Strategy: Information Overload II - Algorithmic Thinking

Another fantastic example of when too much information is simply too much is discussed in Blink in the case of a cardiologist named Lee Goldman.

Dr. Goldman was a research scientist and while working for the Navy faced a problem. It turns out that when diagnosing possible heart attack patients it is very difficult to tell from their history, and symptomology whether they are in fact having a heart attack. Unfortunately, even soldiers on submarines have heart attacks and this becomes a very inconvenient problem when the sub is submerged. A submarine can not just rush a soldier to the hospital every time that soldier has gas pains that they mistake for a heart attack.

Based on studies it turns out that Doctors love to take information about patients. The more information they take the more comfortable they feel with their diagnoses about heart attacks. However, it turns out that not only are Doctors not able to easily diagnose heart attacks, more information does not make their diagnosis more accurate. It just makes them feel more confidant. Many of the diagnostic tools that are used in heart attack diagnosis can give false positives and false negatives so any single diagnostic reading is useless. For example ECG readings can be normal or abnormal in both healthy and heart attack patients.

Dr. Goldman developed a controversial method. He came up with an algorithm that took several diagnostic readings and developed a treatment method for each combination. He looked at the ECG readings, whether there was fluid in the patients lungs, whether there was unstable angina, and whether the patients systolic blood pressure was below 100. Although Doctors balked at this method (no history, no other information taken into account) it turns out that it was significantly more reliable at identifying false positives and 95% correct at identifying the most serious patients and getting them quick medical help.

Reading this reminded me a lot of pivot points. You don't need a lot of information to make a decision in SNG's. The structure of the tournament makes it so that correct decisions are really based off of a formula of Effective Stack Size, Range, and Position. It turns out that the range is the really tricky thing to evaluate quickly, but that by assigning standard ranges we can correctly create pivot points that evaluate and determine the correct decision with a high percentage of accuracy. Plus, pivots allow us to make quick decisions and really filter out the noise of all the other factors at the table that might cause us to feel more confidant about our decision, but that either don't help us at all or that actually might make us second guess our strategy.

I am certain that if Mr. Gladwell were to take a look at pivot point strategy he would concur that it is in line with the basic premises of this section of Blink.

Good Luck at the tables,
Greg