The SANS Institute has released a list of the 25 most dangerous programming errors – here: http://www.sans.org/top25errors/.
What’s interesting about the way they have produced this list is that they are not pitching it as “these are bad, so don’t do them” rather they are pitching it as being a helpful list from four perspectives. The first perspective they pitch is that it gives customers of software intelligence about how to select software that is likely to be more robust. Personally, I think this is quite an interesting way of looking at the problem as I sit on both sides – i.e. I’m generally interested in how people build software, but I am also interested in how people buy and sell software as that’s my day job.
The other perspectives are more classical – i.e. help developers build better software, but also help people coming into the industry know how to build better software from the get-go.