This reminds me of A/B testing. Sure, you may stumble upon a method that leads to a more motivated and productive staff, but do your successes have a reasonable chance to make up for the lost productivity when your social experiments perform worse than current industry practice?

I don't know, maybe that doesn't even worry them. Maybe their real goal is to attract the sort of workers that would be interested in this sort of workplace experimentation. (Edit - I'm noticing a number of comments asking if they're hiring.)