Login
You're viewing the functional.cafe public feed.
  • Nov 14, 2026, 6:01 PM

    When some big, messy, complex system breaks down, there is •never• a single smoking gun.

    See how.complexsystems.fail item 7: “Post-accident attribution to a ‘root cause’ is fundamentally wrong.”

    It’s never enough for just one thing to go wrong, or just one component to fail; if a single point of failure •does• in fact bring everything down, then that in itself is a sign of a multitude of other failures.

    2/

    7. Post-accident attribution to a ‘root cause’ is fundamentally wrong.

Because overt failure requires multiple faults, there is no isolated ‘cause’ of an accident. There are multiple contributors to accidents. Each of these is necessarily insufficient in itself to create an accident. Only jointly are these causes sufficient to create an accident. Indeed, it is the linking of these causes together that creates the circumstances required for the accident. Thus, no isolation of the ‘root cause’ of an accident is possible. The evaluations based on such reasoning as ‘root cause’ do not reflect a technical understanding of the nature of failure but rather the social, cultural need to blame specific, localized forces or events for outcomes.

Anthropological field research provides the clearest demonstration of the social construction of the notion of ‘cause’ (cf. Goldman L (1993), The Culture of Coincidence: accident and absolute liability in Huli, New York: Clarendon Press; and also Tasca L (1990), The Social Construction of Human Error, Unpublished doctoral dissertation, Department of Sociology, State University of New York at Stonybrook)
    💬 7🔄 69⭐ 156

Replies

  • Nov 14, 2026, 6:04 PM

    This warning means that trying to pick out one bogeyman here is a flawed approach, because (for example) “Was it layoffs or not?!” is a meaningless question. It was (as Baldur says) almost certainly “layoffs and….”

    That warning •also• means that companies like Microsoft can try to save face by cherry picking the least PR-damaging causal factor and trumpeting that as the real cause. Beware of such declarations when they arrive. If they name a single cause, it’s a PR stunt.

    I don’t think we know what really caused the MSFT or FAA outage. I doubt we’re going to anytime soon, maybe ever. The answer will be large; it contains multitudes.

    3/

    💬 4🔄 24⭐ 88
  • Nov 14, 2026, 6:09 PM

    Even without knowing all the gory and confidential details, however, and even without looking for a single root cause or smoking gun, we can usefully assess the •landscape•. And if we do, we see serious problems — ones that are alarming, but also ones we can address.

    4/

    💬 1🔄 6⭐ 52
  • Nov 14, 2026, 6:17 PM

    Was it layoffs? Was it reckless vibe coding?

    Ask instead: what kind of industry / socioeconomic environment are we in right now that these things are so very likely to be among the causal factors?

    If orgs are chucking people out the door at the same time they’re rolling codebases they don’t really understand into production, what does that say about the management environment? The financial and political pressures at play?

    I was riffing on this yesterday:

    hachyderm.io/@inthehands/11761

    5/

    💬 3🔄 39⭐ 106
  • Nov 14, 2026, 6:24 PM

    A lot of the riffing happening on engineering social media right now can look like cynicism or gallows humor, perhaps, but I think it’s also something deeper and much more positive: it’s people in the trenches recognizing and resisting an environment that is dehumanizing and psychologically oppressive and broadly harmful.

    When @grimalkina talks about how we treat software creators as “brains in jars” (see her book _The Psychology of Software Teams_), she’s putting her finger on both psychological experiences •and• systemic forces:

    6/

    💬 1🔄 25⭐ 86
  • Nov 14, 2026, 6:24 PM

    “Brains in jars” is how you get management leaping like kittens at laser pointers any time somebody promises a way to commoditize developers.

    “Brains in jars” is how you end up believing that you can ship mountains of code that reflect no careful thought and that nobody actually understands (vibe coded or otherwise).

    “Brains in jars” is how you get Microsoft conducting mass layoffs with no apparent managerial understanding of the cost of lost institutional knowledge and damaged relationships, or the concrete risks that creates.

    7/

    💬 4🔄 28⭐ 83
  • Nov 14, 2026, 6:29 PM

    Mix “brains in jars” with a version of capitalism where extreme wealth inequality means that investors and not customers are the ones with the money and power, and mix that with anti-trust law being effectively nonexistent, and mix that with a right-wing lurch toward corruption, and…well, there you have our present moment.

    A few other things belong on that list, I’m sure. Point is, the •kind• of failures we’re seeing the last few days — particularly the slow and strangely confused recovery — are symptoms of large systemic forces.

    8/

    💬 1🔄 22⭐ 67
  • Nov 14, 2026, 6:31 PM

    So LLMs and layoffs aren’t the root cause, because there is no such thing as a “root cause” here. And maybe they don’t even turn out to be one of the proximate causes.

    But if they’re not causes, they’re comorbidities.

    9/

    💬 2🔄 10⭐ 72
  • Nov 14, 2026, 6:33 PM

    I quipped yesterday:

    When termites are eating the whole structure, can you identify which single specific beam made it fall?

    •That• is the framing I’m advocating in this thread.

    10/

    💬 5🔄 34⭐ 98
  • Nov 14, 2026, 6:36 PM

    So I say yes, do complain about vibe coding, do complain about layoffs, point that finger, mock, cajole, crack those jokes, post those memes — not because everything would be better if only the vibe coding would just vanish, but because it’s a way of drawing attention to the termites in society’s structural wood.

    All the meming and gallows humor is good if it is part of recentering and refocusing on humans.

    Let’s build and do great things by being kind, by being humane, by never losing sight of the value of the humans who touch and make every part of these messy systems that are breaking so badly right now.

    11/

    💬 3🔄 43⭐ 117
  • Nov 14, 2026, 6:36 PM

    P.S. Apologies to termites, they are cool creatures, this is not an anti-termite account, these pernicious societal forces are way worse than termites; the metaphor works only inasmuch as you don’t want termites eating your home.

    💬 3🔄 10⭐ 75
  • 💬 0🔄 6⭐ 10
  • 💬 0🔄 0⭐ 3
  • 💬 0🔄 2⭐ 6
  • 💬 0🔄 0⭐ 1
  • Nov 14, 2026, 6:41 PM

    @inthehands 🐜🐜🐜🐜🐜🐜🐜🐜🐜

    “What happened to Paul?”

    “The termites came for him. Though, to be fair, it is overly simplistic of us to attribute the downfall of Paul’s anti-termite account to this single factor. The reality was a complex interaction of tightly coupled systems led us to throw the least powerful and most disposable parties under the bus: termites.”

    💬 0🔄 1⭐ 12
  • 💬 0🔄 0⭐ 1
  • Nov 14, 2026, 10:34 PM

    @inthehands

    We should teach kids "either/and" rather than "either/or" because rarely do systems (and everything breaks down into systems at some point) produce binary choices.

    Decisions may be yes or no, but we're more complicated: as people, as systems, and as societies.

    💬 0🔄 0⭐ 2
  • Nov 15, 2026, 1:59 AM

    @inthehands But stop trying to be good cute and perfect, be natural, loud and laugh without bbeing PC guilty, bring back NATURALNESS so you can bne funny and not tedious and boring..as WELL was working for GOOD!

    💬 0🔄 0⭐ 0
  • Nov 14, 2026, 6:37 PM

    @inthehands Tangential to what you're saying in this thread, but related: in school, I learned about how individual artisanal crafting of machines gave way to interchangeable parts as things modernized.

    As someone in IT, it became very clear that modern software is in tension between the people who write it -- individual artisanal crafters -- and management, who want it to be (or become) interchangeable parts. "AI"/LLMs are the latest iteration of this un-discussed conflict.

    💬 3🔄 3⭐ 0
  • 💬 0🔄 0⭐ 6
  • Nov 14, 2026, 7:28 PM

    @Unlikelylass @inthehands The key is that machines are really good at producing the same thing repeatably - and that's fine.

    If every line of code, or every method, or class, or program did the same thing, then we would be comparing apples and apples.

    The machine part of producing software was mastering CDs and floppies, now it's making assets downloadable.

    But crafting the assets? That's what we do. We figure out what is needed and then we design and implement it.

    That's what we do.

    💬 0🔄 0⭐ 5
  • Nov 14, 2026, 7:30 PM

    @rhempel @inthehands I think my point is that the management refuses to acknowledge that part. They desperately want it to just all be interchangeable parts, and continue to act like *this time* the silver bullet (LLMs "ai") will make it behave that way.

    💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 1
  • Nov 15, 2026, 12:22 AM

    @rhempel @Unlikelylass @inthehands

    You can automate production, and you can even automate production of the machines that produce, but at some level you run into a thing that has to be built by a conscious being who is motivated to get it right, or else it probably won't be. (...because there are infinitely more ways to be wrong than to be right.)

    The oligarchs and managers always want to believe that whatever piece they're in charge of is one that can be automated.

    ...and maybe it can be -- but first you need someone to make that automation, and to ensure that it works properly... and you need to keep them around to fix it when it breaks, keep it oiled properly, and presumably update it as the need changes.

    (This made sense inside my head; I apologize if the sense got detached on its way out the door.)

    💬 1🔄 0⭐ 3
  • 💬 0🔄 0⭐ 3
  • 💬 1🔄 0⭐ 2
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 3
  • 💬 0🔄 0⭐ 1
  • 💬 0🔄 0⭐ 0
  • Nov 14, 2026, 6:53 PM

    @inthehands Agreed with all this VEHEMENTLY — would however add that while some don’t understand, more don’t care. They aren’t interested in building effective systems to produce good outcomes. Realizing that is mainly what led to me leaving the industry and opening a bar.

    💬 1🔄 0⭐ 9
  • 💬 0🔄 0⭐ 1
  • Nov 14, 2026, 8:12 PM

    @inthehands I think the rot set in when they stopped calling employees "personnel", and instead went with "human resources". Interchangeable disposable slots to be filled in at the whim of a corporation, rather than individuals.

    💬 0🔄 0⭐ 1
  • RobertJacksonRobertJackson58585858@masto.ai
    Nov 15, 2026, 11:25 AM

    @inthehands

    Homeostasis.
    The tendency of things to remain the same or self repair is expensive.
    Think of a fruit bowl with apples and oranges. You can usually nudge an orange without much effect.
    You might make a slightly flatter bowl more cheaply. It'll still work.
    Or even have no bowl at all. Until you nudge an orange and they all roll away.
    Homeostasis is usually people, internal controls, reconciliation processes.
    All costs.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 1
  • Nov 14, 2026, 11:53 PM
    💬 0🔄 1⭐ 1
  • Nov 15, 2026, 2:07 AM

    @inthehands

    One thing is certain: the Trump administration will not tell the truth about what caused the nationwide ground stop.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 1
  • Nov 14, 2026, 7:33 PM

    @sandorspruit @inthehands the most dangerous thing I do regularly is drive on public roads, so I often find myself thinking about what makes driving safe and what makes it dangerous, and I’m convinced that collisions are rarely a result of a single driver’s single error; errors that reduce the margins are very common, and collisions happen when too many drivers make small errors too close together. Close calls are the same phenomenon, compared to a lower threshold

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 14, 2026, 9:03 PM

    @inthehands In industrial safety, the "root cause" is always management system failure, if the parties are honest with themselves.

    💬 1🔄 0⭐ 1
  • 💬 0🔄 0⭐ 1
  • Nov 15, 2026, 12:13 AM

    @inthehands Mentour Pilot has been teaching me a lot about the eternal conflict between (a) the need to see failures of a system as a systemic problem which requires changes to the system versus (b) the ever-present human tendency to want to have a face to be angry at.

    This seems like basically that, except complex systems in general instead of aviation in particular.

    💬 0🔄 0⭐ 0
  • Nov 15, 2026, 4:01 AM

    @inthehands For anyone interested in hours of excruciating detail on famous, infamous, and little known incidents - and the inevitable conclusion of multiple things going wrong but generally accelerated by poor management - this is the podcast for you!

    engineered.network/causality/

    💬 0🔄 0⭐ 0