Two Truths and A Lie

I just tweeted a very interesting two truths and a lie:

• P=NP
• All even numbers greater than 2 are sums of primes
• The logical conjunction of the above two

I'll leave the solution to the reader. It occurred to me while lying in bed, however, that it should be possible to make a list such that any of the three can be the lie. After a little thought and some trial and error, the following fits the bill:

• One of the following two statements is false
• One of the other two statements is false
• One of the previous two statements is false

Simple, but it works. Neat.

This reminds me of the riddle in Inside Man:

Which weighs more: all the trains that pass through Grand Central Station [in NYC] in a year - or the trees cut down to print all U.S. currency in circulation? Here's a hint: It's a trick question.

Once again, I leave the solution as an exercise to the reader... or you can watch the movie.

EDIT: here are a few more humorous Two Truths and A Lie's I came up with:

• Water is wet
• Fire is hot
• The cake

"well, obviously, the cake is a lie."

• You think this one is true
• You think the last one is probably the lie
• You're not sure anymore

What?

• Either you won't answer, or you'll answer correctly
• Either you won't answer, or you'll answer incorrectly
• You will answer

Interestingly, it's possible to correctly answer this last one. You can figure out how.

A Year in My Life

As most of you know, I keep a personal journal where I record my thoughts and significant daily happenings. Unlike most people who journal - but probably typical as a Linux guy - I keep my journal digitally as plain text. Because it's searchable, I often include references to previous events and people that I am reminded of. I almost depend on it now as an integral part of my memory, and will sometimes be surprised when some incident I remember wasn't recorded in it.

Anyway, recently I started playing with graph visualization. I discovered Graphviz, which automatically moves nodes arounds to make the graph understandable. Primarily I needed it to do some visualizations for my research, but I then realized that the back references between journal entries are perfect for visualization.

Here's the graph for 2009. There are some extra edges to entries from other years which are not drawn.

That's 99 vertices and 183 edges, out of 110 entries for the entire year. Because most entries only reference previous dates, the dates are earlier near the bottom of the graph. You can see some seemingly important dates in the year, ones which I reference often: 2009-03-08, for example, and 2009-04-05. They can be identified by the high in-degree.

Out of curiosity, I also made a graph for my entire journal, a small version of which is shown below. The full version is 27917x4667 pixels, and weighs in at 20 mb.

The horizontal lines are not indicative of anything - it's just an artifact from how Graphviz works. This much larger graph contains 940 vertices and 1821 edges, out of 1283 entries I've written in the last 8 years. I don't have anything to say about it though.

BMI on Craigslist

There was a post, w4m, on Craigslist. In it, the author listed that she was 170lbs (I'm not sure if she posted her height), and was looking for someone who's single, under 30, and has no kids.

As tends to happen on any anonymous internet board, someone replied that she was picky. Someone else posted that the requires were not particular at all (which I agree with). Then two other people chimed in to say that 170 is "pretty big" unless "you must be like 7'8?", and that 1"70 is large unless you are REALLY tall".

I was curious, so I looked up the Body Mass Index (BMI), which has the following form:

BMI = (weight * C) / height^2

where C is a constant adjusting for the units.

A BMI in the normal range means it is between 18.5 and 25, with 22 being the average. This gives a height range of [69.14,80.37] inches, with an average of 73.70 inches. The average height for American females above twenty is 64 inches, which results in a BMI of 29.18, firmly in the overweight category and bordering on obese. There is a 3.8% chance of her being within the normal range, as 96.2% of American females are shorter than 5'9.

So, while the Craigslist posters were harsh, they were also, unfortunately, correct.

On a different note, the section of the US female population least likely to be overweight is (sterotypically) the Asian/Pacific-Islander (25.2% overweight/obese). The next least likely (surprisingly) are those with a graduate degree (29.2%). Interesting.