Older editions of which books were better than the new ones? (2010)
mathoverflow.net41 points by susam a month ago
41 points by susam a month ago
In one way, many newer editions of textbooks are worse than older ones: the printing. That's courtesy of modern print-on-demand systems, where instead of sharp offset printing you get glorified inkjet output that's noticably blurry. And instead of a bulletproof Smyth-sewn binding, your "hardcover" book gets you a short-lived perfect-bound paperback glued into a hard cover.
When it comes to buying textbooks, unless a newer printing has corrections that I care about, I'll purchance an earlier printing to get the better book.
I wish there were a service that would unbind cheap books and sew them into a new binding of something bulletproof.
The problem is that the cheap books have already had their signatures cut into individual pages so that they can be perfect bound, basically running a strip of hot melt glue down the binding edge. At this point, you cannot rebind the pages in a way that will compare to the strength and durability of the original signatures.
They exist. Book rebinders are a thing. I recall my school's library had a fair number of rebound books. There's probably one around you.
I find the vast majority of textbooks I have ever used tend to be overly verbose and graphically confused, with lots of inserts that distract from the reading process without adding any relevant information. If you read a textbook from fifty years ago, it notably reads like a book, with a coherent narrative structure. Contemporary textbooks read more like magazines; eliminating text boxes, colors, and columns would make them far easier to read.
There’s also a tendency in contemporary textbooks to try to relate topics to concrete things in the real world. This is a key thing to do when you’re teaching to show students both how they can do something and why they would want to do it, but someone who is reading the book has already been convinced of the subject’s importance - they just need to be told what they need to know about it. This also leads to the book becoming dated faster because the examples draw on trends that go out of fashion, even though the principles of the field likely haven’t substantially changed since the 1960s.
> I find the vast majority of textbooks I have ever used tend to be overly verbose and graphically confused, with lots of inserts that distract from the reading process without adding any relevant information.
Very true. Much of the "extra helpful content" is simply distracting, adds nothing much of value and is just padding. Contrast with old Soviet-era textbooks i point-to here - https://news.ycombinator.com/item?id=43452118
>There’s also a tendency in contemporary textbooks to try to relate topics to concrete things in the real world.
This seems to be a personality thing. I was doing a Linear Algebra course last year and complained at the lack of applications in our textbook.
It seemed so strange to me. Here's this wonderful subject with literally dozens of wonderful, practical, illuminating real-world applications, which make the subject more concrete and easier to understand (and help you understand what's the point of all these obscure manipulations). And the book just... doesn't mention a single one of them? It felt both dishonest and pedagogically unskillful.
I expressed these feelings to my classmates and was met with disgust: "We are pure mathematicians. We do not concern ourselves with applications!"
Lucky for me, their disgust led them to discard several wonderful "Linear Algebra With Applications" books to the "free book pile", which I made excellent use of and greatly enjoyed.
That being said, in most of these books the "applications" are confined to a bonus section at the end of each chapter, and do not interfere with the main text.
I might have explained that poorly. What I’m talking about is not so much demonstrating applications by including practical problems in the problem sets (show, don’t tell), but instead opening every chapter with a paragraph or two about those applications. It often starts with an anecdote and then goes on to make a convoluted case for why the topic matters in the real world. These anecdotes are supposed to intrigue the student into wanting to learn more, but more often than not they just end up being skipped over. If the student had to be motivated to learn about the topic, he wouldn’t be reading the textbook in the first place.
I can usually distill everything I truly need to know in a (densely-printed) 30 page chapter down to less than 3 pages of notes, but that information tends to be spread out across those 30 pages, making reading the book much less efficient than it ought to be. These texts will often have a three-page-long introduction explaining how to read the textbook, and in my opinion most of them are trying to fill a role that can only be meaningfully satisfied by an experienced instructor.
An instructor can assess his students and relate lessons to what they already know. A textbook cannot do this, so it ends up offering several explanations, metaphors, and anecdotes that many students might not even understand. This leads a lot of students to skip over the readings in favor of reading their lecture notes.
1. I’ve read that the first edition of Philip Wadler’s and Richard Bird’s Introduction to Functional Programming is better than the second edition, but having read neither, I can’t give an opinion one way or the other.
2. I have deep respect for older editions of Apple’s Human Interface Guidelines, particularly those from the 1980s and 1990s. Part of it is my bias toward the classic Mac UI (though I love early Mac OS X through Snow Leopard), but another part is due to how well written these older guides are. The older Apple Human Interface Guidelines contain a lot of useful information that all developers should read; much of the advice given is timeless even after 30-40+ years.
https://github.com/gingerbeardman/apple-human-interface-guid...
Courtesy of the great Jef Raskin. Much of his work would still be highly relevant today if anyone actually cared about humane interfaces instead of using it to extract and refine dark patterns.
Thanks for sharing that link!
Through it I was able to find that Apple had design guidelines going back to the Apple 2 era! There's even a fascinating anecdote about a surprisingly challenging UI problem for the Apple //e computer: How can we get users to accurately tell us whether the computer is using a color or B&W monitor? I submitted it to HN here:
Computer books change over time, for obvious reasons. The O'Reilly animal books are always good, but they've removed some useful information in later editions.
For instance, the [sendmail bat book](https://www.oreilly.com/library/view/sendmail-4th-edition/97...) originally talked about all sorts of ways to deal with resource limits, particularly memory, because this was often an issue with servers in the early '90s. Later, memory became cheap and huge, so discussions about this were removed from later editions.
Even when memory ceased being a limitation, though, learning about how to monitor and what to do about heavy resource usage remained important, so the removal was an overall loss. I wish that O'Reilly would make available for free the things they removed from later editions.
I wish I'd've had the money and space to buy the original editions of many of the animal books. I figured they'd live at the bookstore until I did, but then bad things happened, and now bookstores are rare. Perhaps one day I'll find some.
Somewhere in the 80's most basic science textbooks went from clear and concise to overly verbose and ladled to the hilt with colors and distracting "illustration" clutter making most later editions not just worse, but even an assault on learning and understanding.
I was at library some time ago and skimmed upon some old textbooks, written still during the Communist rule here - and they just were so much better in terms of explaining things. (as long as they weren't history stuff, naturally.)
Soviet era Science Books for everybody's enjoyment - https://mirtitles.org/ Concise, Information-dense and to-the-point unlike the majority of books being published today.
Also checkout all "Dover Publications" catalog on Science/Mathematics/Engineering. Most are old classics and very good - https://store.doverpublications.com/
On the first link, I had to scroll for a while to find a single book in English, and then a while for the next one. But they are indeed there.
Perhaps adding tags or sections by language would be helpful.
What are you talking about?
Most books on "Mirtitles" are in English (very good scans and some of them redone via LaTeX) and only a few are in other (mostly Indian) languages.
Just search by name of book (eg. Calculus, Algebra etc.) or name of author (eg. Piskunov, Zeldovich etc.) or book series name (eg. Little Mathematics Library, Science for Everyone etc.) You can also click on the categories listed in blue below the "Follow" button on the right-hand side.
The website is a blog and hence the books are listed in a series of entries. It is a real labour of love and i am grateful to the author(s) that such excellent scientific books are not lost in the mists of time but are available for free to the common public. Every Teacher/Student/Interested Layman should read these books.
Finally, If you (or somebody else) wants to make the site better contact the site author and offer help; i am sure they would appreciate it.
Data and Reality, by William Kent. A timeless classic on the computational representation of, well, anything and everything.
The third and currently only published edition, published posthumously, insists on adding a bunch of mid-2000s era design conceits based on OO paradigms. An absolute travesty.
The author's webpage on the book with excerpts for more info. - https://www.bkent.net/Doc/darxrp.htm
More of the author's works - https://www.bkent.net/catalogsource.htm
Kent’s original book is a masterpiece. He pretty much defined data modelling nearly fifty years ago. It’s depressing how little of real interest has been added since.
For the reasons you rightly pointed out, the current edition is a disaster.
I tried to get through the book, but I was bored out of my mind. Chapter after chapter, it wasn't getting to any coherent point.
I seem to remember that I really long initial chunk of the book consisted of presenting a barrage of self-evident problems in the modeling of real world entities in databases, without offering any solutions.
There's no shortage of people that will fill your ears with problems, so tuning that kind of talk out after a while could actually be an evolutionary self-defense mechanism.
I found out the older edition of the Dragon book was way better than newer editions, which have many sections chopped off.
This is especially true if you consider all the "variations" - the second updated edition (global edition) which should rightfully be a third edition as it has two additional chapters and the two volume precursor texts: The Theory of Parsing, Translation, and Compiling Vol 1 & 2.
The first one was also perfectly sized and not making it a bulkier text like the current CS curriculum texts (CLRS, OS:Concepts, Korth's DBMS etc.)
For CS-related books, I think the original 1979 Hopcroft and Ullman is better than its successor editions.
+1.
To elaborate on this, the original Introduction to Automata Theory, Languages, and Computation, by John Hoffcroft and Jefferey Ullman (The one with the black cover is the 1979 edition). My instructor, even in 2024, made us use that one instead of the newer one.
The other text if i recall right is Tanenbaum's Computer Networks book, where there is a significant portion of bloat that does not really fit there although my source for this is a review for the book.
K&R 2nd Edition was rushed but I personally wouldn't claim the older edition was 'better'.
It would be difficult for the older edition of K&R to be better because it's about a worse dialect of C, which doesn't offer type checking between function declarations and definitions.
May I ask why?
... I just bought the second edition few days ago for my course.
I am hazy on the details but my instructor briefly told us that the 1979 edition had better problems and had topics for the advanced undergraduate. The major thing for us as students was that the 1979 was much shorter to use and concise to the point of being a good handbook.
(We do Theory of Computation/Formal Language and Automata Theory as a dual sem course meant to be both a junior and a senior class and thus the 1st edition suits the course better.)
Googling gives https://cs.stackexchange.com/questions/149072/hopcroft-ullma... (see the question for the links)
IME with technical computer software books, the good ones worth reading and dedicating bookshelf space to anyways: The first editions tend to describe a simpler form of the subject, and contain the most distilled and succinct description covering the minimum required salient points for publication.
I'm pretty sure it's without exception. I have no multi-edition technical books that shrink in size in subsequent editions. So I end up with collections of editions for books of this type I really value, preferring to reach for the earliest editions first, and only after I can't find what I'm looking for, do I proceed to the subsequent editions.
Old programming/operating system design books are such a pleasure to read. Simpler times make for simpler, easier to understand documentation. It sucks to come at a foreign, complex, mature system in reverse chronology.
I have a copy of the CRC Standard Mathematical Tables, 27th Edition, CRC Press which I purchased as a suppliment to the text in a university calculus class I was taking. I should have picked a different year, they had both newer and older, but this one had a discounted price and was shrink-wrapped. It was also missing pages 153 - 184, which I soon discovered when I tried to look up something in the book that was listed in the table of contents, but was among the missing pages.
Lonely Planet. In 2013 we've done a Morocco backpacking trip with just a Lonely Planet guide - no phones, no Internet access, no pre-booked accomodation or transport. The guide contained maps, hostel suggestions, POI, transport options and a lot more. This year we brought the newest LP release to Thailand, and it was mostly useless - no transport or accommodation info, bad maps, generic POI descriptions. We are not buying another LP guide.
I’ve never been convinced by the value proposition of print travel guides.
Would be interested to hear if anyone derives value out of them (beyond what you can find on Google and TripAdvisor for free).
They can be quite good during the trip planning phase, like a helpful friend who makes suggestions that you are totally free to ignore. Having multiple threads of thought is easier with books, where you can riffle around the pages, maintain multiple contexts simultaneously, etc., compared to the very linear, one search at a time model that google search gives you.
The other small advantage is that they work without power and cell service.
Richard Dawkins claims later editions of Darwin’s Origin of Species got worse because of trying to appease critics.
An older edition of Think Java has an example that still amuses me greatly to this day. Appendix B.3, Page 222 (240 in pdf):
In the realm of fiction, they released edited versions of Roald Dahl books that were more politically correct and designed to be less frightening to the fragile children of today.
Straight and Crooked Thinking. New editions aren't by the original author.
Pat Billingsley's Probability and Measure 3e. was reissued in an Anniversary Edition, which is just the same content but with several typographical errors.
Calculus Made Easy is a "thoroughly bad and vicious book."
(I especially recommend the epilogue.)
Essentials of Programming Languages 1st ed. is special
K&R
[dead]
[dead]