Dennis Hamilton's, "Trust but demonstrate" post is a nice formulation of our human / systems interdependencies and interactions. It reminded me of a comment in a paper about proving programs correct. Paraphrasing: we don't prove people correct; we expect people to be responsible.
It's the same with a computer system. I expect (or want to expect) the system to take responsibility for it's part of the interaction. After all, I'm relying on the computer when I'm paying bills online, or buying a book from Amazon, or even posting to my blog and pinging Technorati. And if I'm ever going to be able to feel comfortable with using the computer, then the builders of these systems and applications need to step up and demonstrate trustworthiness. Let's write software that takes responsibility for its own actions. Follow up when things don't go as expected. And when it's appropriate, let me know that all went well. Seriously, I need software that can help me take care of my part of the interactions. Systems are getting more complex and complicated. I think the whole computer-human-computer interaction thing would be better if the systems took some responsibility for their actions.
I know this isn't easy. But with all the recent news about software that can tell if I'm lying, or when I can be interrupted, or ..., it doesn't sound that hard. So this is a request to all systems architects and developers and service providers. Build trustworthy applications -- demonstrate that we all need to be responsible participants in this system we call the Internet.