Possibly interesting: as someone who's worked with compression algorithms, including JPEG, I can tell you the compression in JPEG comes from suiting the output to the limitations of the human eye.

Most blue information is simply tossed (we see it poorly) as are small color changes (we see differences in brightness better than those in color). Then the 32000/64000 levels for each pixel are bucketed into 256 levels. I won't get into linear vs logarithmic and why that means that most of what we'd care about in post gets tossed in that bucketing process.

As a final product, it's good enough. I send JPEGs to my printer, for instance.