Canon have announced
the next crop of PowerShot cameras, at least in the A and S series. As expected, they're still running the megapixel race hard, cramming more and more increasingly tiny pixels into a pocket-sized sensor to give the public excitingly bigger numbers, whilst ignoring the growing chorus of criticism that
more pixels aren't better. So when you buy one and realise that your new 10-megapixel photos look like crap compared to the old A570IS you just replaced (which didn't look that crash hot when compared to the silky-smooth dynamic range from your ancient 4MP PowerShot G2), you'll know why, and wish you had gone and bought a camera with fewer pixels. Well, if they actually still made them, that is.
You can blame the gullible idiots who are convinced that more pixels is better, have no idea of esoteric concepts such as "noise" or "dynamic range", and will rush out to buy anything with a higher pixel count, dooming any compact camera with fewer, decent-sized pixels to certain death in the marketplace. Though you'd think that there'd be enough people with a clue and a demand for a high-quality pocket-sized camera with the right compromises made in its design parameters to give the best (as opposed to biggest) images you can expect from that size, for Canon (or someone) to bother making such a line.
They need a way to summarize the quality of a camera with one number. A quantifiable measure of picture quality. Analogous to how the Transaction Processing Council measures server speeds, so you don't have to decide by engineering specs like "Megaherz". Would a comparison of the photo against the original scene work? Does quality = similarity to reality?