Outliers

May. 14th, 2017 07:27 pm
[personal profile] tigertoy
I was listening to the radio this morning and there was some discussion about the connection between facts and truth. The person on the radio contended that while facts can be checked and verified and reasonably be shown to be correct, but truth is more subjective and hard to pin down. This inspired a train of thought.

I believe in science as a way of understanding the world. Science approaches a problem by forming a hypothesis and collecting facts to see how well it fits. If most of the facts align, the hypothesis is strenthened. If there are facts that don't fit and there's a pattern to those facts, it points to a way to refine the hypothsis, or possibly even throw it out and start over. But in most sets of data, there are a few outliers -- measurements that are way outside the range of most of the data. If the data can be graphed on a scatter plot, there's a very heavy concentration of points along the expected line, but there are just a few apparently randomly around the graph. If there are few enough outliers, we consider the hypothesis valid despite these outliers.

In politics, on the other hand, we focus much more on the outliers rather than going with the general trend. If a debater can present a single fact that disagrees with an opponent's point of view, and that fact is verified, it's considered a valid argument (at least to one side of the question). There is little attempt to see the general pattern and ignore the outliers.

This problem is very much driven by the way the media work. Ordinary events -- those in the main blob of data points on the scatter plot -- aren't newsworthy because they're common. Outliers, because they are novel, receive far more attention than they generally deserve; and this tends to reinforce extreme viewpoints by reinforcing them with outlier examples. For example, consider how many people worry about how dangerous air travel is but never pay attention to how dangerous car travel is. This happens because plane crashes are so rare that every one is going to get media attention, while car crashes are so common that they're hardly ever mentioned.

Truth is somewhat subjective and squishy, but a reasonable view of the truth should be formed by considering the ordinary majority of the facts rather than concentrating on the outliers.

Profile

tigertoy

July 2017

S M T W T F S
      1
234 5678
9 10 1112131415
16171819202122
23242526272829
3031     

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags