Of course I'm aware it's a 25% difference (or 33%). But at this level, when I have done, what, five minutes' of work? -- 25% is huge. It's probably better sometimes than anyone in my field is capable of getting after months of detailed calculation. At any rate it's comparable. If I can land within 25% of the "correct" answer with barely any work at all, I'm incredibly chuffed. And so would any other scientist be.
What this really means is that you don't understand how scientific estimation works, what the word "about", means, and how orders of magnitude are used. To a first approximation, 3 and 4 are the same size, and they are both *not* equal to one. That is the point. To complain that I said "about" four first time (and anyway, with the figures I gave, it was 3.72-ish, and I rounded up), and went for nearer to three with a different set of figures to reflect your suggestion, is to split hairs and to miss the point spectacularly badly.
I'm not interested in what you think of me as a scientist. Your opinion doesn't matter, and as it's wrong anyway. What *is* sad is that you seem to think that my training and experience count for exactly nothing. But that's because you apparently have been badly misinformed about the field, and the standards therein. Obviously, my results above wouldn't make it into a rigorous scientific paper. But I wasn't setting out to write one. I was just trying to make a point that voter power varies quite dramatically between certain states, and gave a rough estimate as to the maximum possible size of this variation. If you are interested, I can go through and provide you with the full table, based on four or five criteria: population of each state, adult population in each state, registered voting population in each state, actual number of voters in each state, and number of voters for the victor in each state. The numbers would no doubt vary, I know not how much by (I'm talking about a day or a couple of days' work to prepare the full table above). But it doesn't affect the argument. Why should you care if one vote in Alaska counts for 2.77 votes in Maine, or 3.12, or 3.4? If the ratio isn't one, shouldn't that be the only thing that matters?
But I bypassed all this, did one calculation, came up with one answer, and stuck with that as a *very* rough guide to the problem. This is how estimation works. The only reason I can see a problem is because it's useful to pick up on a figure I already knew to be wrong (but not wrong by all that much) in order to avoid having to discuss the actual point.