Double Murder, Anti-Semitism in Phoenix, and Bayesian Inference

This is a true story about a double murder I took part in.

In November of 1994 I was a 15 year old Israeli Jewish 10th grader. Back then Israel was much more popular than today, however kids my age were nevertheless sent to the USA in order to give presentations about Israel in US high schools. To make a long story short, I was welcomed in Phoenix’s airport by a warm mother and daughter, all-American and all-blond. After mutual introductions, kisses and hugs, I was led to their car towards what was about to become my home for roughly two weeks.

After we started driving, following an uncomfortable silence of about 10 minutes, the mother said to me “You won’t believe what I saw today. I was walking in the street, and there was this guy who wore a shirt that said” – and here I need to make a short pause. The words she said right then were not clear to me. It sounded like “let the joos loos”.

Now in order to understand the context of the situation, you need to put yourselves in my shoes for a moment. Israeli kids abroad are constantly warned to remain alert about two things: kidnapping and anti-Semitism. Paranoia is the default state of mind.

The context in which I heard “let the joos loos” was: 1) I was in a new country, afraid of anti-Semitism; 2) I assumed, like most kids do, that adults around me know exactly what bothers me; 3) the adult sitting next to me just told me with dismay about something she saw today that related to “joos” and “loos”.

So, it was perfectly clear to me that the mother had some speech impediment, and that this guy’s shirt said “Let the Jews Lose”. I thought it was very disturbing that such anti-Semitism exists in Phoenix, but I was glad the mother showed dismay against it. In addition, although I wasn’t a great English speaker at the time, I thought the phrasing chose for the shirt was very strange: “Let the Jews Lose”? That’s a silly way to convey a Jew-hating message… Lose what? But then again, I thought maybe there’s some trial going on where Jewish people are involved. So although I felt relatively safe at their home, I was afraid for my life on the streets of Phoenix.

Many years have passed, I have been to the US many many times since, and have never encountered anti-Semitism. Still, I knew that in Phoenix people are anti-Semite, and was concerned about the next time I’ll have to go there.

Many years later I saw a documentary about OJ Simpson‘s double murder trial. I was exposed to the fact he was nicknamed “The Juice”, and in that movie they actually showed a guy wearing a shirt saying “Let the Juice Loose”. I was astonished, and ran to Wikipedia to check the dates of the double murder versus the dates of my arrival to Phoenix. It made perfect sense.

To summarize the story, many years ago I heard someone say “let the joos loos” in a certain context. There were two possible explanations.

The first explanation required all of the below to exist concurrently:

  1. The saying was actually “Let the Jews Lose”
  2. The person who told me about it had a speech impediment, that’s why it sounded like “Let the joos loos”.
  3. There was some trial going on against Jews or some other case where it’s relevant to wish that “Jews Lose”

The second option was that the person saying “let the joos loos” was saying something else. For example, the word “Juice” sounds phonetically exactly like “joos”, and the word “Loose” sounds phonetically exactly like “loos”. But, “Let the Juice Loose” doesn’t make any sense (for a kid who doesn’t know there’s someone called “The Juice” that’s on trial).

And here’s the kicker, pay close attention:

From my perspective, the probability that all 3 conditions yielding “Let the Jews Lose” co-exist, was much higher than the probability of that woman said something like “Let the Juice Loose”, which made no sense at the time.

The above is a clear example of Bayesian Inference at play. Bayesian Inference takes place when we encounter an event (e.g. hearing “let the joos loose”), and try to find a theory that would explain that event (e.g. the mother said “let the Jews lose” and had a speech impediment). Bayesian Inference states that (following is a brutal simplification):

The probability that a theory explains an event equals to the probability of the event resulting from the theory, multiplied by the probability of the event.

This fancy-shmancy definition boils down to the following. When you encounter something, you try to figure out what it is, sometimes considering different explanations in your head. You see a shiny spark on the pavement. Is it a coin? Is it a diamond? Did you imagine it? Is it the feces of an alien life form?

When considering each explanation, your answer is a combination of:

  1. Had this been the real explanation, would the result look like that? Both a coin and a diamond will result in the same spark.
  2. What are the odds of this explanation occurring in reality? Coin: high, diamond: low.

Bayesian inference happens naturally in our brain, we’re wired to infer automatically. And sometimes we get it wrong. It plays a major part in our day-to-day lives, and I’ll revisit it in the future.

Read more about Bayesian Inference here.

This entry was posted in General. Bookmark the permalink.
  • ViRUS

    Woah, nice read. I find it amazing how coincidences can change our lives, even if it’s one of the smallest ones, like missing a bus and then finding out someone hijacked it and stole their belongings to the persons on it (personal experience there).

    Also, I find it very intriguing how our brain works and does things for us without we telling it to or sometimes even without realizing what’s going on.

    • Roee Adler

      Yes, thanks. Let that be a lesson for you about busses :)

  • Pingback: Embarrassed by an Answering Machine | Roee Adler's Clown College (aka Blog)()

  • Ophir

    Nice post. There is a famous experiment where a group was shown photos of different people and had to say which profession they were more likely to possess out of two options. One of those photos was of a tall black man, and the group was asked whether he was a teacher or an NBA player. “Naturally”, almost everyone in the group said he was an NBA player. However, this is a really bad guess. While the probability of an NBA player to be a tall black man is very high, the probability of a tall black man to be an NBA player is extremely low. To illustrate:

    Total number of NBA players: 500
    Total number of NBA players who are tall and black men: 450

    Total number of teachers: 6,500,000
    Total number of teachers who are tall and black men: ~65,000

    Chance of the guy being a teacher is >100 times bigger than being an NBA player, but yet almost everyone got it wrong. I agree we have a wired bayesian probability calc mechanism in us, but it doesn’t seem to be working that well…