[copypasterepeat]

The way I think of this is that there is no shame in Jeffries not having as good and intuitive grasp of the problem as Norvig. After all, this kind of problem is almost tailor-made for Norvig's skill set. Like many of us who haven't done a lot of work in this space would do, Jeffries spends a lot of time early on just exploring the problem space. He starts with a seemingly intuitive (naive?) data structure and goes from there. So far so good. No shame in any of that. All of us have been in a similar situation. I think that the key moment comes at blog post number 12. As the author of this post describes, this is where Jeffries basically recognizes that his approach needs to radically change and where a lot of people would throw away what they have, it having served the function of helping one understand the problem, and start from scratch with different assumptions, data structures, etc. Jeffries didn't do that, probably because he firmly believes in the TDD dogma and probably because he felt pressured to make this work, since he way publicly working through this. That's where Jeffries goes wrong. He should probably have admitted that in some cases you should just start from scratch instead of tinkering at the edges of something that's clearly suboptimal.