Failed weaves

[Guest]

Watch what changes when Trump takes over.

[Paigey]

I've posted this repeatedly and I'll say it again - 90-92% is not sufficient. Not only does it fail often - it will streak fail.

On both my Accepted and my Wilder I have 99% in refresh and all damage spells (live or die related spells).

Every time I practice my Accepted as a jack of all trades I remember why 99% matters - because I'm not willing to fail 3 of a spell in a row - which was a regular occurrence for me in my prior build.

Status weaves are less relevant because the most likely reason for failure is the level check.

A few months ago I failed 2 refreshes in a row - it stands out in my memory because that's how uncommon it was for me.

[Sybrina]

Not sure if the random number algorithm in the MUD code was customized or if its using the default from GNU libs. Its not a truly random generator but is rather optimized to achieve a mathematical distribution of numbers over a large population set. It might be possible for it to get a funky run of bad numbers by pure chance in clumps within that distribution. The opposite is also true, having a run of great numbers.

[Sybrina]

This reminds me of a weaves success study I peer reviewed with another Accepted where what was being measured was successes of status weaves under certain variables and controls. At the time it seemed like weaves were failing more frequently than not. When she plotted her numbers in a distribution scatter plot, it was pretty clear that they randomness converged on a specific probability over perhaps a hundred attempts. I did the same thing in the library---ever book I shelved, I recorded whether it awarded a qp or not. Initially the computed probability from the small samples had a large amount of variance. But when the samples reached about 50, it quickly converged (it was about 9.7% of the time so I'm guessing it was coded for 10%.) But I remember in the library going days without getting a qp from shelving a book. Turns out its just bad luck and my assumption that the probabilities were getting worse was just confirmation bias.

Not very reassuring, I know.

[Cass]

yeah, 90-92 aint enough, never have been and ive been playing channies pretty regurarely since 2000.


94%-96% on dmg weaves and expensive weaves.

99% isent needed though, very very little difference between 96 and 99.


but 90-92% on spike and its not rare that you fail 2 in a row.

[cass]

Also, forgot to mention.

I've always thought of it as if u practice 90% theres 10% failrate..

I think thats pretty accurate, and scales pretty accurate too.

[Iskveri]

You only ever remember the fails, not the weaving of 10 in a row. That's the way your brain is wired.

[edard]

the failure rate was uped a few years ago 2? or so i cant remember but it was anounced at the time

[Abraham]

Does opponent level factor into damage weaves landing %? The average level is much higher these days!