DISQUS

Fun book review. You're more funny since you shaved your head. Maybe I should shave mine.

"Errors using inadequate data are much less than those using no data at all." -- Charles Babbage

Nice post. I think many heated arguments could be made more calm and productive if examined in the context of Bayes theorem.

There is a corollary to the Bayesian argument in your post that I often point out: no amount of evidence can convince someone of a proposition to which they assign prior probability zero. This goes back to what Dennis Lindley called Cromwell's rule, the advice to always assign a least a tiny positive prior probability to any proposition unless it is logically impossible.

I doubt I agree with the author's approach to mathematics, but I'll go along with Horatio's Principle if it is meant to be a metaphor. Models are always inadequate, and the larger your world, the more your model must leave out.

Along these lines I like the story in Black Swan about casino actuaries mastering game probabilities but never thinking about possible losses from a tiger attacking a performer.

When people say things like "Look how much money we've spent on testing and it still didn't catch this bug" I bet their testing efforts were poorly allocated, not that they spent too much on testing.

The biggest expense in testing is inefficient allocation of effort. We concentrate on what is easy to test or what we think we need to test. But bugs are more likely to be hiding in code that is hard to test or code that we don't think we need to test. Randomization is a way to compensate for this bias.

That's cool! I've never seen a snow elephant.

One of my kids made a tiny snowman the last time we got about a millimeter of snow in Houston, really more of a very premature snow baby than a snowman, fit in the palm of your hand.

I thought about you and your snowman as I was soaking up some sun in Houston this afternoon. 77 F, 25 C.

Lynne: I'm glad to hear I wasn't the only person who thought 2.7 has its drawbacks. I concur that it is slower in general and some common tasks now require more clicks.

It takes me longer to write blog posts with WordPress 2.7 than it did with 2.6. The new dashboard is very pretty, but it runs much slower. Some things I do now require more mouse clicks, and each mouse click takes longer to respond.

Your comment about making Gmail as ugly as you like made me think of a 37 Signals post from this morning. The post said the Drudge Report was one of the best designed sites on the web even though it's ugly. Here's the post.

Congratulations on 4,000 posts!

I disagree that the Stack Overflow project broke the rules Joel said they broke. For example, according to the Stack Overflow podcast, the programmers were vetted extremely well: Jeff chose people he'd worked with before and had known for years! Also, Jeff mentioned on the podcast how they unit tested their code and he talked at length about the private beta. So according to their own account, they at least followed rules 1 and 5. Except for putting all the developers in one office, I would imagine they followed all of the rules, at least informally.

Now *that* looks like fun!

I find it annoying when I'm teaching and I look out at a sea of laptops with bumper stickers. I'm looking for faces to gauge how well I'm connecting, and instead I see advertising. It would still be a problem if the laptops were clean, but the stickers add insult to injury.