Random Events Don't Look Random
Jul. 9th, 2015 09:59 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
chris_gerribAs you may have heard, yesterday the New York Stock Exchange and United Airlines both had significant computer failures. There were a lot of questions to the effect of "are these events related?" Well, maybe not, because random events don't look random. As explained very clearly at this site (PDF), random events can be subject to a "Poisson burst." Poisson was a French mathematician who was tasked with figuring out why Prussian army officers kept getting killed in riding accidents.
Consider your work day. If you're like me, at various times of the day, people call or stop by your office with questions. But this rarely happens on a regular schedule. I, for example, can go an hour with no calls then in the same 10 minute span I get two calls, three important emails and somebody knocking on my door. This is a Poisson burst - a bunch of random events that randomly happen at the same time.
Poisson developed an equation to predict how often these events burst. So, using an example from the link, consider shark attacks. If we have an average of two attacks per year, using the equation, we have a 1% chance of six attacks per year (or once every 50 years) and a 13% chance of no attacks on a given year (one year out of seven is shark-less).
Another example of a Poisson burst is birthdays. There's a 1 in 365 chance that you have the same birthday as another person. But in a random group of 30 people, there's a 70% chance two people share the same birthday. In a group of 100 people, it's 99%.
The human mind is trained to look for patterns. (The saber-toothed tiger always attacks on Tuesdays.) But sometimes, there's no pattern to detect. (Casinos make money off of this human training. For example, there's an electronic display next to every roulette wheel in a casino telling you what hit recently.) Part of managing a business, or life, is determining if the perceived pattern is real or random.
Consider your work day. If you're like me, at various times of the day, people call or stop by your office with questions. But this rarely happens on a regular schedule. I, for example, can go an hour with no calls then in the same 10 minute span I get two calls, three important emails and somebody knocking on my door. This is a Poisson burst - a bunch of random events that randomly happen at the same time.
Poisson developed an equation to predict how often these events burst. So, using an example from the link, consider shark attacks. If we have an average of two attacks per year, using the equation, we have a 1% chance of six attacks per year (or once every 50 years) and a 13% chance of no attacks on a given year (one year out of seven is shark-less).
Another example of a Poisson burst is birthdays. There's a 1 in 365 chance that you have the same birthday as another person. But in a random group of 30 people, there's a 70% chance two people share the same birthday. In a group of 100 people, it's 99%.
The human mind is trained to look for patterns. (The saber-toothed tiger always attacks on Tuesdays.) But sometimes, there's no pattern to detect. (Casinos make money off of this human training. For example, there's an electronic display next to every roulette wheel in a casino telling you what hit recently.) Part of managing a business, or life, is determining if the perceived pattern is real or random.
