The Great 202 Jailbreak
November 3, 2015 1:19 PM   Subscribe

The Great 202 Jailbreak. 1979: Bell Labs bought a typesetting machine that could do some pretty fancy stuff. There was only one problem: you could only do the fancy stuff that the manufacturer, Mergenthaler, decided you were allowed to do. In one of the first cases of what today would be known as jailbreaking a device, researchers Joe Condon, Brian Kernighan and Ken Thompson reverse-engineered the machine and wrote up their efforts and results in a technical memo [pdf] which was quickly suppressed due to legal concerns.

To understand why this work was considered so sensitive, it's worth recalling how specialized the domain of typesetting was at this point in history.

What used to be a process of melting and shuffling hot metal around (Linotype and You! After School Special) had developed into an even more complex series of mechanical operations, now hidden from view beneath a nice branded box.

How these programmed font glyphs took physical form on paper became closely-guarded trade secret. Mergenthaler was none too pleased to have these mechanisms revealed.

Though only poor-quality photocopies exist of the original memo, digital forensics efforts by Steven Bagley, David Brailsford and Brian Kernighan have faithfully recreated it in a particularly meta exercise resulting in the recreation we can enjoy linked above as a PDF.
posted by odinsdream (57 comments total) 91 users marked this as a favorite
 
That technical memo was an enjoyable read! Are words I never thought I'd type.
posted by sevenyearlurk at 1:39 PM on November 3, 2015 [8 favorites]


Fascinating stuff. I love the idea of reverse engineering. As saurik has said, "Jailbreaking is effectively forging someone a key to a lock they own: I fundamentally disagree with the idea that we should be attempting to protect users from themselves, as that implies that we think we are smarter than they are, and leads to a form of corporate-sponsored [authoritarianism]."
posted by exogenous at 2:00 PM on November 3, 2015 [12 favorites]


For want of chess font, the trade secret was lost...
posted by ChurchHatesTucker at 2:03 PM on November 3, 2015


That technical memo gives me flashbacks to the ecommerce software we bought a few years ago. We simply couldn't understand how it shipped, and in the course of events ended up rewriting many of its modules.

I wonder if they ever heard back from the regional manager.
posted by Nonsteroidal Anti-Inflammatory Drug at 2:03 PM on November 3, 2015


Etaoin shrdlu
posted by unliteral at 2:05 PM on November 3, 2015 [10 favorites]


"This section has no technical content, but may provide some entertainment."
posted by bruceo at 2:08 PM on November 3, 2015 [2 favorites]


Fascinating.

Various Mergenthaler documents list [the vertical resolution of the imaging CRT] as 288 per inch, 576 per inch, and even 1152 per inch. Which is correct? Surprisingly, none of the above. The true value is 486 per inch, a number that occurs in no document that we have seen. 486 is convenient, how- ever, since it divides the horizontal resolution, 972.

Sounds like their particular CRT operated within some of the specs of standard NTSC or 525/60 video monitors, then. I wonder if Mergenthaler's clients in PAL territories had different hardware and thus different software.
posted by infinitewindow at 2:24 PM on November 3, 2015


Hah, of COURSE Donald Knuth is in the references.
posted by supercres at 2:45 PM on November 3, 2015 [2 favorites]


That technical memo gives me flashbacks to the ecommerce software we bought a few years ago.

Yeah, reading this:

The only documentation on the programming of the machine was a manual that the installer told us "applied only to England"

Ed P, a Mergenthaler applications person, determined (as we had already) that our problems were indeed a broken machine

In the meantime, repeated phone calls still had not produced any action on the missing parts of our order.

provides the valuable lesson that technology may change but shitty customer service is forever.
posted by MCMikeNamara at 2:45 PM on November 3, 2015 [6 favorites]


This short doc on the last day of hot type at the NY Times is pretty great. They go through the whole process of the hot type system and then a bit about how the new digital system works:

Farewell - ETAOIN SHRDLU - 1978
posted by wemayfreeze at 2:48 PM on November 3, 2015 [13 favorites]


Love the section of their notes they reproduced.
posted by benito.strauss at 2:53 PM on November 3, 2015


We wanted to do this mainly because it was too hard to defeat their software—it believed very strongly that it was a powerful general-purpose formatter, and heroic measures were needed to disabuse it of this notion. Furthermore, it became clear that it was riddled with bugs, and that we simply could not go into production with it.
posted by a lungful of dragon at 3:01 PM on November 3, 2015 [4 favorites]


That whole computerphile channel is a TVTropes-level timesuck.
posted by ChurchHatesTucker at 3:08 PM on November 3, 2015 [7 favorites]


As you can see from this tale of woe, our particular 202 is not "packed with reli-
ability and convenience features", as your brochures suggest.


Good stuff.
posted by smammy at 3:23 PM on November 3, 2015 [1 favorite]


This is fascinating, odinsdream. Thank you so much for posting.
posted by indubitable at 3:24 PM on November 3, 2015


One of the report's authors, Brian Kernighan, wrote the first C programming language tutorial along with that language's creator Dennis Ritchie.
posted by StickyCarpet at 3:31 PM on November 3, 2015 [4 favorites]


Lyle R, our (by now regular) repairman, suggested wiggling the chips [sic] on the code converter board. This apparently nonsensical suggestion in fact cured the problem for a while.
posted by double block and bleed at 3:55 PM on November 3, 2015 [2 favorites]


If Mergenthaler really wanted to maintain control over the secrets of their hardware, selling any of it to Bell Laboratories was maybe not a smart move.
posted by uosuaq at 4:02 PM on November 3, 2015 [19 favorites]


As a designer, this really brings back memories of that era of typesetting. By 1980, phototypesetting had become compact and affordable enough that just about every shop and free newspaper had their own machine. And, there was already talk on the street about "computers" doing the work in the near future. Postscript will be introduced in just two years after this Bell Labs memo, and become available to the public in 1984.
posted by Thorzdad at 4:07 PM on November 3, 2015 [3 favorites]


One of the report's authors, Brian Kernighan, wrote the first C programming language tutorial along with that language's creator Dennis Ritchie.

Ken Thompson's also known for his work on a little side project.
posted by MikeKD at 4:18 PM on November 3, 2015 [4 favorites]


The machine is highly susceptible to static electricity; the tiniest shock pushes it into catatonia

Ha! This reminds me of an early automated reel to reel tape system we had at a radio station I DJed at in the late 70s. It had a control panel that let you program in the sequence of songs and ad carts to play. Whenever there was a nearby thunderstorm, it would go nuts and several tape decks would play simultaneously along with random advertisements.
posted by CheeseDigestsAll at 4:27 PM on November 3, 2015 [1 favorite]


That linotype video was the most fascinating thing I've seen in YEARS.

I kept having jaw drop reactions and saying out loud "no fucking way!".

And then there was even the part about lead, antimony, and abesotos!

The surprise ending was amazing.

None of this sarcasm. I wanted it to end so I could say how amazing it was and also wanted it to go for on forever.

yes I do like How It's Made. Why do you ask?

I also love books and type and have no idea how I never knew about this before.

Super cool.
posted by sio42 at 4:38 PM on November 3, 2015 [7 favorites]


If my own experience with photosetters is anything to go by, the static electricity problem and the light leak problem are probably closely related.

...and yeah, those damn film canisters.
posted by ardgedee at 4:43 PM on November 3, 2015 [2 favorites]


The good old days.
posted by davidpriest.ca at 4:48 PM on November 3, 2015


Have you tried turning it off, wiggling the chips, and then turning it on again?
posted by ethansr at 5:28 PM on November 3, 2015 [1 favorite]


Metafilter: so far no non-frivolous applications have arisen
posted by ckape at 5:48 PM on November 3, 2015 [1 favorite]


The "wiggle the chips" thing is similar to the Apple /// "drop it from 2 inches to reseat the chips" advice. There's a similar anecdote in Soul of a New Machine in which a tenacious hardware bug is finally fixed by a similar drop of the board.
posted by We had a deal, Kyle at 6:00 PM on November 3, 2015 [2 favorites]


Having been an Apple /// dropper in the past, it wasn't the chips you were resetting. It was a daughter card of RAM that would warp from the heat and pop out if its connectors. I remember the first time the machine wouldn't boot and someone showed me this technique. I was a little shocked that we were shipping this machine...
posted by njohnson23 at 6:11 PM on November 3, 2015 [3 favorites]


sio42: The Baltimore Museum of Industry gave (gives?) live demos of a Linotype machine. (Baltimore was Mergenthaler's adoptive home, and the company was originally an American one.)

There's also a feature length documentary that came about five years ago or so.
posted by ChurchHatesTucker at 6:13 PM on November 3, 2015 [4 favorites]


Meatfilter: has no technical content, but may provide some entertainment.
posted by ArgentCorvid at 6:28 PM on November 3, 2015 [2 favorites]


To understand why this work was considered so sensitive, it's worth recalling how specialized the domain of typesetting was at this point in history.

I'm still waiting for someone to write up the fascinating history of the impact of the IBM Selectric (introduced 1961) and the Selectric Composer (1966) on independent/DIY publishing, for this reason. Almost overnight, it radically deskilled and reduced the cost of producing justified, camera-ready text, and the counterculture, for example, was very quick to embrace it. Stewart Brand has said that, along with the Polaroid MP-3 , they were responsible for the birth of desktop publishing.
posted by ryanshepard at 6:51 PM on November 3, 2015 [1 favorite]


Here's a modern video of a working Linotype that I couldn't figure out how to stick in the FPP.
posted by odinsdream at 6:53 PM on November 3, 2015 [3 favorites]


There's something soothing about the sound of a really complex machine going about its business.
posted by Thorzdad at 7:24 PM on November 3, 2015 [2 favorites]


double block and bleed: " suggested wiggling the chips [sic] on the code converter board. This apparently nonsensical suggestion in fact cured the problem for a while."

This is an extremely common problem and remedy for many electronic devices with socketed chips. It's one of the trade offs for socketed vs. direct soldered assemblies.
posted by Mitheral at 7:25 PM on November 3, 2015 [2 favorites]


That linotype video was the most fascinating thing I've seen in YEARS.

Yes, the linotype machine is steampunk as fuck! The only machine in the world to feature both a blast furnace and an alphanumeric keyboard. As a special bonus, it occasionally spits blobs of molten lead at the operator.

That thing took balls to use.
posted by Naberius at 8:08 PM on November 3, 2015 [13 favorites]


I am having a little tiny technical writing/document design professor stroke over here. This is the people from one movie I show in class hacking into the device from a totally different movie I show in class, and then writing an awesome document about it that I now need to find an excuse to use in class.

I didn't know about the feature-length Linotype documentary until one of my students last year, who was a former print shop operator coming back to school for an engineering degree, loaned me the DVD. It's great.
posted by Tesseractive at 8:18 PM on November 3, 2015 [8 favorites]


So, did all the newspaper guys have lead poisoning? I mean not only were their no gloves; some of the guys were smoking while handling the lead and breathing in the air of open vats of molten lead.
posted by Mitheral at 8:27 PM on November 3, 2015


This was just amazing. I always wanted to know how anybody typeset stuff before laser printers without hot type.

It makes me apreciate LaTeX all the more. Which, BTW, if you are ever writing a thesis or any academic article, you should be using.

I was someone with an arts degree, that ended up in a career in IT, servicing somewhat fittingly the Arts Faculty of a large university. The horrifying travails most grad students had trying to complete their theses in Word prompted me to start LaTeX classes for new and existing grads. Our need to deal with issues dwindled overnight, plus everyone was turning out beautiful looking documents!

If you haven't given it a try, you owe it to yourself to check it out. It's remarkable.
posted by Thoth at 8:32 PM on November 3, 2015 [2 favorites]


\begin{holywar}
LaTeX is great but what is the correct editor in which to write documents for it?
posted by ROU_Xenophobe at 8:53 PM on November 3, 2015 [4 favorites]


\begin{holywar}
LaTeX is great but what is the correct editor in which to write documents for it?


Missing \end. Badness 9000
posted by stevis23 at 9:14 PM on November 3, 2015 [4 favorites]


thatsthejoke.jpg
posted by ROU_Xenophobe at 9:24 PM on November 3, 2015 [2 favorites]


Man, while reading the technical memo PDF I kept wondering to myself how they managed to recreate it as a nice perfect crisp PDF (Did they do it by hand? Did they use actual old data?). And boom, at the end of the FPP, a paper on exactly how they did it. Fantastic post.
posted by zsazsa at 10:35 PM on November 3, 2015


I can't even imagine being the poor beleaguered customer service and tech support reps on the other side of this transaction, unaccustomed to dealing with customers with technical chops who demand their machines consistently and correctly do the work they were purchased to do. I picture them making copious mental notes along the lines of

never sell anything to these guys again unless we can waive the warranty
ok no just never sell anything to these guys again warranty or not
wait no never sell anything to anyone at bell labs again
oh dear god never sell anything to computer scientists again
holy crap maybe never sell anything to academics or professional geeks of any stripe
never sell anything ever ever again why are we even in this business

Either that, or taking extreme delight in watching these hyper-critical, insanely competent nerds sticking it to those arrogant jerks in research and development. Hard to say.
posted by town of cats at 10:42 PM on November 3, 2015 [4 favorites]


14 minutes in "Brian had saved the troff source code"...
posted by mikelieman at 11:39 PM on November 3, 2015 [1 favorite]


People saying "oh, and Kernighan/Ritchie/Thompson also made THIS" is kind of funny to me, because these are famous names in computing, you cannot extricate them from the history of the field. For instance, Ritchie didn't just write the first C "tutorial," but he invented C itself, and the tutorial in question is the book The C Programming Language, that is, the definitive bible of C, the whole reason its original dialect is sometimes called Kernighan and Ritchie, and the book itself is sometimes just called "K&R."

Ritchie and Thompson also implemented the original version of Unix, in assembly.
posted by JHarris at 3:04 AM on November 4, 2015 [3 favorites]


There's an interesting offhand reference to "resurrecting the B interpreter" for their work on the Linotron 202; B was the short-lived predecessor language to C.
posted by orthicon halo at 6:10 AM on November 4, 2015 [2 favorites]


The link says there was not a language called A, which is a shame because it dodges the chance for a Glengarry Glen Ross mashup:
A - Always
B - Be
C - Compiling
posted by exogenous at 6:16 AM on November 4, 2015 [1 favorite]


People talk about Vint Cerf et al as 'The Fathers Of The Internet', which is more than fair enough, but it wouldn't have happened without Ritchie, Thompson and Kernighan. C, Unix and the knowledge to use them, provided the tools to move the TCP/IP stack across different hardware platforms much more easily than any alternative, and then to share utilities and tools higher up the stack. Not that there was a practical alternative; everyone else doing network design and implementation was locked down to a higher power in some way with control over how intellectual property would spread and a different agenda than 'connect all the things'.

What I find particularly thought-provoking is that this all came from AT&T, a highly-regulated monopoly and one heavily invested in maintaining a stranglehold over 'connect all the things'. But AT&T couldn't really go into computing with its corporate might, because of that self-same regulation, so Bell Labs had not just the ability but the duty to share its work with others. Which is why the transistor had spread so quickly with minimal IP restraints twenty years before. (Later network wars between AT&T and IBM - really, telco v computer - were characterised by neither side wanting to create an effective competitor out of the other in their space, nor to step on a regulatory landmine, which gave the genie lots of room to play around in after it had escaped its DARPA bottle.)

There's an extraordinary amount of important history to explore in the interplay of all these mid-late 20th century factors, in the creation of the landscape of the 21st century - see also the two major wireless systems, WiFi and GSM, created out of very lightly and very heavily regulated environments - and I don't think the definitive work has been done yet. I originally wrote 'the technological and commercial landscape' in that previous sentence, but really, it's everything.

The FPP story is a minor yet exemplary skirmish in that war, and leads in all manner of fascinating directions.

I know who my heroes are.
posted by Devonian at 6:28 AM on November 4, 2015 [12 favorites]


(Oh, and if you are going to write the definitive work, get going soon while the sages are still walking this earth.)
posted by Devonian at 6:36 AM on November 4, 2015 [4 favorites]


If you're in the L.A. area and want to see a linotype in action, keep your eyes peeled for the next Los Angeles Printer's Fair.
posted by benito.strauss at 7:27 AM on November 4, 2015


MCMikeNamara: "provides the valuable lesson that technology may change but shitty customer service is forever."

Another thing that is evidently timeless: Printers are the worst.
posted by schmod at 7:35 AM on November 4, 2015 [11 favorites]


A - Always
B - Be
C - Compiling


I've made this joke before, but...

Put those parentheses down. Parentheses are for Clojure.
posted by Mr. Bad Example at 7:45 AM on November 4, 2015 [5 favorites]


Sorry. Could you repeat that? Your LISP makes that rather difficult to understand.
posted by schmod at 7:54 AM on November 4, 2015 [1 favorite]


troff is still such an impossibly powerful typesetting engine. If your output is columnar text, and you can find a troff implementation that matches your use case (baseline: Latin-ish, LTR), it'll create some lovely output. I've known a couple of troff wizards: one ran a dictionary and catalogue house out of a shed in Scarborough, and could make impossibly complex layouts via the creative misuse of footer commands. The other wizard made dynamic documents building from database queries and shell commands embedded in pic macros. Their code looked line line noise, but did beautiful things. Much respect.

There are a couple of more modern implementations of troff that support UTF-8 and do clever things with OTF font features like ligatures, etc. The sort-of-still-in-development-kinda Heirloom troff (demo pdf) is based on Sun and Plan 9 sources. The insanely brilliant neatroff (intro pdf) does RTL typesetting and micro-justification too. They're both a much smaller installation than a modern TeX implementation, which is now many gigabytes of macros that you'll likely never use.

If you need to learn troff, Ralph Corderoy's troff.org is a good resource. The venerable Sun manuals Using NROFF and TROFF and Formatting Documents are also helpful, if a quarter century old.
posted by scruss at 3:21 PM on November 4, 2015 [4 favorites]


one ran a dictionary and catalogue house out of a shed in Scarborough

I always wondered who was accepting that stuff. Makes sense now.
posted by ChurchHatesTucker at 6:30 PM on November 4, 2015


Okay, I returned to this story after a day and went through some of the open tabs from the related browsing, couldn't remember opening half of them. Immediately got that fresh "what the hell am I looking at" feeling.
posted by wwwwolf at 3:27 AM on November 5, 2015 [3 favorites]


Mitheral: "double block and bleed: " suggested wiggling the chips [sic] on the code converter board. This apparently nonsensical suggestion in fact cured the problem for a while."

This is an extremely common problem and remedy for many electronic devices with socketed chips. It's one of the trade offs for socketed vs. direct soldered assemblies.
"

Yes. I wiggled a lot of chips in the 90s as a technician before I saw the writing on the wall for techs who did component-level troubleshooting. That's why I found the "nonsensical" part funny.
posted by double block and bleed at 6:08 PM on November 21, 2015


« Older "I believe that everyone has a life worth saving."   |   Yeah, that's right, you like it, don't you?! Newer »


This thread has been archived and is closed to new comments