I have to figure out the math on it, but I doubt that 2:1 is a good deal.

Follow-up edit:

• In New Jersey, as an example of a safe state for Harris, Fivethirtyeight has Harris winning in 993 out of 1000 simulated elections. Assuming the same turnout as 2020 of 4,549,457 votes, there’s a 0.500546 chance, on average, that a NJ voter will vote for Harris. I figured that out using the BINOM.DIST.RANGE function and the Goal Seek tool in Excel.
• In Michigan, with a turnout of 5,539,302 voters in 2020, Harris wins in only 605 out of 1000 simulations. Using the same tools above, if you randomly picked any Michigan voter, there’s a 0.500059 chance that he or she is voting for Harris.
• Using the BINOMDIST function with the assumed turnouts and the chances we determined that voters in each of the above states would go for Harris, there’s a 3.25986e-4 chance that Michigan is decided by a single vote. Likewise, there’s a 2.47681e-5 chance for the same in NJ. Based on the probability that it could shift electoral college votes, a Michigan ballot is distinctly more powerful than an NJ one.
• If you could could reliably convince one more person to vote like you in NJ, your chances of affecting the NJ outcome only increase to 2.48222e-5.
• For an NJ voter to match their chances of affecting the Michigan outcome, they would have to command about 1,925 votes besides their own. In other words, there’s an almost equal chance of a single vote Harris victory in MI as a 1,926 vote victory in NJ.
• Therefore, if a Michigan voter values their power, they should not trade their vote for anything less than 1,926 New Jersey votes. The rate should actually be greater to account for welching and Michigan having one more electoral vote than NJ.

I just learned about this around 11:45pm ET November 4. I signed up as a swing-state voter, and in 5 minutes got connected with 2 safe-state voters who agreed to vote for Jill Stein as a swap for my (very reluctant) vote for Harris. I feel much better now. :-)