I can personally recommend "Managing Gigabytes". It is very well written and contains a wealth of very valuable information--definitely worth its price!
I think it still is relevant, though the scale has changed by a couple of orders of magnitude. Their showcase system is dated, too. But I think it contains the most important basics for most people who are interested to see how these systems work, in addition to being the only book (at least to my knowledge) that describes how to combine all of the techniques into a search engine.
Obviously, this answer depends on what you're interested in.
As for myself, I'm quite interested in LISP and making compliators/interpreters for it. Therefore On Lisp and Lisp in Small Pieces are books I'd recommend.
(No, I'm not going to tell you that LISP is going to change your life.)
Good recommendations. I have read about 20 books on his lists and have them in my library. The difference for me is that some of these were very difficult for me to work through!
Off topic, but: I am reading "Data-Intensive Text Processing with MapReduce" right now and can strongly recommend it: it characterizes different types of problems that can be solved with map reduce, possible solutions, and tradeoffs for different approaches. Really a useful little book.
It's always very interesting to see the library of other people and especially the one of Peter Norvig... You may find books you don't know or see relationship between topics that you underestimated at some points.
Maybe it would be very interesting to see the bookshelves from HN users? I don't know if a lot of HN users have a LibraryThing account but it's a nice way to see what you share (or not share) in your bookshelves with someone else.
There are already two groups (moderately active) :
I've used my GoodReads (http://www.goodreads.com/) account in the past for this sort of thing. I didn't realize that there was such a selection of these "share what books you're reading" kinds of sites!
And there's even more... My personal favorite is http://readernaut.com/ just because the design is more modern and simple. Although it has the same problem I had with GoodReads: poor consolidation of editions of the same book, so you end up with so many duplicates. Compared to Wikipedia where there is only one page for "To Kill A Mockingbird" but many on GoodReads and Readernaut.
GoodReads has a pretty thorough-looking API (http://www.goodreads.com/api). Don't know if anyone has yet written a tool to just dump your entire account's contents to an XML file, but that seems like something that a motivated person could do. ;)
A lot of books on "agent technology". All I know about this topic is the popular press gave it a lot of attention a few years ago. Anyone care to give a thumbnail sketch on this topic?
I am confused. Have Peter Norvig really read that many books? I believe he won't mark something as favorite if he hasn't read it. I scrolled about 10 pages trying to see how many books are there and then gave up.
How does he find the time to do it? I have read a decent number of books, but compared to his numbers, mine looks like a rounding error.
And to top it, majority of books he has listed are pretty heavy and dense books.
Why is this so surprising? I know many people who've read many thousands of books. I've read thousands of books cover to cover, I have about a thousand books in my library at home (about 300 ish technical books - I gave away another 300 or so) and I am still in my thirties (Fwiw I started reading by myself when I was 3. I buy a book every two weeks or so these days).
Even if you restrict the scope to technical books(assuming they have to be read relatively slowly) reading and completely understanding the content of a few hundred technical books is well within the scope of a human lifetime. It isn't even that hard if you work steadily in a disciplined fashion.
"And to top it, majority of books he has listed are pretty heavy and dense books."
"Density" is a function of what kind of background you already have. Dr.Norvig has a degree in Applied Math (Erdos number 4 I believe) and a PhD in Computer Science. I am sure he can read a "dense" tract faster than most people could. Also, most of those books (I own some of them) are textbooks. They aren't really all that "dense" (compared to a PhD thesis in Pure Mathematics, say). Most advanced material in CS/AI is in conference papers and rarely, if ever, make their way into textbooks till years after publication. I bet Peter Norvig reads a thousand or so papers every year. Every CS researcher I know does.
I've heard him called the no. 1 AI guy, in the world. He's basically Google's head researcher, and used to head search quality, he's been big in NASA, worked in industry, and written the no. 1 AI book for students.
I expect he's had to read a few books to get to where he is.
Also, he probably knew 80% of the content before he read it. Most of those books will have a chapter on vectors, a chapter on probability, a chapter on information (etc), and he's mastered that years ago ...
because the subject of these books is not read, it is learnt
I can also skim through TCP/IP Ill. SICP, Dragon book, and say that I have 'read' them, but the impressive thing here is that Norvig is super smart, and probably has not only read all of these books, but understands them all.
I've started allocating a fixed amount of time for learning everyone week. This mostly consists of working my way through books like these and writing some small scripts that utilizes the material on some data I have.
A bunch of the technical books, especially the textbook-ish and "handbook of X" ones, aren't books you "read" so much as "read the beginning of, then skim and refer to the rest". A common pattern is that there's general material in the first 10-50% of the book, which you probably want to read, then the latter parts of the book are series of specialized chapters and sub-sections on various techniques and variants of techniques. You wouldn't normally read every page of all of those in sequence, unless you were working really deeply in that exact area.
You can also often skip loads of stuff if you're reading for applications-focused reasons rather than looking to directly extend the techniques in question as part of theoretical research. I read a lot of logic papers, because I build tools on top of logic-programming systems, and I find that out of an 8-page paper, there's usually only 1-2 pages of material relevant to me. There'll be 1-2 pages of introducing a new construct, motivating why the construct is useful, sketching how it would be implemented and how it relates to other ideas; and then 6-7 pages of proofs showing that their extensions have all the soundness and whatever properties you want them to have. When I'm reading, I read those 6-7 pages as a single sentence in my head---"and now we prove that it works"---making for a nice 75%+ compression ratio.
If you think you may be missing out on something, then you read that proof. Reading all of the proofs is usually a waste of time. (Not always a waste, sometimes they are all relevant, but usually.)
Usually no. :) The proofs are often pretty mechanical and proceed the way similar proofs do, if you've already read a few in that area (PLs proofs are even worse, a totally mechanical exercise in pages of structural induction). I actually wish there were more areas where they were just saved for tech reports and appendices instead of breaking up the flow of the main paper. When papers are 75%+ proofs, the nice nuggets are parceled out so slowly that it's pretty frustrating.
I do often read English sketches of why something works, if a paper includes them--- text that points to the core insight of why we're able to do something, which theorems or constructions allow it, which problems you might think would exist are actually avoided, etc. But I rarely need to see all the detailed symbol-pushing.
There are duplicates (Witten and Frank's "Data Mining", which is excellent), and some of the books he lists as favorites I would say are utter garbage. Some of them are so bad, I don't see how anybody smart could have read them and concluded they were worth reading.
That said, as far as the idea of reading that many books, I don't see how it's out of line. I read a lot, but not nearly as much as a professional researcher such as Norvig would. I have several hundred CS textbooks I've either read or skimmed pretty thoroughly, and most of my favorite professors had many more which they'd read cover to cover.
He's been reading for years; he may very well have read them all. A book I read (much of) in college, Reading for Power and Flexibility, does point out that you don't always have to read a book cover to cover to get a lot of knowledge out of it. Some books I like a lot I start recommending long before I finish reading them, and I don't know whether or not Norvig might have the same approach to reading some books.
Excellent point! @irahul, I see that you view a library as the storage place for the books that one has read. It should come as a surprise that most researchers/scholars view their private libraries not in this way, they view it as a source of information of mostly unread books.
You should definitely read Eco's essay titled 'How to Justify a Private Library' in his collection How to Travel with a Salmon. In it, he discusses his answer to precisely the same question that you have put forth. Here's how he fantasizes answering the question "Oh, what a large number of books you have! Have you read them all?"
"In the past I adopted a tone of contemptuous sarcasm. ’I haven’t read any of them; otherwise, why would I keep them here?’ But this is a dangerous answer because it invites the obvious follow-up: ’And where do you put them after you’ve read them?’ The best answer is the one always used by Roberto Leydi: ’And more, dear sir, many more,’ which freezes the adversary and plunges him into a state of awed admiration. But I find it merciless and angst-generating. Now I have falled back upon the riposte: ’No, these are the ones I have to read by the end of the month. I keep the others in my office,’ a reply that on the one hand suggests a sublime ergonomic strategy, and on the other hand leads the visitor to hasten the moment of his departure."
It's also worth pointing out that if you read 3 books a year, you would have read 90 books in a 30 year time span. I would suspect that Norvig reads a good deal more than that and that this selection of books for his book shelf does not represent everything he has read or owns.
The man IS the director of research at Google. And before that he was a very highly-respected faculty member at Cal.
Yes, he's probably read all these books, probably most of which after he received his PhD. He's probably taught classes where many of them were required reading. In his own writing, he's probably cited items in this library hundreds of times
Maybe he's read them all or maybe he's read some. A personal library doesn't mean a collection of books you've read or intend to read; they could simply be points of reference, whether it be the whole book, a chapter, a page or a paragraph.
Note that books appear more than once on that page, because they can be in multiple categories. That being said, it is entirely plausible that he has read them all in his career.
What really stood out for you about a mathematical text being typeset in (La)TeX? Were you also surprised by the popularity of No.2 pencils in elementary schools?
I had no idea Google even had a page for listing all of your books! Now if only there was a way to seamlessly import that data from Amazon that didn't involve hand coding a screen scraper script...
