Why is the exterior derivative called exterior derivative

The exterior derivative is a sort of infinitesimal operation. See for example Bachman's notes on differential forms. (See also comments to the question, above).

EDIT: Updated link to Bachman's notes. They are now on the arXiv.

It is precisely an infinitesimal change. But an infinitesimal difference in what?

A differential 1 form measures the flux across an infinitesimal line situated at some point. By extension a differerential 0 form evaluated at a point in space measures the flux across an infinitesimal point (an infinitiseimAl point is the same as a finite point is the same As a point.)

Consider a smooth curve. Zoom in on a point and look at its neighborhood. The exterior derivative of a 0 form measures the difference between the zero forms evaluated at points infinitesimally apart. As the points draw nearer and nearer, the difference becomes more and more a differential. This physically measures the difference in flow that exits one point and enters the other. If one were to add all these infinitesimal differences across a curve, the interior regions cancel leaving the exterior difference behind. For when the flux exits one end, it immediately enters the next.

This is called the exterior derivative because the integral of the "exterior derivative" leaves the difference of the exterior behind. Loosely the word "integral" cancels "derivative" leaving " exterior" behind...hence its name.

This is naturally extended to differential two forms.

Consider a smooth surface. Zoom in at a point and look at the neighborhood. It looks like an element of a plane which when parametrized leads to a small parrallellogram.

a side note: A differential 2 form evaluated at a point measures the flux across the infinitesimal parallelogram (which is defined by two vectors infinitesimal in length). The skew symmetry comes from the linearity in flux (consider w(x,y) + w(y,x). The sum is w(x+y,x+y) which must be zero since the area of the parallelogram is zero...no passage is admitted since none can be admitted.

Back to topic: Considering this surface and having zoomed in, an exterior derivative of a differential 1 form represents the differences in flux across the lines of the infinitesimal parallelogram. This is different than the two form. The two form would measure the flux of air as it slices through the paper. This "flux a across the lines" represents the flux of water as it travels along the paper.

To understand this draw the parellogram. Suppose we've parametrized our surface to r and s. Then one point of the parellogram is (r,s). The two vectors that make up the parellogram are formed by varying r and s infinitesimally. Let's call them dr and ds. There are the two other vectors parallel to these on the other side. I'll call them drr and dss respectively. First, Find the differences in the flux from dss and ds. This is the differences in one forms from dss and ds. Next, Find the differences in the flux from dr to drr (notice this direction of the difference is polar to the first. This small detail is crucial. I'll ask you why its this way? Its because you'll run into a small issue a bit later when integrating across the surface) Then just add these differences up and that's it. That's the exterior derivative of 1-form...the net flux across the lines of a parallelogram.

As it happens the exterior derivative of 1 form is a 2 form.

Adding them all up across the whole surface leads to again a cancellation of the interiors leaving only the integral across the contour behind. Notice that's an integral of a differential 1 form.

This was a summary of the true inner workings of the calculus, I think. All of it can be made more vigorous and generalized, but this is the spirit of it, I think.