Yup. In these puzzles, typically one of two things happens: division by zero, or taking square roots and ignoring the existence of two possible roots. As hdc_bst found, in this case it’s the latter here.
I’ve also seen them based on omitting a constant of integration. IIRC, it made use of logs, since for most integrals, leaving off the constant will give you the integral from 0 to your point, but for the integral of 1/x, you find the integral from 1 to your point (since the integral from 0 diverges).
Also, this line is incorrect. So to move the 2xa over to the other side, you subtract it and that works out.
But to move the 2xb to the other side you ADD it, which means it’s b[sup]2[/sup]+2xb
What would be incorrect?
To move -2xb to the other side, you have to add it on both ends.
Actually, what would happen is you’d get -b[sup]2[/sup]+2xb, which you could convert to b[sup]2[/sup]-2xb but only if you put a -1 on the other side as well. Either way it doesn’t work.
A trivial observation on hdc’s solution: If X is the mean of A and B and A!=B, X must be halfway between A and B, i.e. smaller than one of them but bigger than the other, and by the same amount each time.
Therefore, either: A - X = X - B (and B - X = X - A).
In general, an easy way to find a step where things go wrong: Plug in actual numbers for a and b and walk through the lines till you find the first equality that isn’t actually true with those specific values. If you did this, it would readily have been apparent that the step moving from (a - x)^2 = (b- x)^2 to a - x = b - x was the source of the problem.