Two Envelopes Paradox Explained: Should You Switch?

Quick answer: In the two envelopes paradox, one envelope holds twice as much money as the other. You pick one and see (or imagine) the amount x. A tempting argument says the other envelope holds 2x or x/2 with equal chance, so it is worth 1.25x on average and you should switch, and then switch back, forever. The flaw is treating x as fixed while also treating the 50/50 split as valid for every x. With fixed amounts a and 2a, switching gains a or loses a with equal chance, so it does not help.

Two identical envelopes contain money. One has twice as much as the other. You pick one. Before opening it, you are offered a chance to switch. A simple argument says you should always switch. But by the same argument, after switching, you should switch back. Something is wrong, and finding exactly what is surprisingly subtle.

The Tempting Argument

Call the amount in your envelope \(x\). The other envelope has either \(2x\) or \(\frac{x}{2}\), each with probability \(\frac12\). So its expected value is

\[ \frac12 (2x) + \frac12 \left(\frac{x}{2}\right) = \frac{5}{4}x. \]

That is 25% more than \(x\), so switch! But the envelopes are symmetric, so the same reasoning says to switch back. You would switch forever, which is absurd.

The Correct Calculation

Name the amounts honestly: the envelopes contain \(a\) and \(2a\). You picked one at random.

You pickedProbabilityGain from switching
\(a\) (the smaller)\(\frac12\)\(+a\)
\(2a\) (the larger)\(\frac12\)\(-a\)

The expected gain is \(\frac12(+a) + \frac12(-a) = 0\). Switching neither helps nor hurts.

Where the Paradox Goes Wrong

The tempting argument uses \(x\) for two different things. When you hold the smaller envelope, \(x = a\); when you hold the larger, \(x = 2a\). Writing “the other has \(2x\) or \(x/2\)” mixes these two cases as if \(x\) were the same number in both.

What if you open the envelope and see, say, 100 dollars? Equal odds for 50 or 200 are possible in a particular case: if the host picks the pair (50, 100) or (100, 200) with equal chance, then seeing 100 really does make 50 and 200 equally likely, and switching is good at that observation. But under that same rule, seeing 200 tells you that you hold the larger envelope, and switching loses. The invalid step is assuming equal odds at every possible amount; no probability distribution with a finite average allows that. For such distributions, “always switch” and “always stay” have the same expected payout.

A Twist: When Switching Really Does Seem Good

Mathematicians have built strange distributions (with infinite expected value) where, whatever amount you see, the conditional expected gain from switching really is positive. These cases are related to the St. Petersburg paradox: once expectations are infinite, ordinary reasoning about averages stops working. With any distribution that has a finite average, the paradox disappears.

Common Misconceptions

“Switching is a free 25% gain.” No. With fixed amounts, the expected gain from switching is exactly zero.

“Opening the envelope makes no difference.” It can. Seeing the amount, combined with knowledge of how the amounts were chosen, can tell you whether switching is wise in that case.

Watch: The Two Envelopes Paradox in Video

If you learn better by watching, these videos are worth your time.

The Two Envelope Problem – a Mystifying Probability Paradox (Formant). A careful walk through the paradox and its resolution.
The Two Envelopes Paradox (BriTheMathGuy). A short explanation of where the 1.25x argument fails.

Check Your Understanding

Five quick questions to see how much of this has stuck. Pick an answer and press Submit; you will see a short explanation either way.

Answer each question and press Submit to see your result right away.

1. What expected value does the tempting argument give for the other envelope?
Found a bug? Let us know!
2. With fixed amounts a and 2a, what is the expected gain from switching?
Found a bug? Let us know!
3. What is the main error in the tempting argument?
Found a bug? Let us know!
4. Why can’t equal odds hold at every possible amount (with a finite average)?
Found a bug? Let us know!
5. Which paradox is related to the strange infinite-expectation versions?
Found a bug? Let us know!

Frequently Asked Questions

What is the two envelopes paradox?

One envelope holds twice as much money as the other. An argument based on expected value seems to show you should always switch, and then switch back, forever.

Should you switch envelopes?

If you have no information, it makes no difference: with amounts a and 2a, switching gains a or loses a with equal chance.

What is wrong with the 1.25x argument?

It treats your amount x as one fixed number while assuming the other envelope is equally likely to hold 2x or x/2 for every possible x, which no distribution with a finite average allows. Equal odds can hold at some particular amounts, but not all.

Is the two envelopes paradox solved?

For distributions with finite expected value, yes. Versions with infinite expected value remain subtle and are linked to the St. Petersburg paradox.

Keep Exploring

Sources and Further Reading

Related to This Article

What people say about "Two Envelopes Paradox Explained: Should You Switch? - Effortless Math"?

No one replied yet.

Leave a Reply