Science is a way of thinking much more than it is a body of knowledge.
– Carl Sagan
The rules are simple: always at noon, and never decide the location until you’re standing outside. The former provides structure to what would otherwise be perfectly structureless days, and the latter gives people a chance to light a cigarette while we stand around and debate where we want to eat lunch. This specific day we wandered towards a small sandwich shop owned by the most enjoyable disgruntled man in Virginia.
As I picked up my sandwich from the counter, I caught the eye of a cute girl standing in line. There are lots of disparate groups working in the area around the Patent Office, most of them being miscellaneous professional organizations like law firms. She was dressed professionally but was around my age, and gave me a slightly hesitant but warm closed lipped smile. I smiled back and went to make quick work of my sandwich.
Observation. A constant feature of scientific inquiry.
Several days later, at a different local eatery, wearing the same pair of pants (I need to go shopping), I saw her starting to walk out just as I had opened to door to walk in. I held the door for her and she smiled warmly and said thanks, looking at me just a little bit longer than is strictly necessary.
Description. Information must be reliable, i.e., replicable (repeatable) as well as valid (relevant to the inquiry).
There aren’t a lot of women that work in the same part of the Patent office as I do. It seems computer architecture isn’t that attractive to the fairer sex, so I end up spending a lot of time surrounded by a rag tag band of engineers, nerds and wannabe lawyers. This is normally my absolute cup of tea, but sometimes it’s nice to be around … you know … women.
The following Thursday I was wearing a different set of pants, I had exact change, and just as I walked over to grab my drink from the cooler I spotted her deliberating on the chip selection. She turned as I walked up, I gave a friendly “hi”, and she returned the favor. “Salt and Vinegar is by far the best,” I posited. She furrowed her brow, thought for a couple seconds, then picked the bag up. “I guess we’ll see,” she said with twinkly eyes.
Prediction. Information must be valid for observations past, present, and future of given phenomena, i.e., purported “one shot” phenomena do not give rise to the capability to predict, nor to the ability to repeat an experiment.
The oft-repeated problem with women (or men, depending on how you look at it) is that it’s almost impossible for a man to determine if a woman likes him. This gets harder the more technical your degree is, as you brain has wired itself to the complicated pursuits of circuit design, recursive function debugging and pointer arithmetic. To suddenly switch tasks back into detecting subtle variations in audio frequency distributions emitted by members of the opposite sex proves beyond most men’s capabilities.
This applies to most men, but not all men. I’m a scientist, after all.
Control. Actively and fairly sampling the range of possible occurrences, whenever possible and proper, as opposed to the passive acceptance of opportunistic data, is the best way to control or counterbalance the risk of empirical bias.
I had picked up several non-verbal clues as to what the girl-from-the-local-eateries was thinking, but not quite enough to strictly differentiate between the two obvious possibilities: she could just be friendly, or she could want me. Until I have definitive proof one way or another, I have two possible default strategies to take. One, to assume friendly until proven interested. Two, to assume interested until proven friendly.
Deciding on a default strategy in this case comes down to weighing between the two types of statistical errors that can happen. Type I errors (false positives) are when you incorrectly label someone guilty when they are innocent, and Type II errors (false negatives) are when you fail to label someone guilty when they really are guilty. If I assume the girl is friendly, I eliminate the possibility of committing a Type I error but I open myself up to the possibility of a Type II error.
Falsifiability, or the elimination of plausible alternatives. This is a gradual process that requires repeated experiments by multiple researchers who must be able to replicate results in order to corroborate them.
In the ideal criminal justice system they’ve made the decision to place the burdan of proof on the prosecution so as to prevent innocent people from being convicted of crimes they did not commit. This is a case of the negative consequences of a Type I error outweighing those of a Type II. In every case, a Type I error is caused when an action is taken due to a false conclusion, and a Type II error is an action NOT taken due to a false conclusion. Every statistics class I’ve ever had has drilled into my head that Type I errors are the devil, and that Type II errors are what you laboriously add certainty to your conclusions to avoid.
Accordingly, I assume she’s friendly. I commit to collecting more data.
Causal explanation. Many scientists and theorists on scientific method argue that concepts of causality are not obligatory to science, but are in fact well-defined only under particular, admittedly widespread conditions.
It was several weeks before I saw her again, and when I did she was sitting at a two person table with (in my heterosexual opinion) a very attractive professional looking man. She was recounting some story in a slightly southern accent. From my own observations I’ve found southern girls to be unabashed extroverts when it comes to friendliness. They seem to operate under a default strategy of assuming people are good until they are proven evil.
At least the South understands Type I errors.
Sciencey bits from Wikipedia.
It seems that evolutionary biology and the Economist disagree with my rationale.
Leave a Reply