This morning I've been thinking a lot about precision and accuracy. You know the difference, right?

For example: I'm about 5'6". I've been measured many times with different rulers, over the last 30 years. They all come out to be about 5'6". I've never been measured with a laser or other high-precision measuring device, just marks on the wall or the thing on top of my head.

So 5'6" is an extremely accurate but not very precise measure of my height.

If you go into google and convert 66 inches to cm, it gives you 167.64 cm. Great! That's how tall I am!

No.

I could easily be 166 cm. Hell, I could be 167.52 cm. We have not measured me that precisely.

I'm seeing people who want to dismiss COVID mortality rates throwing around numbers on twitter that have 7 or 8 significant figures--the numbers that indicate precision.

That is a red flag.

Nobody knows the mortality rate for COVID to more than one or two digits.

EVEN IF mortality was a constant, which it is not, we do not have high certainty that we've counted all the cases of COVID *or* all the deaths from COVID. In fact we're very certain we have not. What does that mean in this discussion?

It means that if you see someone claiming that the mortality rate for some group or another is 0.027118479, or 0.017444323 or whatever, RUN. That person is NOT ARGUING IN GOOD FAITH. They are trying to use the precision of that number to look certain, smart--and accurate.

But precision is not accuracy. You can make up that I am 190.5527 cm tall, and that's a very precise number, but it's also wrong; it's about 6'3", and I'm about 5'6".

Sounding more certain than anyone CAN be about a current data set makes that source less trustworthy, not more.

(Incidentally, I never see people doing this with numbers that match at all up to the less-precise-but-more-accurate ones--it's not that they're accidentally overstating. It's that they are MAKING SHIT UP. But this is one way to tell.)