This is a really interesting topic, I think. I read a lot of papers on distributed systems, and while many of the systems are conceptually not terribly complica

This is a really interesting topic, I think. I read a lot of papers on distributed systems, and while many of the systems are conceptually not terribly complicated, there's a huge leap from understanding the concept to making it work. A good example of this is "Paxos made live" - conceptually Paxos is relatively simple, but even a top engineering team at Google had trouble implementing it well. In the comments the point was also made that academics are simply not rewarded for making their work accessible - the theoretical work is everything. I made my first contact with a friend (and now colleague) asking him when he planned to implement a feature in the open source library based on his research. He said he had very little incentive to implement it having proved it correct. Until this changes, I think this barrier is unnecessarily high.