Locklin on science

Managerial failings: complification

Posted in Progress by Scott Locklin on February 19, 2022

“To the engineering mind, a state will probably appear decadent in just the degree that there are numbers of inhibitory or uselessly tabulative persons employed to interfere with, and inquire into the actions of others” -Ezra Pound, Machine Art


The managerial class in America is failing everywhere; it’s obvious, demoralizing and a  dangerous moment in history. It is the same way mandarinates have failed through all of history. Mandinarates don’t necessarily become corrupt in the sense of taking bribes, though this is definitely a factor. More serious is the problem of mental corruption. Mandinarates fail due to complification. Failed mandinarates such as our managerial class can’t simply solve problems any more. They make simple problems into complicated ones and add new and arbitrary problems which literally can’t be solved. The tendency is near universal now, and you can see it and its effects everywhere.

The “managerial elite” of current year aren’t anything like the hard men of old who raised us up from the muck. The actual engineers and managers who pushed the needle lived in contact with the cold reality of matter and existence. They had to do difficult things like milk the cow, bring water up the hill to their homes, deal with bullies, chop wood to prevent their families from freezing, shit in a bucket and carry it to the cesspool. They were mental adults in a way that the present day managerial “elites” mostly can’t be.

Current year managerial “elites” are psychological children who primarily look to others for approval. These are coddled people whose fragile self conception is based on their “cleverness.” They depend on authority figures making their lives comfortable. Like most immature people, they’re also conformists; they wouldn’t dream of diverging from the consensus of experts. Just like kids in school who can’t bear to stand out, a stage in life which they never escape. While there were great innovators who were playful and child-like in their creativity, being praised or feeling clever for being “oh so smart” wasn’t part of their psychological makeup. This tendency is infantile and ineffective at delivering anything useful. It results in outcomes about as helpful as a toddler’s Lego creation, or mummy thinking you are a good boy for getting high marks on a test.

HG Wells saw this psychological tendency back in his day. In his Time Machine novel, he had the childish and degenerate cummies-obsessed “elite” of his day evolve into helpless conformist psychological children; the Eloi. He didn’t anticipate that the Morlocks, aka the people who keep technological society functioning would politely excuse themselves and let the Eloi attempt to run everything. Today’s engineers and managers are almost all worthless Eloi who will eventually destroy everything they’re put in charge of. Eloi engineering and management barbarism is a tremendous danger. If it isn’t stopped, we’re rapidly approaching a time where technological society decays irretrievably, if we haven’t already gone over that waterfall. Eloi are (rightly) terrified of Morlocks, but they’re only dimly aware of the fact that they depend on Morlocks for every facet of their existence.


Consider running a college. In the old days it was a couple of janitors, a couple of accountants and some professors who teach. Now, we have creatures called administrators. These festering pustules have metastasized and made running a college absurdly expensive and difficult.  There are more infantile crackpot administrators than professors in most modern Universities. Mind you, this is a class of “managerial” Eloi bureaucrat that didn’t exist 80 years ago. None of these dingbats can justify their existence. Thanks to Pournelle’s Iron Law of Bureaucracy, these childish meshugganahs have run the educational system into a ditch. Adding them to the College headcount doesn’t add to the institution’s effectiveness; it does the opposite: it makes it more expensive, vastly more ideological and less oriented towards imparting anything resembling knowledge. Headcounts are absolutely exploding in Universities in every category but the people who actually teach things.

Yale for example: more administrators than undergraduates. This is ridiculous; Yale students would be better off if they hired each undergraduate a PhD educated personal tutor and a maid/servant, and it would be cheaper. There is a Yale administrator event horizon at which the mass of administrators at Yale within the confines of the Yale campus will form a black hole from which light cannot escape. If current trends continue, this will happen by the year 3622.

Administrator headcount at Yale doubles every 20 years and there are 5000 of these slugs now. I will approximate their mass (including laptop and papers) at 100kg each; 5×10^5kg. Yale occupies about a square kilometre. A black hole of that size is 0.3 or so solar masses, about 5×10^29kg. Log2 of 10^24 is 80. 80*20=1600 years from now. Long before that, each new administrator added will simply have their electrons and protons fuse together into a neutron blob. This will improve both the productivity of the administrator and the value of the local New Haven real estate by bathing it in lethal radiation. A mere 1200 years from now  the mass of Yale administrators will exceed the mass of the rest of the planet. The present population of the earth will be matched by Yale administrators in 400 years, and the population of South Korea matched by Eloi administrators by the time Yale is twice its present age of 320 years. Lux et veritas.

Yale class of 3622 as the gravitational pull of administrators causes light to orbit New Haven

The “smart city” is a favorite concept of the TED talk/Davos class of Eloi managerial imbeciles.  It is, of course, hilarious that these self proclaimed “elites” have themselves destroyed numerous Western cities already. This is the core philosophy of our parasitic Eloi “experts” -they create tremendous problems with their complexification, then present us with ever more byzantine, intrusive, complicated and oppressive “solutions” requiring their intervention. Western civilization had the most glorious cities in the world for thousands of years. No dystopian hellscape IoT surveillance baloney was required; clean water, strong families, sound money, religion and good civic culture was sufficient. Dystopian hellscape surveillance frippery won’t help the problems we have in current-year cities, which can be summarized as a refusal to enforce any laws or require standards of behavior of any kind; an entirely Eloi concept.

Eloi software goons and engineers compexify things for themselves. Generally for no observable reason. Four hour long meetings instead of one 5 minute standup. Using some retarded object oriented thing to access a database instead of pasting SQL strings together. Using computers where a two wire feedback loop would do. Adding an operating system and wifi to a refrigerator. Most of these chuckleheads have no idea why they’re making things more complicated. Often because they’re so far from the actual enterprise they think they’re being paid to make really sweet solutions that show off how smart they are rather than products which function properly.

Consider the second parity multisig wallet failure. The parity multisig was a multi-authenticated cryptocurrency wallet on Ethereum. You need 2/3, 3/5, 4/5 signatures to make it send funds somewhere. Great use of “smart contracts” (don’t get me started on this simpering descriptor; hence referred to as SCS) assuming it isn’t built into your chain from the get go, which it should have been. This adds commercial bank functionality to otherwise badly designed blockchains which didn’t think of adding this as a core feature. However, this thing was complicated and apparently poorly understood. Some Eloi genius added a “feature.” This “feature” was deploying a “library” which all parity multisigs  depended on. The justification for this was to make it a few dollars cheaper to deploy your personal multisig. “Let’s be modern and use shared objects in blockchain just like the big boys in OS design.”

Well, in return for this twee complification, a complification for which there was no sane justification, the library became a central point of failure. This central point of failure had a self-destruct function that made a sort of deranged sense when the multisig was originally to be deployed as a monolithic smart contract. There’s no reason to have this method, but at least you could only kill your monolithic multisig if the owners agreed on it. When this method was pulled out into a library it had no owners; if it did have owners, it would be a security issue (the concept of “owners” was another complification which was completely unnecessary). Some curious person called the self destruct method they left hanging around in the library and destroyed the $2 billion+ in value locked up in all the parity multisigs. All to save a couple of bucks. Well played, sirs.

At the time this “use a library in a SCS”  idea was very actively pursued. People even gabbling on about writing SCS operating systems. SCS operating systems that could be upgraded if people voted to do so. In fact at one point there was the very bad idea of making the upgrades to your dependencies mandatory.  The reason people use shared libraries in normal computers is you’re worried about it gobbling up too much memory, and it makes it easier to upgrade buggy things without upgrading all the binary dependencies.  Worrying about this on a blockchain is the worst kind of premature optimization. You’re already doing something incredibly wasteful by writing code on a blockchain: own it. Adding dependency graph management to something which is designed to be immutable is barking lunacy.


Why I invested in Immunefi instead of MakerDao

The shared object concept itself is a towering failure. This is little appreciated but undeniably true. The idea of the shared object is simple enough: if you have a computer running lots of code, some of the code used will be the same. Why not just load it to memory once and share that memory at runtime? I’m old enough to remember when this happened to Unix style operating systems back in the 90s. Before that, you’d compile binaries which  contained everything you need statically linked. That actually worked rather well, and allowed you to do things like ship a binary that worked on different versions of the operating system. Mind you, back then, most computers were 32 bit and something like a quarter gig of RAM was considered an enormous extravagence. You could boot a real operating system using a 1.44mb floppy disk. People back then were interested in squeezing a little bit more RAM by using old Multics tricks from the days when “RAM” was a bunch of little magnetic cores knit by old ladies.  When people invented shared objects back in the 1960s, the computer was a giant, rare thing ministered to by a priesthood: there was no such thing as multiple versions of a shared object. You used what the mainframe vendor sold you. Now, when every half-human shambling ape-man in creation has multiple computers of varying vintage, to say nothing of the infrastructure depending on old computers, shared object versioning is an enormous problem. Nobody really thought of this in the 90s when Current Year shared objects were dreamt up and deemed futuristic.

It’s now such an enormous problem there are multiple billion dollar startups for  technologies for dealing with this complexity by adding further complexity. Docker, Kubernetes, various Amazon atrocities for dealing with Docker and Kubernetes and their competitors,  flatpak, conda, AppImages, macports, brew, RPM, NixOS, dpkg/apt, VirtualBox, pacman, Yum, SnapCraft, various app stores and associated applications of varying degrees of completeness, or for just one toolchain (NPM, Rust Crates, go get, OPAM, CRAN, CloJars, Maven, etc etc). The package manager tools that come with the OS generally push the shared object dependency hell back on …. the OS developer. If they support your application and write helpful test scripts, it will get built and you’ll have your application. Good luck getting them to upgrade it if you need a new feature.  As the number of packages grows, this breaks down, and even the OS maintainers are giving up and turning to flatpak, AppImages and Snap files. These are extremely complicated and incredibly wasteful (of memory) ways of literally packaging up a bunch of needed shared libraries with your application and presenting it to you as a crappy simulacrum of a statically compiled binary. Which is a lot more retarded than building a static binary. Mention compiling a static binary to an Eloi developer and you’ll be met with a doltish NPC stare, at which time you’ll be regaled with excuses for why Snap packages or flatpaks or whatever represent the future because reasons .

At the other end of the shared object spaghetti monkeypatch spectrum, you have shitty but simple virtual operating systems like docker. These push the complexity of shared object versioning onto the user. You now have to administer multiple shitty little operating systems, which sort of obviates the concept of having an operating system on your computer in the first place.  Things like conda or brew are vendored mini OS solutions concentrating on making some tool chain run usefully without breaking the OS. They only exist because people want to use shared objects instead of just shipping a statically linked binary. Shipping a statically linked binary; so “early 90s.” What’s the matter with you? Are you a Moorlock?

All of these “solutions” are incredibly complicated and require someone do a lot of work to deal with shared object library issues. Which are only there because in 1995 memory was kind of expensive. Complification. Even if I were nominated Lord of All Computers after the inevitable Butlerian Jihad, I’d have to kill most surviving “software engineers” to fix this, as they’re almost all imbecile complexifying, “playing in the gardens they don’t maintain,” Eloi.

I’m sure some dolt will pipe up and attempt to justify the parlous state of the shared object spaghetti which makes up contemporary operating system design. Or worse, defend one of the monkeypatch things people have cobbled together to make things work. Such people  can’t conceive of a world different from the one they occupy. A world where their “tech skills” are useless because humanity actually made progress. A world where we made progress by remembering something from the past. Complification is a choice, and we can choose to make things better. Eloi never do; they’re too busy playing with themselves.

Eloi complification brings ruin even in little ways. Consider the open source “code of conduct.” The idea of open source is pretty simple; share the code with the world so it doesn’t go away when a company goes out of business. Everyone can fix bugs and share code and make things better. These days you can’t just do that, you need a “code of conduct.” A complicated system involving ad-hoc courts, witch hunts and catering to the whims of emotionally turbulent “danger-hair” dunderheads rather than just ignoring jerks. Somehow people were able to develop significant software before the mid-2010s when this sort of bureaucratic nonsense became a thing. The glorious benefits of not hurting the feelings of debilitatingly fragile contributors haven’t manifested in any observable way.  It’s abundantly obvious that Linux Kernel development has gotten shittier and more bug ridden since it sprouted useless witch hunters and began catering to perpetually offended people. I don’t have a time series of number of critical kernel security patches over time, but I used to have the same kernel version on my machine with uptime measured in months. Now I’m lucky to go a week without a hard reboot being required because nobody told Danger Hair Daphne that she’d be better off pursuing a job as an HR administrator than a kernel developer.  Bring back Mean Linus. Morlock code is better code.

The job of management and engineers is ultimately solving the problem, and the problem almost always gets solved faster and better by making things simpler. When you’re given the opportunity to be clever ask yourself what the Morlock would do. Eating the Eloi should be included in your spectrum of possible actions.

76 Responses

Subscribe to comments with RSS.

  1. Brent said, on February 19, 2022 at 9:58 pm

    I was just thinking this morning it would be interesting to see a tree of package managers for a typical LInux distribution – u,e, how much crap gets installed and managed by pip or texlive or whatever after those tools are installed by apt/dpkg. Not that I think there should be one tool to rool them all, but gosh a little consistency would be nice.

    Don’t even get me started on the other stuff.

  2. Brün said, on February 19, 2022 at 11:08 pm

    Kids that study Business Management around my university have the kind of mindset you talked about, spoiled, brattish and imature. Well, their rich parents will get them a good job, so it does not matter to them.

    I study Economics (yes, that sucks), but i hope i won’t turn into a braindead (or bad economist, although i am really fighting against the odds). My parents raised me on a farm, so i got dirty everyday with poo or blood, that gave me some reality checks, good times.

    So, any tips for an economics undergraduate so he doesn’t turn into an Eloi? I just want to have a family, to be happy and to keep my sanity, is this possible in this rotten world?


    • Scott Locklin said, on February 20, 2022 at 2:02 am

      Suffering. Sports the usual path.

      • Brün said, on February 20, 2022 at 1:03 pm

        I have to start doing sports, i’m being too much of a lazy ass when the matter is exercising. The Bill Starr book you recommended looks like a good option for me.

        But, even though i’m just akwardly aware of the state of things nowadays (education, science, economics and society) i just find it really hard to keep on going forward, although i do it everyday.

        Thanks for the tip.

  3. Corsair said, on February 20, 2022 at 12:20 am

    AFAICT there are basically 2 distinct categories of managerial complification

    Type-A complification is essentially productivity enhancements by specialization: good in moderation, but often carried to obscene extremes. Every process, organization, machine has a division of labor, and generally spreading it out improves efficiency at the margin. Just like financial markets or engineering, people naturally tend to apply too much leverage and underestimate correlations between parts when something goes wrong. There’s also a Type-A complexity arms race in private sector winner-take-all games, where you have to constantly outdo competitors in day-to-day efficiency just to stay afloat, and hope the rising tail risks take them out first. Ergo supply chains spread across 6 continents and 20 corporate entities with JIT inventory all the way down.

    Type-B complification is purely parasitical complexity rent-seeking. Lawyers writing opaque laws that only other lawyers can untangle, HR commissars writing new wrongthink codes they can police, Yale administrators creating new administrations to administer administratively, the list goes on. Obviously there are far too many of these people. These types all benefited from American; they collected much of the 1990s-2020 global labor arbitrage spread, as evidenced by this chart: https://twitter.com/pmarca/status/1494894081217945605 . Now thanks to WFH these type-B complifiers are suddenly much more replaceable with the developing-world equivalent, which hopefully will shake things up some, although this probably won’t be good for general social stability.

    • Sprewell said, on February 20, 2022 at 2:12 pm

      Was going to say the same about what you call Type-B complication, that much of this is intentional creation of labyrinthine bureaucracies by bureaucrats so they can steal more for themselves (my only quibble is that you attribute the labor spread in that great chart you linked primarily to globalization, whereas I’d guess automation was more important). Conformity is a great tell for this, bureaucrats flock to it when faced with any challenge, as we saw with the current Covid debacle.

      This is merely another rise in the cycle of bureaucracy that’s been going on for millenia of human civilization, where the success of a culture like the Roman empire or colonial Europe inevitably leads to masses of these human barnacles that bring that society down, with escape pods then leaving on their Mayflower to start over someplace else.

      However, I can’t get too worked up about this current iteration, as most of these organizations they’re taking over are seriously outdated and should be destroyed, whether the university or Big Tech corporations. The Internet presents the next frontier and the way out. While I was hopeful the crypto guys would spearhead this, they seem mostly clueless so far.

      Let me be the one to “pipe up” and point out that shared libraries allow distributing security fixes quickly and efficiently so it’s not the best example, though dll hell is a real problem. I guarantee Scott would be pitching a much bigger hissy fit if he had to download GBs of package updates every time his preferred static binaries had some core component need a security fix. Both static and shared have their pros and cons, as this long analysis points out while actual running some measurements too.

      • asciilifeform said, on February 20, 2022 at 3:58 pm

        > every time his preferred static binaries had some core component need a security fix

        Modest proposal: suppose that every time an 0day is found, the author of the bug were to lose a finger; and the language designer — a limb. Still think “security fixes” would be a regular thing?

        • chiral3 said, on February 20, 2022 at 4:30 pm

          At that point we’d innovate keyboards.

        • Sprewell said, on February 20, 2022 at 4:39 pm

          Heh, there’s no doubt the languages and systems are the root cause here, but I don’t think excessive punishment will help. Software is still in the Turkish building contractor stage of engineering, it’s going to take focused effort by people like you and me to elevate it into a real engineering profession.

          • asciilifeform said, on February 20, 2022 at 6:25 pm

            Punishment — works. And the Turks made at least an elementary effort in the right direction.

            Recall old man Hammurabi:

            If a builder builds a house for a man and does not make its construction firm, and the house which he has built collapses and causes the death of the owner of the house, that builder shall be put to death.
            If it causes the death of the son of the owner of the house, they shall put to death a son of that builder.

            • Sprewell said, on February 20, 2022 at 7:33 pm

              Heh, that quote is a great pull. 😀

              I tend to focus much more on carrots than the stick. The fundamental problem is that we’re forty years into the software revolution and we still don’t have a great business model for software. Gates and Brin certainly made billions off the current flawed models, but the software itself is generally not very good.

              Specifically, most of the packaging tools Scott lists suffer from an open-source tragedy of the commons, while the few paid ones get ever more complex because of the bureaucratic incentives we started off talking about. As I’ve pointed out in earlier comments on other posts, we’re moving toward a blend of open and closed source, which I hope will prove to be the magic middle.

              As for the stick, I definitely agree that we need to start cracking down on white-collar crime much more, which software malfeasance is just one manifestation of. But if you’re serious about long jail terms and the like, while that would work for those carefully considering crossing the line, I question their effectiveness to dissuade the real psychopaths and the wantonly reckless, as I think has been shown for violent crime too. It is emotionally satisfying as punishment or revenge, but I don’t think it’s much of a deterrent for those willing to do real damage.

              Perhaps the best we can do is throw the book at the building inspectors who looked the other way to abet those reckless contractors, knowing some contractors are just going to go off the rails from time to time.

              • Verisimilitude said, on February 20, 2022 at 9:01 pm

                It’s unacceptable that software as we know it has existed for at least seventy years, but has effectively been destroyed by corporations and other bands of fools. I agree with asciilifeform, but personally believe the people responsible should be buried up to their necks in ant mounds.

                Society depends on computers, and they must work correctly, but now the very idea that software should be correct in all cases is considered odd. I don’t recall the old priest classes of societies being quite so utterly worthless, because at least they could make people feel better about things, whereas now some code artisans, particularly of the C language dipshit ilk, are proud of their incomprehensible code that does so little and doesn’t work anyway.

                • Sprewell said, on February 21, 2022 at 6:15 pm

                  Well, I mean architecture has been around for millenia, didn’t stop those building contractors I linked from wantonly risking people’s lives. You can dream up all the tortures you want, I don’t think they faze such psychopaths, as there were reports of lynch mobs hunting them after the earthquake. Better to focus on methods to verify their work, which keep getting cheaper.

                  As for software being correct, how can it be when the hardware is often buggy too? The old priests were even more worthless, but what makes the current priests so dangerous is that they can actually deliver on some simple things, so when they inevitably bait and switch, they don’t get called out by a credulous public.

                  C code artisans? That’s a contradiction, though to be fair, most of them would never use such a hipster term. Programmers are starting to replace that old tech with modern languages like Go, Rust, and Swift, all of which do array bounds checking by default, so things are slowly improving.

                  • Verisimilitude said, on February 21, 2022 at 9:42 pm

                    Many of those older buildings, and all of the valid math, still stand. Software is math, so there are no such earthquakes. See Software Does Not Fail.

                    “As for software being correct, how can it be when the hardware is often buggy too?”

                    Don’t give me that. Most “programmers” don’t even try. If the software were only as flawed as the hardware, we’d be in a great position, and we know this.

                    It’s a contradiction, sure, but it’s not unfitting; instead of a coffee shop, they’re more likely to be found in a toy shop. Ada has existed for decades, and actually focusses on correctness, unlike those C-flavoured ALGOL clones.

                    “so things are slowly improving.”

                    Have a joke or other such thing I’ve written: “The only way to write good software is to have it be correct and able to be fast and then have it made to actually be fast. Most software isn’t correct, but is able to be fast, and is then made to be slow in an infinitesimal stepping towards correctness.”

                    • Sprewell said, on February 22, 2022 at 8:22 am

                      > Many of those older buildings… still stand.

                      Sure, and you can still run old games like Pong on your smartphone if you choose, your point is? My point was that even in fairly well-understood and much older fields like architecture, the age of the profession is irrelevant because there’s always new monkeys out there who will screw it up.

                      > all of the valid math

                      Bit of a tautology isn’t it, what about all the non-valid math? 😉 Perhaps you haven’t look into it too closely, but there are holes in math all over the place, such as the original and long-used infinitesimal formulation of calculus, which I don’t think the current limit formulation has really solved.

                      > Software is math, so there are no such earthquakes.

                      No, it isn’t math but even if it were, that wouldn’t stop it from being initially used but eventually ditched for better versions, just like math is. Earthquakes hit math all the time too, as our understanding of its uses and limitations are always changing.

                      > See Software Does Not Fail.

                      Started reading it but there were so many wrong statements in his initial premises that I couldn’t go any farther once he started talking about software objects.

                      Software does wear out and decay, precisely because it runs on hardware that he admits wears out and becomes outdated. Saying that it is only the spec that can be wrong, useless, or dangerous is simply defining away the problem, as no user cares about the arbitrary distinction of whether it was the project manager who wrote a bad spec or the programmer who wrote code that didn’t follow that spec. All they have is the software, and it can be any combo of right, useful, incomplete, or safe at any particular time and, since the world and people are constantly changing, will eventually fail all those criteria till the end of time without constant upkeep.

                      Let me present an alternative and much more realistic view of software and technology, from a guy you’ve probably heard of, Linus Torvalds:

                      “You know what the most complex piece of engineering known to man in the
                      whole solar system is?

                      Guess what – it’s not Linux, it’s not Solaris, and it’s not your car.

                      It’s you. And me.

                      And think about how you and me actually came about – not through any
                      complex design.

                      Right. ‘sheer luck’.

                      Well, sheer luck, AND:
                      – free availability and _crosspollination_ through sharing of ‘source
                      code’, although biologists call it DNA.
                      – a rather unforgiving user environment, that happily replaces bad
                      versions of us with better working versions and thus culls the herd
                      (biologists often call this ‘survival of the fittest’)
                      – massive undirected parallel development (‘trial and error’)

                      I’m deadly serious: we humans have _never_ been able to replicate
                      something more complicated than what we ourselves are, yet natural
                      selection did it without even thinking.

                      Don’t underestimate the power of survival of the fittest.

                      And don’t EVER make the mistake that you can design something better than
                      what you get from ruthless massively parallel trial-and-error with a
                      feedback cycle. That’s giving your intelligence _much_ too much credit…

                      And I know better than most that what I envisioned 10 years ago has
                      _nothing_ in common with what Linux is today. There was certainly no
                      premeditated design there.

                      And I will claim that nobody else ‘designed’ Linux any more than I did,
                      and I doubt I’ll have many people disagreeing. It grew. It grew with a lot
                      of mutations – and because the mutations were less than random, they were
                      faster and more directed than alpha-particles in DNA…

                      And I will go further and claim that _no_ major software project that has
                      been successful in a general marketplace (as opposed to niches) has ever
                      gone through those nice lifecycles they tell you about in CompSci classes.
                      Have you _ever_ heard of a project that actually started off with trying
                      to figure out what it should do, a rigorous design phase, and a
                      implementation phase?

                      Dream on.

                      Software evolves. It isn’t designed. The only question is how strictly you
                      _control_ the evolution, and how open you are to external sources of

                      And too much control of the evolution will kill you. Inevitably, and
                      without fail. Always. In biology, and in software…

                      And no, I’m not claiming that the rest is ‘random’. But I _am_ claiming
                      that there is no common goal, and that most development ends up being done
                      for fairly random reasons – one persons particular interest or similar.

                      It’s ‘directed mutation’ on a microscopic level, but there is very little
                      macroscopic direction. There are lots of individuals with some generic
                      feeling about where they want to take the system (and I’m obviously one of
                      them), but in the end we’re all a bunch of people with not very good

                      And that is GOOD.

                      A strong vision and a sure hand sound like good things on paper. It’s just
                      that I have never _ever_ met a technical person (including me) whom I
                      would trust to know what is really the right thing to do in the long run.

                      Too strong a strong vision can kill you – you’ll walk right over the edge,
                      firm in the knowledge of the path in front of you.

                      I’d much rather have ‘brownian motion’, where a lot of microscopic
                      directed improvements end up pushing the system slowly in a direction that
                      none of the individual developers really had the vision to see on their

                      And I’m a firm believer that in order for this to work _well_, you have to
                      have a development group that is fairly strange and random…

                      You’re giving the human mind too much credit, and giving too little credit
                      to the selection process.

                      The selection process (whether in evolution or in ‘intelligent design’) is
                      what creates the design in the end. The only thing intelligence does is to
                      try to cut down on the cost of trial. And that exists in evolution too: why
                      do you think people (and animals) find healthy specimens ‘sexier’ than sick
                      ones? It’s all just a way to direct the trials so that you don’t get pure
                      randomness and increase the efficiency of the process..

                      And the thing is, most engineering is _not_ ‘intelligent design’ in the way
                      we like to think of ourselves. The number of major scientific and
                      engineering breakthroughs that have happened as the result of outright
                      accidents or just pretty random tinkering is _staggering_. Very few things
                      indeed have been ‘thought out’ by those Intelligent Designers.

                      Humans are better at rationalizing (after the fact) what happens than at
                      actually thinking things through. Which means that every time a discovery
                      or new technique gets invented, it ends up being rationalized as ‘deep
                      thought’, regardless of how the heck it came to be. It’s just another form
                      of the ‘winners’ rewriting history (by being remembered), and all the
                      millions of losers being totally forgotten.

                      But some people have an agenda to think otherwise. Either the religious one,
                      or just the idealization of intelligence…

                      Darwinian evolution ends up able to create more complex things ‘by accident’
                      than we currently can really even envision to design, and may never be able
                      to design as ‘engineers’ in the sense that we understand everything that is
                      going on. Evolution can (and does – very naturally) take into account
                      emegent behaviour that was unintentional and comes about from secondary or
                      tertiary effects of some particular design decision.

                      If you don’t see technology as a primal jungle of competing species, you’re
                      not getting the picture.”

                      I highly recommend you read this entire set of emails from which I’ve just snipped some highlights, as it was written to open the eyes of someone precisely like you. You want to elevate software to some platonic ideal, which you believe math represents, but the reality is that it’s all contingent and bug-ridden and thus eligible to be swept away by change.

                      He probably goes a bit too far in elevating evolution over design, but there is no question all technology, indeed human society itself, is a complex interplay of the two.

                      > Most “programmers” don’t even try. If the software were only as flawed as the hardware, we’d be in a great position, and we know this.

                      Why should most programmers try? Most are writing trivialities like Quake 57 or javascript modules for Instagram, hardly things to care much about.

                      I agree that those dealing with more important software haven’t tried hard enough, but that just leaves an opening for others to pass them by. Those are pretty bad hardware bugs I linked, so I disagree that that’s a “great position,” but you’re probably right that software is usually worse.

                      Your joke rings true, but if the market values speed over correctness, I don’t blame them for obliging.

                    • Verisimilitude said, on February 22, 2022 at 5:08 pm

                      I can’t reply normally now, due to this comment system:

                      “Software does wear out and decay”

                      I don’t see how I can continue a discussion with someone who believes things which are so clearly not true.

                      “Saying that it is only the spec that can be wrong, useless, or dangerous is simply defining away the problem”

                      Consider actually reading the article rather than making false claims about it.

                      I don’t respect Linus Torvalds, an intelligent idiot who managed to get plenty of people behind his terrible software. His disrespect to RMS means I don’t really feel sympathy now that he’s been eaten alive by the crowd. He’s proud of “his” kernel that does so little with so many millions of lines, rather than ashamed he doesn’t do it with fewer. They’ve built their shitty little tower, and haven’t questioned it in all these decades.

                      I believe in IA, Intelligence Amplification, rather than AI, Artificial Intelligence. Linus may praise mankind however much, but someone with a man-made gun is still better off, the same way a man with language and math is; he should feel free to leave the Internet for what I figure is a primitive lifestyle suiting some kinds of people.

                      Amusingly, academics do use evolution-like methods in some automatic program synthesis, rather than simulating it with C language programmers writing shitty software.

                    • Sprewell said, on February 23, 2022 at 8:55 am

                      >> Software does wear out and decay

                      > I don’t see how I can continue a discussion with someone who believes things which are so clearly not true.

                      I explained what I meant by that, which you conveniently snipped out. Since you need more explanation, it “wears out” because the hardware it was written for becomes obsolete, so nobody runs an old app like WordStar any longer. Even though you could still run it on an emulator on new hardware, nobody is going to bother because our expectations and needs have changed since that early software, so it “decays” in that its old features don’t fit new times.

                      You can keep an old car like a 1920’s Duesenberg running as it was when it was newly made, but nobody is going to want to drive it on modern roads today, other than a few nostalgic car enthusiasts.

                      > Consider actually reading the article rather than making false claims about it.

                      I was summarizing this moronic passage from your link:

                      “The specification for either hardware or software is abstract and therefore soft. The specification does not do what software does, however. So the specification cannot be said to work or not work.

                      The specification, which is derived from functional requirements, can be dichotomized many ways:

                      – right or wrong,
                      – complete or incomplete,
                      – appropriate or inappropriate,
                      – useful or useless,
                      – safe or dangerous.

                      To go from one to another, the specification itself would need to be ‘revised.’ Or the functional requirements would need to be ‘amended.’

                      Only one dichotomy applies to software: either it works or it does not work, which is determined by whether the software complies with the specification…

                      For the record, then, complying with a specification that is…
                      – wrong,
                      – incomplete,
                      – inappropriate,
                      – useless, or
                      – dangerous
                      …does not mean failure of either hardware or software.”

                      I summarized this as “Saying that it is only the spec that can be wrong, useless, or dangerous is simply defining away the problem, as no user cares about the arbitrary distinction of whether it was the project manager who wrote a bad spec or the programmer who wrote code that didn’t follow that spec. All they have is the software, and it can be any combo of right, useful, incomplete, or safe at any particular time”

                      If you have an alternate interpretation of what he wrote, go ahead and provide it, else you’re wrong to say I was “making false claims about it.”

                      > I don’t respect Linus Torvalds, an intelligent idiot who managed to get plenty of people behind his terrible software.

                      You don’t have to respect him, but he has written and now manages the most widely deployed software ever in the linux kernel- largely open source, so much more successful than RMS- so there are things you can learn from him. You haven’t read what he wrote, as he explicitly says it’s not “‘his’ kernel.”

                      I agree with you that one could design a much better kernel, but you ignore or downplay his emphasis on evolution too at your peril.

                      > I believe in IA, Intelligence Amplification, rather than AI, Artificial Intelligence. Linus may praise mankind however much, but someone with a man-made gun is still better off, the same way a man with language and math is; he should feel free to leave the Internet for what I figure is a primitive lifestyle suiting some kinds of people.

                      I don’t know what triggered this, I don’t think Linus is involved with the current “AI” bubble. If you think Linus is against guns or language, you’re completely misreading what he wrote, or based on your earlier mistake, not really reading it all.

                    • anonymous said, on February 26, 2022 at 8:27 pm

                      “And I will go further and claim that _no_ major software project that has
                      been successful in a general marketplace (as opposed to niches) has ever
                      gone through those nice lifecycles they tell you about in CompSci classes.
                      Have you _ever_ heard of a project that actually started off with trying
                      to figure out what it should do, a rigorous design phase, and a
                      implementation phase?”

                      In the hardware world, the James Webb space telescope seems like it might end up being a happy counterexample. I’m honestly very surprised, since the “design precarious complexity straight out of Platonic space into not-very-testable hardware” is a design philosophy that I regard as brittle and untrustworthy. Happily, it seems not to have bit us in the ass this time.

                    • anonymous said, on February 26, 2022 at 8:45 pm

                      “I don’t respect Linus Torvalds, an intelligent idiot who managed to get plenty of people behind his terrible software. His disrespect to RMS means I don’t really feel sympathy now that he’s been eaten alive by the crowd. He’s proud of “his” kernel that does so little with so many millions of lines, rather than ashamed he doesn’t do it with fewer. They’ve built their shitty little tower, and haven’t questioned it in all these decades.”

                      RMS got eaten by the same crowd that Torvalds did. The guy who worked on his kernel security, and cryptography was also being targeted by this Ade-Ehmke creature. As far as I can tell, Torvalds also has a lot of very nasty family pressures on him that lead to him (temporarily) caving. (In an old old biography I read of the guy, he has some pretty ideologically fanatical commies in his family, which in Finland is really saying something.)

                      Rather than getting pissed off at these accomplished people who never should have been subjected to this witch-hunting lunacy, we should be pissed off at the witch-hunting lunatics. No one who actually ships a product (whether we agree with their design philosophy or not) should ever have to answer to the likes of some CoC red guard.

                      As for Linus’s insistence on C: After working with an “architecture astronaut” a few years back, I get it. You can’t have an architecture holy crusade in a language that simply doesn’t facilitate architecture crusades. Also, C forces a very pedantic focus on what the hardware is doing, and gives you handles to manipulate the hardware which is what you honestly need in something like a kernel, which deals with the hardware. The kernel guys do have to care in at least general terms what the processor is actually doing, what the stack looks like, etc.

                    • anonymous said, on February 26, 2022 at 9:07 pm

                      I used to work with someone who, while the rest of the team was a little exasperated by some of his personality, worked in a way that I appreciate more and more with time. He was an older guy who used his “set-in-his-ways” persona to sidestep and ignore a lot of interfering bureaucratic noise and nonsense. He wrote a lot of scientific simulation software for our team. He would talk about what he was doing in the weekly meetings, and then work and ignore all the JIRA nonsense throughout the week. He would not reference “user stories”. He would not make a thrice-daily update on his Jira tickets, or remember what their numbers were supposed to be. And he would deliver his software … in Fortran. If you wanted his computational reletavistic electromagnetics code you would take delivery in idomatic Fortran-90, which doesn’t care about your style-guide, and doesn’t facilitate lambda-functors, or whatever the flavor-of-the-week was. He would email it to you in a tarball whenever you wanted the latest version, and you could post it to the secure sharepoint if you wanted to. I would wrap all that stuff in various layers for the various users so they could call it from matlab.

                      I almost aspire to be that guy. 😛 He was very productive because he could (and could get away with) ignoring a lot of irrelevant-to-the-function micromanagement.

            • chiral3 said, on February 20, 2022 at 8:16 pm

              ascii I see your point but disagree that it would work in modernity. Punishment needs to be linked to a rule – if this than that. In Hammurabi’s time that would fit on a stele. Today it would take 2000 pages and contain covenants and cross-covenants making the actual offense impossible to not argue around. Scott used a word I like – “clever” – there’s a bunch of clever people out there. What happens is the more specific the rule the more easy it is to break it and avoid punishment, especially as it applies to corporations and government, the denizens of which are the managerial class.

              Case(s) in point: the financial crisis is maybe 70% the result of regulation that were very rules-based (the Basel capital accords). Nobody was punished. Fast forward to the Libor crisis and the person singled out and punished was a low-level retard, but not one other person was punished other than this spergo. Goldiestix and 1MDB? Iraq? Ruby Ridge? Waco? Not one person in government or business punished (ok, goldie stix got a fine). Regulation, explicitly worded post crisis, facilitated one of the largest migrations of assets to offshore entities for the sole purpose of tax and regulatory arbitrage. This transferred vast sums from the public to the private space. These people are clever. They have the best lawyers. A head of credit I know made about 50 sticks last year. These people are making 40, 50, 60 million dollars a year not even standing up business but re-investing in leveraged assets that aren’t called CDOs anymore but function the same. This is created by the government and run by corporations. The money largely comes from retirement, taxes, insurance, and public shareholders. When this implodes, and it will, the asshole I know that made $50M last year will probably be worth $200M and you’ll never hear his name. But his grandkids will go to Yale. Rules are completely interpretive these days which is why punishment is really reserved for the dumb and poor.

              So as it relates to the managerial class I really can’t imagine anything within the boundaries of the system fixing the issue. It’ll take a violent revolution.

              • Verisimilitude said, on February 20, 2022 at 9:04 pm

                The solution to vermin working around rules is to have a society in which people interpret vague rules. The “Do no harm” is simple, although still isn’t followed. The solution to pilpul has always been killing and exile.

                Now, some may claim philosophical nonsense means humans can never really be certain that others understand. Such “people” are an acceptable loss.

              • T said, on February 20, 2022 at 9:43 pm

                Reminds me of the thoughts I had when studying for the FINRA SERIES exams. Every rule/regulation was supposed to protect the little guy, but the end result was it took opportunities away from them(for fear they’re too stupid to know what they’re doing) and only allowed the big players to partake. The same things play out with the regulators trying to stop retail options trading (https://seekingalpha.com/news/3778492-the-risk-of-regulators-restricting-options-trading-is-rising-and-that-could-be-a-problem-for-robinhood-and-others), but allow the FED to print unlimited amounts of money.

                • Scott Locklin said, on February 20, 2022 at 11:03 pm

                  Its one of the reasons crapto thrives. Crowdsales were an run around rules on money raises put in place after 99. The rest is inventing an unregulated parallel financial system. The technology is just a totem: it’s actually causing more problems than it solves, but the mystification of my technology allows it to continue to exist while gibbering numskull Eloi regulators like Clayton stick their hands out like the bum on the street corner.

                  • chiral3 said, on February 21, 2022 at 3:52 am

                    Re the exams, they are interesting. I’ve done some. I hardly have the alphanumeric soup many people have after their names. My experience is they are largely memorization of rules and indoctrination. In your example, Finra, it’s ways to make you, and not your firm, culpable. It’s also indoctrination. Fees are good, fees are fair. Active management is better. Complex products with higher fees benefit investors. Etc.

                    Crapto, and I am really not one to talk about it, but it’s like any nascent and inefficient thing: there’s alphas everywhere, if you want to even call it alpha. (Remember when alpha was also called “skill”? As in “the strat returned 300bps, 120bps of skill.”) That’s what reg lite emerging markets have always been. Why trade options in the US when I can trade on the same tech in highly illiquid and mispriced SE Asia, leveraged by US capital? Why do you think half the dry powder in US private markets right now is Saudi linked?

              • Scott Locklin said, on February 20, 2022 at 11:04 pm

                Iron gardener.

                • remnny said, on February 22, 2022 at 6:30 am

                  Your taste is immaculate. God, I love Uncle Boyd.

    • LM said, on February 24, 2022 at 9:43 am

      “There is a Yale administrator event horizon at which the mass of administrators at Yale within the confines of the Yale campus will form a black hole from which light cannot escape. If current trends continue, this will happen by the year 3622.”

      This made my day. Thank you.

  4. Corsair said, on February 20, 2022 at 12:21 am

    *These types all benefited from American deindustrialization

  5. asciilifeform said, on February 20, 2022 at 12:56 am

    FWIW I discovered in 2015 how to repair the static linking which GCC “maintainers” had deliberately sabotaged and have been happily building and deploying statically-linked binaries — without any “blue hair” idiocies whatsoever — ever since.

    But no, you won’t hear about it on e.g. Idiot Combinator or via Retarddit, or via Google search, and there are no prizes for guessing why not.

    Exactly for the same reason you won’t hear about the frozen-for-all-time systemd-free Gentoo, the working 100%-P2P encrypted WOT network, and many other similarly IMHO interesting things.

    • g.b. said, on March 11, 2022 at 11:53 pm

      A link to search results of a log of chats is the least clear possible way of explaining what you found in gcc, and how you get around it.

  6. Verisimilitude said, on February 20, 2022 at 1:39 am

    My first draft of this reply wasn’t good enough, apparently.

    I like how this article bemoans current managers, but not those of the past. Could it be the case that managers have always been unnecessary? It’s certainly the case that a gaggle of managers likely won’t be able to make anything of worth, as this article demonstrates, so it’s clear only the engineers are truly necessary.

    Universities are unnecessary now, if only due to the Internet and wide availability of good books. It doesn’t help that corporations have strangled education into becoming job training, but that’s not brought up here.

    I’m learning Latin, so I understand “Lux et veritas.” means “Light and truth.”, but what’s the purpose of this in the article?

    While these “smart cities” are stupid, how do they meaningfully differ from Fordlândia, that nice Brazilian failure from Henry Ford? A company town is not a new concept.

    Dynamic linking is no evil, and Lisp is an example. I prefer to create my metaphors, so I call Eloi Intelligent Idiots. If anyone ever demanded I have such a code, I’d firmly grip my Code of Conduct Keeping in hand, but since I simply avoid the cesspits, I’m automatically spared from this nonsense. Of course, I don’t get much attention that way, so it’s a trade-off.

    I don’t really feel these stingings, but I’m isolated, which is preferable to me.

    • Scott Locklin said, on February 20, 2022 at 2:07 am

      Lux et veritas is Yale’s motto.
      If you can’t see the difference of Davos hell holes to Fordlandia, well, I guess we can exclude “learning Latin” as a meaningful filter.

      • Verisimilitude said, on February 20, 2022 at 6:31 am

        Well, feel free to educate me about how different Ford telling employees not to drink or do other things in their homes is so very different from a technologically-enhanced version of the same shit, Scott. Have I merely ruined the illusion that these things aren’t solely the results of homosexuals and others, and in fact the mere result of people who have power over others?

        • Scott Locklin said, on February 20, 2022 at 12:41 pm

          You should stick around here as a sort of warning to others.

          • Anonymous said, on February 21, 2022 at 4:56 am

            Versimilitude is basically right. The current elois haven’t appeared out of nowhere with guns blazing supported by an alien expeditionary force. They were gradually let into the ruling class by the old elites. The problem however is deeper: ever since its inception the ruling class wanted to have as many slaves as possible, as obidient as possible. This has been a major success, gotta give it to them. Nowadays when a rare freedom-loving person attracts goverment attention, he’s immediately labelled as a terrorist, goons are dispatched to kill him, and the crowd either supports the goverment or doesn’t meaningfully interfere. And all these ‘pro-freedom’ revolutions, like the one currently happening in Canada or the various color revolutions, are explicitly about swapping a dirty boot for a clean and shiny one, because degeneral public can’t imagine not licking one. The domestication game however became self-defeating with the advent of mass armies. While it is true that a mass army can be more efficient than a professional one, having one puts you in a precarious position because now your slaves know how to fight. So you have to put on some show (democracy) to appease them but that just delays the inevitable when the domesticated start infiltrating the ruling class bringing their subhuman genes with them. I mean, if you’re so vile that breeding obidient subhumans seems like an ok thing to do, then at least maintain hygiene, for fuck’s sake.

        • asciilifeform said, on February 20, 2022 at 3:47 pm

          One obvious difference between the new gulags and the old is that the latter (the Lowell Mills; Ford’s, Stalin’s, etc. hells) actually produced something other than “more studies needed” reports.

          I do however find it difficult to imagine that e.g. Ford wouldn’t have installed cameras in his barracks if period tech had permitted it. Just as Stalin would have been happy to bug every flat, if it had been physically possible with the available tooling.

          Today’s totalitarians are very much ideological descendants of the 20th. c. variety; the operative difference isn’t even the tech, however, but that “the snake has lost its legs” : the new ones are not particularly concerned with building up the physical economy or imperial expansion, but rather with social control for its own sake, and in fact with organized impoverishment as an instrument thereof.

  7. […] on Science discusses one inherent fault in the managerial systems we now live and work under: Complification. I agree, and would say it applies to human organization in workplaces, as well as software […]

  8. seanvanderlee said, on February 20, 2022 at 4:06 pm

    Fine article Scott. It seems particularly relevant timely in light of how the bizarre, to put it mildly, government response in Ottawa. I don’t see an obvious course correction however

    • Scott Locklin said, on February 20, 2022 at 7:08 pm

      Trudeau is living embodiment of Eloi mentality, from his recitation of quantum computing BS like a toddler that shit his pants, to his smug smirk, to the fact that we all know his last name is actually Castro. A friend of mine punched him in the face at college in a brawl. This friend of mine is a hero.

  9. chiral3 said, on February 20, 2022 at 4:16 pm

    F’ing gold Scott, just perfect. I think you need to spitball the many-campus problem, though. Harvard is about a km^2 and a bit over a hundred miles away. A school like NYU has to have at least 1.5x the number of administrators versus Yale. These schools and their ilk would combine early on accelerating all of this much faster. Many excessively woke schools would probably violate a weak energy condition, for obvious reasons; but, assuming it didn’t, the hole would radiate microaggression and the area of the hole would decrease proportionally. Wheeler was wrong, this black hole will have hair.

    I can’t agree more, and I have nothing but hatred for the managerial class you describe. It’s a pregnant class of apparatchiks who’s only skill, apart from being excessively cinematic (called “executive presence” kids, learn it), is that of locking arms with other apparatchiks and sequestering wealth and power for their progeny and co-conspirators. But there’s much to unpack here….

    I’ve experienced the following in the context of software development, strategy development, and operations/systems (trading capital mgmt, etc.) Before I mention specifics, I’ll say that, imo, we need three archetypal proto-layers of players in order to be efficient and be successful in an organization. I’d also argue, using Klein, that this probably applies to Klein-2,3,4 organizations, but not Klein-1, as the mission is so well defined and everyone is mission oriented and bought in to a singular idea. These layers are roughly 1) builders, 2) designers, and the 3) management layer. The builders, well, build. In my world these are quants, engineers, coders; largely PhD-level but also includes the special types that just know how to get shit done. The designer class architects shit: a high-end version of project management. The management layer provides strategic direction (what to do and why), blocks and tackles (secures funding, deals with solutions issues that impact success but aren’t native to the project, deals with other feckless brahmins, regulators, lawyers, and other panjandrums of corporate husbandry). Maybe examples will work better: I am a builder by pedigree. The only reason why I am up to my eyeballs in other things is because I also like to make decisions and have a hard time doing things I think inspire idiots. I also have a psychological revulsion towards inefficiency and systems that aren’t well designed. As such, by my own design, I’ve always had very vertical responsibilities. The issue with builders having autonomy independent from designers and some semblance of a management class is, and I’ve seen this more time than I can count, it doesn’t work out, arguably ever, but it’s admittedly a function of the project’s ambition. By analogy, lets say I am the boss and decide to go from point A to B (this is setting strategic direction, as in #3). Left to their own devices the builder class will set about building an airplane. There will not be any discussion of whether a boat or a train is better. It’s an airplane. Further, nobody thinks about how the airplane should look or perform. Not the payload, the flight consoles, the layout of the cargo hold. Everyone decides to go off and build propellors and then fight over who is smartest. After a year I have a collection of propellors from the coders (in the form of a class that extends a thing called propellor that the quants can use), the experimental physicists (actual propellors), the engineers (more actual propellors of all different sizes and specs); and plans for propellors from the theoretical physicists and mathematicians, who are still fighting over vorticity and reynolds numbers and the solutions to the Korteweg–deVries equation. There’s a war over who’s propellor is better but there’s no airplane. In fact, if we used a designer, a boat would have made more sense. And all this started because someone decided we need to get form A to B.

    In my experience we need all three layers. The problem is that the management layer has become bloated and leveraged, that people exclusively aspire to this layer, and these people (largely) can’t do anything themselves, unfortunately including loading a weapon and eating a bullet. It’s astonishing to think of the arc, at least in the US, from maybe 1950 to today, the role of education, student loans, management consulting, executive compensation, and the population demographics/dynamics. The US has incumbency and balance sheet, and that’ll last a while. We’ve dodged bullet, though. It sounds horrible, and after coming through the pandemic I keep thinking about this, but the Spanish flu wiped out a third of the planet. Primarily kids that never had kids. Many ways to load the calculation but any way you do you have a thought experiment of “what if we came out of the last century with 11B people on earth”. I am not sure the Haber process and penicillin and the internal combustion engine was enough innovation to support that number. Even more unnerving is the thought experiment for the next 100 years.

    Mentioning the Haber process reminded me of something I wrote probably 15 years ago. I always zeroed in on the food to support all these mouths and assholes. It reads sophomoric now, but it always made me laugh. I meant to develop the story more but never got around to it. Here’s a copy-paste of a snippet:

    The Haber process or, more precisely, the Haber-Bosch process, is a high-pressure marriage arrangement between a hydrogen and a nitrogen. Haber had the idea and Bosch figured out how to industrialize it on a large scale. Haber was the brain and Bosch was the company-man. Both were German chemists and both got the Nobel for their work. To Bosch’s credit he had a soul: he hated the Nazi’s so much that, while he was great at climbing the perfectly spaced rungs of the German corporate ladder, he thought the Führer was a genocidal asshole. Wehrmacht GmbH won and Bosch lost. Jobless and full of despair he drank himself to death five years before it all ended, before Mussolini was shot and hung from a meat hook in a piazza in Milan. Five years before Hitler and Braun consummated their union in the Führerbunker. (It’s easy to imagine, as a gesture to help assuage his sexual anxiety, a little red button mounted next to their marriage cot, a wedding gift from the schutzstaffel, linking the Führer’s turgid digit to a battery of wunderwaffe.) Haber, an academic, distinguished himself by greatly advancing chemical warfare during the war. To paraphrase Jünger: “a half-witted [academic] can cause more damage in a second than Frederick the Great in three Silesian campaigns”. Beware the danger a researching academic poses: they can be brilliant careerists, yet dim pragmatists, not realizing the diabolical potential for their constructions.

    So, on one hand, Haber melted people’s faces off and, on the other, he fed them. Maybe in his logical German mind these cancel. After all the death, the miracle of the Haber-Bosch process is it can facilitate the production of an awesome amount of fertilizer precursor that can be used to grow food in soil that would otherwise be depleted of nitrogen by agricultural demand. So large is the benefit that the earth can sustain more people and flatulating animals to feed the people than it could ever naturally support otherwise. Entire industries owe their existence (and salt) to the industrialization of ammonium salt production.

    There’s another way to make ammonium nitrate on a large scale involving nitric acid, but it never got industrial legs. The inventor was another German who, ironically still, was associated with a pacifist movement. He died one year before the fall of Weimar and thirteen years before it all ended. Timing is everything.

    It is worth noting that ammonium nitrate lags another fertilizer in popularity, primarily due to the economics of transportation: urea has more nitrogen per unit of volume than any other compound and was first synthesized by the German chemist Frederick Wöhler in 1928. Another German. They must have had the same inspirational teachers in their respective gymnasiums. Urea nitrate can be used as an explosive. The six individuals that perpetrated the 1993 World Trade Center bombings had handlers that knew this fact. They were reportedly pissed off about US-Israeli sympathies and figured out how to make bang-bang without blowing their dicks’ off. Despite avoiding blowing themselves up their efforts were not without comedy. Mohammed Salameh, the esteemed driver of the truck housing the explosive devices, failed his New Jersey state driver’s test no less than four times leading up to the attack, arguably protracting the timeline by not being able to rent a car with a valid NJ state driver’s license. So mystified was he by the complexity of the modern automobile that he almost killed his passenger and better-known co-conspirator, Ramzi Yousef, just thirty days prior to the attack in a car accident. Then, just two weeks before the attack, he got into another fender bender totaling yet another rental car. Logically, when assessing the team’s talents, one would assume Salameh was deemed the best driver of the lot. Without chariot the team almost had to roll the fertilizer drums by hand to the World Trade Center. In Dinkins’ 1993 New York this might not have raised suspicions in places like Times Square, but the financial district would have taken notice. Khalid Sheik Mohammed, the money and brains behind the operation, perhaps chalked this misfortune up to the price one pays for recruiting the dumb and impressionable. When General Mao wrote of one training ten training a hundred training a thousand… he never said the resistance had to be smart. Only the first guy has to be smart. The rest need to be able-bodied. Smarts are a sufficient but not necessary condition: whether you’re killed by a genius or a dullard doesn’t matter, you’ve lost at that point and all the IQ points in the world aren’t going to bring you back. I often think about the NJ Motor Vehicle Commission employee that finally capitulated on Salameh’s fourth driving attempt. Salameh was finally caught on his second attempt to get his deposit back from the rental company.

    • Scott Locklin said, on February 21, 2022 at 12:01 am

      I’m building something new where I’m all 3 things right now. I’ve been thinking about how to build it without hiring a half a dozen goons who build me a propeller when I ask for a steam traction engine. I’m considering starting a religious cult based around worship of Ken Iverson.
      I think I got it figured out how to hire management: men who played team sports. This used to be a standard feature of American management when American management was still helpful. Maybe I’ll be the architect, dunno. Right now it’s more or less divided between deal makers and innovators, both of whom are me.

      • asciilifeform said, on February 21, 2022 at 2:13 am

        Seriously consider whether you even need to hire anyone. Historically, the best softwares had 1 (occasionally 2-3) authors.

        • Scott Locklin said, on February 21, 2022 at 2:35 am

          Well someone has to deal with hollering at lawyers and exchanges anyway.

      • Anonymous said, on February 23, 2022 at 1:52 am

        >I’m considering starting a religious cult based around worship of Ken Iverson.
        I have personally been planning the creation of the Brotherhood of Steel irl for when our civilization inevitably collapses. We need to preserve our knowledge and technology and not let it fall into the hands of primitive barbarians (which I think has already happened with current-year mandarins). Perhaps we can come together and meld these cults into one…

  10. chiral3 said, on February 20, 2022 at 5:13 pm

    I spent some time ferreting around my shelves and I don’t think I have that Pound? I certainly never read Machine Art. What a great quote. I’ve mentioned this before here: the artists see shit first, are often the most correct futurists, and offer their vision in their books, movies, poems, and paintings.

  11. Igor Bukanov said, on February 20, 2022 at 11:59 pm

    In “The Utopia of Rules” David Graeber argued quite convincingly that US became rather bureaucratic already in fifties. But it was OK since most bureaucrats were in the private sector in the form of middle management and administrators and still needed to care about workers.

    The turning point was beginning of seventies and in particular unpegging of USD from the gold standard. That in turn opened a way for complex financial products and financialization of capitalism. And bureaucrats needed to care about investors and Wall Street, not those who made things.

    • chiral3 said, on February 21, 2022 at 2:43 pm

      I haven’t read The Utopia of Rules – I’ve read Debt – so I may not appreciate Graeber’s point around the gold standard. When I think of complexification of financial products, and all that entailed, I think of the response to 80’s FED and 14% inflation. There’s a whole slew of reasons why we had that inflationary period dating back to the war through the 70’s, and why it was special, versus post war inflation. Conventional thinking would be that that the standard was far from a stabilizing force as it related to modern inflation – gold went to 800USD per oz in 1980 – that it made sense during war time, but was largely moribund by the 70’s, where we happened to have two inflationary periods. People old enough at the time would have had a special hatred for Roosevelt and executive order 6102 (a good reminder for us today that really fucked up communistic things happened before we were born too). Obviously central banking and it’s interaction with the Treasury was distasteful to these people also. The inflationary period in the 1980s created rate derivatives and mortgage products, though. These clever fucks needed something to do when rates came back down and equities shit the bed (=vol) in 1987 and thus began twenty years of innovation. Interesting to think about which complexification begat which complexification, public v private, vice versa, et seq.

      • Igor Bukanov said, on February 21, 2022 at 8:40 pm

        I personally do not buy Graeber’s theory that financial institution changes were a grand plan of managers/bureaucrats to take control from the middle class. Still it did pave way to inflation and all the complexities of financial products and the rise of Wall Street.

        But Graeber has another observation that I agree with. Big private business in US already post-war was much stronger bureaucratized than in Europe and this continues to present day. I can relate to that personally. In nineties I worked as a programmer in a small oil company in Norway. Most engineers there worked before in big companies or continued to work there as consultants. From their stories it became apparent US Big Oil internally looked like Soviet government departments even in their offices in Norway. Shell was significantly better in that regard. Even Norwegian Equinor controlled by the state was better.

    • Corsair said, on February 21, 2022 at 5:23 pm

      I think “managers” and “bureaucrats” are too often conflated. As long as we’re discussing the private sector, I like James Burnham’s 4-type taxonomy of capitalists: managers, executives, financiers, and shareholders. Later theories of Managerialism, not entirely wrongly, shifted the focus onto bureaucrats generally, some of whom would be managers and others some form of worker in Burnham’s still quasi-Marxist formulation. I would say that modern-day China is more of a managerial state in Burnham’s sense than the United States. Some of this is ideological, some of it is just industrial economies of scale.

      What the 50s were was the peak of managerial dominance here in America, with the 70s as a sort of weird interregnum before the new era of financialization/globalization Chiral3 discusses better than I could below. Managers who formerly dominated American capitalism were thrown back into competition with superstar executives, financiers, and shareholders. Michael Eisner, Milken, and Jensen might be the archetypal characters here. Many of these managerial positions were outright eliminated in the 1990s, and I suspect more were eliminated after the GFC. Bureaucratization in the broader sense continued more or less apace the whole time.

      You could tell a parallel story in the public sector with post cold-war military reform and general slimming-down (managerial power), the emergence of Unitary Executive type ideas (executive power), Washington consensus fiscal policy, behold the bond vigilantes (financier power) and general loss of state capacity against other lobbying/interest groups (shareholder power). The last analogy is a stretch, but I don’t think the others are.

      • Corsair said, on February 21, 2022 at 6:27 pm

        Post-war managerialism wasn’t all bad, as any reader of this blog knows, small, efficient risk-taking Klein type-1 organizations like Bell Labs and Skunkworks existed within big lumbering organizations like AT&T and Lockheed.

        Ben Rich was Kelly Johnsons #2 at Skunkworks; his autobiography is a great read. In the late 50s Rich posed a as physicist “Ben Dover” to acquire info from a Bureau of Standards lab about liquid hydrogen, then created explosions large enough to knock workers off of scaffolding 500 yards away. This was in the heart of Burbank, jammed between the Bob Hope Airport and the I5. There’s now an Amazon warehouse and some kind of office building. SR-71 pilots buzzed Kelly Johnson’s house in Santa Barbara, timing their sonic boom appropriately. Legend has it an SR-71 pilot blew out the windows of Presidents Nixon’s SoCal home. He called the Air Force to complain. This all seems unimaginable today.

        By the early 80s, the bureaucracy was crippling. Rich was forced to buy 2% of his materials from minority-owned businesses and harassed for not hiring Latino engineers. This is while Reagan was president! OSHA demanded 65 different types of masks, but bearded employees could not wear masks. Ergo no bearded employees. An inspector found 7000(!) violations, and fined accordingly. Coincidently Skunkworks accomplishments trickled off about the same time, although the F-117 was a glorious Parthian shot at the decline of the West and all that.

        Kelly’s rule #13: “Access by outsiders to the project and its personnel must be strictly controlled by appropriate security measures.” I doubt this was just for NatSec reasons; managerial complifiers take cruel delight in reducing everyone to their efficiency levels.

  12. William O. B'Livion said, on February 21, 2022 at 4:42 am

    Golang has moved back to static binaries.

    But developers don’t use libraries to save memory–they use libraries to avoid reprograming the wheel, unless it strokes their ego to do so.

    So whether we have static binaries (like everything in /sbin SHOULD be) or dynamic libraries we’re still going to have Chinese fire drills when some dipshit publishes example code with his description of a significant flaw in a popular library.

    The only difference is that with static binaries we’re going to have to test *every* binary on *every* system (in some fashion) to find where that particular flaw may have been incorporated.

    Golang, being a product of Google, is of course perfect and without flaw, so nothing you write in Go can be a security risk, so there’s no worries there.

    Also don’t mis-underestimate “Putting more useless shit on my resume” as a significant force behind complificating things.

    • Scott Locklin said, on February 21, 2022 at 10:40 am

      I like golang a lot.

    • Igor Bukanov said, on February 21, 2022 at 8:54 pm

      The problem is that different applications use different version of libraries. So one is going to patch a lot in any case. And then consider that this problem with flaws in widely used statically linked libraries mostly does not exist on Android or IPhone. Apple and Google scan apps for known security vulnerabilities and developers better to patch their apps or risk being booted from stores.

  13. anonymous said, on February 21, 2022 at 8:43 pm

    Re: Static linking and DLL hell, appimages, etc.

    I also came to the conclusion that shared libraries were (mostly, in the original-flavor concept) a mistake. DLL hell was a solved problem on Windows, in the early 2000s up through the end of the Win7 era. It was *solved*! The solution was forgotten. The solution basically ended up being abandoning the concept of saving some paltry amount of hard drive space by sharing code – it was a bad idea, it didn’t work, it was a road straight to DLL hell. Unless it was a core, rock-solid, never changing operating system feature (in which case it lived in system32, and had an ABI backwards compatible to cuneiform tablets), all the dependencies were bundled with the application in the application folder. If it made sense for them to be broken out into a dll (some other library maintained by someone else), every application had its own, so that it didn’t step on the others.

    That’s why software in Windows “just worked”. Linux, on the other hand, appeared to try to manage dll hell with package managers. Appimages and flatpaks seem to be Linux finally discovering (albiet in an awkward opaque-package way) what Windows developers knew for decades. (Wish it was just a directory you could browse and do things with).

    Docker is *insanity*. What python does is *insanity*. Having one stack of libraries that all grind each others gears, which requires multiple separate instances of the scripting language to deconflict is nuts. Having an OS for every application is cancer.

    (BTW, I once managed to *accidentally* privilege escalate on a server because I was trying to figure out how to mount a data directory in docker, because no one would give us sufficient access to run our software in the ludicrously locked down environment we were supposed to develop in. Security! :-/ It’s a bit nuts to try to give an engineer only part of a computer – if you don’t trust your own employees, that’s not a problem that IT is going to help you fix. Also, they could buy me a brand new laptop every day and let me set the old one on fire for the cost of my time they were wasting with security bullshit. )

    I’m not a software engineer – I’m an aerospace engineer. But the solutions that software engineers come up with to the problems of software seem actively nuts and harmful to me. “Complification” is a good word for it.

    • anonymous said, on February 21, 2022 at 8:54 pm

      So many things repeated from every direction as “best practices” in software engineering seem like “worst practices” to me. If you want to succeed, you pretty much have to ignore the canon in the field and go rogue. “Cowboy coding”, etc. It’s probably not just software development.

  14. Bill in Texas said, on February 22, 2022 at 4:52 am

    I use a fork of Arch and it basically gives me no problems for my very basic computing usage, but for routinely reinstalling Spyder, but worry about the future of Linux when the kids in school today take over. You could extend this worry to any field. What happens when these ruined gender freaks who spent 3 years wearing surgical masks are responsible for maintaining projects? We will be reliant on Eastern Europeans and Japanese, I imagine. At least we have them to keep software puttering along in some fashion. Civilizational decline will become acute as Gen X ages out.

    • Scott Locklin said, on March 2, 2022 at 11:12 am

      I dunno, seems pretty acute now. GenX were last generation who remembers functioning society (and society without internet), and there were never very many of them.

    • volatilityjunkie said, on March 6, 2022 at 11:49 pm

      The only group of people that haven’t been infected with the Western Civilization (TM) brainrot are Chechens — and they’re not going to waste their lives away making sand think.

      Eastern Europe and Japan are no different than the Anglosphere; the formers’ disease simply at an earlier stage than the latter’s.

      Who will be responsible for maintaining projects? The technologically-illiterate. Once everyone of any capability (and perspective) has rocketed off into the stratosphere by delivering real value (and demanding their fair share), those who couldn’t must now keep their better’s creations alive.

      It’ll be one grand jobs program. Give it 30 years.

    • Verisimilitude said, on March 8, 2022 at 8:47 pm

      Maintaining the Linux kernel is already a fool’s errand. It turns out the real idiots were inside the house all along. Why shouldn’t software be finished at some point? Now, the stupid bastards piling up their tower will never be finished; so what?

      The technologically-illiterate are already in control, because they never understood for why the machines exist, and likely never will.

      • sigterm said, on March 11, 2022 at 10:35 am

        The Waldorf school guy, a hundred years ago, claimed he could see spirits, and those behind every piece of technology were shockingly ugly, and all servants of Satan. If nothing was done to control them instead of the opposite, future Earth would be covered in “a web” of “spider-like creatures”, all “very wise, but only in a materialistic way”. Quoting from memory. As Scott’s post hints at, what’s happening is deeper than computer science.

        Also, interesting to see that Dijkstra and Kay, both from opposite sides, had/have the same contempt for modern computing.

  15. remnny said, on February 22, 2022 at 6:38 am

    The creator of one of the most egregious CoCs isn’t even a woman: https://kiwifarms.ru/threads/coraline-ada-ehmke-corey-dale-ehmke.31003/

    I can’t believe that we allow men dressed as women to shove their CoCs in people’s faces without repercussion. Whatever happened to common decency?

  16. Walt said, on February 23, 2022 at 11:44 pm

    Now I’m lucky to go a week without a hard reboot being required because nobody told Danger Hair Daphne that she’d be better off pursuing a job as an HR administrator than a kernel developer

    Danger Hair Daphne does much worse damage in HR, but arguably she’s part of the distributed HR dragnet as a developer especially when she transitions. I’m hearing CEOs of small companies say they don’t want to hire Millennial womxn because they’re too much trouble. This could be self-righting in that way.

    Anyways, good riff on Casey Muratori’s “30 Million Line Problem” and subsequent talk by Jonathan Blow.

  17. […] Jack Baruth links to a piece entitled Managerial failings: complification. […]

  18. anonymous said, on March 16, 2022 at 12:55 pm

    I am so damn sick of the world. The Western world in particular. I won’t be able to ever leave here, obligations and family are here. But basically, there is no future.

    I hope you are doing well, wherever you ended up, and have found somewhere where sanity still exists.

  19. Sprewell said, on April 2, 2022 at 3:55 am

    Btw, Scott, Boston Dynamics is going into production now. You appeared to be a skeptic based on their demo-heavy and productless past, care to predict how they do?

    I personally feel we are on the verge of a robotics revolution, a la computing circa 1980. But it all depends on if the tech is ready yet and I’m not immersed in it, maybe it’s really 1950 or 1970 for robotics.

    • Scott Locklin said, on April 2, 2022 at 9:17 am

      Sure: they will fail utterly, probably while remaining around as someone’s face saving vanity project periodically doing ridiculous publicity stunts a la dwave. Considerable PR achievement; if you can’t tell the difference between a technological achievement and a public relations achievement by now, well, I can’t help you.

      People have been “feeling” we’re on the verge of a robotics revolution literally since they invented personal computers.

      • Sprewell said, on April 2, 2022 at 6:08 pm

        As I said before, I maintain an open mind on BD. I’m not willing to call it “PR” or not, or a complete joke like D-Wave or Magic Leap, without actually examining their code. What makes you so certain it’s not “a technological achievement?” It seems like you’re simply speculating.

        As for whether we’re on the verge, I remember reading a blog post by two excitable mobile guys in the early 2000’s saying that we were on the verge of a mobile computing revolution, and I left them a comment saying the hardware wasn’t ready yet. Well, smartphone sales surpassed the PC in 2010, so while I was technically correct, it happened much sooner than I anticipated.

        Wrong predictions by others in the past are irrelevant: Apple was too early with the Newton, but eventually got it right with the iPhone. Someone is going to crack the robotic puzzle, maybe BD, though it’s not my field so I can’t say when.

        • Scott Locklin said, on April 4, 2022 at 11:50 am

          There’s no reason to be open minded about a “company” that has been around for 30 years and not yet produced a single useful product. In 3 decades. I have grown-ass adult friends who are younger and have produced far more value in the world than Boston Dynamics.

          The simplest model in the world is AR(1) -what they did last year, aka Potemkin press releases and nothing of importance, is probably what they’ll do next year. If they manage to excrete a single useful product it will one of the greatest corporate turnaround stories of human history.

          Internet in your pocket was dead-ass obvious: I was working on asics for G3 MEMS accelerometers in 2000. Robotics revolution? Lol; just wishful thinking by lizard-men from Davos. Maybe some day. Not some day soon.

          • Sprewell said, on April 5, 2022 at 10:08 am

            Unlike D-Wave, what BD is doing is actually feasible and they have some demonstrated competence so I remain more optimistic than you. As usual, you offer a lot of cynical speculation for why you distrust them and not much else, whereas I would actually examine their software before deciding.

            As for mobile, nuclear-powered cars were also considered “obvious” once, yet it never happened. The question under discussion wasn’t if mobIle computing would take off, as even Kubrick had hand-held tablets in 2001, the trillion-dollar question I raised is when? As ever, that is the crucial question.

            As for robotics taking off, I’m not talking about mobile robots like BD is starting to sell and which I’ve said can come later, I’m talking about a lot more stationary robots like these, say starting to replace cooks in industrial kitchens. It’s happening a lot sooner than you apparently notice.

            • Scott Locklin said, on April 5, 2022 at 10:30 am

              They’ve demonstrated nothing but a very accomplished PR effort.

            • shmee said, on December 21, 2022 at 5:11 pm

              Industrial robots were used in car production for decades before Boston Dynamics was founded.

  20. Eric said, on November 6, 2022 at 5:27 pm

    Reblogged this on Calculus of Decay .

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: