semi-related ... some of my colleagues wrote a research paper about 4chan (but I don't think they posted it ON 4chan!):
2,356 karma · joined September 28, 2009
Next up: "4chan flame comments are a terrible way to incrementally advance the state-of-the-art in a narrow academic field, presenting theorems, experimental data, and citations to closely-related work."
there might be some value in "forced" non-academic activities, as long as people don't take it too seriously. iirc the P.E. teachers were often grad students who loved their sport, not jaded old fogies from middle school nightmares.
So the point of programming language research is to come up with ideas for the practical programming languages of the next 50 years to steal.
More generally, one point of research is to come up with ideas for practical technologists of the next 50 years to steal.
some 50+ year old tenured professors co-found companies and serve in advisory roles, but i highly doubt that they're staying up late setting watchpoints in gdb and punching their monitors.
Reminds me of Greg Wilson's http://software-carpentry.org/
Hmmm, I doubt that taking an OS project course or reading some virtualization research papers gives anyone the amount of insight into the nitty-gritty details needed to make a production-quality VM product. There is simply no substitute for thousands of hours of hard work pushing the boundaries on a field in order to get that required expertise.
Yes, perhaps today someone could improve upon VMware's products and form a competitor, but when VMware was first forming, I doubt that anyone else had the required expertise besides the few research groups who were actively working on virtualization for PC-class machines.
One alternative option is to be a founder who raises money and then hires someone who has actually done that research.
(Of course, I'm only talking about startups founded on research technologies ... you don't need a Ph.D. to found the majority of software startups)
I hope that's tongue-in-cheek, because compilation is definitely non-trivial for any sophisticated piece of software with lots of external dependencies, especially if they're not available in the package manager that happens to come with the distro you're running at the moment (and it's even harder if you don't have sudo/root access on your box).
I think people in the software world are inherently biased since it's possible to start dazzlingly-profitable companies at age 19 ... in many other fields (e.g., medicine, chemistry, biology, physical engineering fields), it takes years of research experience to be able to make a breakthrough that can lead to a profitable disruptive technology. In those worlds, you rarely see 19-year-old entrepreneurs who have just started college.
Shameless self-promotion ... over the past year, I've been working on a project that aims to solve this dependency hell problem: http://www.stanford.edu/~pgbovine/cde.html
All you need to do is find a distro where jdee works, package it up with CDE, and then transport it to your own machine. email me if you have questions.
I also noticed that someone recently posted a list of publications by Microsoft Research (MSR), presumably to contrast with Google's publication list. MSR does some amazing academic research, but it's not a fair apples-to-apples comparison to compare their publications with Google's, since the primary output of MSR is academic research papers (that's why the lab was founded!), while the primary output of Google is free web-based consumer products :)
Publishing in academic conferences is not the only way to be "non-secret", nor is it the fastest way to get early feedback on usability and systems design issues. In fact, it's a pretty slow way to get feedback. The sorts of conferences that Googlers submit papers to have an acceptance rate of 10% to 25%, which means that most papers get rejected; moreover, it takes a ton of effort to write up and present a paper to an academic audience using writing conventions that are very different than those required to write engineering design documents. Given those low rates of publication "success", it's not surprising that Googlers put their efforts elsewhere.