A few weeks ago I DiscoBlogged about a thoroughly dumb Mayo Clinic press release and BBC news article on a "vertical workstation"—a standard treadmill Elmer's-ed to an office desk for giants—that was supposed to help pudgy people lose weight as they worked. Since then I've found myself thinking that there must be a lot of lazy, inaccurate, misleading, or otherwise just bad science reporting out there, and DiscoBlog could do a great service to its readers by pointing out those articles that should be taken with not a grain of salt but an entire Salar de Uyuni. I forgot about that idea until hitting this AP article about how some parts of the world will be helped by global warming. Reading it convinced me that DiscoBlog should, nay, must take on this task of defending the world from crimes against science journalism. I hereby begin that mission by naming it the Worst Science Article of the Week. What's wrong it? First off, the headline: "Surf's up, Buffalo: The good side of global warming." You'd think by reading the first half of this headline that global warming just might bring surfing to Buffalo. But toward the end of the article we read, "So ... surf's up, Buffalo? Probably not. While oceanfront cities might have to build seawalls to hold back the ocean in a warming world, some researchers believe the freshwater Great Lakes will evaporate a bit." I'd say that this basically makes the headline a lie. Blatantly false. Contrary to the truth. The other big problem is here: "[Canada] would see a 220% increase in international tourist arrivals by the end of the century, followed by Russia with a 174% jump, and Mongolia, up 122%." How the hell do these researchers know how many tourists are going to go to Canada—let alone Mongolia— in 2100? No, seriously. How. I'd love to know. This is obviously an extremely complicated system that depends on millions of un-trackable variables, and there's no reason to crank some silly result out of a silly algorithm purporting to forecast the number of tourists a hundred years from now. You can add some caveats about not being sure, maybe put some error bars to the graph, but this is straight-up hoodoo. No one should try to pass this off as science, and no one should report it as such either.