Pacific Rim Labs
22
Jul
2019

Binomial distribution (aka the "Sh-t happens rule")

July 22nd, 2019 | by Matthew MacLennan | in Learning

While he was explaining string theory and alternate universes, I was dumb enough to ask Pat for his thoughts about botched experimental results and human error. He promptly introduced me to his "Sh-t happens" rules. 

1. You can't change

a. The laws of physics

b. Statistics

2. Statistics: "Data points need to be erroneous and unexplainable sometimes"

a. Just like how one of your electrons has a probability of being around Jupiter right now

b. Just like how you (don't) fall in love

3. All analytical data is assumed to be under the binomial distribution

a. We use the student's t-distribution with a built-in uncertainty. The distribution gives confidence limits to the data, but also says we can't have 100% correct data.

i. There's always human error

ii. 99% certainty is the best we can do.

A. SOME event has to occur where we can't explain the outcome.

1. Sh-t happens = rule of the universe.

4. So, the next question, when looking at a pile of data:

a. If nothing is out of control, has the data been "clarified"?

i. Leads to unfair bias in the results

Pat quickly returns to explaining string theory and alternate universes. He turns to me and remarks, "You are in my universe right now. Get out."

I think Pat needs a nap.

~ Filtered by Dave Hope ~

Pat nap - Copy.jpg