## Thursday, 9 March 2017

### Born to do Math 2 – Bayesian Statistics and Bars

Born to do Math 2 - Bayesian Statistics and Bars
Scott Douglas Jacobsen & Rick Rosner
March 9, 2017

[Beginning of recorded Material]

Rick Rosner: Bayesian statistics is statistical inferences based on putting people in subsets. I use Bayesian logic when trying to catch people with fake IDs of people trying to get into bars.  I worked in bars for 25 years. I caught about 6,000 people trying to use fake IDs. I caught another 6,000 people who snuck into bars through some other means. At some places, I was in the inside guy like Anthony’s Garden.

It is five acres and holds up to 10,000 people inside the Harvest Hotel in Boulder, Colorado. One way I would catch underage girls who snuck in is I would look for a cluster of lame guys who came to the bar a lot, never went home with anybody. In the cluster of guys, there would be an underage girl who hadn’t yet learned how to fend off lame-os. When checking IDs at the door, I have ways to find about 1 person in 90, which is the number of people who have fake IDs.

You can’t accuse 100% of the people coming through of having a fake ID. You have to narrow it down and concentrate. You can have to slice away everybody who doesn’t have a fake ID within 10 seconds. You should be able to deal with most customers within 10 seconds. You can do that by dividing people into subsets. If somebody comes up to you, you look at them. If they look old enough, look over 30, then they probably don’t have a fake ID.

That leaves people under 30. Say half of the people coming in are under 30, and half of the fake IDs are under 30, so you’ve brought the number down from 1 in 90 to 1 in 45. If all of the features match, not a fake ID, now you’re left with, say, 10% of the people coming through and you’ve concentrated the fake IDs by 10-fold. Instead of 1 person in 90 coming at you with a fake ID, it is 1 in 9, then you ask them questions.

What’s your Zodiac sign—Taurus, Leo, Gemini?  Very few people don’t know their sign, so the concentration of fake IDs in the group of people who don’t know their sign has gone up to 2 out of 3. Then you ask them, “What year did you graduate from high school?” It is almost 100%. The criteria of having a fake ID, the various criteria, and you increase your likelihood that somebody’s bullshitting you to 100%, just by drawing Venn diagram circles and putting people in the various circles until you’ve got one intersection of they look young, don’t know their sign, don’t know the year they graduated high school, and their features don’t match.

They’re in the center of these four intersecting circles. So you’ve concentrated the subset of them that is highly likely to be lying to you. Bayesian logic is useful and dangerous. In that, it encourages you to stereotype. At the same time, some of the stereotyping and typing can be really powerful. Odds that a random person is having a kid out of wedlock. Say the odds for the general population are 10%, or giving birth to a kid of out wedlock, the odds for general population are 10%.

If you limit your population to women, you’ve double your odds because men can’t give birth, so 20%. It is something that we subconsciously, unconsciously, maybe, do for good or for ill in a lot of situations, it is helpful to know how to do it, to know the dangers of it. Stereotyping rests on this stuff. So that’s an evil it, but it can be helpful.

[End of recorded material]

Authors[1]

Rick Rosner
American Television Writer
RickRosner@Hotmail.Com
Rick Rosner

Scott Douglas Jacobsen
Editor-in-Chief, In-Sight Publishing
Scott.D.Jacobsen@Gmail.Com
In-Sight Publishing
Endnotes
[1] Four format points for the session article:
1. Bold text following “Scott Douglas Jacobsen:” or “Jacobsen:” is Scott Douglas Jacobsen & non-bold text following “Rick Rosner:” or “Rosner:” is Rick Rosner.
3. Footnotes & in-text citations in the interview & references after the interview.
4. This session article has been edited for clarity and readability.
For further information on the formatting guidelines incorporated into this document, please see the following documents: