Expected Value and the Two Envelope Problem

September 4th, 2011
expected_value
Someone puts $X in one envelope and $2X in another. They hand you one at random. Before you open it and keep the contents, they offer to let you switch your envelope for the other one. You can't detect any differences between the two envelopes. Should you accept their offer to switch?

Common sense says no: the two are identical, you have even odds of getting $X and $2X. Expected value, however, says yes. Whatever envelope you have has some amount of money $Y in it. The other envelope has, with even odds, $2Y and $0.5Y. The expected value of switching, then, is the average of $2Y and $0.5Y, or $1.25Y, minus the $Y in the envelope you have. That would be $0.25Y as the expected value of switching, so you should switch. Unfortunately, now that you have switched, you would repeat all this logic, determine that the other envelope had more money, and switch again. Forever. Does expected value trap us in eternal loops of envelope cycling?

This is often considered to be a demonstration of how expected value can give unreasonable answers. But perhaps a solution to this is to frame things differently. We know one of the envelopes has $X and the other has $2X. In switching you are either going from $X to $2X or the reverse; either gaining $X or losing $X. So the expected value of switching is the average of $X and -$X, or zero.

The second framing gives a more reasonable answer, but is there any other reason to prefer it to the first?

Referenced in: Internal Statistics

Comment via: google plus, facebook, substack

Recent posts on blogs I like:

The benchmarkpocalypse

There's been a lot of talk about the vulnpocalypse, to which I don't have much to add because I'm not a security person, but I haven't seen much discussion on the closely related (and to be fair, less serious, issue), the benchmarkpocalyps…

via Posts on August 17, 2026

You can't join the secret inner circle of cool effective altruists

I, and a lot of other people I know, seem to believe something like this:

via Thing of Things August 11, 2026

Microfictions 2

A few more microfictions, still no LLMs.

via Evan Fields August 7, 2026

more     (via openring)