Do you think the Chinese government hackers that are attacking us spend time on a) banned in China b) hacking *is* their job c) this is a wash d) they aren't allowed time for any of those things, except possibly 4 hours of sleep a day?

Show me a great physical specimen and I'll show you a crappy hacker who will always be behind the curve vs. the bad guys. By definition, the people who are good truly skilled at hacking and cyber warfare spend every waking hour doing nothing but geeking out/researching/testing/commenting on Gizmodo, etc. If you have…

"It was the first time in life I had a chance to date, since I wasn't allowed to growing up with my mom"

In your example you have a second, different test of 99% accuracy applied to a group with a 50% incidence of actual drug use. The size of the pool doesn't really matter. And you are correct that with those numbers only 1 out of 100 would throw a false positive. But even in your scenario 4.75 users would test negative…

I went full human civetpede. Cut out the middleman, amirite?

As I and others have shown, a 98% accurate test applied to a group of test subjects with a low enough actual incidence of the thing being tested for (5% in my example, although you could argue that the incidence of what they're testing for is actually higher in the baseball player pool) will result in only 50% true…

Wait, wait, wait. A steel cannonball is too dangerous to use? Because it might leave his territory and is really hard to stop? I think this myth needs to be busted...

Always a good idea to lead with your face.

So that would make it Tan In Knots Oi. So now we should be blaming Mossad or MI6?

This just in: Amazon plans a major update to improve the functionality of their shopping suggestions, so that stupid people won't be so confused by their site. They're calling it the Brangelina Algorithm.

The craziest thing I notice is that by the time the shadow appears on the wall behind the target object, the front of the object is no longer lit. Wild.

He's spinning in his grave. We just don't know in which direction.

False positives as a result of test accuracy aren't a function of "error by the instrument or user".

The accuracy is a limitation of the test itself. If it generates a false positive on an individual sample, it's always going to generate a false positive on that same sample (and in this case the samples are the same).

Trekker was adopted as a more acceptable descriptor vs. the derogatory Trekkie a very long time ago.

You are correct that using a more accurate test will yield better than 50/50 TP/FP results, but even in your example, it's only an 84% chance that the guy is actually using. Careers shouldn't hinge on that kind of accuracy.

Use these numbers:

You are correct. In my example, the generally accepted actual drug use statistic amongst the general population is 5%.

What is the accuracy of the testosterone test? 95%? 98%? A real false positive would show high testosterone when there actually wasn't, and would do so as a result of the limitations of the test. Unless it's a 100% accurate test, false positives will occur (and they would occur on both samples tested).

So, not only does this McClintock dude have insane pogo chops, but he's got editing chops, too. Well played, sir.