Right. You’ve identified a couple of interesting ideas there, and let me try to play them back. So you talk about running code and you talk about sandboxes, and the observation being that in a technological system the theory is not as useful as the actual practice where you can see the data, you can see the results, and you can see the outcome. And the notion of having a sandbox is that it’s a small, somewhat constrained part of the environment where you can afford to take a little bit of a gamble because it doesn’t necessarily affect your entire business. In the case of Facebook, Taiwan is 20 million out of 2 billion users so it’s big enough to matter to do a good test, but it’s not big enough to take down the whole business.