3 ms·
Ulikelieness is hard to reason about though. Suppose we were to list all the different metrics that describe our world and solar system. How long would the list
by simonh 11y ago
Ulikelieness is hard to reason about though. Suppose we were to list all the different metrics that describe our world and solar system. How long would the list be? Then think of all the possible synchronicities there might be between all those attributes. There might be many millions or billions of possible 'magic' ratios and equivalences. So if the chance of any particular synchronicity randomly occuring is one in a million, you'd expect to see many synchronicities in the data set, just by random chance.
It's one reason I'm skeptical of probabilistic evidence in court cases. Suppose the chance that soem particular piece of evidence linking a suspect to a crime has a 'one in a million chance' that it would be randomly associated with someone other than the perpetrator. Sounds convincing, looks like this person probably did it. Well, there are 60 million people in the UK. Does that mean we would expect about 60 of them to be incorrectly incriminated by this evidence? Suddenly it doesn't sound so convincing.
It's also a problem in 'big data' analysis. If you have many thousands of items of data about people, if you try to find correlations in that data with some characetristic of interest, such as tendency to some medical condition or insurance risk, as the number of factors expands the chance that you'll find completely spurious correlations also goes up until it becomes near-certain.