Salvage is BAD for skillups

[Shismar]

This goes back to my above question. How many separately seeded RNG's do you think there are? I suspect there is at least one per zone. Beyond that...

It has been stated long time ago by a dev (sorry, no source), that there is one RNG per server. If this means a physical server, which may host several zones, or the virtual server I do not know. Much of the insides of EQ are not known to us players.
Also they ran some tests themself and found that the randomness was ok but only with a very large number of samples. Sadly, measures of "streakyness" are tricky business as even very good random sources can be streaky. Some software random generators try to reduce streaks but than they are not exactly putting out a random number anymore.

The idea of the OP is, of course, nonsense because there is no predetermined random table for your tradeskills and even if there was, the salvage check may as well skip over bad sections as well as good sections.

[Boleslav]

Salvage is fine for skillups. Your mileage will (and does) vary.

The randomness of the RNG comes up again and again and again. In one fairly recent thread (http://mboards.eqtraders.com/eq/showthread.php?t=19749) someone was suggesting that the streaky nature of the RNG meant that observations were not independent of one another. I suggested a simple test, and the person actually ground out quite a few thousand /randoms to try it. No surprise, it came looking pretty random.

In another thread http://mboards.eqtraders.com/eq/showthread.php?t=17015 , Tanker mentioned that there is one RNG thread (his term) per zone, and that there are many things that require a random number check. His opinion was that no matter how fast you clicked your /random, that there would be at least one other call to the RNG between your calls. This means that you aren't really experiencing sequential calls to the RNG at all. Tanker also mentions that he is not a statistics expert, and is not intimately familiar with how the RNG works.

I'm sorry that you believe Salvage has ruined your skillups. I don't believe it has.

Bolelsav Forgehammer
Paladin of Brell in his 67th Campaign
E'ci – Sacred Destiny

[Neebat]

Thanks for the information! Satisfies a large chunk of my curiousity.

[Kakg]

OK, one thing that seems to be lost to everyone when they talk about the RNG, streaks and tradeskills is......... the RNG is generating numbers for everyone in the game all the time.

Everything that happens in EQ uses the RNG for something. Whether it is a skillup check for tradeskills, a resist check for a mezz or a critical check for a combat melee. The RNG is throwing out numbers to everyone.

So when you get one skillup immediately after another its NOT a streak. The RNG didn't allocate a specific set of numbers just to you. In the gap between the first press of the combine button and the second press several thousand if not millions of RNG numbers have been created and used....... It's just total coincedence if one skillup follows another, nothing more.

To suggest that the Salvage AA uses up your precious streak of RNG numbers is just preposterous in the extreme.

You might as well stand on one leg, stick a celery stick in one ear, dip your left hand in cold ice cream and hum the star spangled banner out of your right nostril. Then claim that the reason you didn't get a skill up was because the ice cream wasn't Ben and Jerry's and then celery had started to wilt. It makes as much sense as blaming the slavage AA for a lack of skillups.

Random numbers are exactly that - Random. Yes, anything which depends on machinery to generate a number will never be completely random however when you get a human being to randomly hit a button to randomly pick a single number out of a steady stream of millions of psuedo random numbers generated by a computer then the result of that one number will be as close to completely random as makes no odds whatsoever.

So long as humans are involved in the combine process and they have no way to see or select the numbers to be used then the RNG is perfectly random.

[Bolas]

Very few people have sufficient mathematical background to accurately comment on this.

Actual random numbers are inherently streaky.

Have several people either flip a coin 100 times or write down what they think might happen if they flip a coin 100 times.

You'll be able to tell who actually flipped the coin and who made up the list of results. Why is that?

Somewhere in that 100 you'll have either 6 tails in a row or 6 heads in a row if it was an actual list generated by flipping a coin. But you won't have that if someone just guessed what the results would be. Actual random events are more streaky than most people realize or admit to themselves.

-Bolas

[Bolas]

One more thing... if there's a different random number generator per zone... some of the instanced zones are having problems with their random numbers. Try doing tradeskills of trivial or near trivial combine in draniks' hollows and you'll see what I mean.

Failed a bunch of jewelry cuts on flow stone augs in dranik's hollows. Enchanter with max skill & aa's should not be failing every combine.

I stick to PoK for pricey combines now.

-Bolas

[chrisian]

Try doing tradeskills of trivial or near trivial combine in draniks' hollows and you'll see what I mean.

Failed a bunch of jewelry cuts on flow stone augs in dranik's hollows. Enchanter with max skill & aa's should not be failing every combine.

*gasp* you mean on a random event you got a bad streak?!?
should take your own advice and not comment on the issue without sufficient understanding.

correlation does not always imply causation.

i'd assume draniks hollows has nothing to do with it... if you look at the other threads here there are a lot of people reporting issues with certain combines (like aug cuttings) where the success rate formula seems to no longer be accurate, and they're doing it with sufficient sample sizes to raise suspicion (generally reporting 75ish percent success rate on 100+ flowstone cuts that should have been 95% success rate).

[Kakg]

Very few people have sufficient mathematical background to accurately comment on this.

That's never stopped anyone commenting before.

[Phantron]

From what the devs said before they can certainly adjust the min chance to fail on any recipe up or down. That'd seem to be a far more plausible explanation (that some recipes have higher than normal min fails) than having RNG that varies depending on the time of the day, your EQ name, which container you're using, and the zone you're in.

[Neebat]

From what the devs said before they can certainly adjust the min chance to fail on any recipe up or down.

Careful! I got flamed by a moderator for suggesting that very idea. Here...
http://mboards.eqtraders.com/eq/show...t=21927&page=3

I only point this out because I'm think I saw a dev say that too, but I'm too lazy to track it down in my own defence. I'm hoping you'll do the search.

[Phantron]

Well devs have said they can decrease min chance to fail dynamically (on a post going over min fail rate with regard to raw skill points but not related to mastery AAs). It seems to me if they can fine tune a reduction in min chance to fail it'd be extremely unlikely that they can't even raise the min chance to fail on certain recipes if they wanted to.

[Boleslav]

Backing Phantron up here...

In this thread: http://mboards.eqtraders.com/eq/showthread.php?t=21468

Tanker wrote:
"Ok, I looked into the code, and it's working normally. The way it works:

- For normal recipe, min failure chance is 5%.
- For recipe where your skill (unmodified) is more than than the trivial, the min failure chance is reduce by 1% for every 40 points over the trivial your skill is. e.g. your skill = 250, recipe trivial is 100, diff = 150, 150/40 = 3 (almost 4, but not quite), so min failure chance is 2% (5-3). NOTE: this is an integer divide. It will only take off whole percentages... which is why this is 2% instead of 1.25%.
- This min failure chance can be overridden in data by the designer on a per recipe basis, replacing it with a higher value."

Looks pretty definitive to me.

Boleslav Forgehammer

[Neebat]

This min failure chance can be overridden in data by the designer on a per recipe basis, replacing it with a higher value.

So they actually said specificly the failure rate can be RAISED, but didn't say anything about lowering it. That's a handy replacement for the 300+ trivials.

Thanks for finding the link, Boleslav. I didn't think I was imagining it.

[Zanaspus]

Very few people have sufficient mathematical background to accurately comment on this.

Actual random numbers are inherently streaky.

Have several people either flip a coin 100 times or write down what they think might happen if they flip a coin 100 times.

You'll be able to tell who actually flipped the coin and who made up the list of results. Why is that?

Somewhere in that 100 you'll have either 6 tails in a row or 6 heads in a row if it was an actual list generated by flipping a coin. But you won't have that if someone just guessed what the results would be. Actual random events are more streaky than most people realize or admit to themselves.

-Bolas

Well, I have sufficient mathematical background, and I can say that your statement is false. If I know beforehand that 6 ones or 6 zeroes will be generated randomly from a random process, then the process is not random.

I'm not trying to flame, I understand precisely what you're implying, I have just always been a stickler for minutia.

[Phantron]

It'd appear to me that the no fail combines would be a good example of where the minimum fail chance can be lowered. It is possible they just have some special no fail flag but that'd be a pretty convoluted way of doing things consider they can just change the min fail chance.