How does the scroll system work?

Is the % that they tell you the ACTUAL %?
Or is it just a rough amount?
Or is it not that amount at all?
How can we ever verify?


I'll tell you why I ask this question.

I can say, without a doubt, whenever I have used 10% scrolls on something unneeded or a junk item, it has almost ALWAYS worked.
I can say, without a doubt, that I have used 16 50% scrolls on something good, like a unique item, and only 4 of them succeeded, which means that is a 20% success rate.

Of course, the more you scroll in a row, the more accurate the actual % would be, i.e. using 100 of 50% scrolls.

I know that Nexon gave out a reason as to why you shouldn't expect 50% scrolls to work exactly 50% of the time, and that is called "the gambling falacy". Where probability does not affect the genuine outcome. This means that even if you flip a coin 50 times, you will always have a 1/2 chance to have heads or tails, even if you flipped 49 heads prior to the last flip.
However, that is only slightly true. It's accurate to say, that flipping 49 heads/1 tail is equaling flipping 50 heads has the same statistical SEQUENCE outcome. But it is VERY false to say that flipping 50 heads in a row is more probable than something like 35 heads and 15 tails. This is the OVERALL outcome.

The gist of this means that even though flipping the following: heads, heads, heads, tails heads, heads, tails = heads, tails, heads, tails, heads, tails...

Does not mean it is equally likely to have an outcome of: heads, heads, heads, heads, heads, heads

Especially when getting into more consistent and large amounts of flips.


I also did a physical test, I flipped a coin 50 times - 27 heads, 23 tails. Pretty even, but I only did it once for now.
However, this also made me realize that IRL variables (such as how hard you flip, the consistency of how you flip, where the coin lands, etc) makes it so that it isn't a genuine 50/50 chance.

Which further puts me in a position to question whether or not these "% scrolls" reveal their true percentage.

In-game programming SHOULD allow for more precise/accurate/consistent chances since there are no outside variables that will effect the outcome of fail or success; heads/tails.
Yet, it is somehow less accurate, and less precise than probable theory/hypothesis and physical testing.

As a player, all you really need to know is that the rates are more or less what they say they are.
Even if they were not, there is no method that you could employ to alter the process in you favour. At least by use of any normal or legitimate means, anyway.

If this more of an inquiry in programming and math, then by all means keep spinning away.

Really, unless Nexon itself releases all of their RNG systems on how their percentages are generated, we will never know for certain. From a semi legal/ethical standpoint, the percentages should be a fix percent they work at, regardless of any outside variables: item being scrolled, type of scroll being used, location of character when scrolling, etc.

With that being said however, it is totally possible that the servers could house within their item databases a "rarity" rank or whatever you would care to call it that impacts the probability that a scroll will fail/pass. To take things further you could even say that those probabilities are more than likely directly linked to drop chance, seeing as how most people tend to claim the more rare the item, the less likely the scroll seems to work. This would best chart out as an inverse variation, as the item rarity goes up, the actual scroll chance to work goes down. This chart, of course, would only apply to scrolls under 100%, and to not make things too unfair or noticeable would be a fairly small curve to it. Adding in event and quest item's "rarity" as seen fit by Nexon.

But, that is total speculation and will never be verified without the raw equations from their own servers.

Essentially, yes. Also, 100% scrolls always work.
(i think one potential scroll was glitched causing it to always boom a while back?)

Seed theory bro.

There's some problems with the RNG. I had over 90% success rate across samples larger than 100 with 70% potential scrolls when I scrolled as fast as I could using Legendary Spirit. This happened on 3/3 samples and never produced items above Rare. I was potting Phantom cards after Phantom release.

Quote:

---

Originally Posted by MountLag

There's some problems with the RNG. I had over 90% success rate across samples larger than 100 with 70% potential scrolls when I scrolled as fast as I could using Legendary Spirit. This happened on 3/3 samples and never produced items above Rare. I was potting Phantom cards after Phantom release.

---

Sadly all that can really definitively prove is that you had fairly decent luck/results when doing those. Without being shown the formula they use to generate these numbers, any set of data values, no matter how many there are, cannot, beyond a reasonable doubt, prove anything. You can make assumptions as to if their system stays true to the claimed percentages or not, but that is about all you can do.

Quote:

---

Originally Posted by LeoDirk

Sadly all that can really definitively prove is that you had fairly decent luck/results when doing those. Without being shown the formula they use to generate these numbers, any set of data values, no matter how many there are, cannot, beyond a reasonable doubt, prove anything. You can make assumptions as to if their system stays true to the claimed percentages or not, but that is about all you can do.

---

I had success rates closer to the normal 70% when scrolling normally though.

The way I've understood it is this: whenever you use a scroll (or any other action that can have different effects at random) a "ticket" (my own word for it) is sent to the server, telling it what you did. Now, when you use a scroll (let's say it has a 60% chance to add +1 stat, 40% to do nothing) it sends a specific "ticket" to the server (let's say the "ticket is the number combination 1006723920). Whenever you use a new scroll it sends a different "ticket" (http://en.wikipedia.org/wiki/Random_number_generation) to the server. The server then decides what "ticket" to send back. The original ticket will have told the server what item it is, so the server knows what combinations are possible and hence which tickets could be sent back. In this case, out of all the tickets possible, 60% of them will say +1 stat, and 40% of them will say do nothing).

So the result is completely up to the server's (p)RNG and your client has no control over the result, except for what ticket is sent to the server. In order to prevent someone from figuring out which tickets result in something good, and hacking the client to send those tickets, Nexon would have the server periodically reset the table that assigns a result to a specific ticket (so say at time A the ticket 1374701470 is a pass, but at time B 1374701470 is a fail).

It's impossible to determine anything that goes on server side*, such as what % of tickets do what (it says in the item description sure, but it only matters what the server thinks, which could be different as you said) or which tickets combinations do what, or how often the tickets are reset, etc etc

I have more hypothetical inquiries into the system but I've spewed out enough potential bullpomegranate already, as I said the server is a mystery. In the end all you have is to trust that what the client says is what the server thinks.

Btw in regards to what you said about the outcome of the coin depending on how you flip it,etc. if you keep going down that logic you'll have to deal with the uncertainty principle limiting your prediction (http://en.wikipedia.org/wiki/Laplace's_demon) You'll always only have a probability of something happening

- The only way you could actually know something like this is by looking at their code. No one here, or anywhere else who has not seen the code will factually know the answer to this question.