Check out any high-quality font recently released by Adobe, like Garamond Premier Pro. It comes in different "optical sizes", one for small print, one for slightly larger print, one for regular body text, one for large display sizes, etc. This is in addition to light, regular, semibold, and bold weights for each optical size. Type designers have been doing this since 500 years ago, and digital types are just beginning to catch up. You can't have one font that looks exactly the same in all media, all antialiasing algorithms, and all sizes and resolutions.

Most people don't even notice, but serious type designers care about these subtle differences. Besides, antialiasing alone won't solve this problem as long as screens vary greatly in pixels per inch. No matter how good your antialiasing algorithm is, text at 96ppi is going to look different from the same text at 326ppi. Just like text printed on coarse paper will look different from text printed on glossy paper.

Paper coarseness variants will generally be designed around the physics of ink, to trap it in the junctions of small letterforms and prevent ink bleeding. Optical sizes tend to be more about the relationship between headline and body, with things like tighter standard kerning, taller x-heights, and so on. A DPI variant should only concern itself with aliasing, and aliasing is a render-time issue. You can design a typeface to counter this from the foundry side, but a more sustainable solution would be to counter it from the software side.

Anyway, as long as such differences exist, I don't think we'll be able to convince type designers to disregard them. They want their fonts to look perfect now, they're not going to wait for smartphone manufacturers to catch up, and they already have the skills and tools to produce subtle variations of their typefaces.

A truly sustainable solution would be for every reading device to have 300ppi+ screens so that it doesn't really matter how you shade the last subpixel. Until we have that, leaving optimization to perfectionist type designers seems to me like a better stopgap measure than trying to standardize on one or another antialiasing algorithm across vastly different platforms.

Different vendors like Microsoft and Apple use antialiasing algorithms that are not only technically different but also based on different UI paradigms. (MS emphasizes crispness whereas Apple emphasizes preserving shapes.) Neither is obviously superior to the other, and the choice is to some extent a matter of taste. So it should be the designer's prerogative to decide whether to follow either paradigm or to disregard both and enforce his own aesthetic tastes.