swatch. It is a no fail combine. At 0 skill you should be able to make the combine. I personally failed it ONCE... (just once in many thousands of combines...)

The code is actually really set to lead to a 0% chance if you are 200 above the trivial. Just a minor random bug somewhere makes that 0% chance a 0.01% chance (or some such very small chance, but not 0) to fail.

[KyrosKrane]

Ngreth has posted somewhere that there is a bug with the no-fail combine code. When the bug surfaces, it causes a no-fail combine to fail. However, the bug is so rare and so obscure that most likely they'll never get around to hunting for it, let alone finding it and fixing it. This likely accounts for the odd reports of no-fail combines failing.

We know for sure that combines of trivial 15 or less are supposed to be no fail. We know for sure that if your skill is 200 or more above the trivial, the combine should be no fail. However, due to that bug, there is a very, very, very low chance you'll still fail.

Edit: Bah. Between the time I opened this window (yesterday) and the time I made my post (today), Ngreth snuck in his reply. Ignore me. =)