News for hackers. Stuff that matters.
Re:The Harry Potter fallacy by wongarsu on Thursday July 28, 2022 @08:22AM (#32263329)
The first is Zeno's Paradox, or a variation thereof. The solution is that passing through infinitely many midpoints is fine, since the sum of an infinite number of positive numbers can still be finite. The example that solves the original Zeno's Paradox (you run half the distance, then one quarter, then one eight, etc, and never get there) is the fact that 1/2 + 1/4 + 1/8 + ... sums up to 1, since it's a Geometric Series.

The second comment touches on how just because the limit of something approaches 0, you can't treat it as 0. If you divide a line into infinitely many segments, the length of those segments approaches 0, but if you treat them as if they were really 0 you get wrong results when summing them back up. Usually that gets relevant with integrals, but in this example it happens with a simple 1-dimensional sum.

Re:The Harry Potter fallacy by sltkr on Thursday July 28, 2022 @12:47PM (#32266874)
> The second comment touches on how just because the limit of something approaches 0, you can't treat it as 0.

I think this one is a little more subtle. You can definitely treat the limit of something as an exact value.

If you split a segment of unit length into n segments, then each segment has length 1/n. As n goes to infinity, 1/n goes to zero. But at the limit, 1/n = 0, exactly. Not approximately, really exactly, and there is nothing wrong with defining lim[n->inf] 1/n = 0 and using that in a calculation. It's no different than saying lim[n->inf] sum[k=1..n] 2^-k = 1 as you did in the first paragraph.

The problem is slightly different. The person conceives of a sequence of `n` copies of `1/n`. For any finite value of n, these add up to 1 of course; it's just the result of calculating `n × 1/n`. But at the limit, they pull a trick: they say lim[n->inf] 1/n = 0 (which is correct) therefore any sum of these values is 0 too. But we never had lim[n->inf] 1/n, we had lim[n->inf] (n × 1/n) instead!

If you put it in mathematical terms, the argument is:

   lim[n->inf] (n × 1/n) = 
   (lim[n->inf] n) × (lim[n->inf] 1/n) = 0
Which has two problems: first, you can't split up a limit like that, so the first step is already invalid. Second, even if it were valid, lim[n->inf] n = infinity, which is not a number, so you can't multiply with it, and the conclusion that the result is 0 is wrong. You might think `x × 0 = 0` for any x, but that's only true if x is actually a number; if x is not a number the answer is undefined.

This is a similar kind of fallacy that underlies the proof that pi=4, here: http://www.kevinhouston.net/blog/2011/03/pi-is-4-video/, where the author (intentionally) confuses the perimeter of the limit (pi) with the limit of the perimeter (4).

Re:The Harry Potter fallacy by causi on Thursday July 28, 2022 @10:38AM (#32264932)
If you divide a line into infinitely many segments, the length of those segments approaches 0

Hence the real world being quantized. You can't even hypothetically divide a string into infinite points because there is a smallest distance, below which it really does round to zero, i.e., a singularity.

Re:The Harry Potter fallacy by lamontcg on Thursday July 28, 2022 @03:15PM (#32268917)
We have no solid evidence of space-time itself being quantized.
Re:The Harry Potter fallacy by clairity on Thursday July 28, 2022 @02:23PM (#32268302)
alternatively, you realize the runner/observer is not infinitely small and that effectively puts a finite, lower bounds on the steps, because most of that approach to zero is too small to matter, practically.

learning that infinite things could be finitely bounded was mind-blowing back in the day however.