Login
You're viewing the mstdn.social public feed.
  • May 29, 2026, 7:20 AM

    What a lot of people do not understand is the fragility of supply chains.

    Setup

    Let us assume there is a part X that is used by a large number of other companies. It is not glamorous, it is not expensive, and nobody outside a small circle of specialists ever thinks about it. But it is needed. Without it, other products cannot be finished.

    Everything is fine

    Part X is made by a small company with a few dozen employees and a machine that is several decades old.

    Everything runs smoothly. The company knows how to maintain the machine. They know how to operate it. They know its strange noises, its moods, its undocumented rituals. They know which lever needs a bit of persuasion on cold mornings and which replacement part must be machined by hand because nobody has made the original since 1987. They also know how to train new employees, because the knowledge still exists inside the company.

    Crisis

    Then, suddenly, a few large customers run into a crisis.

    Orders slow down. The warehouses fill up with part X. Prices fall.

    The company downsizes. Then it downsizes again. But nobody really notices, because stocks are still full. Customers are not yet affected. Purchasing departments can still get part X from inventory. Balance sheets still look fine. The problem has not yet reached the spreadsheet.

    If this goes on for long enough, the company goes broke.

    Again, nobody really notices. Stocks are still full. Some people may worry, but as long as no current production line is stopping and no quarterly number is visibly bleeding, nothing urgent is done.

    The employees move on. They retire, change industries, start new careers. The old machine is sold for scrap. The building is repurposed. The knowledge evaporates.

    Recovery

    A few years later, demand picks up again.

    The warehouses slowly empty. Stocks run low. People start ordering part X again, only to discover that nobody is offering it anymore.

    Now someone decides to restart production.

    The first thing they discover is that there is no machine. Building a new one would be prohibitively expensive, assuming anyone still knows how to build it properly. So they desperately search for an old one.

    They get lucky. In a scrapyard, they find a machine that used to produce part X. It is rusted, incomplete, and dysfunctional. Naturally, they buy it.

    Now they try to get it working again.

    But there is another problem. There are no people left who know how to maintain it. So they hire someone who understands industrial maintenance in general, but has never worked on this specific machine. That person does their best. They improvise. They read old manuals. They reverse-engineer undocumented fixes. They keep the machine alive with skill, patience, and increasing amounts of despair.

    But it breaks down every few hours. Output is abysmal.

    Bottlenecks

    And now that one poor maintenance person is overworked. They need help. But training help requires time, and the only person who can train others is the same person needed to keep the machine barely running. Every hour spent teaching is an hour not spent preventing the next failure.

    Very few businesses survive this phase.

    There is no institutional knowledge anymore. New people are hired, begin training, look at the state of the machine, the chaos of the process, and the constant emergency mode, and conclude that the business is doomed anyway. Then they quit.

    Churn becomes terrible. Even if the company survives financially, it remains fragile. It is always just one or two people quitting away from disaster.

    At the end the world decides it needs to get rid of part X as the supply is too fragile.

    Summary

    This is still very much simplified. The reality is more complex, more ugly.

    And that is the part many people miss: a supply chain is not just warehouses, contracts, prices, and transport routes. It is also people, habits, obsolete machines, informal knowledge, and boring little skills that nobody values until they are gone.

    💬 46🔄 456⭐ 564

Replies

  • 💬 3🔄 4⭐ 22
  • 💬 5🔄 0⭐ 1
  • 💬 1🔄 2⭐ 29
  • 💬 0🔄 0⭐ 1
  • May 29, 2026, 10:34 AM

    @eckes @jt_rebelo @masek Yes that's inevitable, but not weeks, sometimes a year or more, and not cheap either. Also, people are starting to realize that resources are not endless. The lack of a specific plastic machine part 10 years ago was a real problem. Now you can solve it with a 3d printer and maintain that factory line working another 10.

    💬 1🔄 0⭐ 1
  • May 29, 2026, 11:03 AM

    @rcosta @eckes @jt_rebelo @masek
    Maybe for documented parts that can be 3D printed. The 3D printing is over-hyped. Esp. plastic printing ones even to replace plastic. You might need to get a mould made. Also easier.
    You're missing the point. The supply chain is as good as weakest link & replacement relies on docs & expertise that may not exist. It could take years to replace.
    There are products that are desired & no current equivalent exists because the parts not made. New design isn't viable.

    💬 1🔄 0⭐ 15
  • May 29, 2026, 11:30 AM

    @raymaccarthy @eckes @jt_rebelo @masek I'm not missing the point. Just saying there's nuances. Somethings now can be solved, others still can't, that's an improvement. If the market truly demands a product that product will be produced, replacing the link of the chain or all the chain if necessary. Some parts also end because they are dangerous to the environment or the raw matters used are now more expensive (ie gold). Big companies understand and prepare themself for this things.

    💬 3🔄 0⭐ 0
  • May 29, 2026, 11:36 AM

    @rcosta @eckes @jt_rebelo @masek

    Any link not solvable is a disaster. That's the point.

    It's more common than people imagine and sometimes the reason for a product sold out and 2 years later no replacement. Then later the supposedly replacement product is inferior, or never appears.

    See Kobo Sage 8″ ereader.
    Yaesu attempted to replace FT817ND with FT818ND, but had to cease that too due to lack of parts.

    There are many more examples of inferior or no product replacement. It will get common.

    💬 1🔄 0⭐ 7
  • May 29, 2026, 12:03 PM

    @raymaccarthy @eckes @jt_rebelo @masek Covid19, remember all those medic ventilation machines off-use because the lack of a single small plastic part? That's a real tragedy...solved thanks to 3d printing. All this is essentially an environmental issue. Products are created to have short lives by design, parts production count with cheap raw materials and workforce, etc. This is changing but we still can't cope. That's admit our current capitalist system failed and the elites can't do that.

    💬 1🔄 0⭐ 1
  • May 29, 2026, 12:10 PM

    @rcosta @eckes @jt_rebelo @masek
    That's a rare example of 3D printing to the rescue. Helped due to low volume.

    No, it's not primarily environmental. It's lack of investment in training, lack of open source documentation (or even any under NDA) and lack of second sources.

    Yes, building to last, repair and reuse are important. But only tangential to supply chain risks.

    💬 1🔄 0⭐ 5
  • May 29, 2026, 12:51 PM

    @raymaccarthy @eckes @jt_rebelo @masek "lack of open source documentation", I agree and that can be fixed without a big investment is more a matter of political will. But I maintain my views to systemic causes, specially the ones related with predatory economics. Greedy CEOs scrapping machines, ending internships and factory repair workshops, etc. Lack of parts is old as industrialization itself, is inevitable. We only can delay it and choose how to deal with it when finally happens.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 4
  • May 29, 2026, 3:58 PM

    “ If the market truly demands a product that product will be produced”

    This is way more faith in “the market” than is supported by facts. Monopolies, limited runway, financialization that reduces forward planning… the people who study markets know how far the ideal is from reality!

    @rcosta @raymaccarthy @eckes @jt_rebelo @masek

    💬 2🔄 0⭐ 6
  • May 29, 2026, 4:59 PM

    @clew
    You are being kind. The "market" doesn't decide. Rich guys decide, often on a whim.

    A really wanted thing might never be produced because the big companies can't see the value and too much startup capital is required by smaller companies.

    So called social housing is an example. Lower tax, lower interest, buy to let, laxer planning permission etc don't work. What does work is a Government simply building them. That's the only way to bring down rents too. A rent cap simply cuts supply!

    💬 1🔄 0⭐ 2
  • 💬 0🔄 0⭐ 1
  • May 29, 2026, 7:29 PM

    @clew @raymaccarthy @eckes @jt_rebelo @masek "faith in “the market”"...The things you mentioned aren't exclusive to a specific economic system. You find them in USSR/Russia, USA, UK past and present. But in some places you have (or had😞 ) more tools to demand changes (ie right to repair laws). Things inevitable advance. New machines, new parts, new demands. Small companies can't continue producing marginal requested products. Those things demand resources they don't have and clients can't pay.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 2
  • May 29, 2026, 1:12 PM

    @eckes @jt_rebelo @masek I'm sure anyone could compress decades of research and experience into a few months... oh wait, instead we get product shortages and inferior solutions until the decades needed have passed.

    💬 0🔄 0⭐ 5
  • May 29, 2026, 7:37 PM

    @eckes This post is not about steam engine parts. Btw, we had a steam engine at school that we learned to service because they were still used in industry and probably still are in some places. I'm long out of that line of work because of nickel allergy and lots of other stuff so have lost touch.

    Ofcourse there are things that will get replaced by more modern parts but this post is not about those. But don't worry, you will know when the non-replaceable parts break in your product chain. I'm pretty sure it is much wider and deeper than you can imagine, whatever you use.

    💬 0🔄 1⭐ 1
  • May 29, 2026, 8:42 AM

    @jt_rebelo @masek AI certainly does not help, but from the "end" side of my career, I can tell that, even without AI, very few companies, and indeed very few people in those few companies, know how important it is to *preserve and transmit knowledge* within the company.

    💬 3🔄 2⭐ 17
  • 💬 0🔄 0⭐ 2
  • 💬 0🔄 0⭐ 4
  • May 29, 2026, 4:05 PM

    @aaribaud @jt_rebelo @masek Used to have that problem a lot in my university. Yes, a place supposedly dedicated to teaching and knowledge had a huge problem with repeated institutional amnesia. And we were very much not unique.

    We're gradually fixing that (at whole-sector scale), but it's *really difficult* and time consuming because it's so utterly tied to old employment practices and habits.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 1⭐ 4
  • May 29, 2026, 7:25 AM

    @masek ... standardisation of parts and second source comes into thought ...

    In the good-old-days you could buy a TTL7400 from a list of semiconductor vendors - or a M4 screw - or ...

    💬 3🔄 0⭐ 6
  • May 29, 2026, 8:14 AM

    @lobingera @masek Buuuut - 2nd source is mostly done at a point, when a single source can't scale with you any more. It is rarely done for supply chain safety.

    And most of the time it is used to have pressure on prices (until they go out of business), rather than having a loving relationship, that respects the dependency.

    💬 0🔄 0⭐ 6
  • May 29, 2026, 8:15 AM

    @lobingera @masek

    You know what company I hold a fiece grudge on? Analog Devices.

    They've been gobbling most if not all of my second source suppliers for precision electronics circuits.

    These days I rather spend weeks designing circuits from discrete parts, around wide operational ranges, and if I really have to put an IC in there, it's going to be a part that has hundreds of similar performing replacements.

    💬 1🔄 3⭐ 11
  • May 29, 2026, 9:31 AM

    @datenwolf @lobingera @masek "You know what company I hold a fiece grudge on? Analog Devices." That's very worrying because a huge swath of products is held together by their products! Audio, RF, SDR, etc., etc. My least-favourite is Qualcomm for many of the same reasons.

    💬 1🔄 1⭐ 5
  • May 29, 2026, 9:47 AM

    @vp9kf @lobingera @masek

    Yes, I know. ADI's actions over the past decade (and a half) should have triggered several cartel regulation authorities (FTC, etc.) to spring into action.

    ADI is already holding monopolies on several key component classes. There might be a couple of Chinese replacements in existence, but they're more or less invisible to the western markets.

    The EU is asleep at the wheel. They should have incentivized creation of European counterparts decades ago.

    💬 1🔄 1⭐ 7
  • 💬 0🔄 0⭐ 0
  • May 29, 2026, 6:33 PM

    @lobingera @masek And when you have the second source, some fresh-from-college manager joins and wonders if maintaining that partnership is worth it.

    After all, What Did The Second Source Ever Do For Us?

    By the time things come tumbling down, that manager already left for their next gig, pointing out how they saved their previous employer tons of money by optimizing supply chains.

    💬 0🔄 1⭐ 0
  • 💬 1🔄 0⭐ 2
  • 💬 2🔄 0⭐ 3
  • 💬 0🔄 0⭐ 2
  • May 29, 2026, 3:54 PM

    round me ending a post with “/s” means “that was sarcastic”; close enough? how many languages have a relevant word starting with s?

    @af @kritzelkatze @masek

    💬 0🔄 0⭐ 1
  • May 29, 2026, 7:58 AM

    @masek And the people who don’t know this the most are the coddled wealthy - ignorant of all the tiny things known and done that keep our collective world turning. Unfixed potholes on jet runways will be
    the harbinger of their doom.

    💬 0🔄 5⭐ 18
  • May 29, 2026, 8:20 AM

    @masek this. Marquetry saw blades come to mind. At some point we were down to one machine worldwide still able to produce them.

    Since then stuff got a bit better and we even got some good innovations in this domain but yeaaaah

    💬 0🔄 0⭐ 6
  • May 29, 2026, 8:25 AM

    @masek @tante
    This is exactly why the Toyota Productio. system emphasises two things (among others):
    1. total supply chain knowldge. Not just you immediate suppliers but all the eay back to raw materials.
    2. The whole of TPS, including JIT is of . apiece. Trying to institue things like JIT without kaizen, valuing workers, and total supplychain knowledge, produces brittleness.

    💬 1🔄 1⭐ 14
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • May 29, 2026, 8:49 AM

    @masek I am in a manufacturing industry and I see this kind of thing all the time at all scales. I’m aware that the product we make is also part of a supply chain that, if we were to stop manufacturing, could put an entire state’s economy into recession. The glib replies from people who just don’t get it are demoralizing.

    💬 2🔄 4⭐ 30
  • 💬 0🔄 1⭐ 2
  • Jun 1, 2026, 6:43 PM

    @futzle @masek

    > The glib replies from people who just don’t get it are demoralizing.

    From my POV, many people in IT and economics, have no real sense of how deep and complex fields like materials science and mechanics can get.

    And have little or no exposure to those areas, so they wrongly extrapolate from their own experience in IT. They also underestimate how much knowledge, judgment, and complexity can be involved even in jobs that are often seen as simple ... and/or are ignorant.

    💬 0🔄 0⭐ 0
  • May 29, 2026, 9:12 AM

    @masek Counter-example: In early 2020, there was a lot of demand for FFP2 masks, exceeding production capacity by a large margin. BYD's management decided that they go into this business quickly, but had no experience whatsoever with FFP2 masks. They did have ample experience with batteries, and spare machines for just about anything you might need for battery or car production. And a huge team of skilled workers to reconfigure machinery of that kind.

    They built machines to produce FFP2 masks within a week, and within months, they were the world's leading FFP2 mask producer. Actually, the machines they had were much higher quality than the machines they needed. They managed to assemble new face mask production machines at a rate of 5-10 new machines per day (!).

    sustainable-bus.com/news/byd-n

    Unlike the story you told, this was not an uphill battle task, but it was downhill: The requirements to make face masks are easy compared to what BYD regularly does.

    💬 1🔄 1⭐ 3
  • May 29, 2026, 9:29 AM

    @forthy42 Sometimes you can transfer KnowHow and in the end an even better product emerges. But I see enough cases where things go wrong.

    Yesterday evening a friend was complaining: for their work they need measurement instruments. They have 30 year old ones which work. But they have only so many to go around. They ordered new ones. The original manufacturer has now three times in a row failed to deliver a functional one.

    💬 1🔄 0⭐ 5
  • May 29, 2026, 10:27 AM

    @masek Often, the whole product category moved over to some Chinese who cloned the functionality decades ago and sells it for a fraction of the price. The original manufacturer is broke, and therefore, don't buy there. Buy the clone.

    Fierce competition can crush the original manufacturer, and it's what free market is about. Sometimes, the competition comes with degrading quality. Many hobbyist labs use Rigol scopes. Do they need to be as good as a Tektronix? Nope. They need to be good enough, which they are. Tektronix is fine, they survived by providing the Ferrari equivalent of scopes. If you need the multi-GHz, a Rigol is out of question. If you need to see if a signal is there, a Rigol is enough. And Rigol already expanded into the high end range…

    💬 0🔄 0⭐ 2
  • May 29, 2026, 10:02 AM

    @masek
    A similar scenario. In happy days, the world came to depend on a particular part. Three companies in one country learned to make it better, faster and cheaper than anyone else. They competed, guaranteeing the lowest price. The world bought from them to their satisfaction.

    Then one day, in a fit of pique, one country applied huge tariffs, in hopes of moving production home. Soon the world had two standards and two prices. Everyone was worse off.

    It's happening today in WiFi routers

    💬 1🔄 0⭐ 4
  • May 29, 2026, 10:09 AM

    @Stinson_108 Disturbances for supply systems are many: tariffs is one of them, pandemics are another, we also had natural disasters, political upheaval.

    But I observe that we build every day longer, more complex supply chains and care less and less about redundancy.

    This will cost us ...

    💬 1🔄 0⭐ 6
  • May 29, 2026, 10:24 AM

    @masek

    As long as sea lanes remain open, #tariffs stable, and no currency restrictions, the market will figure out the proper complexity of supply chains. Pandemics and ships sideways in the #Suez can be overcome. Like everything else, it's the idiocy of our leaders we should fear.

    💬 3🔄 0⭐ 1
  • 💬 0🔄 0⭐ 1
  • May 29, 2026, 11:59 AM

    @Stinson_108
    No. I've seen this happen in the company I worked for until 2012. Back then there was no problem with shipping lines, tarrifs, etc.
    It took them a few years and a lot of money to restore damage. Knowledge AND image...

    💬 0🔄 0⭐ 3
  • May 29, 2026, 12:42 PM

    @Stinson_108 Iow markets are fantastic at finding an optimal (i.e. most profitable) solution unless you expose them to the real world.
    @masek

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 2
  • May 29, 2026, 10:17 AM

    @masek

    Sounds a bit like the story of COBOL.

    I wonder how many financial institutions still rely on COBOL somewhere in their systems and have nobody in-house who understands it?

    Instead relying on outside contractors employing the few remaining and ageing COBOL specialists.

    What happens when they finally retire and/or die?

    💬 4🔄 5⭐ 7
  • May 29, 2026, 12:24 PM

    @gwentlarry @masek The Dutch government. In a frenzy of neoliberal market think they left many things to “the market”, getting rid of civil servants with domain knowledge. “Small government”, you know?

    It got so bad that they couldn’t even put out a proper tender or manage an infrastructure project. And don’t get us started on government IT projects.

    💬 1🔄 0⭐ 3
  • 💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • May 29, 2026, 5:39 PM

    @mossyfoot @gwentlarry @masek

    And, obviously, very few people understand that COBOL is not the problem. COBOL is a fairly simple and easy to learn language, really. I've done some pretty good work in it.

    But due to limitations in its design, the overall system, of many thousands of programs, is inevitably quite brittle, with a great many hidden dependencies.

    And that's why most programmers do not *WANT* to do COBOL.

    💬 2🔄 0⭐ 3
  • 💬 1🔄 0⭐ 0