The industry as a whole makes its choices on the basis of folklore

This is a question it is reasonable to ask any blogger endorsing a product. It is also a question any blogger should clearly and explicitly answer, if asked.

>The relationship between the speaker and the statement has no bearing on the truth of the statement.

It has no bearing on the objective truth of the statement. However, we have no Oracle we can check for objective truths. So with incomplete information, we reason subjectively and probabilistically.

Books that stand the test of time, and are still popular in our community, are more likely to be good than an arbitrary new unknown book. We apply a different prior, and there's nothing wrong with that.

P(Book_good|classicInCommunity) > P(Book_good)

When we see three books endorsed, two respected classics, and one new to the market, its natural to wonder whether there's a great new book on the topic that has just come out. But we must ask ourselves, how often does this happen? And how often do people shill on blogs?

There's nothing wrong with reasoning about these relative probabilities, from our perspective of incomplete information - and it'd be foolish not too.

Anyway; I also believe: P(shill|HNreader) < P(shill), so I'd give you the benefit of the doubt. Theres nothing wrong with applying such a belief, even though it obviously doesn't change whether you ARE a shill or not.

But it'd be nice if you answered bugsy's question. Because P(book_good|endorsed,interest) < P(book_good|endorsed)

Your statement of defense "The relationship between the speaker and the statement has no bearing on the truth of the statement." does imply as I suspected that there is a relationship between the author of the blog article and someone involved with the just released book he is recommending as an instant classic. What is the nature of this relationship?

First off, everything on raganwald's blog is in his personal interest directly or indirectly. He isn't a journalist, he doesn't have to answer questins like "Do you own shares in Apple" whenever he mentions OS X, because he doesn't have a "Conflict of Interest." That's because he isn't employed by the publisher, he is the publisher. So his interests are the publisher's interests. He doesn't represent his writing as objective or news. He represents it as his own opinions advanced for his own interests.

So I get why he won't answer whether he has an interest in anything he writes about. The answer is "Yes," whether he says so or not. In my mind, a better question is, "Nickieben, what made you suggest that book?" Note the neutral tone :-) And indeed, I can and will answer that question.

Both raganwald and Greg Wilson spoke at Stack Overflow's Dev Days in Toronto. Here are a few reactions to their talks:

http://meta.stackoverflow.com/questions/27151/devdays-review...

http://blog.techscene.ca/2009/10/24/in-review-stack-overflow...

Greg's slides are available:

http://www.slideshare.net/gvwilson/bits-of-evidence-2338367

Flash forward to last week, and raganwald published a post claiming that programming wasn't an empirical science.

I wrote a post correcting raganwald's error, and named three books. Why three? Well, it's a weird "magic number." You know, you say "There are books treating programming as an empirical science," and someone retorts "Oh yeah? Name three!"

Two are indeed "classics," although quite honestly I don't treat old books as somehow magic, especially if there aren't a lot of others to read. I wanted the most current example of a new book, and since Greg's book is (a) decent, and (b) doesn't have cobwebs, that's the choice I made.

But honestly, don't sweat Greg's book. The important message is really that there is research, but not enough of it. We need so many that the "classics" are the ones that have stood the test of time by being better than the 100s of other books published since they came out.

That isn't the case with empirical research on programming. My "endorsement" would mean a lot more if Greg's book was one of twenty published every year backed by research, and you want to know why you should read his instead of the other nineteen.

If you want me to retract that endorsement, I could just as easily say Read every book that is backed by actual research. The lack of competition is the real problem here, not whether there's some cozy Toronto cabal of shills in operation.