Chainmail in DN.

dice_t.gif
RNG Now more random!
dice_t.gif
 
i did that place for about.... 40 or so hours total, about 4 each afternoon while doing work, and never got a silk chest to drop, and coincidentally, never got a leather boots to drop. i arranged silk/chain/plate/leather, and got every single piece for a set of each, but never one of those two.


its random. =(
 
It's not really strange, flipping one coin once you are more likely to see results like "all heads" than when you flip 100 coins once.

If the old system had /ran 100 with fields of 1-10, 11-20... and the new one is /ran 1000 with fields of 1-100, 101-200... then the likelihood of any specific result is the same as before, yet will provide a better dispersion of total outcomes.
Click to expand...

Uh no. if you doo a 100 million times /ran 100 and only count 1-10 you will get around 10 million, same as when you do /ran 10000 and only count 1-1000.

This is the basic of chance and random.
 
Noktar said:
Uh no. if you doo a 100 million times /ran 100 and only count 1-10 you will get around 10 million, same as when you do /ran 10000 and only count 1-1000.

This is the basic of chance and random.
Click to expand...

the point was that if you flip a coin only 4 times, the coin may be heads up 4/4 times, regardless of what the odds are.. if you flip a coin 100 times, you're going to get a more even spread over time.. resulting in a more fair RNG
 
Noktar said:
Uh no. if you doo a 100 million times /ran 100 and only count 1-10 you will get around 10 million, same as when you do /ran 10000 and only count 1-1000.

This is the basic of chance and random.
Click to expand...

There aren't 100 million trials here, there are a much smaller number, making the rng seem less random. It's easier to get streaky results with a smaller range.

Increasing the field size will make it appear "more" random. There isn't any overall change, with enough trials it should balance itself out regardless, but over the short term and limited samples every player will see, this should increase the "randomness."
 
robopirateninja said:
There aren't 100 million trials here, there are a much smaller number, making the rng seem less random. It's easier to get streaky results with a smaller range.

Increasing the field size will make it appear "more" random. There isn't any overall change, with enough trials it should balance itself out regardless, but over the short term and limited samples every player will see, this should increase the "randomness."
Click to expand...

Tyrone is a statistics wiz imo.
 
robopirateninja said:
There aren't 100 million trials here, there are a much smaller number, making the rng seem less random. It's easier to get streaky results with a smaller range.

Increasing the field size will make it appear "more" random. There isn't any overall change, with enough trials it should balance itself out regardless, but over the short term and limited samples every player will see, this should increase the "randomness."
Click to expand...

was i rite? i have no f'ing clue when it comes to these types of things
 
Noktar said:
that has nothing to do with what was changed though
Click to expand...
if you take 200 people, and have 100 do 4 coin tosses, and 100 do 100 tosses, i guarantee you that you'll get more people who get 4 of the same out of 4 than 100/100.
 
Aaubert said:
if you take 200 people, and have 100 do 4 coin tosses, and 100 do 100 tosses, i guarantee you that you'll get more people who get 4 of the same out of 4 than 100/100.
Click to expand...
You're talking about large sample sizes, though. That's a very different beast than the size of the ranges for different items.

As n increases, (number of coin tosses, mobs killed, whatever) the results will get closer and closer to reflecting the actual percentages. n=100 will be far more diverse than n=4.

What Wiz did, if I understand it correctly, is increase the ranges so that instead of this:

/ran 10
0 = result A
1 = result B
2 = result C
And so on

It would look more like this:

/ran 1000
0-99 = result A
100-199 = result B
200-299 = result C
And so on

I don't remember my statistics well enough to be able to analyze this properly (it's been years since I skipped every lecture of that class), but my impression is that this makes it so that as n increases (number of kills), the results will more rapidly approach the probabilities than with the smaller ranges. I could just be imagining things, though.

Someone who actually knows statistics please feel free to tear this post to shreds.
 
Hasrett said:
You're talking about large sample sizes, though. That's a very different beast than the size of the ranges for different items.

As n increases, (number of coin tosses, mobs killed, whatever) the results will get closer and closer to reflecting the actual percentages. n=100 will be far more diverse than n=4.

What Wiz did, if I understand it correctly, is increase the ranges so that instead of this:

/ran 10
0 = result A
1 = result B
2 = result C
And so on

It would look more like this:

/ran 1000
0-99 = result A
100-199 = result B
200-299 = result C
And so on

I don't remember my statistics well enough to be able to analyze this properly (it's been years since I skipped every lecture of that class), but my impression is that this makes it so that as n increases (number of kills), the results will more rapidly approach the probabilities than with the smaller ranges. I could just be imagining things, though.

Someone who actually knows statistics please feel free to tear this post to shreds.
Click to expand...
you explained the principle i failed to, but we meant the same thing. ty.
 
robopirateninja said:
It's not really strange, flipping one coin once you are more likely to see results like "all heads" than when you flip 100 coins once.

If the old system had /ran 100 with fields of 1-10, 11-20... and the new one is /ran 1000 with fields of 1-100, 101-200... then the likelihood of any specific result is the same as before, yet will provide a better dispersion of total outcomes.
Click to expand...

No, it's actually related to the number seeder we use for our random and how it doesn't do well at low ranges.
 
robopirateninja said:
oic because /loc's seed values are too large :toot:
Click to expand...

/loc is not seeded. This thread would be done a tremendous favor if you stopped assuming things.
 
it was a joke ok, some gm said something about using /loc to seed the rng once and you told him it was the dumbest thing you'd ever heard of

has: a computer doesn't really have a way to choose a "random" number because computers can only follow a preset program. to "seed" a generator you give it some value that won't always be the same as a reference, and then it will perform an operation on it to find the value to be used.

this leads to a lot of stupid conjecture, like (in the above reference) someone thinking that if you stand at a certain angle to a mob you will crit on it more often, or something equally retarded.
 
Trust me, fight there enough and you'll have three or four before you figure that there's no point in collecting them anymore.
 
I thought most rngs use the time to derive value looking down a large amount of place values, of course.
 
You must log in or register to reply here.
Back
Top Bottom