Rather than funny analogies about signals spinning on poles, I think it's relatively easy to understand what the DFT is with just two things that many people already know: 1) knowledge of the complex exponential definition of Fourier series [1], and 2) how to approximate integrals with the left rectangle rule [2].
Take your continuous signal and represent it with a Fourier series. Since the Fourier series is a linear decomposition into integer frequency sinusoids, the coefficients of the series tell you the amount of each frequency contained in the signal. The DFT gives you an approximation of these.
The coefficients of the Fourier series of a function are integrals. Approximate these integrals with a left Riemann sum. Integrals turn into sums ... sums turn into a linear system ... the linear system turns into a matrix ... bingobango there's the DFT matrix [3].
Take your continuous signal and represent it with a Fourier series. Since the Fourier series is a linear decomposition into integer frequency sinusoids, the coefficients of the series tell you the amount of each frequency contained in the signal. The DFT gives you an approximation of these.
The coefficients of the Fourier series of a function are integrals. Approximate these integrals with a left Riemann sum. Integrals turn into sums ... sums turn into a linear system ... the linear system turns into a matrix ... bingobango there's the DFT matrix [3].
[1]: http://users.wpi.edu/~goulet/Matlab/overlap/efs.html
[2]: https://en.wikipedia.org/wiki/Riemann_sum#Left_Riemann_Sum
[3]: https://en.wikipedia.org/wiki/DFT_matrix