[wspride]

"start building things from papers."

Really liked what you wrote and found this particularly interesting. Can you elaborate on why you think this is important and how you generally go about it (finding papers, choosing what to code, etc)

[tptacek]

There are rapidly diminishing returns for picking up new languages for the sake of having multiple languages.

There are accelerating returns for knowing one or two languages so well that you can build nearly anything in them.

There aren't diminishing returns for learning new concepts, new algorithms, new distributed systems protocols, new problem domains. You can keep implementing things from papers indefinitely and you will keep getting better and, potentially, more valuable.

[yen223]

How should one go about finding interesting white papers to implement?

[tptacek]

Go to the websites of well-respected university computer science programs and browse their research groups; for instance, go to MIT's site, and then check out PDOS. University research group sites usually have indexes of projects, each of which has a bunch of papers backing it. Read the web pages to see if the project is interesting, then read the first paper, then decide whether you should get your hands dirty with the concepts.