What is the difference between digital high ISO noise and film grain? Why does one "eat detail" and the other does not?
The size of the grains in the film varies depending on the film sensitivity. The more sensitive the film, the larger the grains. Digital noise is always the size of a pixel, regardless of the ISO setting.
Film grain is color neutral, as it consist mostly of luminance differences. Digital noise consists of both luminance and color differences, and is most visible in the blue color channel.
In the more recent digital cameras the digital noise is quite even. In earlier models the noise had more banding and patterns. The film grain doesn't have any banding or patterns, so it's seen as pure noise. If the digital noise has any banding or pattern, the brain can easily pick that up, and that is more disturbing than pure noise.
Neither grain nor noise eats detail. It's noise reduction that eats detail, as it can't tell the difference between small details and noise. Noise reduction is used on digital noise, but it can also be used to reduce film grain.
There is an example below. On the left is the film grain from a Kodak Gold ISO 200 film. On the right is the digital noise of a Canon EOS 5D Mark II @ ISO 3200. Notice the blue noise in the dark areas in the right image.
There's also the issue of chroma (color) vs. luma (brightness) digital noise. People generally find chroma noise more objectionable because it appears less natural; this is why noisy photographs sometimes work better in a B&W conversion. The better noise reduction algorithms can also address one or the other selectively.
I believe that film has pretty much only luma noise, which is one reason people sometimes find it less objectionable than digital noise.
The biggest difference is the patterns in the noise. Film grain is caused by the grains of silver present in the film, and are not in a consistent pattern.
ISO noise is caused in the digital sensor and is pixel based, and therefore in a pattern.
Some feel that film gran is more pleasing because of the inconsistent pattern in which the noise occurs.
Higher ISO film tended to have more grain; and higher ISO digital shots exhibit more noise - a similar cause, but the visual appearance is different.
Digital ISO noise is related to the size of each pixel, as the noise is per-pixel (so the more pixels you have, the less obvious noise is when viewed the same size), whereas with film, the noise is per crystal - you need larger crystals for higher sensitivity.
One analogy I've used in the past to demonstrate this is to ask several people to time with a stopwatch how long it takes a car to drive around a car park, and then to time how long a person takes to do the same journey - because the person is slower, the margin of error is a smaller in proportion to the overall figure, even though different people will give timings to within a few seconds of each other.
One interesting difference is that digital camera noise to a good degree is quantization noise related to the imaging process. This can go as far as to "ISO invariant" sensors where the ISO setting does not affect the raw image capture anymore but instead determines what exposure the camera aims for and how the raw values are then interpreted. For such a sensor, the noise level depends on the actual exposure but not on the set ISO value.
In contrast, the choice of ISO/ASA level for film determines the film grain. An ASA100 film underexposed by 3EV retains its fine grain compared to an ASA800 film but loses its contrast. If you bring up the contrast in enlarging, you get more of a dot-dithered look than a coarse-grained one.
Some digital cameras have the ability to use higher ISO values at reduced resolutions: that is a bit more like what higher ASA films do with their coarser grain even though the digital camera "grain" is a rectangular pattern while the film grain is irregularly spaced.
Film grain is more like a texture that brings to life the images.
Noise is a defect of the sensitivity of the camera to the light. Noise is the result of the interpretation of the light in an image that the camera can't capture. The camera can't capture light in low light situations so it invent the colours that the sensor can't caught. So this noise is not the representation of a real image is an interpretation of what the sensor thinks the could be there.