Is there any hope we will see substantial improvements in software quality? There are no silver bullets. However, we have seen progress. In the past five to eight years, we've seen much more use of automated tools in the software development process. Tools do more and more semantic analysis [of code]. That requires annotating the software so the tools have a better understanding of what the code represents. Eventually, we may be able to more automatically generate the annotations. The tools will become smarter, but the research challenges are still there.

What worries you most in computer security? I've been advocating to the research community to think about the threats of the future. The threats of today come from things like coding vulnerabilities -- buffer overruns and so on. We have been pretty fixated on code-level vulnerabilities, but we should be thinking about vulnerabilities at a higher level, at the component level. So you might use one component, like your browser, to interact with another component, like the DNS server, and all of a sudden there is a gap that can be exploited in an attack. We may in the future see more and more of these composition flaws, where even though components may be individually deemed secure, the ensemble may not be.

What kinds of projects do you fund with your [US]$535 million budget? The -- for example, "What is computable?" -- whose impact may be far in the future but which could be truly revolutionary. We also fund fundamental research driven by societal grand challenges, such as climate change, energy, environment and health care.

There are projects in bio-inspired computing, where individual molecules are considered a machine. People have built molecular machines, and the research challenge now is to get them to communicate chemically.

Another hot trend in computer science is in economics. For example, ad placement on Yahoo. And is all about auctioning keywords. There's a whole new field called computational macroeconomics. And there is algorithmic game theory. My mantra in computational thinking is that it will really influence the way people think, whether they are scientists, engineers, economists or musicians.