Return to Answer

The javascript code used to make the picture frames of the gif follows at the bottom. The code can be used as a starting point to make your own improved gif/animation or just single png frame. [may try to clean js code up later on as well as make more running time efficient]. I then clicked through to each pic, carefully screen captured the same bordered region for each pic, and saved to file. I integrated them into a gif using http://gifcreator.me/ (most frames got 250ms delay, but the first and the last of each of the 6 sequences got 750ms). I took that final gif and uploaded to stackexchange http://meta.stackexchange.com/questions/75491/how-to-upload-an-image-to-a-post

In case the above very short explanation + pic is not enough, here is a longer re-explanation (leveraging pic):

Simple Geometrical Explanation:

[To get a simple explanation, we have to have a simple approach. A circle is a simple, easy-to-make shape, and this problem was studied ages ago with simplified reasoning.]

The question posed is why can't we approximate the length of a circle [PI = the length of a circle of diameter 1] by measuring the length of a "staircase" path that hugs the circle tightly?

The answer is simple:

If we aim to find the length of some near straight object from point A to point B, we want to measure as closely as possible to a straight path from A to B (see green/red quasi-overlap). We won't get the correct answer if instead, like the staircase approach above (purple), we measure from A to a point far off to the side and then from that point to B. This is very intuitive.

Now, to approximate the length of a circle, we replace the whole circle with many little straight paths following closely the shape of the circle (green). We do use a single direct connecting (green) piece between every two adjacent points A and B (A and B, not pictured, would be where adjacent gray lines intersect red circle) instead of using the inaccurate 2-piece (purple) step. Do observe a key point that makes this work out: any little arc of a circle, as with any small section of any simple curve, becomes nearly indistinguishable from a similarly sized line segment when these are short enough.

[Recap:] So, at any angle around the circle, for large N, a small green line segment ≈ small red arc. Meanwhile around most of circle 2 right angled purple line segments are clearly > matching red arc, no matter N. This is why the green approximation gets very close to π while the purple approximation is way off at 4. [Note: green π = N sin (pi/N) and is easily derivable from basic geometry by summing 2*N pieces that are opposite radial triangles with hypotnuse .5 and central angles 2π/(2N).]

[Finally, I apologize if you can't discern green from red. I may change colors later but found these convenient and generally easy to differentiate.]

The javascript code used to make the picture frames of the gif follows at the bottom. The code can be used as a starting point to make your own improved gif/animation or just single png frame. [may try to clean js code up later on as well as make more running time efficient]. I then clicked through to each pic, carefully screen captured the same bordered region for each pic, and saved to file. I integrated them into a gif using http://gifcreator.me/ (most frames got 250ms delay, but the first and the last of each of the 6 sequences got 750ms). I took that final gif and uploaded to stackexchange https://meta.stackexchange.com/questions/75491/how-to-upload-an-image-to-a-post

In case the above very short explanation + pic is not enough, here is a longer re-explanation (leveraging pic):

Simple Geometrical Explanation:

[To get a simple explanation, we have to have a simple approach. A circle is a simple, easy-to-make shape, and this problem was studied ages ago with simplified reasoning.]

The question posed is why can't we approximate the length of a circle [PI = the length of a circle of diameter 1] by measuring the length of a "staircase" path that hugs the circle tightly?

The answer is simple:

If we aim to find the length of some near straight object from point A to point B, we want to measure as closely as possible to a straight path from A to B (see green/red quasi-overlap). We won't get the correct answer if instead, like the staircase approach above (purple), we measure from A to a point far off to the side and then from that point to B. This is very intuitive.

Now, to approximate the length of a circle, we replace the whole circle with many little straight paths following closely the shape of the circle (green). We do use a single direct connecting (green) piece between every two adjacent points A and B (A and B, not pictured, would be where adjacent gray lines intersect red circle) instead of using the inaccurate 2-piece (purple) step. Do observe a key point that makes this work out: any little arc of a circle, as with any small section of any simple curve, becomes nearly indistinguishable from a similarly sized line segment when these are short enough.

[Recap:] So, at any angle around the circle, for large N, a small green line segment ≈ small red arc. Meanwhile around most of circle 2 right angled purple line segments are clearly > matching red arc, no matter N. This is why the green approximation gets very close to π while the purple approximation is way off at 4. [Note: green π = N sin (pi/N) and is easily derivable from basic geometry by summing 2*N pieces that are opposite radial triangles with hypotnuse .5 and central angles 2π/(2N).]

[Finally, I apologize if you can't discern green from red. I may change colors later but found these convenient and generally easy to differentiate.]

Analysis goes further than Euclidean geo ever could.. but you are still getting at the same basic thing

To get π, you use a straight line path connecting the many points on circle. Tiny circular arcs straighten out and approach that path ("difference" bounded below only by 0) any other poly path that clearly does not approach straight line ("difference" bounded below by number higher than zero) will thus not approach circle. Problem is that there is no real definition of length of a curve we are going on. Euclidean geom defines length more loosely. It gives definite values for some shapes, including lines, circles, etc. These match the physical world notion of length. Analysis (and there are different variations, some of which go even further) goes further than Euclidean Geo and define more generally a distance definition for arbitrary curves. To prove using those tools, you have to first know precisely how length is defined there and then build the formal argument upon it. With Euclidean (intuitive) view and not diving more formally than that, you are limited to a certain amount of hand waving. You really must define length for a curve precisely if you want a precise argument.

We can see at low zoom how the (purple) staircase hugs circle, but higher zoom shows it always remains a crude approximation to the circle's shrinking matching segments except near 0, π/2, π, and 3π/2. [In contrast, the (green) inscribed polygon is an increasingly good approximation and equally good at all angles.]
-- see Update 1 at bottom and see "Simple Geometrical Explanation" below.

The javascript code used to make the picture frames of the gif follows at the bottom. The code can be used as a starting point to make your own improved gif/animation or just single png frame. [may try to clean js code up later on as well as make more running time efficient]. I then clicked through to each pic, carefully screen captured the same bordered region for each pic, and saved to file. I integrated them into a gif using http://gifcreator.me/ (most frames got 250ms delay, but the first and the last of each of the 6 sequences got 750ms). I took that final gif and uploaded to stackexchange http://meta.stackexchange.com/questions/75491/how-to-upload-an-image-to-a-post

In case the above explanation + pic is not enough, here is another explanation:

Simple Geometrical Explanation:

[To get a simple explanation, we have to have a simple approach. A circle is a simple, easy-to-make shape, and this problem was studied ages ago with simplified reasoning.]

The question posed is why can't we approximate the length of a circle [PI = the length of a circle of diameter 1] by measuring the length of a "staircase" path that hugs the circle tightly?

The answer is simple:

If we aim to find the length of some near straight object from point A to point B, we want to measure as closely as possible to a straight path from A to B (see green/red quasi-overlap). We won't get the correct answer if instead, like the staircase approach above (purple), we measure from A to a point far off to the side and then from that point to B. This is very intuitive.

Now, to approximate the length of a circle, we replace the whole circle with many little straight paths following closely the shape of the circle (green). We do use a single direct connecting (green) piece between every two adjacent points A and B (A and B, not pictured, would be where adjacent gray lines intersect red circle) instead of using the inaccurate 2-piece (purple) step. Do observe a key point that makes this work out: any little arc of a circle, as with any small section of any simple curve, becomes nearly indistinguishable from a similarly sized line segment when these are short enough.

[Recap:] So, at any angle around the circle, for large N, a small green line segment ≈ small red arc. Meanwhile around most of circle 2 right angled purple line segments are clearly > matching red arc, no matter N. This is why the green approximation gets very close to π while the purple approximation is way off at 4. [Note: green π = N sin (pi/N) and is easily derivable from basic geometry by summing 2*N pieces that are opposite radial triangles with hypotnuse .5 and central angles 2π/(2N).]

[Finally, I apologize if you can't discern green from red. I may change colors later but found these convenient and generally easy to differentiate.]

We can see at low zoom how the (purple) staircase hugs circle, but higher zoom shows it always remains a crude approximation to the circle's shrinking matching segments except near 0, π/2, π, and 3π/2. [In contrast, the (green) inscribed polygon is an increasingly good approximation and equally good at all angles.]
-- see "Simple Geometrical Explanation" below for longer but still simple explanation. The Updates at bottom add more insight once the simple geometrical explanation is not good enough for you. [Need to add more pics to clarify some aspects better.. ultimately potentially leading into something approaching a formal proof.]

The javascript code used to make the picture frames of the gif follows at the bottom. The code can be used as a starting point to make your own improved gif/animation or just single png frame. [may try to clean js code up later on as well as make more running time efficient]. I then clicked through to each pic, carefully screen captured the same bordered region for each pic, and saved to file. I integrated them into a gif using http://gifcreator.me/ (most frames got 250ms delay, but the first and the last of each of the 6 sequences got 750ms). I took that final gif and uploaded to stackexchange http://meta.stackexchange.com/questions/75491/how-to-upload-an-image-to-a-post

In case the above very short explanation + pic is not enough, here is a longer re-explanation (leveraging pic):

Simple Geometrical Explanation:

[To get a simple explanation, we have to have a simple approach. A circle is a simple, easy-to-make shape, and this problem was studied ages ago with simplified reasoning.]

The question posed is why can't we approximate the length of a circle [PI = the length of a circle of diameter 1] by measuring the length of a "staircase" path that hugs the circle tightly?

The answer is simple:

If we aim to find the length of some near straight object from point A to point B, we want to measure as closely as possible to a straight path from A to B (see green/red quasi-overlap). We won't get the correct answer if instead, like the staircase approach above (purple), we measure from A to a point far off to the side and then from that point to B. This is very intuitive.

Now, to approximate the length of a circle, we replace the whole circle with many little straight paths following closely the shape of the circle (green). We do use a single direct connecting (green) piece between every two adjacent points A and B (A and B, not pictured, would be where adjacent gray lines intersect red circle) instead of using the inaccurate 2-piece (purple) step. Do observe a key point that makes this work out: any little arc of a circle, as with any small section of any simple curve, becomes nearly indistinguishable from a similarly sized line segment when these are short enough.

[Recap:] So, at any angle around the circle, for large N, a small green line segment ≈ small red arc. Meanwhile around most of circle 2 right angled purple line segments are clearly > matching red arc, no matter N. This is why the green approximation gets very close to π while the purple approximation is way off at 4. [Note: green π = N sin (pi/N) and is easily derivable from basic geometry by summing 2*N pieces that are opposite radial triangles with hypotnuse .5 and central angles 2π/(2N).]

[Finally, I apologize if you can't discern green from red. I may change colors later but found these convenient and generally easy to differentiate.]