Monday, September 29, 2008

Predictions

OK, here’s a quick thought experiment for you to try. Imagine two Highly Trained Professionals (HTP) are asked to make a prediction about whether Event A will occur.

HTP #1 says there is a 30% chance the event will occur.

HTP #2 says there is a 70% chance the event will occur.

The event occurs.

Question – which one was correct? 

As far as I can tell, they both were correct, insofar as they both recognized the event was possible. And since they were both correct, was there any real difference between the quality of their predictions? Is HTP #2 40% more correct? If there’s no difference, what’s the point? 

I think this is why weather forecasters don't advertise their "success rate" at predicting rain. It would probably be 100% success for every single weather guy - either that, or their success rate would be basically unmeasurable.

I wonder - do we work extra hard to avoid a negative event that has a 70% chance of occurring, and slack off if there's only a 30% chance? Do those numbers really mean anything at all about the future?

5 comments:

Mark said...

Slightly different thought experiment...

There are 1,000 opportunities for an event to occur. At each opportunity, HTP #1 says there is a 30% chance of it occurring, while HTP #2 says there is a 70% chance.

At the end of the 1,000 opportunities, 290 actual events were noted.

Now who was more correct?

Probability is all about implying the properties of a whole and relating that to the properties of the individual.

Kim said...

If a girl (G) has two brothers (B), and B1 has a 30% chance of being nerdy and B2 has a 70% chance of being nerdy, but they both end up nerdy, have they used up all 100% of the family nerdiness?

Mark said...

Hey - I resemble that remark!

Unknown said...

You guys crack me up - and don't worry, Kim, there's plenty of nerdiness to go around!

Remind me again who works in a science lab at a university?

Gabe said...

Dude, my answer is, what are the HTP's trained in?

Maybe neither one knows anything about the subject matter in which the event will occur. So just becasue they are HTPs, doesn't mean they know everything.

This happens to me A LOT. Many HTPs making predictions about stuff they just don't know about.

Maybe a kindergarten should actually have been predicting the event