That is all true if each trial (skinning) is independent, but some are speculating that they are not. While I'm inclined to think the trials are indeed independent, I wouldn't put it past ZOS to have something in there to correlate trials with other variables - a pure independent RNG may not achieve whatever goals they have (e.g., modern slot machines don't always use independent trials, as other distributions have been shown to be more effective at keeping people feeding in the $$$). For example, they could just keep a ratio of the XP scrolls used to the XP drinks used, and feed that into the roe probability to "auto correct" any imbalance from the desired ratio.DaveMoeDee wrote: »You can't really find a ratio for this, because it's n% per fish, not n% per x number of fish skinned. Each event is independent.
What you'd have to do is have 1,000 different people skin one fish and tabulate the results. If three out of that 1,000 get roe, then your sample success is 0.3%. That's still not a drop rate. That's just a survey of yes over no.
Then you'd have to run 1,000 people many more times to get an observed low-high range. For example, in a sample of 1,000, between 2 and 11 people found one roe off of one fish. Or 0.2 to 1.1%.
Which is a pretty crappy range for something prized and sought after yet worth absolutely nothing to a vendor.
But that's all right. Fishing is a nice, relaxing in-game hobby. And it's strange how Ambrosia is one of the most interesting puzzles in the game so far. In some ways, the search is going to be more fun than actually using it.
You are making this way more complicate than it actually is.
The odds of a particular outcome are universal in this case. Just like flipping a quarter. If it is a balanced quarter, the odds of getting heads are 0.5 or 50%. If I toss twice and get both heads, that does not change the underlying probability to 1.0. The sample size is too low for us to say much about the probability in that case.
To figure out the unknown probability of an even, we need a large sample. It does not matter if the sample is all from a single person or from multiple people. The same goes for perfect roe since each skinning is an independent even with a set probability of getting the row.
Now, your approach of using multiple people can get you additional information. What it would show is the distribution of outcomes for that probability. The things is, that will not give any information that is as useful as just adding up all the outcomes and computing a percentage. If this is just a RNG with a set probability, than we already know the distribution without going through that extra effort.
We also know the low and high range. The low is 0% success. The high is 100%. Those are just insanely improbably for large sample sizes. Again, there is no need to test specifically for that info since math can tell us all that once we have a large enough sample to be confident of our probability that a skinning will result in perfect roe.
DaveMoeDee wrote: »That is all true if each trial (skinning) is independent, but some are speculating that they are not. While I'm inclined to think the trials are indeed independent, I wouldn't put it past ZOS to have something in there to correlate trials with other variables - a pure independent RNG may not achieve whatever goals they have (e.g., modern slot machines don't always use independent trials, as other distributions have been shown to be more effective at keeping people feeding in the $$$). For example, they could just keep a ratio of the XP scrolls used to the XP drinks used, and feed that into the roe probability to "auto correct" any imbalance from the desired ratio.DaveMoeDee wrote: »You can't really find a ratio for this, because it's n% per fish, not n% per x number of fish skinned. Each event is independent.
What you'd have to do is have 1,000 different people skin one fish and tabulate the results. If three out of that 1,000 get roe, then your sample success is 0.3%. That's still not a drop rate. That's just a survey of yes over no.
Then you'd have to run 1,000 people many more times to get an observed low-high range. For example, in a sample of 1,000, between 2 and 11 people found one roe off of one fish. Or 0.2 to 1.1%.
Which is a pretty crappy range for something prized and sought after yet worth absolutely nothing to a vendor.
But that's all right. Fishing is a nice, relaxing in-game hobby. And it's strange how Ambrosia is one of the most interesting puzzles in the game so far. In some ways, the search is going to be more fun than actually using it.
You are making this way more complicate than it actually is.
The odds of a particular outcome are universal in this case. Just like flipping a quarter. If it is a balanced quarter, the odds of getting heads are 0.5 or 50%. If I toss twice and get both heads, that does not change the underlying probability to 1.0. The sample size is too low for us to say much about the probability in that case.
To figure out the unknown probability of an even, we need a large sample. It does not matter if the sample is all from a single person or from multiple people. The same goes for perfect roe since each skinning is an independent even with a set probability of getting the row.
Now, your approach of using multiple people can get you additional information. What it would show is the distribution of outcomes for that probability. The things is, that will not give any information that is as useful as just adding up all the outcomes and computing a percentage. If this is just a RNG with a set probability, than we already know the distribution without going through that extra effort.
We also know the low and high range. The low is 0% success. The high is 100%. Those are just insanely improbably for large sample sizes. Again, there is no need to test specifically for that info since math can tell us all that once we have a large enough sample to be confident of our probability that a skinning will result in perfect roe.
Sure, it could in theory not be independent.
Though the people who think these things aren't independent are usually the ones who poorly understand distributions and expect everything they do to be at the mean and are surprised by clumpiness. In fact, the frustration many feel could be a result of making each skinning independent which (as expected) leaves some people as outliers at the futility end of the distribution.
All of the people who don't like fishing hate how much time it takes to get perfect roe, while those of us who like fishing are just walking around with big grins on our faces
How? I lose my bait just out of pure boredom at not really watching the most terribly designed fishing animation in history of all history of games.
Taylor21554 wrote: »Saw this thread, thought I would give it a try. Bought 10 simple bait and fished outside else root. Skinned 3 fish got a perfect roe.. lol. I bet my next 200 won't get a thing.
HaymakerXtc wrote: »i have a dumb question... I dont even know hot to Fillet the fish...
in my inventory when i have fish.. i have 1 option.. When i click Use... my option is to destroy...
what am i missing.. iwas told to do it from inventory..
12 fish from foul water, 1 roe
Would happily swap it for a four eye grog recipie though
.All of the people who don't like fishing hate how much time it takes to get perfect roe, while those of us who like fishing are just walking around with big grins on our faces
How? I lose my bait just out of pure boredom at not really watching the most terribly designed fishing animation in history of all history of games.
All of the people who don't like fishing hate how much time it takes to get perfect roe, while those of us who like fishing are just walking around with big grins on our faces
How? I lose my bait just out of pure boredom at not really watching the most terribly designed fishing animation in history of all history of games.
DaveMoeDee wrote: »Just use the addon that lets you know when to pull in the fish.