Most recently, there was this book. It was supposed to be a guide to Natural Language Processing using a programming language I'm aware of, and have read books about, but have never used. Since Natural Language Processing is both interesting (if you like computers and psychology) and also of interest to some stuff I'm presently doing, I thought this book would be good for me to read.

I should point out at this stage, that I got the book for free as part of my Amazon Prime membership, which allows me to borrow books. So far, I've borrowed a few books via this route. The last one I borrowed was fairly unreadable, but I chose not to review it because although it wasn't to my taste, it was written the way such things should be written and I kind of know the author, and I've no reason to want to publish an opinion on something that I can just walk away from. The NLP book I just read, and I pushed myself to read it all, was a disaster area.

I'll link to and quote the review.

I was curious about this author and curious about whether I'd made a mistake, either in my interpretation of the book, or in my decision to read it in the first place. I had posted my 1 star review, a sister to the existing 1 star review, because I'd come to the conclusion that this book needed a warning to other fellow travellers on the path to knowledge. Do not read this - it's at best pointless, at worst misinformation. However, could I have been wrong? Who was Guermo Isaac, the book's author. I did some Googling.

I found Guermo's blog here. In particular, on my birthday this year, Guermo under the pseudonym on his blog (and I think his name is already a pen name), berated the writer of the scathing review. "They're doing it for their own purposes - they wreck good business" - seems to be the sentiment.

I, like the prick I am, made a comment on his blog, which I kind of regret and kind of don't. My fears that this man may be a little quirky and not a reliable source were intensified in the discussion and I think he's best left alone. What worries me slightly is that he's probably going to republish the book with a different edition/title/identity to try to distance himself from the reviews as an unreviewed book might, internationally, pick him up a single sale a day - maybe $5 worth, maybe $1500 per year's worth of income. To him, it's receiving his dues for his hard work in writing the books and knowing what he thinks should be in them.

The problem is this - who is right? Am I right, as a consumer, to warn others off this product, which I am convinced is terrible. Or is he right to say "if you don't like something, keep quiet about it"?

I think the thing which hurts this man's ego the most is that the review, which looks like 5 minutes' work, is so short and yet so damaging. I pointed out that it take 5 minutes to write the review and a couple of hours to read the book in the first place to be able to form the opinion behind it. He considers the star rating to be rude - "Say it's bad, but don't give 1 star". My view is that shit is shit. It's the same with Edinburgh shows, if you think it's shit, don't go 2 stars to be nice - put it down as 1, especially if you're trying to say that nobody would like this.

I didn't write a character assassination of the individual and his work on Amazon. Two reasons. Firstly, I was writing on a mobile device in the middle of the night without my glasses, so couldn't be bothered. Secondly, I reckoned it wouldn't get past the moderators. I was more respectful. Here, however, I can say what I like and I'm using a keyboard. In short, here are a few of the things which I noticed while reading a book I cannot recommend enough to avoid. I'm write and he's wrong. Sorry.

• A lot of his explanations of the Python programming language come from a position of ignorance - "I don't know what it is precisely, but I call it..." - just do the research
• He regularly declares for loops as inefficient without explanation - when he comes up with an alternative, it's probably less efficient
• He describes tricks and workarounds which kind of work as they are, but doesn't seem to know why they work - in some cases, they seem to be voodoo, rather than a simple solution
• He doesn't understand exceptions and uses them incorrectly
• He writes "therefor" instead of "therefore"
• The book has a few cases of apostrophe-s to make a word plural - that's wrong
• His description of the sciences required for natural language processing is plain weird and misrepresented - he thinks you need quantum physics to parse sentences
• He uses words he's made up, or made up new meanings for
• He claims that parsing and parsimony are related and uses parsimony when talking about parsing - this is bullshit.
• He describes the Turing test in terms that he's made up
• He claims to have a chatbot that passes the Turing test - he quotes a conversation with it - it's rubbish
• He describes a "best practice" for keeping track of versions of code and experiments you've tried which is to save lots of versions with different file names - so he's never heard of source control then.

And on that bombshell, I'm outta here!