Is it possible to go from 1 - 250 on 250 combines?

[Praxx]

That's probably true for 250 consecutive numbers on the same sequence on the RNG, but some people do 10-40 combines at a time...

Right, my point was that by design it really isn't random. Upon further thought, depending on how the RNG actually works, if you were able to estimate the outcome then you could conceivably obtain 250 skill within 250 combines.

[Nairn Niteraven]

Possible?

Most of what has been said is true. Is it theoretically possible? Of course. Is it possible Britney Spears will call me tonight and beg me to spend the night with her and give me 1 billion dollars for doing it? of course.

Coming back from fantasy land, this is purely academic. It ain't gonna happen in either case.

[Harzel]

Well, yeah, but I would take the 250 straight skill ups. However, for Britney, I think I would charge 2 billion dollars, minimum. :shock:

[Harzel]

I think the ability to "make" an item is random based on your skill. But then again, if it's based on skill, it's not completely random.

They want us to believe that skill-ups are random, but they are clearly not. There are percentages applied and possibly limits imposed.

Although I know sometimes people use it that way, "random" does not mean something has a 50-50 chance of happening. Whether I get a 1 when I roll a 6-sided die is "random", even though it has only a 1 in 6 chance of happening. "Random" just means nondeterministic, or not completely predictable.

Although it seems clear that the probability of a skill up is usually not 50%, getting a skill up does seem to be random, that is, it is a matter of chance. The thing that I, at least, found surprising is that the results on successive attempts seem to be so obviously correlated. (In other words, I sort of expected that successive attempts would be independent, like dice rolls, where the probabilities remain the same from attempt to attempt.) However, for both success/failure and skill-ups, the results seem very streaky. Either VI is using a pitifully bad PRNG, or else they have deliberately introduced dependence because they thought, well, who knows what they thought.

[Yalum]

Either VI is using a pitifully bad PRNG, or else they have deliberately introduced dependence because they thought, well, who knows what they thought.

There is a third option, that human brain's pattern matching algorithm is prone to false positives. This option also has the benifit of being absolutely true and well documented.

[Gello]

What about a lucky sob that happens to get mayb 10 straight skillups in the 200 - 250 range? That's not too much to ask eh? But no one has, to my knowledge, ever had a great run luck that.

That's because we don't hear them. The evil RNG summons a dozen tailors with coldain-temper-stained hands and quietly removes the evidence.

[Mystiqie]

Well, I posted in the Jewel Craft forum just now about the luck I had last night, I went from 232-236 in 6 combines And the first of 6 was a failure... so go figure.

I know I'm gonna pay for that though... please don't drag me off, I will scream :? Really, don't hurt me, I cant afford it!

[BuilderOfCastles]

Maddeningly Unrandom

I have done a great deal of studying the results of the pRNG; Both my numbers and others. From the analysis, I claim that:

The pRNG is maddeningly Unrandom

The reasons for this:

I believe all EQ operations are integer (1,2,3...) not floating point (1.23e-6). This is because integer operations are at least 3x faster. (All numbers in EQ are stored as an integer; item #014534, smithing 206, 408pp,...)

If you stay with only integer operations, then your case statements (a program switch board, ie, if 1 do this, if 2 - 4 do that, 5 or more do something else) are very easy to construct.

if you use a floating point number anywhere in your code, you will need to convert it from an integer and then back to an integer. Two extra operations.

Integers are like buckets balls fall into. If you have a plinko machine with a slightly bent nail, the randomness gets skewed. A couple of bent nails, and things get really skewed.

When tweaking a pRNG algorithm for speed, you end up bending nails. (excuse me for the untechnical term ^_^) Tweaks that are likely to be in the system: If things are too busy, return 0. If things are too busy, return the last number again. Use a smaller seed number. (greater speed, less randomness) Take the number returned and split in half, because you really need two numbers but for speed only want to call rand() once. (this produces numbers that are high coordinated).

Because of seeing signs of these, the answer to the forwarded question is, "No it is not possible."

Further thoughts:

I believe that in the range 190-210, for one of the harder skill up items, that there is a 1 in 64 chance of getting a skill up. Fortunately, the pRNG is streaky, so you are likely to get results of 64, 63, 64. (Thus 2 skill ups close together)

A very weird oddity: In smithing, I have had an inordinant amount of skill ups on my first combine of that day. Explain that one?

[Iustus]

Depending on exactly how the EQ RNG works, it is likely that it is possible to skill up 250 times in a row, just so unlikely it will almost certainly never happen.

The RNG has been changed a few times as I understand it. One common aspect that is often used to achieve more random numbers is to use something about user input in the calculation. Something like when you click the combine button as one of the variables in the calculation.

Of course this is tricky in an online game like EQ. At one time, the number that was used was local to the player's clock. I have heard that someone reverse engineered the code and used a hack that would wait to press the combine button until the computer clock was such that it would always be a successful combine. A friend of mine said he actually saw this working the first year of EQ.

They changed the RNG and many other things to be more secure, using the server, not the client, so such things are no longer possible. So now, it really depends on how they changed it. If the current RNG still uses some truely random factor as an input, then there is some tiny chance to get that many in a row. If it is a pure computational RNG, then almost certainly not. However, if some outside truely random input is used as a seed, then there is some slight chance.

On the 1000 non-trivial combine without a skill up, I have not heard of that one, but I did about 700 non-trivial combines without a skill up at about 198 smithing before the next skill up.

-I

[ialaman]

Possible?

Well if you zone after each combine, and/or wait a while between combines, you can try to make it more truely "random"

However, let's be generous and say that, overall, your chances of getting a skill up are 1 in 10. (it's a nice round number)

That means you have to be that one in ten. 250 times in a row. In other words, the odds of it happening are 10^250

Now for ease of computing, let's say there's 500,000 subscribers. Now let's start getting extreme and say every one of those has 8 characters. (multiple accounts = multiple subscribers) That's 4,000,000 characters. Now let's say that every one of those characters tries to GM all 7 tradeskills, that's 28,000,000 attempts at going up to 250, which is 28 * 10^9, which is indescribably lower than 10^250.

Possible? Yes, though I think your odds are far higher of getting hit by a rogue comet colliding with the earth.

[Linea]

took me well over 600 combines on faydwer shakers to go from 183 to 184 in brewing... took me six casts to land 197 198 199 and 200 in fishing /shrug