It might be considered bad, because linear lookup is slow when compared to other alternatives. It can be replaced either with binary search which could be a couple of times faster, or, even better, constant time subtraction.
That would be theory, actual performance difference would have to be measured and verified.

Could you, please, explain, why it is a bad solution to use "the indexOf on the alphabetical-order literal"?
I'm new in programming, and it is an obvious solution for me as I don't know and don't understand many of the JS methods.
Thank you.

Easily the quickest and greatest example.

This is definitely neither a 'best practice' nor a 'clever' solution. Using the indexOf on the alphabetical-order literal is a VERY, VERY, VERY bad practice. Javascript can already implicitly convert chars to numbers and their offsets from eachother are already correct.

If I saw someone use this solution in the real world, I would kindly show them the next solution in this list - this solution is simply atrocious.

Best practice: no, because this takes a bit of interpolating to understand. Clever? 100%, yes, absolutely.

Because each element is ensured to be repeated an even amount of times except for one, the XOR operation works perfectly because... (A ^ A) = 0 and (A^A)^A = A.