If you want to implement anti-erase as a layer mask, then for antierase to be available, this layer mask (not shown to user) has to be present all the time (if not, the information needed for anti-erase would be lost).
But how this situation differs from separate alpha channel -- except for storing the same data in a much more complex way than necessary?
Because alpha is coverage. Zero coverage means there is nothing there. The pixel is not part of the image, there is no information conveyed by the pixel. It means literally that "there is no coverage of this point in space". Also "this pixel is not a sample of the image."
What I mean when I say alpha is coverage is that alpha was modeled and incorparted into the forumlas we use as the amount that the image covers the pixels. This is a very real life, "I just took a photo" of something concept. The image it not equivilent to the pixels in this context. The pixels represent a digital sampling of the image. The real image has nigh-infinite resolution. To save space, we choose a sample size and fill in the color that is closest to what the image represents. Sometimes, though, especially when compositing, an average color per pixel is not good enough. You need a number that describes how well the pixel is covered. This is alpha as described by computer scientists.
Now, alpha means transparency to most graphic artists, so this is how it gets used, and in most cases this is ok. But there is a big real world difference between partially obsuring a pixel with opaque objects and putting a piece of colored glass in front of a pixel, even though the result is usually the same when sampled as RGBA.
Although back on the topic of anti-erase, I think that the only way to do anti-erase correctly is with another layer. Once alpha goes to zero, the pixel no larger part of the sampled image.
