Why does software matter to scientists? It may seem obvious to people who read this blog, but that's like saying that the answer to, "Why opera?" is obvious to the sort of person who pays a month's rent to get a decent seat at Covent Garden. Why does software matter? And why does it matter whether it's written well?
"It's the only way to tackle today's big questions" is a popular answer to the first question, while "We need to know if we can trust it" is a common response to the second, but I think both miss the point. Open source, open science, open access, open data... they're just enablers. The real prize is massive, open collaboration. It's meeting people you would otherwise never have met who can extend your work in ways you never thought of and then share their results so that others can take it even further.
But even that is just a means to an end. Software and science may be what we collaborate on, but they're also what brings us together, just as cleaning up a park can be what brings a neighborhood together. That's why I think groups like the Software Sustainability Institute matter so much. They are helping scientists (and through them, everyone else) build the technical and cultural muscle they will need when climate change, mass extinction, resource depletion, drug-resistant diseases, and all the other problems we're so resolutely not addressing right now can't be avoided any longer.