Hacker News new | past | comments | ask | show | jobs | submit login
The Tears of Donald Knuth (acm.org)
180 points by mrry on Dec 25, 2014 | hide | past | web | favorite | 55 comments

I find that most history books on technology appear to focus on everything but the technology. For example, books on the Wright bros. show very little understanding of the technical achievement of the Wrights. They enumerate places, dates, and trivia of their personal lives. History books on locomotives contain arcana on the numbers painted on the side, production figures, schedules, and very very little on the engineering advances. Same with airplanes. Etc.

Exceptions are rare enough to get called out for praise. A few:

Digital Apollo, by David Mindell (development of Apollo guidance and avionics generally)

The Dream Machine by Mitchell Waldrop (technical development leading up to the Arpanet, thence to the Internet)

Colossus, the secrets of Bletchley Park's codebreaking computers, by Jack Copeland et al.

Unfortunately, there's a lot more of the technically unsophisticated junk that Knuth is complaining about, and the excuses offered here rather badly miss the point. ("The history of software is not the history of computer science". Yes, and so? How does a history of software from which all technical detail has been drained away contribute anything of value?)

The original article also calls out such praises in the paragraph starting with "The truth is that regrettably little history of computer science, whether dumb or deep, has been written by trained historians".

Let me reiterate one recommendation in particular, Mechanizing Proof, by Donald MacKenzie. This is a superb introduction to the technical subject of formal methods as well as a superb history! My only regret is that it was published in 2001 and a lot happened in the last decade.


Are there any other works like this? ... I just bought it, and I want to learn what has happened since 2001.

"Ignition" by Clark is another exceptional exception.

Ignition is fantastic and I'm continuously surprised how unknown it is amongst the engine folk at MSFC.

Wow, it's going for $3,115.00 (+ $3.99 shipping & handling) on Amazon. Put this next to Sled Driver on my list of books to buy when I win the lottery.

Would you happen to have a copy of a similarly priced-out book, "How to Avoid Huge Ships"?


Google is your friend and Bob's your uncle.

Seems the bots are getting damn creative. Perused the first few pdf links and they were all "spam" pages in PDF form, with embedded download links i don't dear follow.

Isn't algorithmic pricing fun?

Worldcat indicates that the book is available from various libraries: https://www.worldcat.org/title/ignition-an-informal-history-...

I can't help but link to a classic fight between two algorithmic book bots: http://www.michaeleisen.org/blog/?p=358

While fun to watch on Amazon, this is a cruder version of what HFT does on the stock market.

Sure, but Amazon pricing is more amusing: the results are visibly and clearly absurd. HFT done right is much subtler.

Also http://www.amazon.com/The-Making-Atomic-Bomb-Anniversary/dp/...

The science was so good that I got a bad grade on my History class paper, because I focused too much on the science and not so much on the "storytelling" of history. Which apparently puts me in good company slongside Knuth :-)

I have noticed the same thing in a different context. When I was still in college, many art classes were essentially about historic trivia, paying absolutely no attention to the intent of the artists, the ideas they tried to express, and what they said about their own work. It was 99% about context and maybe 1% about content. The art was treated as a kind of societal byproduct.

Now that you have mentioned it, inventors and technology are treated in a similar fashion. Not sure why. Maybe because historic biographies adopt their methodology from general history where kings and generals are simply dots in a larger picture that does try to describe the society of the time?

Another reason may be that humans are in the end social animals...

The problem begins to happen a lot in school where history is taught the way it is. At least back home in India the way history books read, you needed to know only two things:

1. A timeline where you ought to remember what happened before and after that thing.

2. Objectify events based on causal reasoning presented in the books itself. For example "America got dragged into WWII because of an unprovoked attack by Japan that caused a massive loss of life and military resources"

And the share volume of world history covered in a year or so makes it so hard to do any justice to a topic.

Mostly related to #2, that's what most annoyed me about school-taught history. Sure, the Japanese, just for the hell of it, spend a bunch of money sending planes to Hawaii and that's how poor ol' innocent United States ended up in WWII. C'mon, why did the Japanese do it? It wasn't until years later that I learned anything about a blockade in the Pacific. I mean I can understand a little "America: fuck yeah!" attitude, but no one thought to mention that little tidbit?

The entire planet gets into a war in the early 1900s because some two-bit archduke of something-or-other got shot? WTF? Woah, back up there, Teach; you've got some explaining to do. (And when I read the explanation years later, I still go "WTF?")

Point is, if the cliche is correct then we're doomed to repeat history again and again because there seems to be little effort to explain why things happened. What the hell was the U. S. doing in some dot on the map in Southeast Asia in the 60s? Domino theory? C'mon, I doubt anyone believed that even at the time if a ten-year old version of myself is saying, "how dumb do you think I am?"

But, as you mention, it would seem that there's little enough time to teach the timeline, let alone explain the treaty situation of early 1900s Europe such that WWI makes sense.

I guess we had a pretty good history teacher. She drilled into our minds that any important historical event usually had an underlying cause, a corroborating cause and a triggering cause (and often more than one of each category). So for instance on WWI, we learnt not only about Archduke Ferdinand and the Serbian "Black Hand" movement, but also about the geopolitical frictions that already existed between Austria-Hungary and Russia over the Balkan situation, etc.

That's brilliant! How much more useful than arguing over the 'great man' theory etc. Just lay out the causes and understand how they contribute.

Btw I was at the WWI memorial in Kansas City (created just a few years after the war) and they list Darwin as a cause of the war! The leaders of the world had read his "Origin of the Species". It was variously interpreted politically to mean 'the strongest culture will win' and used by the German leadership to justify war. War was natural, it would winnow the nations down to just those worthy of survival, and Germany was the most worthy.

from wikipedia: "The attack on Pearl Harbor was intended to neutralize the U.S. Pacific Fleet, and hence protect Japan's advance into Malaya and the Dutch East Indies, where it sought access to natural resources such as oil and rubber."

The cause of war is not usually taught. It's the events that led up to war that is. http://en.wikipedia.org/wiki/Events_leading_to_the_attack_on...

Wow, you're right. I've never noticed as such, but in retrospect very little info on history that is designed to be mass consumed actually contains historical notes that include the technological advance itself.

Does anyone know of resources (books, websites, etc) on historical advances that does actually cover the technology?

Racing the beam. About the atari 2600. Even the title is a technical treatment of the subject.

I had a delightful time reading "A History of Engineering and Science in the Bell System".

For the Wright bros. I recommend "Inventing flight" by John D. Anderson, Jr.

By dumbing it down, it makes the book readable by a wider audience == more sales.

[1] seems to be the paper Knuth is discussing. [2] is the link for the IEEE, but you're not getting the full text from them. (Weirdly, the top hit for Cambell-Kelly himself is [3], which only has a link to track his research proposals. Warwick's keeping their eyes on the ball, I guess.)

[1] http://www.thecorememory.com/THTHS.pdf

[2] http://www.computer.org/csdl/mags/an/2007/04/man2007040040-a...

[3] http://www2.warwick.ac.uk/fac/sci/dcs/people/martin_campbell...

[1] does seem to be the paper of interest. It's not clear what Knuth is unhappy about. Campbell-Kelly's big complaint seems to be that the early history of computing mentions machines, languages, and algorithms too much, and doesn't discuss applications and users enough. That's a reasonable comment from a historian.

When Knuth wrote Vol. I, "Fundamental Algorithms", there were almost no computer books published other than manufacturer manuals and a few introductory programming language books. Knuth's series is really a history of algorithms, with each one traced back to its originator. The early history is on the record.

We're now inundated with corporate histories, mostly of applications companies. There's a glut of CEO biographies. There are academic papers on the history of the spreadsheet. There's no longer a lack of history on the user-facing side.

This comes with progress in the field. A book-length history of the electric motor, from Henry to Edison to Sprague to Tesla and up to modern brushless motors, would be interesting - to a very small number of people. Popular articles on motors can be found in mainstream publications from 1880 to 1925 or so. After that, they were just routine items, not newsworthy. Low-level algorithms now belong in that category. A few people still have to study how to do hash tables, just as a few people have to study rotating electrical machinery theory and design new motors.

The important thing is not to lose the history. Campbell-Kelly says "we can't save everything". Well, we can. Disk space is cheap. Future generations (of humans and machines) can re-summarize it.

What I've seen is that history is saved, and even accessible, but mythology is often stronger than history. Here on HN I've seen people say that Apple was the origin of the garage startup, when that goes back to at least HP, or the Apple was the inventor of profit sharing, when that was a Valley practice predating even Intel. Elsewhere I see people say that PARC invented the mouse, when that came out of SRI, or that Xerox flubbed on the laser printer, when laser xerography was making them over a billion dollars a year in the late 1980s. I see people sell Agile using waterfall as the strawman opponent, as if those were the only two possibilities. (See http://www.infoq.com/articles/bossavit-agile-ten-years-on for one commentary.)

I personally think there's a rebranding of Linked Data as a way to rescue the real-world failures of the Semantic Web dream by saying that all interlinked databases are part of Linked Data and ergo the Semantic Web, and my limited research of the historical record backs that up.

But this knowledge doesn't really affect how most people do business now. (Though we end up with highly paid consultants who peddle a false mythos rather than paying historians for the real information.) It's mostly relevant for those points that Knuth enumerated, which are superb humanistic goals that few beyond the academically secure or retired can afford to advance.

While much can be saved, many things are only in people's heads, or in their personal effects. Doing that history now, rather than waiting, means that more of those can be recorded for the future.

And there's other examples of that, less obvious but more pervasive. For example, every single 'first' is mired in definitional issues which all serve to recast previous things in the light of more modern technology. It does a disservice to those older things to only view them through the lens of what we have now.

It's hard to fight because it is rarely based on outright lies, per se, just pervasive misinterpretations which cast things in a false light. Disputing it makes you look like a nitpicking fool picking at things which barely even exist, but accepting it makes history seem like a teleological process, inevitably aimed at today like an arrow aimed at a target, with all previous processes culminating in what we have now.

Thanks for the link. The section on emulation and simulation struck me as patronizing and narrow.

Exactly. To me that section was a total demonstration of narrow minded personal opinion.

"So why is the history of computer science not being written \(in the volume it deserves\), or \(the manner favored by Knuth\)?"

\\1 Too many principals of said history are still alive, still influential, and still kicking too much to allow much in the way of historical analysis. It is much easier for a historian to stick with a subject where those who remember events are safely dead.

\\2 It would be improper to expect anyone to write informatively about ideas they cannot themselves understand. And it would be improper (as the author of the article writes) to expect a historian to understand an unrelated technical field.

This is unfortunate, because "history", the preferences of historians notwithstanding, is the story of technology. I do not believe there is any great difference between any of us and some goober fingerpainting on the walls of a cave.

Have to take issue with your analysis here. For an appropriately empty conception of technology, we can perhaps view history as "the story of technology". Whether this has much explanatory power, however, is another question. Really, though, I suspect this is just a popular contemporary bias mostly among technologists, just as in the Christian world of the middle ages history was biblical history.

Goober had to have been making good, even crucial, use of his mental abilities or we would not have ours.

So, what the heck was Goober doing with his abilities that were crucial? He found the cave, kept out animals that could hurt him, built a fire to keep it warm, built families, found food and water, make tools, weapons, and clothes, etc. Apparently found good uses for his brain.

And found the leisure time, materials and learned the skills required required to do some fingerpainting on the side (assuming the subject of the painting did not have practical or religious purposes).

Every history sees through a limited historiographical lens and centers certain questions. If we set out to answer questions like "why did Britain industrialize first?" then a deterministic model of technological progress(the "Sid Meier's Civilization" view of history) does not suffice for our history, and we end up with complex stories that weave together geography, politics, and social factors.

Certainly true, in fact it's kind of my point. Take a look at the "key sentence" from Harold Perkins' review of From Airline Reservations to Sonic the Hedgehog:

"The computer and its software nervous system brought a revolution in human development as significant as the steam engine, the automobile or the aeroplane, and even more effective in shrinking the planet"

What would you get from a history of recent revolutions in human development from someone who doesn't know why steam-powered airplanes never really caught on? It doesn't matter what kind of geo-political story they can put together; the idea just doesn't work.

A Margaret Gowing told C-K about his history of EDSAC, "What you have written is clearly very good. I know practically nothing about computers, but I can tell that what you have written is good history, so far as it goes. However, let me urge you to look beyond programming technology to consider the kinds of people who were using computers and the problems that they were solving." Which is excellent advice, certainly, but part of what he comes up with years later, in his embarrassment over not following the advice the first time, is, "Jim Wilkinson investigated errors and stability in digital numerical methods and developed advanced matrix programs. These were to prove vital in understanding and preventing “flutter” for the British aircraft industry, which was still reeling from the de Havilland Comet air disaster of 1954," which makes me suspect he knows about as much about numerical methods as I do.

It looks like a kind of inverted physics envy, a cultural cringe reaction.

For scans of much of the historic software source code that Knuth mentions near the end of the video, see:


What would it take to get Computer Science departments to view the history of the discipline as an important issue to be dealt with inside the department? Surely the author's right that if the technical history of a discipline is to be done in the way Knuth wants, and I think that is incredibly valuable, it needs to be done by those who understand the discipline.

But can C.S. departments manage that as an institutional matter?

Here is the web/non-mobile version (OP posted mobile version):



Non-mobile version has readable font size and color.


Thank you!

That link just redirected to the mobile version on my Galaxy Note 3, but you reminded me to try the "Request desktop site" option which worked fine.

It's funny that the mobile site is completely unreadable on a mobile device, where the desktop version formats cleanly on mobile and is much easier to read!

Holy crap, thank you. The article linked by OP was hurting my eyes. Who creates a website where the body has no margin? And light-gray text on a white background? What were they thinking?

Lack of margin actually helps readability on older mobile devices. Not sure why they'd go with those colors, though.

I would recommend my tool which have a readable view.


This redirects to the mobile version for me :|

I use Firefox with Noscript. To see the desktop version I had to 'temporarily allow all this page' in the Noscript options.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact