Anhrez

Plentiful 1, rank 5 stats and outcome

Recommended Posts

So I am fresh into the patch and wanted to share my first set of stone harvesting with you. need to get the 12 Premium for the Laborer Stone built

Slate Plentiful 1  - 8 Nodes harvested

8 Grey/Red

2 White 

16 Green

4 Blue 

Critical Chance 40.3087 | Critical Amount 4 

 

Its a small sample but an odd one right? 

 

Share this post


Link to post
Share on other sites

We had 9 legendaries drop from one node. Unless someone smarter can explain it to me how can 9 out of our 10 crit amount rolls hit legendary when the chance of that happening is only 2%? I tried to engage Blair on how the system does RNG and how it seems to follow a pattern and it gets gets stuck. He didn't seem to think there was anything wrong.

 

 

Share this post


Link to post
Share on other sites
17 minutes ago, Tark said:

We had 9 legendaries drop from one node. Unless someone smarter can explain it to me how can 9 out of our 10 crit amount rolls hit legendary when the chance of that happening is only 2%? I tried to engage Blair on how the system does RNG and how it seems to follow a pattern and it gets gets stuck. He didn't seem to think there was anything wrong.

Did Blair explain how their RNG works? The only way i see that happening (did you record that? That is basically seeing the impossible happening right in front of you!) is if they are calculating what drops by steps. First check the one with the highest drop then re-checking the other drops one by one and if it gets to the last one (the legendary drop) drop it.

Lets imagine we have W: 50, G: 25, B: 23 and L: 2. Doing it by steps we would have:

  • 50% drops White
  • 12.5% Green
  • 10% Blue
  • 27.5% Legendary

This is probably wrong and has nothing to do with it but this is all the sleepy me can think of. To the bed we go....

Share this post


Link to post
Share on other sites
4 minutes ago, BarriaKarl said:

Did Blair explain how their RNG works? The only way i see that happening (did you record that? That is basically seeing the impossible happening right in front of you!) is if they are calculating what drops by steps. First check the one with the highest drop then re-checking the other drops one by one and if it gets to the last one (the legendary drop) drop it.

Lets imagine we have W: 50, G: 25, B: 23 and L: 2. Doing it by steps we would have:

  • 50% drops White
  • 12.5% Green
  • 10% Blue
  • 27.5% Legendary

This is probably wrong and has nothing to do with it but this is all the sleepy me can think of. To the bed we go....

He touched on some points in the ACE forum. We didn't get into intricate detail other than they've done simulations and the results come out as expected. But I suspect that maybe there is a lack of understanding on how tricky RNG really is. Its a you don't know because you dont know kinda thing. You can't recognize that there might be an issue because you lack the understanding to identify it.

Of course he might understand it just fine and I'm an idiot. Very real possibility.

Share this post


Link to post
Share on other sites
Posted (edited)
2 hours ago, BarriaKarl said:

Did Blair explain how their RNG works? The only way i see that happening (did you record that? That is basically seeing the impossible happening right in front of you!) is if they are calculating what drops by steps. First check the one with the highest drop then re-checking the other drops one by one and if it gets to the last one (the legendary drop) drop it.

Lets imagine we have W: 50, G: 25, B: 23 and L: 2. Doing it by steps we would have:

  • 50% drops White
  • 12.5% Green
  • 10% Blue
  • 27.5% Legendary

This is probably wrong and has nothing to do with it but this is all the sleepy me can think of. To the bed we go....

He said they built a tool to run 10,000 random events to see if there was a problem, and numerically the found a very small and expected variance <.00001% I think was what he quoted.

He did not however say that the tool tests specific interactions of events. It is trivial to build a Random.Range(0, 100)  call in unity that cycles itself 10,000 times and records the values.  What you have answered with something like that is "does unity the function Random.Range generate each number in range within expected variance values". Just because in 10,000 tries of Random.Range (1,100) you end up with roughly 100 hits on each number, does not mean your RNG is not creating predictable patterns.

Some things you have not tested are.

  • Does Unity have a sequence pattern based on seed (initial state)?
  • Are you using a predictable seed (initial state), so end up generating the same series of numbers? This needs to be checked when and how seeds are being used to create a new initial state.
  • Are you making the mistake of re-initializing state mid program? Re-seeding seems like a good idea, but can end up simply putting you back into the same state you just used.
  • Does the algorithm create frequent predictable patterns in large sets? (Short period).

For example the Mersenne Twister RNG, which is a pretty good RNG with a period of 219937 − 1. and is the one used by PHP with the mt_rand() function, but still has a few well know problems.

Quote

It can take a long time to start generating output that passes randomness tests, if the initial state is highly non-random—particularly if the initial state has many zeros. A consequence of this is that two instances of the generator, started with initial states that are almost the same, will usually output nearly the same sequence for many iterations, before eventually diverging. The 2002 update to the MT algorithm has improved initialization, so that beginning with such a state is very unlikely.

Recognize the behavior we are seeming to experience in that bolded sentence?  

Frequent occurrences of the same patterns, that for us just happen to be strings of success and fail events? With as few events as we generate, it would take quite some time to get to the point of divergence, if the Unit Random algorithm has the same problem. If they are reseeding, then it's entirely possible to never reach the point of divergence before being put onto the same pattern. It would be easy to think you had different seeds, but output nearly the same sequence for many iterations.

RNG is hard, and it's easy to get wrong, even when you think, and it seems to test correctly.

2 hours ago, Tark said:

We had 9 legendaries drop from one node. Unless someone smarter can explain it to me how can 9 out of our 10 crit amount rolls hit legendary when the chance of that happening is only 2%? I tried to engage Blair on how the system does RNG and how it seems to follow a pattern and it gets gets stuck. He didn't seem to think there was anything wrong.

If the 2% is correct (2/100), even getting 5 of them in a row would have odds of 1/312,000,000 and then 4 in a row, 1/6,250,000 for a total of (1.95313E+15).  As a comparison, Lotto Max with at potential 60 million jackpot in Canada has 1/28,633,528  chance per $5 ticket with three plays.

9 in a row or 9/10 (the numbers are almost mathematically identical) would have odds of 1/ 1,953,125,000,000,000 chance's.  1.95313E+15

Numbers that big are basically "that never happens" kinds of numbers.

If you have video of that happening, that is proof positive that there is something wrong with the RNG.  

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites
2 minutes ago, KrakkenSmacken said:

He said they built a tool to run 10,000 random events to see if there was a problem, and numerically the found a very small and expected variance <.00001% I think was what he quoted.

He did not however say that the tool tests specific interactions of events. It is trivial to build a Random.Range(0, 100)  call in unity that cycles itself 10,000 times and records the values.  What you have answered with something like that is "does unity the function Random.Range generate each number in range within expected variance values". Just because in 10,000 tries of Random.Range (1,100) you end up with roughly 100 hits on each number, does not mean your RNG is not creating predictable patterns.

Some things you have not tested are.

  • Does Unity have a sequence pattern based on seed (initial state)?
  • Are you using a predictable seed (initial state), so end up generating the same series of numbers? This needs to be checked when and how seeds are being used to create a new initial state.
  • Are you making the mistake of re-initializing state mid program? Re-seeding seems like a good idea, but can end up simply putting you back into the same state you just used.
  • Does the algorithm create frequent predictable patterns in large sets? (Short period).

For example the Mersenne Twister RNG, which is good RNG with a period of 219937 − 1. and is the one used by PHP with the mt_rand() function, but still has a few well know problems, well known enough to end up on Wikipedia.

Recognize the behavior we are seeming to experience in that bolded sentence?  

Frequent occurrences of the same patterns, that for us just happen to be strings of success and fail events? With as few events as we generate, it would take quite some time to get to the point of divergence, if the Unit Random algorithm has the same problem. It would be easy to think you had different seeds, but output nearly the same sequence for many iterations.

RNG is hard, and it's easy to get wrong, even when you think, and it seems to test correctly.

If the 2% is correct (2/100), even getting 5 of them in a row would have odds of 1/312,000,000 and then 4 in a row, 1/6,250,000 for a total of (1.95313E+15).  As a comparison, Lotto Max with at potential 60 million jackpot in Canada has 1/28,633,528  chance per $5 ticket with three plays.

9 in a row or 9/10 (the numbers are almost mathematically identical) would have odds of 1/ 1,953,125,000,000,000 chance's.  1.95313E+15

Numbers that big are basically "that never happens" kinds of numbers.

If you have video of that happening, that is proof positive that there is something wrong with the RNG.  

Not sure in a row is the right word. It was 9 from one node on the final hit. I wasn't there so I don't have a video of it.

Share this post


Link to post
Share on other sites
Posted (edited)
11 minutes ago, Tark said:

Not sure in a row is the right word. It was 9 from one node on the final hit. I wasn't there so I don't have a video of it.

I think 10 is the cap we can hit for critical amount with the pots isn't it?

With the base odd of 2/100 for a single doober, on a number set of 10, the odds are practically identical for nine is a row, or as I put it with the highest possible outcome. 5 in a row, then miss, then  4 in a row. 1/  (1,953,125,000,000,000-50) OR 1/ 1,953,124,999,999,950

Once you get into scientific notation, as spreadsheets do, you can lose some resolution, but even accounting for the most extreme case of scientific notation rounding error, the chance would still be 1/1,953,120,000,000,000.

A Mathematically insignificant difference.

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites

10 is the stat sheet cap. Haven't confirmed if it can surpass the stat sheet cap. 

When I crit my crit amounts are all the same color. Yet I could have sworn blair said they are supposed to be rolled independently of each other. So if my crit amount is 9 and i hit a purple i will get 9 purples. This has happened too many times so I'm not sure what is going on behind the scenes. Either way I'm rolling in blues and purples. At this juncture its way too many. That opinion might change once more crafting gets put into the game and you have to spread out your materials more.

Did a wood session and ended up with 182 blue 29 purple in 48 minutes. Purple came out lower than usual but that is most likely just bad luck on RNG.

Share this post


Link to post
Share on other sites
7 hours ago, Anhrez said:

So I am fresh into the patch and wanted to share my first set of stone harvesting with you. need to get the 12 Premium for the Laborer Stone built

Slate Plentiful 1  - 8 Nodes harvested

8 Grey/Red

2 White 

16 Green

4 Blue 

Critical Chance 40.3087 | Critical Amount 4 

 

Its a small sample but an odd one right? 

 

This doesn't seem that odd, actually!  Seems appropriate with having gotten 5 crits (50% instead of 40%, not a big deal for a sample of 8).  The only thing that seems odd is I was given the impression that crits rolled (crit amount) different ores out of the crit table, where these numbers seem to imply that it actually gives you (crit amount) of the single ore you roll.  This explains tark's 9 legendaries.  Hit the 2% and hit the jackpot.  Not sure if bug or just weird.

The weird thing is that you got 24 small destro events and 8 large ones but only 10 plentiful-based ores.  I'd expect to see something more like 20? 

Share this post


Link to post
Share on other sites

This kinda confirms that gathering skills are pretty meaningless after the basic tree, you can get the same stuff done with the right gear and no investment in the specializations.

Share this post


Link to post
Share on other sites
1 hour ago, Tark said:

10 is the stat sheet cap. Haven't confirmed if it can surpass the stat sheet cap. 

When I crit my crit amounts are all the same color. Yet I could have sworn blair said they are supposed to be rolled independently of each other. So if my crit amount is 9 and i hit a purple i will get 9 purples. This has happened too many times so I'm not sure what is going on behind the scenes. Either way I'm rolling in blues and purples. At this juncture its way too many. That opinion might change once more crafting gets put into the game and you have to spread out your materials more.

Did a wood session and ended up with 182 blue 29 purple in 48 minutes. Purple came out lower than usual but that is most likely just bad luck on RNG.

I am sure they are supposed to be rolled independently, that is why that result is impossible. The critical amount stat function is to give us extra rolls on the crit table, so each drop should be a different event.

If it is possible make a video of some crits you got and post it along with your relevant stats. 10 crits should be enough to determine what is wrong.

Share this post


Link to post
Share on other sites
5 hours ago, KrakkenSmacken said:

He said they built a tool to run 10,000 random events to see if there was a problem, and numerically the found a very small and expected variance <.00001% I think was what he quoted.

He did not however say that the tool tests specific interactions of events. It is trivial to build a Random.Range(0, 100)  call in unity that cycles itself 10,000 times and records the values.  What you have answered with something like that is "does unity the function Random.Range generate each number in range within expected variance values". Just because in 10,000 tries of Random.Range (1,100) you end up with roughly 100 hits on each number, does not mean your RNG is not creating predictable patterns.

Some things you have not tested are.

  • Does Unity have a sequence pattern based on seed (initial state)?
  • Are you using a predictable seed (initial state), so end up generating the same series of numbers? This needs to be checked when and how seeds are being used to create a new initial state.
  • Are you making the mistake of re-initializing state mid program? Re-seeding seems like a good idea, but can end up simply putting you back into the same state you just used.
  • Does the algorithm create frequent predictable patterns in large sets? (Short period).

For example the Mersenne Twister RNG, which is a pretty good RNG with a period of 219937 − 1. and is the one used by PHP with the mt_rand() function, but still has a few well know problems.

Recognize the behavior we are seeming to experience in that bolded sentence?  

Frequent occurrences of the same patterns, that for us just happen to be strings of success and fail events? With as few events as we generate, it would take quite some time to get to the point of divergence, if the Unit Random algorithm has the same problem. If they are reseeding, then it's entirely possible to never reach the point of divergence before being put onto the same pattern. It would be easy to think you had different seeds, but output nearly the same sequence for many iterations.

RNG is hard, and it's easy to get wrong, even when you think, and it seems to test correctly.

If the 2% is correct (2/100), even getting 5 of them in a row would have odds of 1/312,000,000 and then 4 in a row, 1/6,250,000 for a total of (1.95313E+15).  As a comparison, Lotto Max with at potential 60 million jackpot in Canada has 1/28,633,528  chance per $5 ticket with three plays.

9 in a row or 9/10 (the numbers are almost mathematically identical) would have odds of 1/ 1,953,125,000,000,000 chance's.  1.95313E+15

Numbers that big are basically "that never happens" kinds of numbers.

If you have video of that happening, that is proof positive that there is something wrong with the RNG.  

If i got it right you are looking for a error in the RNG function they used. I dont think the problem is advanced enough for that. What we want is which of 4-5 options happens given each option chance, that is so simple that i can use 5 lines to code that. And honestly i think i can make it foolproof, that is how simple the situation is.

All i would need is a function to select a number RAMDOMLY between 1-100, if Unity cant even get that right i must wonder how can so many people use it. That is why i think it is more of a coding error than a problem in the functions used.

What i mean is that knowing that "does unity the function Random.Range generate each number in range within expected variance values" should be enough for this situation.

Share this post


Link to post
Share on other sites
6 hours ago, Tark said:

Not sure in a row is the right word. It was 9 from one node on the final hit. I wasn't there so I don't have a video of it.

Even if it wasn't "9 in a row," if it was 9/10 then it can still be broken down using the same math that dictates a sequence of random events, and if you do so there is no way for that math to involve a sequence of less than 5 in a row, which as KrakkenSmacken pointed out, gives you better odds of winning the Powerball (1/292,201,338).

Not impossible, but pretty damn close.

Share this post


Link to post
Share on other sites

Just going to chime in here and say this:

In my experience, whenever I get a "color" crit, it is never mixed.  That means, if I crit blue, there is no green or purple to be found.  I will only ever get 1 color from a crit.  Not sure if that is part of the code or a bug, but it has been my experience.  Similar to what Canvox said above.

Share this post


Link to post
Share on other sites
Posted (edited)
2 hours ago, Kirchhoff said:

Just going to chime in here and say this:

In my experience, whenever I get a "color" crit, it is never mixed.  That means, if I crit blue, there is no green or purple to be found.  I will only ever get 1 color from a crit.  Not sure if that is part of the code or a bug, but it has been my experience.  Similar to what Canvox said above.

Could be built into the code that they only do one check for all the amounts, which frankly would be fine with me.  Makes the value of crit amount worth even more, and it is certainly a better experience to pop 10 legendaries in one shot, than it is to get them 1/50 doobers.

The volume math works out the same, but with the 10, you can at least immediately consider building a bar out of them, or bootstrapping 3 bars.

4 hours ago, BarriaKarl said:

If i got it right you are looking for a error in the RNG function they used. I dont think the problem is advanced enough for that. What we want is which of 4-5 options happens given each option chance, that is so simple that i can use 5 lines to code that. And honestly i think i can make it foolproof, that is how simple the situation is.

All i would need is a function to select a number RAMDOMLY between 1-100, if Unity cant even get that right i must wonder how can so many people use it. That is why i think it is more of a coding error than a problem in the functions used.

What i mean is that knowing that "does unity the function Random.Range generate each number in range within expected variance values" should be enough for this situation.

 

It's not necessarily Random.Range.  If the software was to call Random.Seed(Time) for each doober.  Each doober would have a high probability of starting at the same initial state, and therefore pull the exact same value for Random.Range(0,100).

 

This even happens with Millitime, which is supposed to be to the millisecond but can be expressed with "2008-01-01T00:30:45.1250000".

See those four end zeros?  Remind you of anything earlier in the thread. 

Quote

if the initial state is highly non-random—particularly if the initial state has many zeros.

When running my own RNG testing, I found that many systems (Hardware/OS/Libraries) have a clear pattern to the microtime value, and that they don't actually report thousandths of a second, but rather pick between a few dozen seemingly accurate options within the millisecond range.

See now how it could be the seed and not the actual functionality of Random.Range?  Ironically, by attempting to make random more random using microtime, and possibly excluding the rest of the date string, it's entirely possible that we are seeing the results of a very limited set (max 999 with just milliseconds) of initial states.

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites
4 hours ago, KrakkenSmacken said:

Could be built into the code that they only do one check for all the amounts, which frankly would be fine with me.  Makes the value of crit amount worth even more, and it is certainly a better experience to pop 10 legendaries in one shot, than it is to get them 1/50 doobers.

I agree with this, the math comes out the same in the end, and when I hit the jackpot I wanna hit the jackpot.

Share this post


Link to post
Share on other sites
10 hours ago, Tark said:

10 is the stat sheet cap. Haven't confirmed if it can surpass the stat sheet cap. 

When I crit my crit amounts are all the same color. Yet I could have sworn blair said they are supposed to be rolled independently of each other. So if my crit amount is 9 and i hit a purple i will get 9 purples. This has happened too many times so I'm not sure what is going on behind the scenes. Either way I'm rolling in blues and purples. At this juncture its way too many. That opinion might change once more crafting gets put into the game and you have to spread out your materials more.

Did a wood session and ended up with 182 blue 29 purple in 48 minutes. Purple came out lower than usual but that is most likely just bad luck on RNG.

I've capped it out at 10 with purple crafting armor. Won't go higher. 

Share this post


Link to post
Share on other sites

It's kind of insane how relatively powerful gear is vs. actually investing time in the damn tree.  Compare to crafting where you can't even use many crafting recipes without a skill investment.

Share this post


Link to post
Share on other sites
Just now, canvox said:

It's kind of insane how relatively powerful gear is vs. actually investing time in the damn tree.  Compare to crafting where you can't even use many crafting recipes without a skill investment.

Well, gear you actually have to earn in some way. Training just happens passively. I don't mind gear being more important than passive training.

Share this post


Link to post
Share on other sites
2 minutes ago, Jah said:

Well, gear you actually have to earn in some way. Training just happens passively. I don't mind gear being more important than passive training.

Then get rid of factory & recipe gating for crafting skills, let anyone with the gear be whatever kind of crafter they like.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.