# Humble Pi

A book about all kinds of math mistakes.

## Contents

### Losing track of time

Math problems with time. Many issues with computer’s limited capacity to track time (e.g., storing milliseconds using a variable with fixed number of bits) and different calendars, etc.

### Engineering mistakes

### Little data

Issues caused by missing data and excel failures.

There are genes with names that resemble a date, such as “MARCH5”, which is automatically converted to a date when read by Microsoft Excel. See Spreadsheet problems in Biology.

### Out of shape

As a general rule, doors should open in the direction they would need to in an emergency. Because of the location of the hinges, a door opens easily in only one direction; every doorway has a bias one way or the other.

In 1883, there was tragedy with many children due to this issue at the Victoria Hall Theatre in Sunderland near Newcastle. Same issue with the Apollo 1 tragedy.

There are many shapes – that are not a circle – with multiple identical diameters. Thus measuring diameters is not enough to guarantee a circular shape.

### You can’t count on it

counting mistakes regarding the inclusion (exclusion) of the starting point. Why the musical interval is weird.

### Does not compute

Overflow (e.g., radiation machine), truncation (patriot missile), …

### Probably wrong

Miracles happen when there are enough trials.

### Put your money where your mistakes are

The story of The Making of a Fly - algorithmic pricing that went far to make it $23,698,655.93.

### A roundabout way

Rounding only up or down can make a huge difference long term.

### Too small to notice

Airplance bolt story (June 8, 1990; British Airways in Birmingham Airport)

### Units, conventions, and why can’t we all just get along?

Stories about unit conversion problems.

### Stats the way I like it

There is no “average person” if you consider many aspects of demographic characteristics. This is related to the unintuitive phenomenon of high-dimensional space where there’s almost no probability mass near the origin of a normal distribution in a high-dimensional space.

If you create a uniform by taking average of all possible characteristics, it will fit no one.

### Tltloay Rodanm

### Does not compute

### So, what have we learned from our mistakes?

“Hot” Swiss cheese model