WEBVTT 00:00:00.000 --> 00:00:02.580 Welcome back to the Deep Dive. Our mission here 00:00:02.580 --> 00:00:05.700 is always the same. We take your complex source 00:00:05.700 --> 00:00:08.759 material, all the research, the books, the technical 00:00:08.759 --> 00:00:13.060 notes, and we turn them into accessible, unforgettable, 00:00:13.259 --> 00:00:16.980 and, well, actionable knowledge. And today, the 00:00:16.980 --> 00:00:19.100 figure we're looking at is someone whose work 00:00:19.100 --> 00:00:22.710 is just. It's so fundamental. It really is. I 00:00:22.710 --> 00:00:24.530 mean, if you're using any kind of digital device 00:00:24.530 --> 00:00:27.469 to listen to this right now, you are standing 00:00:27.469 --> 00:00:31.250 directly on the foundation he helped pour. We're 00:00:31.250 --> 00:00:33.210 talking about Brian Kernighan. It's genuinely 00:00:33.210 --> 00:00:35.429 impossible to overstate his importance. And yet 00:00:35.429 --> 00:00:38.130 he's probably the least recognized giant of that 00:00:38.130 --> 00:00:40.250 early computing era, at least among the general 00:00:40.250 --> 00:00:41.670 public. Right. When you list the foundational 00:00:41.670 --> 00:00:44.350 figures, you hear Ken Thompson, Dennis Ritchie, 00:00:44.509 --> 00:00:46.950 you know, the big industry titans later on. But 00:00:46.950 --> 00:00:49.229 Kernighan is the quiet architect. He's the essential 00:00:49.229 --> 00:00:51.869 guide. He didn't invent the C language, for instance, 00:00:52.030 --> 00:00:54.770 but his initial, the K, is synonymous with its 00:00:54.770 --> 00:00:57.429 defining book, KNRC. And he didn't invent the 00:00:57.429 --> 00:01:00.090 operating system itself, but he coined the term 00:01:00.090 --> 00:01:02.250 unit. That's it. That's the perfect distinction. 00:01:02.549 --> 00:01:04.390 So that's our goal for this deep dive. We're 00:01:04.390 --> 00:01:06.670 going to unpack the legacy of this man, focusing 00:01:06.670 --> 00:01:10.670 on the tools he built at Bell Labs and the programming 00:01:10.670 --> 00:01:13.090 philosophy he codified and really taught to the 00:01:13.090 --> 00:01:15.829 world, a philosophy that fundamentally changed 00:01:15.829 --> 00:01:18.250 how programmers think. Exactly. I mean, the source 00:01:18.250 --> 00:01:20.530 material shows this incredible career. He's 83 00:01:20.530 --> 00:01:24.049 now, born in 1942, and he's still an active professor 00:01:24.049 --> 00:01:26.930 at Princeton. Since 2000. That's incredible. 00:01:26.989 --> 00:01:29.810 He's a living bridge. He connects the very birth 00:01:29.810 --> 00:01:32.590 of modern computing in the 60s with the wild 00:01:32.590 --> 00:01:35.409 digital ecosystem we're all swimming in today. 00:01:35.650 --> 00:01:39.069 That bridge analogy is perfect because his work 00:01:39.069 --> 00:01:42.010 spans both the tech innovation side and the teaching 00:01:42.010 --> 00:01:44.590 side, the pedagogy. It's not just academic history. 00:01:44.689 --> 00:01:47.569 It's the operational DNA of massive, massive 00:01:47.569 --> 00:01:50.349 systems. Absolutely. Just think about Unix, a 00:01:50.349 --> 00:01:53.150 system he helped name and popularize. It powers 00:01:53.150 --> 00:01:55.370 everything from your phone to massive servers. 00:01:55.659 --> 00:01:57.439 scientific computing clusters. Everything. But 00:01:57.439 --> 00:02:00.060 beyond the systems themselves, Kernemann provided 00:02:00.060 --> 00:02:03.260 the methodology, the clear, practical approach 00:02:03.260 --> 00:02:05.799 to writing good code, which he championed in 00:02:05.799 --> 00:02:08.719 his software tools series. He is, above all, 00:02:08.879 --> 00:02:12.159 a synthesizer, a communicator. He takes these 00:02:12.159 --> 00:02:15.539 pioneering, complex tools and makes them usable 00:02:15.539 --> 00:02:18.180 for millions of people. Okay, so let's unpack 00:02:18.180 --> 00:02:21.099 this incredible journey. Let's start right at 00:02:21.099 --> 00:02:22.960 the beginning. at the genesis of his career. 00:02:23.219 --> 00:02:25.099 We'll look at his academic roots in that pivotal 00:02:25.099 --> 00:02:27.319 environment at Bell Labs because we need to find 00:02:27.319 --> 00:02:29.520 out where the K factor in modern computing really 00:02:29.520 --> 00:02:32.719 began. So Kernighan is Canadian, and he starts 00:02:32.719 --> 00:02:34.659 his academic path at the University of Toronto, 00:02:34.860 --> 00:02:37.099 getting a degree in engineering physics from 00:02:37.099 --> 00:02:40.120 1960 to 64. And then he moves south to Princeton 00:02:40.120 --> 00:02:42.259 for his Ph .D. in electrical engineering, which 00:02:42.259 --> 00:02:44.879 he finishes in 1969. And I pulled the title of 00:02:44.879 --> 00:02:46.759 that dissertation. It is a bit of a mouthful, 00:02:46.780 --> 00:02:49.840 but it's so important. Some graph partitioning 00:02:49.840 --> 00:02:53.259 problems related to program segmentation. That 00:02:53.259 --> 00:02:55.580 title might sound completely esoteric, but honestly, 00:02:55.740 --> 00:02:58.060 it's the Rosetta Stone for his entire career. 00:02:58.219 --> 00:03:00.259 Graph partitioning. It's all about efficiency 00:03:00.259 --> 00:03:02.639 and structure. OK, can you ground that for us 00:03:02.639 --> 00:03:06.180 a bit? Why is breaking up a graph relevant to 00:03:06.180 --> 00:03:09.080 writing a computer program? So think of a graph 00:03:09.080 --> 00:03:11.979 as a network, right? a set of nodes and connections. 00:03:12.379 --> 00:03:15.680 A huge computer program can be modeled as a graph 00:03:15.680 --> 00:03:18.080 where you have functions and data flows all connected. 00:03:18.879 --> 00:03:21.539 Partitioning is the art of breaking that massive, 00:03:21.599 --> 00:03:24.259 messy structure into smaller, more manageable 00:03:24.259 --> 00:03:27.340 pieces. And the key is you want to do it while 00:03:27.340 --> 00:03:29.860 minimizing the number of connections, the cuts 00:03:29.860 --> 00:03:32.219 between those pieces. So you're making it modular. 00:03:32.539 --> 00:03:34.719 Exactly. And in practical terms, this translates 00:03:34.719 --> 00:03:38.280 directly to efficiency in, say, laying out a 00:03:38.280 --> 00:03:41.560 microchip or organizing. software like segmenting 00:03:41.560 --> 00:03:44.680 memory or breaking a big program into modules 00:03:44.680 --> 00:03:47.139 that can run more or less on their own. So even 00:03:47.139 --> 00:03:48.819 before he was writing the books that taught the 00:03:48.819 --> 00:03:51.139 world how to program, his mind was already focused 00:03:51.139 --> 00:03:54.039 on optimization and structure, on how to take 00:03:54.039 --> 00:03:55.860 a complex problem and break it down into the 00:03:55.860 --> 00:03:58.889 most efficient parts possible. Precisely. His 00:03:58.889 --> 00:04:01.349 mind was just built for systems analysis. And 00:04:01.349 --> 00:04:03.530 that focus is the perfect setup for his next 00:04:03.530 --> 00:04:06.930 move to Bell Labs, which in the late 60s and 00:04:06.930 --> 00:04:09.729 early 70s was, I mean, it was arguably the most 00:04:09.729 --> 00:04:12.409 important place on the planet for shaping the 00:04:12.409 --> 00:04:14.750 future of IT. And he's working there right alongside 00:04:14.750 --> 00:04:18.009 the titans, Ken Thompson and Dennis Ritchie. 00:04:18.089 --> 00:04:20.310 Can you imagine? The environment must have been 00:04:20.310 --> 00:04:23.629 absolutely electric. Thompson and Ritchie are 00:04:23.629 --> 00:04:26.250 building what will become the world's most influential 00:04:26.250 --> 00:04:29.170 operating system. But our sources point to a 00:04:29.170 --> 00:04:31.550 huge, and like you said, often overlooked contribution 00:04:31.550 --> 00:04:34.269 Kernighan made right at the very beginning of 00:04:34.269 --> 00:04:36.870 that project. Yes. While they were building the 00:04:36.870 --> 00:04:38.970 technical backbone, Kernighan was doing the essential 00:04:38.970 --> 00:04:42.470 work of definition and promotion. And this is 00:04:42.470 --> 00:04:45.310 the key fact. Kernighan coined the term Unix. 00:04:45.449 --> 00:04:47.649 So what was it called before? The system started 00:04:47.649 --> 00:04:50.069 as Unisys, which stood for Uniplexed Information 00:04:50.069 --> 00:04:52.810 and Computing Service. It was sort of a satirical 00:04:52.810 --> 00:04:55.550 jab at the huge, unwieldy, multi -TS project 00:04:55.550 --> 00:04:58.300 they'd worked on before. So Unisee's was clunky, 00:04:58.420 --> 00:05:01.720 maybe a bit of an inside joke. It was. Kernighan 00:05:01.720 --> 00:05:04.139 saw an opportunity for clarity, for branding 00:05:04.139 --> 00:05:07.339 even. He shortened it, refined it to Unix. It 00:05:07.339 --> 00:05:09.759 gave the system this memorable, strong identity 00:05:09.759 --> 00:05:11.720 that set it apart. But it was more than just 00:05:11.720 --> 00:05:14.000 a name change, wasn't it? Oh, much more. He was 00:05:14.000 --> 00:05:16.399 also instrumental in helping Thompson and Ritchie 00:05:16.399 --> 00:05:19.139 articulate and popularize the Unix philosophy. 00:05:19.439 --> 00:05:22.439 The idea that programs should do one thing, do 00:05:22.439 --> 00:05:25.790 it well. and work together exactly using common 00:05:25.790 --> 00:05:28.889 interfaces like pipes that whole philosophy became 00:05:28.889 --> 00:05:32.449 the um the operational ethos for an entire generation 00:05:32.449 --> 00:05:35.709 of computing he didn't just write code he effectively 00:05:35.709 --> 00:05:38.790 branded the entire concept of structured modular 00:05:38.790 --> 00:05:41.779 small tool computing That role of popularizer, 00:05:41.839 --> 00:05:44.040 of communicator, it seems to be his specialty. 00:05:44.240 --> 00:05:46.839 It absolutely is. He saw the revolutionary ideas 00:05:46.839 --> 00:05:49.240 underneath and helped articulate them both in 00:05:49.240 --> 00:05:51.920 memos and crucially to the wider world. He was 00:05:51.920 --> 00:05:54.180 the perfect guide to introduce these new ideas 00:05:54.180 --> 00:05:56.180 to an audience that had grown up on, you know, 00:05:56.180 --> 00:05:59.439 giant mainframe computers. And that guiding role 00:05:59.439 --> 00:06:02.139 leads us directly to the book that defined a 00:06:02.139 --> 00:06:04.800 generation of programming. The C programming 00:06:04.800 --> 00:06:07.699 language, which he co -authored with Dennis Ritchie, 00:06:07.800 --> 00:06:11.019 the book that gave us the term K &R C. That book 00:06:11.019 --> 00:06:13.639 wasn't just a book on C. For over a decade, it 00:06:13.639 --> 00:06:16.199 was the definitive specification for the language. 00:06:16.579 --> 00:06:19.600 This was before any formal standardization. So 00:06:19.600 --> 00:06:21.819 if your compiler could build code, according 00:06:21.819 --> 00:06:24.519 to the K &R book, it was considered a legitimate 00:06:24.519 --> 00:06:27.360 C compiler. That was the standard. It was, in 00:06:27.360 --> 00:06:30.069 essence, the C language bible. And it was responsible 00:06:30.069 --> 00:06:32.930 for C's explosive adoption in universities and 00:06:32.930 --> 00:06:35.769 industry. But here's a crucial nuance that just 00:06:35.769 --> 00:06:39.329 spins volumes about Kernighan's humility. Despite 00:06:39.329 --> 00:06:41.449 his initial being right there on the cover, he 00:06:41.449 --> 00:06:43.970 explicitly said, and I'm quoting, that he had 00:06:43.970 --> 00:06:47.029 no part in the design of the C language. He was 00:06:47.029 --> 00:06:49.670 crystal clear. It's entirely Dennis Ritchie's 00:06:49.670 --> 00:06:52.209 work. And that's vital context. It really highlights 00:06:52.209 --> 00:06:54.410 the nature of their collaboration. Ritchie was 00:06:54.410 --> 00:06:56.670 the linguistic architect. He created this low 00:06:56.670 --> 00:06:59.110 -level, powerful, but remarkably simple language. 00:06:59.350 --> 00:07:01.629 So what was Kernighan's genius then? It was in 00:07:01.629 --> 00:07:03.889 the pedagogy, the articulation of how to use 00:07:03.889 --> 00:07:07.430 C effectively. He provided the clear examples, 00:07:07.689 --> 00:07:10.389 the structure, the style guides, all the things 00:07:10.389 --> 00:07:12.839 needed to take C from a Bell Labs project. and 00:07:12.839 --> 00:07:15.220 turn it into the global standard for systems 00:07:15.220 --> 00:07:19.379 programming. So the K in KNRC and later in AWK, 00:07:19.439 --> 00:07:21.959 it stands for Kernan of the Great Communicator, 00:07:21.980 --> 00:07:24.120 not Kernan the language designer. Perfectly put. 00:07:24.259 --> 00:07:26.899 Yeah. And in that role as communicator, he gives 00:07:26.899 --> 00:07:29.160 us maybe the most ubiquitous piece of code in 00:07:29.160 --> 00:07:31.899 history. Hello, world. We all just take it for 00:07:31.899 --> 00:07:33.839 granted now it's the first thing you type. But 00:07:33.839 --> 00:07:36.259 that specific ritual comes directly from his 00:07:36.259 --> 00:07:39.120 work. It's a foundational cultural artifact of 00:07:39.120 --> 00:07:42.180 computing. And the origin goes back to 1972, 00:07:42.540 --> 00:07:45.439 even before C really took off. He wrote a document 00:07:45.439 --> 00:07:47.639 called A Tutorial Introduction to the Language 00:07:47.639 --> 00:07:50.800 B. B being the immediate predecessor to C. Right. 00:07:50.879 --> 00:07:53.259 And he wasn't trying to create some grand ritual. 00:07:53.360 --> 00:07:55.720 He was just trying to provide a clear, minimal 00:07:55.720 --> 00:07:57.899 example. Of what? What was he demonstrating? 00:07:58.240 --> 00:08:00.540 Memory management and strings. He just needed 00:08:00.540 --> 00:08:03.740 a simple, unambiguous way to show how to declare 00:08:03.740 --> 00:08:07.060 a string, store it, and print it out. So he used 00:08:07.060 --> 00:08:10.370 the phrases hello and world. as simple variables 00:08:10.370 --> 00:08:12.910 to illustrate it and that just stuck it stuck 00:08:12.910 --> 00:08:15.750 it became the iconic starting point the minimal 00:08:15.750 --> 00:08:17.990 proof that your entire development environment 00:08:17.990 --> 00:08:20.949 your compiler your os was all set up correctly 00:08:20.949 --> 00:08:23.389 and ready to go it became the universal right 00:08:23.389 --> 00:08:26.870 of passage for every new programmer it's the 00:08:26.870 --> 00:08:29.850 simplicity that made it so enduring every programmer 00:08:29.850 --> 00:08:32.009 knows that little feeling of triumph when they 00:08:32.009 --> 00:08:34.429 see those two words print out absolutely and 00:08:34.429 --> 00:08:36.450 here's a fascinating little anecdote from the 00:08:36.450 --> 00:08:38.669 source material that shows how much status that's 00:08:38.789 --> 00:08:41.350 simple phrase has, Kernighan's original 1978 00:08:41.350 --> 00:08:44.129 implementation of the program, the actual code, 00:08:44.350 --> 00:08:46.929 was deemed so significant that it was sold at 00:08:46.929 --> 00:08:49.230 the algorithm auction. Wow. They sold the code 00:08:49.230 --> 00:08:52.720 for... Think about that. That's recognizing a 00:08:52.720 --> 00:08:54.899 piece of code as a true historical artifact, 00:08:55.240 --> 00:08:58.580 like a famous painting or a manuscript. It just 00:08:58.580 --> 00:09:00.720 underscores that these aren't just dry technical 00:09:00.720 --> 00:09:03.179 concepts. They are the bedrock of our digital 00:09:03.179 --> 00:09:06.460 civilization. And it all came from Kernighan 00:09:06.460 --> 00:09:09.539 trying to clearly explain string handling. The 00:09:09.539 --> 00:09:11.559 man gave us the name for the operating system, 00:09:11.720 --> 00:09:14.039 defined his programming language, and gave us 00:09:14.039 --> 00:09:16.100 the introductory phrase for every developer. 00:09:16.639 --> 00:09:18.960 Okay, so now let's move from those iconic roles 00:09:18.960 --> 00:09:21.220 to the specific tools he engineered that made 00:09:21.220 --> 00:09:24.600 the Unix ecosystem so productive. We're transitioning 00:09:24.600 --> 00:09:27.440 now from the big picture systems to the specific 00:09:27.440 --> 00:09:30.480 languages Kernighan co -authored. And let's start 00:09:30.480 --> 00:09:34.509 with AWK. Right, AWK. The name is an acronym 00:09:34.509 --> 00:09:37.309 for the three authors, Alfred Aho, Peter J. Weinberger, 00:09:37.490 --> 00:09:39.450 and of course, Brian Kernighan. It was developed 00:09:39.450 --> 00:09:42.629 in 1977. And it is the quintessential Unix utility. 00:09:43.009 --> 00:09:45.149 It's a fantastic example of that Unix philosophy 00:09:45.149 --> 00:09:47.289 and practice. It's a specialized language, a 00:09:47.289 --> 00:09:49.590 domain -specific language, optimized for one 00:09:49.590 --> 00:09:52.009 thing, text processing, data extraction reporting. 00:09:52.289 --> 00:09:54.450 It's a little powerhouse. So if you have a massive 00:09:54.450 --> 00:09:57.370 log file or a huge CSV, and you need to pull 00:09:57.370 --> 00:09:59.649 out specific columns and summarize the data. 00:10:00.220 --> 00:10:02.940 AWK is your best friend. A lot of people today 00:10:02.940 --> 00:10:06.440 might default to, I don't know, Python or maybe 00:10:06.440 --> 00:10:08.299 even try to load it into a spreadsheet. Which 00:10:08.299 --> 00:10:11.340 would probably crash. So why does AWK, a language 00:10:11.340 --> 00:10:14.340 from 1977, still matter so much? What makes it 00:10:14.340 --> 00:10:17.659 unique? Its power is in its simplicity and its 00:10:17.659 --> 00:10:21.200 implicit structure. AWK automatically reads your 00:10:21.200 --> 00:10:24.600 input line by line. For every single line, it 00:10:24.600 --> 00:10:27.259 automatically breaks the data into fields based 00:10:27.259 --> 00:10:30.580 on a delimiter like a space or a comma. So you 00:10:30.580 --> 00:10:32.139 don't have to write all that boilerplate code 00:10:32.139 --> 00:10:34.399 for looping and splitting? None of it. You just 00:10:34.399 --> 00:10:36.399 define patterns and actions. You can just say, 00:10:36.519 --> 00:10:38.899 for every line where the first column says error, 00:10:39.120 --> 00:10:41.860 print the third and fifth columns. That's it. 00:10:42.220 --> 00:10:44.240 It's incredibly fast and efficient for slicing 00:10:44.240 --> 00:10:47.100 and dicing structured text, often much faster 00:10:47.100 --> 00:10:49.120 than firing up a more general -purpose language. 00:10:49.399 --> 00:10:52.220 It perfectly embodies that idea of a small tool 00:10:52.220 --> 00:10:54.809 doing one thing incredibly well. Absolutely. 00:10:55.009 --> 00:10:57.370 The book, The AWK Programming Language, came 00:10:57.370 --> 00:11:00.350 out in 88, and it's so relevant that it was reissued 00:11:00.350 --> 00:11:03.269 in 2023. But here's the fact that just defines 00:11:03.269 --> 00:11:05.389 Kernighan's commitment. I think I know what you're 00:11:05.389 --> 00:11:07.070 going to say. Our source material notes that 00:11:07.070 --> 00:11:10.269 as recently as 2022, he stated he was actively 00:11:10.269 --> 00:11:13.370 working on improvements to AWK. The man is in 00:11:13.370 --> 00:11:16.759 his 80s. And he's still maintaining and curating 00:11:16.759 --> 00:11:20.139 a 45 -year -old programming language. That is 00:11:20.139 --> 00:11:23.120 just astonishing dedication. It's stunning. Most 00:11:23.120 --> 00:11:25.620 software from that era is completely obsolete. 00:11:25.799 --> 00:11:27.820 The fact that he is still actively improving 00:11:27.820 --> 00:11:30.980 AWK shows you how durable and elegant, robust 00:11:30.980 --> 00:11:33.899 tool design can be. It proves that simplicity 00:11:33.899 --> 00:11:37.039 and focus in software can endure far longer than 00:11:37.039 --> 00:11:39.820 these massive, complex frameworks built on whatever 00:11:39.820 --> 00:11:42.059 the trend of the day is. That structural thinking, 00:11:42.200 --> 00:11:44.879 that focus on optimization. It didn't stop at 00:11:44.879 --> 00:11:47.320 text processing. In the mid -80s, he co -authored 00:11:47.320 --> 00:11:49.659 AMPL, the Mathematical Programming Language. 00:11:49.899 --> 00:11:51.779 And this takes us right back to his PhD dissertation, 00:11:52.039 --> 00:11:54.379 doesn't it? It really does. Back to graph partitioning. 00:11:54.659 --> 00:11:58.139 AMPL, from 1985, is designed for describing and 00:11:58.139 --> 00:12:01.120 solving large -scale mathematical problems, optimization 00:12:01.120 --> 00:12:04.740 problems, linear programming, nonlinear integer 00:12:04.740 --> 00:12:08.039 programming, all that heavy -duty stuff. So what's 00:12:08.039 --> 00:12:09.820 the real difference between a traditional language 00:12:09.820 --> 00:12:12.960 like C and a mathematical language like AMPL? 00:12:13.320 --> 00:12:15.379 A traditional language tells the computer how 00:12:15.379 --> 00:12:18.730 to get a result, step by step. AMPL is a modeling 00:12:18.730 --> 00:12:21.710 language. You describe the problem. You describe 00:12:21.710 --> 00:12:24.129 the constraints, the objective, what you want 00:12:24.129 --> 00:12:26.769 to maximize or minimize, and the language then 00:12:26.769 --> 00:12:29.250 interfaces with specialized solvers to find the 00:12:29.250 --> 00:12:31.370 optimal solution for you. So instead of writing 00:12:31.370 --> 00:12:33.990 the optimization algorithm yourself, you're just 00:12:33.990 --> 00:12:35.950 modeling the problem and letting the system figure 00:12:35.950 --> 00:12:38.409 out the best path. Precisely. It lets the experts, 00:12:38.549 --> 00:12:40.830 the mathematicians, the engineers focus on the 00:12:40.830 --> 00:12:42.830 structure of their problem, not the boilerplate 00:12:42.830 --> 00:12:44.809 code needed to manage memory or loop through 00:12:44.809 --> 00:12:46.899 data. It just shows his incredible... range. 00:12:47.179 --> 00:12:49.580 He's equally comfortable writing elegant prose 00:12:49.580 --> 00:12:52.480 to teach C and designing these precise formal 00:12:52.480 --> 00:12:55.120 languages for high level optimization. Speaking 00:12:55.120 --> 00:12:57.700 of optimization, let's pivot to his foundational 00:12:57.700 --> 00:13:00.360 work in algorithms, especially his collaboration 00:13:00.360 --> 00:13:03.940 with Shen Lin on these really hard NP -complete 00:13:03.940 --> 00:13:06.799 problems. Right. And their algorithms bear their 00:13:06.799 --> 00:13:09.820 names, but with this fascinating example of professional 00:13:09.820 --> 00:13:12.440 equity in how they're named. Let's get into that. 00:13:12.500 --> 00:13:14.259 But first, can you give us a quick refresher 00:13:14.259 --> 00:13:16.929 on NP -complete problems? Sure. They're famous 00:13:16.929 --> 00:13:19.090 in computer science because they are computationally 00:13:19.090 --> 00:13:21.570 hard. Things like the traveling salesman problem 00:13:21.570 --> 00:13:25.149 or graph partitioning. For any really big version 00:13:25.149 --> 00:13:28.129 of these problems, finding the absolute, provably 00:13:28.129 --> 00:13:31.649 perfect solution is often impossible in a reasonable 00:13:31.649 --> 00:13:34.870 amount of time, even for a supercomputer. Can 00:13:34.870 --> 00:13:36.490 you give us a quick, tangible example of the 00:13:36.490 --> 00:13:38.690 traveling salesman problem? Absolutely. Imagine 00:13:38.690 --> 00:13:41.809 a delivery driver. They have to visit, say, 50 00:13:41.809 --> 00:13:44.750 cities. Start at the hub. Visit every city exactly 00:13:44.750 --> 00:13:47.429 once and return to the hub. The goal is simple. 00:13:47.889 --> 00:13:50.570 Find the shortest possible route. Sounds simple 00:13:50.570 --> 00:13:53.529 enough. But the number of possible routes grows 00:13:53.529 --> 00:13:57.070 factorially. For 50 cities, the number of routes 00:13:57.070 --> 00:13:59.470 is astronomical. You could never check them all. 00:13:59.549 --> 00:14:01.909 This is where heuristics come in. You need a 00:14:01.909 --> 00:14:03.850 clever way to find a really good answer, even 00:14:03.850 --> 00:14:06.289 if it's not provably the single best one. And 00:14:06.289 --> 00:14:08.509 that's where Kernighan and Lin stepped in with 00:14:08.509 --> 00:14:12.259 practical, efficient solutions. Yes. Two major 00:14:12.259 --> 00:14:15.240 contributions here. First, we have the Kernighan 00:14:15.240 --> 00:14:18.279 -Linn algorithm from 1970. This goes back to 00:14:18.279 --> 00:14:21.480 his PhD. It addresses graph partitioning, dividing 00:14:21.480 --> 00:14:24.340 a graph into two pieces of roughly equal size 00:14:24.340 --> 00:14:26.840 while minimizing the connections between them. 00:14:26.919 --> 00:14:29.000 Which is vital for things like designing a circuit 00:14:29.000 --> 00:14:31.200 board. Right. Minimizing wire length. Exactly. 00:14:31.620 --> 00:14:35.500 Then, a few years later in 1973, they gave us 00:14:35.500 --> 00:14:38.379 the Linn -Kernighan heuristic. This is a powerful 00:14:38.379 --> 00:14:40.700 method for tackling the traveling salesman problem 00:14:40.700 --> 00:14:42.820 itself. Okay, so why the different name order? 00:14:43.100 --> 00:14:45.419 Kernaghan -Lynn versus Lynn Kernaghan. It's subtle, 00:14:45.580 --> 00:14:48.419 but it's a powerful display of, you know, authorial 00:14:48.419 --> 00:14:50.899 equity, professional courtesy. The Cunningham 00:14:50.899 --> 00:14:52.899 -Lynn paper came first and addressed the problem 00:14:52.899 --> 00:14:55.299 directly related to Kernaghan's own PhD focus. 00:14:55.500 --> 00:14:57.299 That's likely why his name is first. And then 00:14:57.299 --> 00:14:59.460 for the second paper. For the second, equally 00:14:59.460 --> 00:15:01.860 famous paper on the traveling salesman heuristic, 00:15:02.039 --> 00:15:04.039 they reversed the order. They put Lynn's name 00:15:04.039 --> 00:15:06.399 first. So even though they were working on similar 00:15:06.399 --> 00:15:09.000 problems, they made sure to balance the credit 00:15:09.000 --> 00:15:11.379 instead of, say, always listing the better known 00:15:11.379 --> 00:15:14.320 name first. Precisely. It's a wonderful reflection 00:15:14.320 --> 00:15:16.700 of the intellectual generosity in their collaboration. 00:15:16.960 --> 00:15:20.059 And both algorithms are foundational. The Lynn 00:15:20.059 --> 00:15:22.279 Kernighan heuristic, in particular, is still 00:15:22.279 --> 00:15:24.379 one of the most effective methods for quickly 00:15:24.379 --> 00:15:27.779 finding near optimal solutions to the TSP. It's 00:15:27.779 --> 00:15:30.639 essential for logistics and routing today. Okay, 00:15:30.679 --> 00:15:32.820 moving beyond the core languages and algorithms, 00:15:33.139 --> 00:15:35.559 Kernighan was also instrumental in building the 00:15:35.559 --> 00:15:37.980 tools for documentation and publishing within 00:15:37.980 --> 00:15:41.179 Unix, the stuff we barely think about now. These 00:15:41.179 --> 00:15:43.740 tools were absolutely essential. Remember, Unix 00:15:43.740 --> 00:15:46.440 was built by scientists and engineers. They needed 00:15:46.440 --> 00:15:48.840 to write papers, manuals, technical reports. 00:15:49.200 --> 00:15:52.220 And they needed to include math, diagrams, precise 00:15:52.220 --> 00:15:54.240 formatting. And they needed to do it all without 00:15:54.240 --> 00:15:56.919 leaving the command line. Let's break down the 00:15:56.919 --> 00:15:59.899 three main tools he co -created. First up is 00:15:59.899 --> 00:16:04.100 DTROF. Right. DTROF, from around 1973, stands 00:16:04.100 --> 00:16:07.740 for Device Independent TROF. TROF was the original 00:16:07.740 --> 00:16:09.820 text formatting program, but it was hard -coded 00:16:09.820 --> 00:16:13.179 for specific output devices. So a specific printer 00:16:13.179 --> 00:16:16.059 or a typesetter. Exactly. By making it device 00:16:16.059 --> 00:16:18.460 independent, Kernighan made it so the same source 00:16:18.460 --> 00:16:21.159 file could produce identical presentation quality 00:16:21.159 --> 00:16:24.580 output on virtually any device. It was a revolutionary 00:16:24.580 --> 00:16:27.240 concept for portability at a time when hardware 00:16:27.240 --> 00:16:29.700 compatibility was just a nightmare. And then 00:16:29.700 --> 00:16:31.399 there's Equin, which he created with Lorinda 00:16:31.399 --> 00:16:34.299 Cherry in 1974. Equin was the typesetting language 00:16:34.299 --> 00:16:36.799 for mathematics inside of Trough. Before this, 00:16:36.860 --> 00:16:38.960 getting a complex equation into a document was 00:16:38.960 --> 00:16:42.159 just a manual, tedious, messy process. Equin 00:16:42.159 --> 00:16:44.259 let scientists write expressions in a simple, 00:16:44.419 --> 00:16:47.440 almost plain text way, like a over b plus c sub 00:16:47.440 --> 00:16:49.620 2. And the system would handle the precise layout 00:16:49.620 --> 00:16:51.639 of the fractions, the superscripts, everything. 00:16:51.919 --> 00:16:54.360 It completely streamlined scientific publishing 00:16:54.360 --> 00:16:56.799 at Bell Labs. And finally, a tool for visuals. 00:16:57.000 --> 00:17:00.720 Pick. in 1988. PIC was for diagrams. If you needed 00:17:00.720 --> 00:17:03.120 a flowchart, a system schematic, any simple diagram 00:17:03.120 --> 00:17:04.920 for your manual, you used PIC. You described 00:17:04.920 --> 00:17:07.140 it with text and it drew it. So if you put them 00:17:07.140 --> 00:17:10.359 all together, DITRA for formatting, ECM for math, 00:17:10.740 --> 00:17:14.180 PIC for diagrams, Kernan helped create a self 00:17:14.180 --> 00:17:17.380 -contained command line desktop publishing system 00:17:17.380 --> 00:17:20.160 decades before most people had a desktop computer. 00:17:20.339 --> 00:17:23.220 Precisely. It shows a holistic approach. A perfect 00:17:23.220 --> 00:17:26.140 system needs good code, but it also needs good 00:17:26.140 --> 00:17:29.359 documentation. And the tools he built ensured 00:17:29.359 --> 00:17:32.119 that the documentation was as rigorous and professional 00:17:32.119 --> 00:17:35.319 as the software itself. And that brings us from 00:17:35.319 --> 00:17:37.960 the tools themselves to the philosophy behind 00:17:37.960 --> 00:17:40.240 them. Because Kernighan wasn't just building 00:17:40.240 --> 00:17:43.140 things. He was teaching people how to think about 00:17:43.140 --> 00:17:44.920 programming, no matter what language they were 00:17:44.920 --> 00:17:47.920 using. The core ethos of that Unix world was 00:17:47.920 --> 00:17:50.339 really spread globally through Kernighan's teaching, 00:17:50.480 --> 00:17:52.759 especially his software tools series, which he 00:17:52.759 --> 00:17:55.119 co -authored with P .J. Plogger. What was the 00:17:55.119 --> 00:17:57.180 central mission of that series? The mission was 00:17:57.180 --> 00:17:59.599 all about portability and methodology. In the 00:17:59.599 --> 00:18:01.940 70s, a lot of places, universities, corporations, 00:18:02.460 --> 00:18:04.619 they were heavily invested in older languages 00:18:04.619 --> 00:18:07.500 like Fortran. And Fortran lacked a lot of the 00:18:07.500 --> 00:18:09.430 structure. programming features we just take 00:18:09.430 --> 00:18:11.769 for granted today. Kernighan and Plager realized 00:18:11.769 --> 00:18:14.710 that the power of Unix, the modularity, the structured 00:18:14.710 --> 00:18:17.849 control flow, the reusability, that power wasn't 00:18:17.849 --> 00:18:19.950 tied to the C language. It was a way of thinking. 00:18:20.250 --> 00:18:22.670 So the books didn't just teach C, they taught 00:18:22.670 --> 00:18:25.309 the style of C thinking, even if you were stuck 00:18:25.309 --> 00:18:27.509 writing in something older and less flexible. 00:18:27.710 --> 00:18:31.349 Exactly. The software tools books were like blueprints. 00:18:31.819 --> 00:18:34.140 They provided adaptations so that programmers 00:18:34.140 --> 00:18:36.680 using these older languages could adopt that 00:18:36.680 --> 00:18:39.240 structured, modular approach that was pioneered 00:18:39.240 --> 00:18:42.420 in Unix. They made versions for BASIC, FORTRAN, 00:18:42.440 --> 00:18:44.880 Pascal. And the most famous example of this was 00:18:44.880 --> 00:18:47.599 RATFOR, right? Oh, RATFOR was a revelation. It 00:18:47.599 --> 00:18:50.579 stands for Rational FORTRAN. Standard FORTRAN 00:18:50.579 --> 00:18:52.680 at the time was notorious for spaghetti code 00:18:52.680 --> 00:18:55.240 because it relied so heavily on JOTO statement. 00:18:55.599 --> 00:18:58.109 Jumps all over the place. A total mess. RAT4 00:18:58.109 --> 00:19:01.109 was a preprocessor. It let programmers use clear, 00:19:01.190 --> 00:19:03.890 modern, structured concepts. Things like if and 00:19:03.890 --> 00:19:06.430 else, while loops, break statements, things that 00:19:06.430 --> 00:19:08.910 were natural in C but really clumsy or impossible 00:19:08.910 --> 00:19:11.089 in standard Fortran. And then what happened? 00:19:11.230 --> 00:19:13.869 The RAT4 program would then compile those structured 00:19:13.869 --> 00:19:17.349 statements down into legitimate, portable Fortran 00:19:17.349 --> 00:19:19.809 code that any standard compiler could understand. 00:19:20.750 --> 00:19:23.210 So it was a way to introduce good software engineering 00:19:23.210 --> 00:19:26.569 principles without telling a huge company they 00:19:26.569 --> 00:19:29.009 had to throw out millions of lines of Fortran 00:19:29.009 --> 00:19:32.589 code. It was a brilliant Trojan horse for structured 00:19:32.589 --> 00:19:35.029 programming. And they put it in the public domain. 00:19:35.130 --> 00:19:37.069 They made it freely available. So they spread 00:19:37.069 --> 00:19:39.450 these disciplined coding practices all over the 00:19:39.450 --> 00:19:42.569 world, promoting portable, readable code. It 00:19:42.569 --> 00:19:44.710 just shows his priority was always the methodology, 00:19:44.950 --> 00:19:48.029 not language tribalism. But despite that, he 00:19:48.029 --> 00:19:50.630 clearly had a favorite. Which brings us to a 00:19:50.630 --> 00:19:54.970 great anecdote, the C -Island test. Yes. He famously 00:19:54.970 --> 00:19:57.569 asked the question, if you were stranded on a 00:19:57.569 --> 00:19:59.690 remote island and could only bring one programming 00:19:59.690 --> 00:20:02.170 language with you, which one would it be? Kernan's 00:20:02.170 --> 00:20:05.490 answer was C. It was C. So why C? Why not something 00:20:05.490 --> 00:20:07.910 newer, safer, or even a high -level tool like 00:20:07.910 --> 00:20:10.930 his own AWK? The choice goes right back to C's 00:20:10.930 --> 00:20:13.430 foundational role. It's a combination of simplicity 00:20:13.430 --> 00:20:16.609 and raw power. C is what we call close to the 00:20:16.609 --> 00:20:19.319 metal. It gives the programmer explicit, direct 00:20:19.319 --> 00:20:22.680 control over memory, over system resources, over 00:20:22.680 --> 00:20:24.720 hardware. It's the language you'd use to build 00:20:24.720 --> 00:20:27.779 a system from absolute scratch. Precisely. If 00:20:27.779 --> 00:20:30.000 you're bootstrapping a new digital world on this 00:20:30.000 --> 00:20:32.279 metaphorical island, you need a language that 00:20:32.279 --> 00:20:34.519 can write the operating system kernel, the compiler, 00:20:34.640 --> 00:20:36.839 the fundamental tools themselves. C provides 00:20:36.839 --> 00:20:39.799 that minimal, powerful tool set to do exactly 00:20:39.799 --> 00:20:42.819 that. It's the language of creation. His choice 00:20:42.819 --> 00:20:46.059 just reflects this deep engineering -driven understanding 00:20:46.059 --> 00:20:48.259 of what makes a system work from the ground up. 00:20:48.420 --> 00:20:50.819 Okay, so moving from the practical to the more 00:20:50.819 --> 00:20:53.619 philosophical, we get to a term Kernighan coined 00:20:53.619 --> 00:20:56.579 that is just a brilliant, biting critique of 00:20:56.579 --> 00:20:59.859 modern interface design. W -Y -S -I -A -Y -G. 00:21:00.039 --> 00:21:02.700 What you see is all you get. This is a phenomenal 00:21:02.700 --> 00:21:04.559 piece of critical design philosophy, and it's 00:21:04.559 --> 00:21:06.720 maybe more relevant today than it was when he 00:21:06.720 --> 00:21:08.559 coined it. We're all familiar with Y -S -I -Y 00:21:08.559 --> 00:21:11.269 -G. What you see is what you get. That describes 00:21:11.269 --> 00:21:13.309 things like a traditional word processor, where 00:21:13.309 --> 00:21:15.170 the screen is supposed to look exactly like the 00:21:15.170 --> 00:21:19.069 printed page. Right. But W -Y -S -I -I -Y -G, 00:21:19.230 --> 00:21:21.809 his sarcastic version, is the critical contrast. 00:21:22.170 --> 00:21:24.589 He's making a point about what's missing, what's 00:21:24.589 --> 00:21:26.390 being stripped away. So what's he getting at? 00:21:26.509 --> 00:21:28.710 What's been thrown away? He's criticizing the 00:21:28.710 --> 00:21:32.049 tendency of purely visual interfaces to hide. 00:21:32.640 --> 00:21:36.079 Or worse, completely discard the underlying structural 00:21:36.079 --> 00:21:39.319 or semantic information that gives a document 00:21:39.319 --> 00:21:42.579 its utility in other contexts. Can we illustrate 00:21:42.579 --> 00:21:44.920 that with a modern example? Sure. Think about 00:21:44.920 --> 00:21:47.140 writing a document. If you write it in a structured 00:21:47.140 --> 00:21:50.039 format like Markdown or LaTeX, you explicitly 00:21:50.039 --> 00:21:52.519 define your sections, your lists, your headings 00:21:52.519 --> 00:21:55.140 using clear source code. That's pure structure. 00:21:55.400 --> 00:21:57.740 You can easily export it, index it, transform 00:21:57.740 --> 00:22:00.160 it into a dozen different formats. And the W 00:22:00.160 --> 00:22:03.299 -Y -Z -Y -G alternative. If you write the same 00:22:03.299 --> 00:22:06.380 thing in a purely visual editor, say Google Docs, 00:22:06.380 --> 00:22:08.240 you might just highlight some text and click 00:22:08.240 --> 00:22:10.579 the bold button. You see the result, bold text, 00:22:10.779 --> 00:22:12.559 but the system underneath might just be storing 00:22:12.559 --> 00:22:14.980 this text is visually emphasized. It doesn't 00:22:14.980 --> 00:22:17.160 know it's a heading. Exactly. It might have thrown 00:22:17.160 --> 00:22:19.660 away the semantic distinction between a section 00:22:19.660 --> 00:22:23.140 heading and just some large, bold text you used 00:22:23.140 --> 00:22:25.000 for emphasis. You only get the visual appearance. 00:22:25.099 --> 00:22:27.660 You lose the structural definition. What you 00:22:27.660 --> 00:22:32.119 see is all you get. And what you get isn't enough 00:22:32.119 --> 00:22:34.660 for any kind of automated processing. That's 00:22:34.660 --> 00:22:37.880 it. Krigman's critique forces us to ask, are 00:22:37.880 --> 00:22:40.599 we prioritizing visual ease right now at the 00:22:40.599 --> 00:22:43.279 expense of structural utility and flexibility 00:22:43.279 --> 00:22:46.140 down the road? And that insight is crucial for 00:22:46.140 --> 00:22:48.339 everything from document management to modern 00:22:48.339 --> 00:22:51.259 web development. On the topic of critique, we 00:22:51.259 --> 00:22:53.279 have to mention his famous takedown of another 00:22:53.279 --> 00:22:56.240 major language of the time, Pascal. It was in 00:22:56.240 --> 00:22:59.720 a 1981 technical report called Why Pascal is 00:22:59.720 --> 00:23:01.900 Not My Favorite Programming Language. This is 00:23:01.900 --> 00:23:05.700 a very popular and definitely a provocative piece 00:23:05.700 --> 00:23:07.220 of writing. It really stirred things up in the 00:23:07.220 --> 00:23:09.640 academic world because Pascal, designed by Nick 00:23:09.640 --> 00:23:11.960 Lessworth, was praised academically for its rigor 00:23:11.960 --> 00:23:14.579 and for promoting structured programming. The 00:23:14.579 --> 00:23:16.740 very thing Kernan was advocating for. So if it 00:23:16.740 --> 00:23:18.480 champions structure, why was he so critical? 00:23:18.660 --> 00:23:21.000 His critique was all about its limitations in 00:23:21.000 --> 00:23:23.680 the real world, in the large -scale systems environment 00:23:23.680 --> 00:23:27.180 of Bell Labs. He pointed to rigidity in how it 00:23:27.180 --> 00:23:30.000 handled strings, a clumsy approach to input and 00:23:30.000 --> 00:23:32.740 output, and its limits on building large portable 00:23:32.740 --> 00:23:35.299 systems that had to interact cleanly with the 00:23:35.299 --> 00:23:38.269 OS. All the things that C was really good at. 00:23:38.490 --> 00:23:41.269 Precisely. He basically argued that while Pascal 00:23:41.269 --> 00:23:43.329 was a great language for teaching computer science 00:23:43.329 --> 00:23:46.329 concepts, it was often inadequate for production 00:23:46.329 --> 00:23:48.490 engineering, for building real things that had 00:23:48.490 --> 00:23:51.069 to work in the wild. Was the criticism just an 00:23:51.069 --> 00:23:53.309 intellectual jab or did it have an effect? It 00:23:53.309 --> 00:23:55.809 was constructive critique rooted in engineering 00:23:55.809 --> 00:23:59.490 reality. And it did have an effect. The sources 00:23:59.490 --> 00:24:01.450 note that some parts of his criticism actually 00:24:01.450 --> 00:24:03.789 became obsolete because the ISO standard for 00:24:03.789 --> 00:24:06.349 Pascal was created after his report came out, 00:24:06.410 --> 00:24:08.710 and it addressed some of the flaws he'd highlighted. 00:24:08.990 --> 00:24:11.990 So his expert critique helped spur the evolution 00:24:11.990 --> 00:24:14.150 of the language. It's a great example of how 00:24:14.150 --> 00:24:16.869 strong, informed opposition can actually lead 00:24:16.869 --> 00:24:19.730 to progress in a technical field. It's clear 00:24:19.730 --> 00:24:21.650 that whether he was building a tool, writing 00:24:21.650 --> 00:24:24.690 a book, or issuing a critique, his central goal 00:24:24.690 --> 00:24:26.880 was always the same. Improve the efficiency, 00:24:27.200 --> 00:24:29.619 the clarity, and the structural soundness of 00:24:29.619 --> 00:24:32.380 programming. He is the ultimate engineer communicator. 00:24:32.440 --> 00:24:35.140 That's the perfect summary. Let's move now to 00:24:35.140 --> 00:24:37.299 the latter half of his career, when he moves 00:24:37.299 --> 00:24:39.500 from the industrial research powerhouse of Bell 00:24:39.500 --> 00:24:42.740 Labs to the world of academia. He's been a professor 00:24:42.740 --> 00:24:44.619 of computer science at Princeton since 2000, 00:24:44.900 --> 00:24:47.059 and he's the director of undergraduate studies. 00:24:47.279 --> 00:24:49.660 And Princeton gained an immense resource. His 00:24:49.660 --> 00:24:51.700 role as an educator is maybe the most enduring 00:24:51.700 --> 00:24:54.880 part of his later legacy, especially his commitment 00:24:54.880 --> 00:24:58.359 to outreach beyond computer science majors. Every 00:24:58.359 --> 00:25:00.960 fall, he teaches this highly influential course 00:25:00.960 --> 00:25:04.700 called Computers in Our World, COS 109. Teaching 00:25:04.700 --> 00:25:06.660 the fundamentals of computing to non -majors 00:25:06.660 --> 00:25:09.470 feels more important now than ever before. Oh, 00:25:09.549 --> 00:25:12.609 it's absolutely crucial for digital literacy. 00:25:13.369 --> 00:25:16.670 The course focuses on the core concepts. How 00:25:16.670 --> 00:25:19.410 does the internet actually work? What does privacy 00:25:19.410 --> 00:25:22.230 really mean? The mechanics of cryptography? Security 00:25:22.230 --> 00:25:24.710 risks? All the stuff we interact with daily but 00:25:24.710 --> 00:25:26.410 don't really understand. Right. And to have a 00:25:26.410 --> 00:25:28.589 course like that taught by a foundational figure, 00:25:28.809 --> 00:25:31.450 someone who can give an authoritative, non -technical 00:25:31.450 --> 00:25:34.730 explanation, it's just invaluable. And the course 00:25:34.730 --> 00:25:37.200 material isn't static either. It's not like he's 00:25:37.200 --> 00:25:39.380 been giving the same lectures since 2000. Far 00:25:39.380 --> 00:25:41.980 from it. The sources note that while the core 00:25:41.980 --> 00:25:43.859 structure, the fundamental principles of computing 00:25:43.859 --> 00:25:46.400 stays the same, the lectures and the examples 00:25:46.400 --> 00:25:48.900 are updated every single year according to what's 00:25:48.900 --> 00:25:50.720 happening. So if there's a major data breach 00:25:50.720 --> 00:25:54.059 or a new development in AI. He's weaving it in. 00:25:54.119 --> 00:25:56.539 He's constantly relating these decades -old fundamentals 00:25:56.539 --> 00:25:59.539 back to the immediate contemporary world. It 00:25:59.539 --> 00:26:01.740 keeps the material grounded and relevant. And 00:26:01.740 --> 00:26:04.099 his influence as an educator extends far beyond 00:26:04.099 --> 00:26:06.619 Princeton, reaching millions through a connection 00:26:06.619 --> 00:26:09.019 to probably the most famous introductory course 00:26:09.019 --> 00:26:12.400 in the world, Harvard's CS50. This connection 00:26:12.400 --> 00:26:15.539 is really fascinating. It links Kernighan's teaching 00:26:15.539 --> 00:26:18.539 style directly to the modern era of massive online 00:26:18.539 --> 00:26:22.039 learning. So back in 1996, Kernighan actually 00:26:22.039 --> 00:26:25.480 taught CS50 at Harvard. And why is that specific 00:26:25.480 --> 00:26:28.160 year so critical? Because one of the people who 00:26:28.160 --> 00:26:29.920 was deeply familiar with his teaching approach 00:26:29.920 --> 00:26:33.019 then was David J. Mallon. Millon later took over 00:26:33.019 --> 00:26:36.140 teaching CS50 and built it into the global phenomenon 00:26:36.140 --> 00:26:38.720 it is today. The one that's offered online for 00:26:38.720 --> 00:26:41.880 free to millions. The very same. It runs at multiple 00:26:41.880 --> 00:26:43.960 universities. It's become the de facto standard 00:26:43.960 --> 00:26:46.160 for introducing people to computational thought. 00:26:46.359 --> 00:26:49.119 And Kernighan helped define the pedagogical DNA 00:26:49.119 --> 00:26:51.920 of that course. So it's a direct continuation 00:26:51.920 --> 00:26:55.079 of that synthesizer and popularizer role we saw 00:26:55.079 --> 00:26:57.440 at Bell Labs, just applied now to mass education. 00:26:57.960 --> 00:27:00.160 Exactly. He defined the model for how to teach 00:27:00.160 --> 00:27:02.299 the fundamentals in an effective, engaging way. 00:27:02.440 --> 00:27:04.960 And of course, his contributions have been formally 00:27:04.960 --> 00:27:07.440 recognized. He was elected to the National Academy 00:27:07.440 --> 00:27:10.440 of Engineering in 2002 and the American Academy 00:27:10.440 --> 00:27:13.630 of Arts and Sciences in 2019. These honors really 00:27:13.630 --> 00:27:16.630 underscore that his work transcends any one technology. 00:27:17.049 --> 00:27:19.950 It's about engineering methodology, about establishing 00:27:19.950 --> 00:27:22.109 foundational principles. And we have to talk 00:27:22.109 --> 00:27:24.450 about his later books. His bibliography is practically 00:27:24.450 --> 00:27:27.470 a history of computing in itself. Right. After 00:27:27.470 --> 00:27:29.910 the early foundational books, he had this very 00:27:29.910 --> 00:27:32.769 fruitful collaboration with Rob Pike. another 00:27:32.769 --> 00:27:35.490 unix giant they wrote the unix program and environment 00:27:35.490 --> 00:27:38.630 in 84 and then the practice of programming in 00:27:38.630 --> 00:27:41.690 1999 and he remained so current i mean think 00:27:41.690 --> 00:27:45.930 about this in 2015 he co -authored the go programming 00:27:45.930 --> 00:27:48.710 language go the modern systems language developed 00:27:48.710 --> 00:27:50.869 to google developed by rob pike and others yes 00:27:51.400 --> 00:27:53.440 The fact that Kernighan was involved in writing 00:27:53.440 --> 00:27:55.819 the standard text for a brand new modern systems 00:27:55.819 --> 00:27:58.619 language just shows his enduring relevance. But 00:27:58.619 --> 00:28:01.079 his focus also shifted dramatically toward explaining 00:28:01.079 --> 00:28:03.700 technology to non -specialists, reflecting that 00:28:03.700 --> 00:28:05.680 teaching role at Princeton. His later books are 00:28:05.680 --> 00:28:08.339 all about digital literacy, D is for Digital 00:28:08.339 --> 00:28:11.319 in 2011, Understanding the Digital World in 2017, 00:28:11.619 --> 00:28:14.180 and one of my favorites, Millions, Billions, 00:28:14.200 --> 00:28:16.619 Zillions, Defending Yourself in a World of Too 00:28:16.619 --> 00:28:19.740 Many Numbers. These books really cement his role 00:28:19.740 --> 00:28:22.609 as a publican. intellectual. He's trying to demystify 00:28:22.609 --> 00:28:25.849 data and technology for everyone. He's using 00:28:25.849 --> 00:28:28.289 his authority not just to teach programmers how 00:28:28.289 --> 00:28:31.049 to code, but to teach everyone else how to critically 00:28:31.049 --> 00:28:34.250 understand the world that code created. And then, 00:28:34.269 --> 00:28:36.650 in a fitting culmination, he brought the story 00:28:36.650 --> 00:28:41.670 full circle. In 2019, he published UNAX, a history 00:28:41.670 --> 00:28:44.210 and a memoir. His own personal account from the 00:28:44.210 --> 00:28:46.390 epicenter. It's the final piece of the puzzle. 00:28:46.529 --> 00:28:48.789 The quiet architect reflecting on the massive 00:28:48.789 --> 00:28:51.549 structure he helped name, define, and popularize. 00:28:51.809 --> 00:28:54.529 It's a body of work that spans five decades and 00:28:54.529 --> 00:28:57.089 covers every single facet of computational practice. 00:28:57.470 --> 00:28:59.650 So if we synthesize this deep dive for you, the 00:28:59.650 --> 00:29:01.789 listener, we started with the birth of Unix and 00:29:01.789 --> 00:29:04.329 C. We moved through foundational algorithms like 00:29:04.329 --> 00:29:06.829 Kerning and Lin, explored that critical programming 00:29:06.829 --> 00:29:09.309 philosophy from the software tool series, and 00:29:09.309 --> 00:29:11.069 then we examined his incredible influence on 00:29:11.069 --> 00:29:13.950 education. He defined the tools, the standards, 00:29:14.029 --> 00:29:15.849 and really the style of thinking that enabled 00:29:15.849 --> 00:29:19.109 the internet age. Kernighan's legacy really exemplifies 00:29:19.109 --> 00:29:22.470 a crucial truth in technology, true foundational 00:29:22.470 --> 00:29:25.569 contributions. They're often not in the biggest, 00:29:25.730 --> 00:29:28.130 flashiest breakthroughs. They're found in the 00:29:28.130 --> 00:29:30.470 clarity of communication, the strength of methodology, 00:29:30.910 --> 00:29:34.329 and the elegance of small, reusable tools. He 00:29:34.329 --> 00:29:36.950 is the ultimate expert guide for the field. And 00:29:36.950 --> 00:29:39.529 the thing that stands out most to me is the commitment. 00:29:40.329 --> 00:29:42.789 That constant curation and maintenance. It's 00:29:42.789 --> 00:29:44.829 the single most enduring statement about the 00:29:44.829 --> 00:29:47.329 value of simplicity and structure. I mean, just 00:29:47.329 --> 00:29:49.390 look at the key facts again. The person who helped 00:29:49.390 --> 00:29:51.470 invent the concept of the Unix operating system 00:29:51.470 --> 00:29:54.910 and wrote the definitive book on C is, right 00:29:54.910 --> 00:29:58.009 now, in his early 80s, still actively fixing 00:29:58.009 --> 00:30:01.029 and improving a 45 -year -old programming language, 00:30:01.230 --> 00:30:04.210 AWK. So what does that tell us about the current 00:30:04.210 --> 00:30:06.289 state of software? Well, it raises a really important 00:30:06.289 --> 00:30:09.410 question for modern developers and users. What 00:30:09.410 --> 00:30:11.529 does it mean when these fundamental systems that 00:30:11.529 --> 00:30:14.829 started it all, like AWK, still possess an elegance 00:30:14.829 --> 00:30:17.730 and reliability that our modern, often bloated, 00:30:17.789 --> 00:30:20.329 incredibly complex frameworks struggle to match? 00:30:20.609 --> 00:30:23.130 It suggests the guiding principle, small tools 00:30:23.130 --> 00:30:25.809 doing one thing well, has a longevity that just 00:30:25.809 --> 00:30:28.910 strips rapid iteration and massive scale. It 00:30:28.910 --> 00:30:31.440 really does. And finally... Building on his brilliant 00:30:31.440 --> 00:30:34.740 YSI OAG critique, what you see is all you get 00:30:34.740 --> 00:30:36.539 here is something for you to mull on this week. 00:30:36.920 --> 00:30:39.759 Consider the tools you use every day, your document 00:30:39.759 --> 00:30:42.279 editor, your web framework, your CRM system. 00:30:42.420 --> 00:30:45.579 What vital piece of structural information or 00:30:45.579 --> 00:30:48.039 semantic definition or just underlying clarity 00:30:48.039 --> 00:30:50.720 might you be missing in your daily work? What 00:30:50.720 --> 00:30:53.140 might have been thrown away by that modern, glossy 00:30:53.140 --> 00:30:57.000 WSIYG system just to prioritize immediate visual 00:30:57.000 --> 00:31:00.140 appearance over long -term utility and flexibility? 00:31:00.559 --> 00:31:03.240 Are you sacrificing future portability for immediate 00:31:03.240 --> 00:31:05.730 ease of use? That's a powerful thought. Think 00:31:05.730 --> 00:31:07.769 about that tradeoff, that fundamental choice 00:31:07.769 --> 00:31:10.710 between structure and veneer. That is the essence 00:31:10.710 --> 00:31:13.029 of Brian Kernighan's philosophy. That's it for 00:31:13.029 --> 00:31:14.470 this deep dive. We'll see you next time.