Ungar's thesis is one of the most magnificent tours de force it's ever been my pleasure to read. It's mostly a series of "Here's why this technique is impossible to implement efficiently, but never mind that; here's how we implemented it efficiently anyway."