I understand how pseudo-random number generators work & that 'dice' have no memory but frequent runs of 4, 5, 6 or 7 fails against a 75% chance of success?
It is frequent to see batches of 4 or more fails in a row, but rare to see the same successes in a row, this I am sure is in part due to:
you retry after a fail,
but have to find another node to try again after a success!
but it does grate to fail, fail, fail again and again on something that should be 3 out of 4 successes...
Is your pseudo-random number generator code:
well seeded
close to random over a large batch
using the same pass/fail rates shown to us
?
search general for a long thread on this topic, where someone logged their very many fusing attempt results. Their results did not match the advertised success rates.
My worst is 15 failures in a row on 40% chance.
This may sound HAMSTER, but I don't think they use a RNG. I swear they just have things like fail/succeed or win/lose running on a loop that is the same for everyone. Winning/losing and failing/succeeding seems to happen in groups all the time.
If you open a lot of lockboxes quickly in a row you will notice you get the same prizes quite a few times until it changes and then you get the next one multiple times. I have seen one player win three mounts in quick succession. I used to pay attention to the announcement when players win mounts it seemed to always happen in groups. Player "X" would win and their announcement would quickly be followed by two more. This has slacked off a bit lately because I only see one winner every so often. They may have tweaked it a bit or there are less people opening boxes.
When it comes to fusing enchants I bet if you and a friend fused the exact same enchants at the same time you would both fail at the same times and if you tried to fuse again quickly you would continue to fail until the "Fail" timer switched back to succeed. The higher level the enchant the quicker it switches back and forth. I swear I get more successes when I get into a rhythm with fusing. It's almost like fuse...wait for fusing to be available...wait one more beat...fuse. I was fusing rank 5 and I had very few failures last night when I did that. When I fail I waited an extra beat and I did not have any back to back failures. I actually had more failures on lower level fusing because I just kept spamming the button and would fail multiple times in a row.
It is the same with skill nodes. When I fail, I will continue to fail multiple times in a row unless I stop and wait a few moments. It seems like the fail time is just a larger block of time.
I am probably just off my rocker and it is all in my head.
It is not possible to generate random numbers at all. At best, they are appropriately uniformly distributed values that seldom give the same value twice in a row and almost never 3+ times in a row (possible but unlikely). The problem is that to get quality uniform distribution, the algorithm is orders of magnitude slower than the fast algorithm (which is just a single, simple modulus operation). Because of the sheer number of values needed in a game like this one, the fast versions are usually used; creating millions of values adds up fast to a lot of cpu time. It is well known that the modulus generated values contain streaks of similar values and give a less than ideal distribution. Also, most built-in (c++ rand() for example) algorithms use the faster approach -- its "good enough" for most ordinary tasks.
So, basically, its a well known problem that can be fixed but may not be worth fixing due to side effects and time spent etc.
Comments
My worst is 15 failures in a row on 40% chance.
If you open a lot of lockboxes quickly in a row you will notice you get the same prizes quite a few times until it changes and then you get the next one multiple times. I have seen one player win three mounts in quick succession. I used to pay attention to the announcement when players win mounts it seemed to always happen in groups. Player "X" would win and their announcement would quickly be followed by two more. This has slacked off a bit lately because I only see one winner every so often. They may have tweaked it a bit or there are less people opening boxes.
When it comes to fusing enchants I bet if you and a friend fused the exact same enchants at the same time you would both fail at the same times and if you tried to fuse again quickly you would continue to fail until the "Fail" timer switched back to succeed. The higher level the enchant the quicker it switches back and forth. I swear I get more successes when I get into a rhythm with fusing. It's almost like fuse...wait for fusing to be available...wait one more beat...fuse. I was fusing rank 5 and I had very few failures last night when I did that. When I fail I waited an extra beat and I did not have any back to back failures. I actually had more failures on lower level fusing because I just kept spamming the button and would fail multiple times in a row.
It is the same with skill nodes. When I fail, I will continue to fail multiple times in a row unless I stop and wait a few moments. It seems like the fail time is just a larger block of time.
I am probably just off my rocker and it is all in my head.
RNG is RNG but I noticed this trend a couple months ago.
So, basically, its a well known problem that can be fixed but may not be worth fixing due to side effects and time spent etc.