This is one of the best articles linked from HN in the last few days. Well worth the read. I can't count the number of times I've hit a wall developing something and then as I explain it to someone else we simultaneously realize the solution. Or other times where a developer meeting (yes, they can occasionally be useful) does a better job of refining some new feature than days of development iteration.

If you get the chance also read this HN article, "Richard Feynman: Do We Live in A Cargo Cult Society?" ~ http://news.ycombinator.com/item?id=1369781 and this one, "The War of Art: An Interview with Steven Pressfield" ~ http://news.ycombinator.com/item?id=1370128 Each have a degree of overlap. The first goes into detail why being truthful to experimental process should be mandatory. The second why we personally screw up.