If you've got a great designer, you need to let him try what he believes in for an extended period of time. The problem with A/B testing is that they tend to h

If you've got a great designer, you need to let him try what he believes in for an extended period of time. The problem with A/B testing is that they tend to have a short horizon. Facebook's famous newsfeed change pissed off/confused a bunch of people-- but in the long term it was a good thing for Facebook. I wonder if they'd done a 1-2 day A/B test how it would've performed? And, clearly-- on tiny things like shades of blue and pixel widths, just launch something. With infinite resources it'd be great to test all permutations-- but surely they have more inspiring things to try/test that might move the needle.