OK, so this shot of English Bay, Vancouver, doesn't look all that weird. Here's how it appeared in today's Independent on Sunday, though:
That building on the right there looks as if it could topple over at any moment. What's happened is that the camera was fitted with a fish-eye lens, which has basically mapped the whole scene onto the outer surface of a sphere. To correct it, all you have to do is to map it onto the inside surface of a sphere. Coreldraw 12, the graphics software I use, can do this (albeit not in a way that allows you to define the radius of the sphere, which is mildly annoying). It took me about 30 seconds to fix up TIOS's picture, and it's the first time I've ever used the feature. Someone who actually knew what they were doing should have been able to correct the image so it looked perfectly natural, and they should have been able to do it very quickly.
This leads me to conclude, therefore, that the image was deliberately left uncorrected. Why, though? What is it about having bendy buildings that is aesthetically pleasing? Maybe it's not the aesthetics; maybe they're leaving the fish-eye effect in place so as to signify that it's such a wide vista they had to use a fish-eye lens? I've actually only shown 80% of the image, as there's another 20% of it spread across onto a page to the left, but that still leaves the woman in the kayak some way left-of-centre.
Photography is one of those art forms I've never read up on, mainly because it's too tempting to want to apply it. "Anyone can take a photo, so that means I can!". After all, it's just point and click. Then again, so is drawing pictures in Coreldraw, and I'm hopeless at that... I'd rather take photos as memories or curiosities, not as statements; if I had the vocabulary to do the latter then it would doubtless interfere with my ability to do the former.
Still, it does mean I end up wondering why photographs in newspapers often look so deliberately weird.