Alternatives to string theory first of all cause escalating human stupidity but as your blog shows, the former is far from being a necessary assumption for the latter. If you read carefully and click the links, you would be able to see evidence for all statements I wrote and you could have avoided writing this silly "article". Thanks, Lubos
If you read the "Not Even Wrong" blog post which David cites, you would note that the original poster of the story mentions about how *many* universities with lackluster security are getting broken into. Certainly a sizable portion of them had sufficient security resources prior to the break-ins, but various user errors (login system programming, social engineering, using untrusted systems for trusted work, etc.) were the ultimate cause of security breakdowns.
In the case of an unsustainable cluster, I don't believe it would be terribly difficult to find an undergrad with time on their hands who would be willing to gather requirements, construct, and document cluster maintenance over the course of a semester. Throw the student some independent study credits (but only if it was combined with a parallel programming course), a written recommendation when they apply for jobs or grad school, and you've got what you need without needing to pay anyone.