The data typically now is attached to the process. You have something you want to do, you want to count widgets, and you collect the data for counting those widgets. You process it, and you have a very effective system for collecting and using that data to count widgets. That data may be perfectly valid for counting glasses, but because it's over here, we go and collect another set of data, that repeats the other set of data, and does it again.
We need to get away from this idea that data is in this repository. It's in a warehouse, and we can't get at it. That's the change in mindset. Yes, there are structural issues, but Amazon doesn't care about the location of the data. In fact, with all these big entities, the data is in a cloud. They go and get it. People who are given permission to the data can get the data, and people who aren't can't get the data. They build in security systems that are probably better than what our banks are using, and they're very effective at doing that.