Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
Windows

'The People Stuck Using Ancient Windows Computers' (bbc.com) 131

The BBC visits "the strange, stubborn world of obsolete Windows machines." Even if you're a diehard Apple user, you're probably interacting with Windows systems on a regular basis. When you're pulling cash out, for example, chances are you're using a computer that's downright geriatric by technology standards. (Microsoft declined to comment for this article.) "Many ATMs still operate on legacy Windows systems, including Windows XP and even Windows NT," which launched in 1993, says Elvis Montiero, an ATM field technician based in Newark, New Jersey in the US. "The challenge with upgrading these machines lies in the high costs associated with hardware compatibility, regulatory compliance and the need to rewrite proprietary ATM software," he says. Microsoft ended official support for Windows XP in 2014, but Montiero says many ATMs still rely on these primordial systems thanks to their reliability, stability and integration with banking infrastructure.
And a job listing for an IT systems administrator for Germany's railway service "were expected to have expertise with Windows 3.11 and MS-DOS — systems released 32 and 44 years ago, respectively. In certain parts of Germany, commuting depends on operating systems that are older than many passengers." It's not just German transit, either. The trains in San Francisco's Muni Metro light railway, for example, won't start up in the morning until someone sticks a floppy disk into the computer that loads DOS software on the railway's Automatic Train Control System (ATCS). Last year, the San Francisco Municipal Transit Authority (SFMTA) announced its plans to retire this system over the coming decade, but today the floppy disks live on.
Apple is "really aggressive about deprecating old products," M. Scott Ford, a software developer who specialises in updating legacy systems, tells the BBC. "But Microsoft took the approach of letting organisations leverage the hardware they already have and chasing them for software licenses instead. They also tend to have a really long window for supporting that software."

And so you get things like two enormous LightJet printers in San Diego powered by servers running Windows 2000, says photographic printer John Watts: Long out of production, the few remaining LightJets rely on the Windows operating systems that were around when these printers were sold. "A while back we looked into upgrading one of the computers to Windows Vista. By the time we added up the money it would take to buy new licenses for all the software it was going to cost $50,000 or $60,000 [£38,000 to £45,000]," Watts says. "I can't stand Windows machines," he says, "but I'm stuck with them...."

In some cases, however, old computers are a labour of love. In the US, Dene Grigar, director of the Electronic Literature Lab at Washington State University, Vancouver, spends her days in a room full of vintage (and fully functional) computers dating back to 1977... She's not just interested in early, experimental e-books. Her laboratory collects everything from video games to Instagram zines.... Grigar's Electronic Literature Lab maintains 61 computers to showcase the hundreds of electronic works and thousands of files in the collection, which she keeps in pristine condition.

Grigar says they're still looking for a PC that reads five-and-a-quarter-inch floppy disks.

'The People Stuck Using Ancient Windows Computers'

Comments Filter:
  • Paging Adrian Black (Score:5, Interesting)

    by Mspangler ( 770054 ) on Sunday May 18, 2025 @05:09PM (#65385637)

    Grigar doesn't know about him?

    Adrian's digital Workshop?

    http://m.youtube.com/watch?v=... [youtube.com]

  • lab equipment (Score:5, Informative)

    by awwshit ( 6214476 ) on Sunday May 18, 2025 @05:09PM (#65385639)

    I work in a test lab. We have a number of pieces of old test equipment, that we use but not often and it does not make sense to replace, that run on XP. I think we finally eliminated the Windows 2000 ones. The manufacturers have no interest in writing new drivers or software, they want to sell you a new one. So, we keep them off the network and eBay spare parts ahead of time.

    • Re:lab equipment (Score:5, Insightful)

      by dbialac ( 320955 ) on Sunday May 18, 2025 @07:07PM (#65385831)
      Too many people still don't understand that, "If it ain't broke, don't fix it." Newer does not mean better. Just look at Windows 11.
      • Re:lab equipment (Score:4, Informative)

        by 4im ( 181450 ) on Monday May 19, 2025 @03:03AM (#65386361)

        Too many people still don't understand that, "If it ain't broke, don't fix it." Newer does not mean better.

        That's certainly right, but only up to a point. When you're e.g. running a mainframe platform and losing the last people that still know it, and all the code running there, to retirement, and even the maker of that (non-IBM) platform obviously has less and less people who know how to keep that platform running, it's definitely time to move on. Unlike Elon / DOGE want to make you believe, no, it's not a 5 month job to migrate, even 5 years won't suffice to do it properly. It's not simply about knowing COBOL.

      • Too many people still don't understand that, "If it ain't broke, don't fix it." Newer does not mean better. Just look at Windows 11.

        Too many people don't understand the many caveats that come with this.

        1. Obsolescence is a "broke" issue. You don't want to be stuck in a situation where you're keeping old equipment running to the point where a failure could mean very extended downtime due to lack of parts, or lack of upgrade support, or the need to do major engineering as part of an upgrade. "If it ain't broke don't fix it" is a great precursor to "It's broke and now we're royally fucked!"

        2. Support is also a "broke" issue. In some case.

      • Yup. It's not "people stuck on ancient versions of Windows", it's "people who have something that works just fine and don't want spend a fortune to replace it with a pile of bloated, buggy enshittified crap that breaks every time a non-disable-able automatic update takes place".
    • I worked in a Failure Analysis lab for a major computer company. Some of our equipment used Windows XP and for a reason. If we were to upgrade our OS on some machines, the newer OS would be incompatible with the software that controlled some of our test equipment. The manufacturer of the equipment either wasn't around anymore or they didn't upgrade their application to the newer OS. Such was the case with our XRF and FTIR equipment. Same with our electron microscope.

    • Re:lab equipment (Score:5, Interesting)

      by innocent_white_lamb ( 151825 ) on Monday May 19, 2025 @12:33AM (#65386223)

      I run a movie theatre.

      My projector runs on Windows XP.

      The cinema server (that reads the movie files and feeds the data to the projector and sound system) runs on some version of Linux; the previous cinema server I had up to a few years back ran on Fedora Core 5 but I can't really tell what version the current one runs on since it's all hidden behind a javascript frontend.

      It's all just "magic boxes", locked down to the n-th degree so no even the service techs can deal with some of the internals because the movie companies are so paranoid about piracy.

      • Indeed, the important part about these machines is that the service techs keep coming. The baseline requirement for the OS is that the vendor is willing to support whatever hardware it is controlling and has a replacement part if it breaks.

        The moment your considerations turns to keeping something important running on parts bought on ebay your business has lost control of its own risk.

    • A lot of the time, people are unwilling to try. Just because the ancient software says, required Windows XP, does not necessarily mean that it will not run on on something newer. Windows does attempt to maintain backwards compatibility, so it is not usual for the software to work. They will say the vendor does not support it, but the fact that the vendor doesn't offer an updated software means their support is pointless anyway, they effectively no longer support it, if they require you to use a long obsolet
      • While that's true for usermode software, it's pretty common for hardware drivers to not work on future systems.
    • by mjwx ( 966435 )

      I work in a test lab. We have a number of pieces of old test equipment, that we use but not often and it does not make sense to replace, that run on XP. I think we finally eliminated the Windows 2000 ones. The manufacturers have no interest in writing new drivers or software, they want to sell you a new one. So, we keep them off the network and eBay spare parts ahead of time.

      Had this years ago with an old MRI scanner at a hospice (so not exactly rolling in cash to buy a new one despite the parent church literally rolling in it, gold for the temples, gruel for the dying nuns). What we did was virtualise the XP, put it on a standalone ESXi box running modern HW and connect it to the scanner. You could export files by burning them to CD. It was an imperfect solution but good enough that we didn't have to worry about HW failures.

      This was 14 odd years ago when VMware was pretty m

  • "Many ATMs still operate on legacy Windows systems, including Windows XP and even Windows NT," which launched in 1993, says Elvis Montiero, an ATM field technician based in Newark, New Jersey in the US. "The challenge with upgrading these machines lies in the high costs associated with hardware compatibility, regulatory compliance and the need to rewrite proprietary ATM software," he says.

    Long out of production, the few remaining LightJets rely on the Windows operating systems that were around when these printers were sold. "A while back we looked into upgrading one of the computers to Windows Vista. By the time we added up the money it would take to buy new licenses for all the software it was going to cost $50,000 or $60,000 [£38,000 to £45,000]," Watts says. "I can't stand Windows machines," he says, "but I'm stuck with them."

    Behold, the true costs of closed source software.

    • by jonwil ( 467024 )

      Even if it was running Linux, the ATM would still need proprietary software to do the actual ATM things and it would still need to have any updates approved through the regulatory process.

      • by Gravis Zero ( 934156 ) on Monday May 19, 2025 @12:25AM (#65386215)

        Even if it was running Linux, the ATM would still need proprietary software to do the actual ATM things and it would still need to have any updates approved through the regulatory process.

        If it's a POSIX program then it doesn't need any updates. At most, the libraries and possibly the kernel might need an update.

        • by jonwil ( 467024 )

          I suspect the entire software stack running on the ATM is what needs certification by regulators.

          • I suspect the entire software stack running on the ATM is what needs certification by regulators.

            Not if it's minor security fix. However, any major structural changes requires certification. Backporting security fixes can be done indefinitely so long as no structural changes are needed.

        • As far as I know, POSIX doesn't support any sort of graphical interface. Only stdout. An ATM requires a user interface that isn't a command line. This means that some non-POSIX code will be present.

        • Even if it was running Linux, the ATM would still need proprietary software to do the actual ATM things and it would still need to have any updates approved through the regulatory process.

          If it's a POSIX program then it doesn't need any updates. At most, the libraries and possibly the kernel might need an update.

          What does that mean, there are no POSIX touch screen interfaces, POSIX graphics interface, POSIX wtfever physical layer network an old ATM uses, etc etc and it's a custom machine from the 90s.

          "Update the libraries" bruh, seriously.

          The people running really old hardware don't want to own the software stack running it. They just want the hardware to function. This hand wavy open source stuff... YES, open or not, if you yourself own the software to your very old machine you can do whatever you want with it. No

    • by Moryath ( 553296 )

      Linux is dropping kernel support for older hardware too. [theregister.com] This isn't a "cost of closed source."

      It's two opposing forces: first that these systems don't NEED increased hardware power to do what they were designed to do, and second that building with off-the-shelf hardware means that hardware will eventually be out of production, though probably not as quickly as if you'd designed or bought a completely bespoke solution.

      Ripping out the guts of the system AND rewriting the software for a newer OS is costly.

    • I saw an ATM a few years ago where the software had crashed for some reason.
      Naturally I was curious, and it didn't take long to realize that it was running OS/2.
      Banks of course just LOVE IBM software.

  • Just the cost of not maintaining your own infrastructure with incremental upgrades as various pieces become obsolete. No one (or business) is "stuck" using old hardware. They backed themselves into a corner and then complain about it as if they didn't do it to themselves.

    • by dbialac ( 320955 )
      I strongly disagree. COBOL is still used all over the place on somewhat newer hardware on an OS that few people know. The benefit? How many people know how to hack into such a system. Replace the system with a Java/C#/etc. system and watch that level of security drop fast.
      • by Bahbus ( 1180627 )

        Disagree all you want. Your argument boils down to security through obscurity, which isn't a valid security policy.

        • Re:This is (Score:4, Insightful)

          by cstacy ( 534252 ) on Sunday May 18, 2025 @10:42PM (#65386119)

          Disagree all you want. Your argument boils down to security through obscurity, which isn't a valid security policy.

          The systems that I imagine he is referring to DO happen to have good (and modern) security. Better than just Unix, for example. The fact that there is an additional barrier to attack (having a clue) is another valid layer. Much of good security includes secrets. It's only a bad approach when (a) that's all there is and (b) that's not enough (it usually would not be enough).

          An understandable example would be in the domain of physical (e.g. building and facility) security. Much of it comes from the secret internal procedures, like the guard schedules. Of course, you probably need actual locks on the doors, too!

          Hiding the Christmas presents in a corner of the closet is quite effective against a certain class of attackers. Even though they are very curious attackers. It is the height of the shelf and lack of device for reaching, that is critical in protecting the asset (or even knowing it is up there). Later, you need to resort to the attic as the threat actors become more sophisticated... Obscurity is relative, and everything about security is about understanding the capabilities of the potential threats.

          The main thing about security through obscurity is that once it isn't obscure one day, the Ho Ho Ho will be on you. The relative value of what these particular systems are typically protecting is high, and the highly motivated attackers have already incorporated the obscure material into their packages. Because of their investment, those packages are not widely available. The obscurity factor makes the development of those attacks expensive enough that the field of actors is much smaller than for most systems. (But as with anything "secret", this is a fragile situation.)

          • by Bahbus ( 1180627 )

            Sure. When other security is also in play, it can help, but it just wasn't what he was saying. He basically said COBOL is secure because of it's obscurity (which isn't true), and that (because of popularity) Java/C#/etc. cannot be secure.

      • COBOL is still used all over the place on somewhat newer hardware.

        Admiral Grace Hopper made a major effort to verify vendor compilers on standard Cobol and to verify government code was sticking to the standard and not using vendor extensions. This ensured easy migration of the code from one generation of computer to the next, from one vendor to the next.

    • It's easy to spend someone else's time and money.

      I've dealt with some small businesses with obsolete hardware/software/OS. They don't have an IT department, often it's one non-IT person doing extra duty. (And yes, things get messy when they leave/transfer). There's no one to handle the upgrade, because everyone's time is 100% committed.

      Sometimes they -can't- upgrade easily. The software doesn't run on the latest OS. Licensing costs or hardware requirements have changed. Retraining staff will take the busine

      • by Bahbus ( 1180627 )

        I've dealt with some small businesses with obsolete hardware/software/OS. They don't have an IT department, often it's one non-IT person doing extra duty.

        Yeah, I've been that guy. And I've made them spend the money to update their shit. I told them I'm not maintaining this pile of shit. Then I made them hire a legit full time IT person. Money was saved in the long run. Not having at least one semi-dedicated IT person is a mistake on the businesses part - unless the business literally needs next to no technology to operate.

        Sometimes they -can't- upgrade easily.

        This, I don't believe actually happens as often as implied. I can think of a handful of extremely niche examples - and it's precisely why

  • Notes (Score:5, Insightful)

    by Artem S. Tashkinov ( 764309 ) on Sunday May 18, 2025 @05:31PM (#65385673) Homepage

    1. The people in question are not stuck. They live happily.
    2. Those systems work like clockwork. They don't usually need replacing.

    • Re:Notes (Score:5, Interesting)

      by Ol Olsoc ( 1175323 ) on Sunday May 18, 2025 @06:31PM (#65385763)

      1. The people in question are not stuck. They live happily. 2. Those systems work like clockwork. They don't usually need replacing.

      This one get it.

      Business is not generally in the Replace everything to get nothing more than we already have mode.

      IT guy goes to see the accountants...

      "We need to replace all our computers for a cost of A million dollars." CAO "Oh goodness, did something happen?"

      IT - Oh no, there are just new models coming out."

      CAO - "But did the old ones quit working? What are we getting for that million dollars?"

      IT - "Well ,they'll just do the same thing we've been doing for a couple decades now - then again, we're going to need about 200 K to rewrite some software that won't run on the new ones."

      CAO - So we're actually going to spend a million to get less functionality, and have to rewrite functioning software to get it to function on the new computer system? That's a really hard sell!"

      IT - "Well, when you put it that way."

      A lot of business savvy people still aren't used to the perpetual upgrade, nothing added cycle. It might seem like a black hole you throw money into.

      • Re: (Score:3, Informative)

        by dbialac ( 320955 )
        Somebody needs to share this idea with Elon Musk before he turns his IRS release into a giant hack-a-thon.
      • Every time I get the urge to upgrade my pc I ask myself, "Does it run everything you need to and reasonably well?" Yup.

        • Every time I get the urge to upgrade my pc I ask myself, "Does it run everything you need to and reasonably well?" Yup.

          My most recent upgrade was pretty recent - I bought a Mac mini M4. The speed increase was significant, and the form factor was a big part as well. Plus Apple had a good trade-in for my old Mac.

          On the Windows side, My laptop is 5 years old, and fits your description of running what I need it for. It is edging ever closer to getting Linux on it.

      • Depends on IT. I'm of the "ain't broke, don't fix it" mold. If users are getting stuff done effectively, you're not getting new stuff unless you can prove why we should spend time and money on this. Because that's what it is: time and money to replace things. I have enough on my plate that I don't need to add pointless tasks to my list just so you can have a new shiny.
        • If it is not internet facing and not locked to old hardware: sure just let it be.
          If it is one of the above it becomes a danger.
      • Now the reality:

        IT: We need to replace system X very soon

        CAO: "Why? It has doing its job for 20 years"

        IT: "It is getting impossible to find spare parts. I don't know if we can even get it up and running again the next time it breaks down"
        • Now the reality: IT: We need to replace system X very soon CAO: "Why? It has doing its job for 20 years" IT: "It is getting impossible to find spare parts. I don't know if we can even get it up and running again the next time it breaks down"

          That's a valid reason, when there are no more parts.

          Speaking of reality, there are still places using Internet Explorer 6 with a lot of Web apps.

          What is their reality?

    • by cstacy ( 534252 )

      1. The people in question are not stuck. They live happily.
      2. Those systems work like clockwork. They don't usually need replacing.

      I think it's the hardware problem that usually gets them. But it is amazing the lengths they will go to before upgrading. Even before trying software emulation of ancient hardware. Like manufacturing your own vacuum tubes.

    • 1. The people in question are not stuck. They live happily.
      2. Those systems work like clockwork. They don't usually need replacing.

      1. Is a guess. You have no idea if they are happy or not.
      2. Is a risk. Your key word is "usually". The difference between replacing a windows 11 PC vs replacing an MSDOS PC is the difference between picking up the phone to your vendor and having DHL on your door on the next day, vs running a multiple month long upgrade project that attempts to replace obsolete hardware from scratch against a statement of requirements.

      "If it ain't broke don't fix it" should really be "If it ain't broke you better have a very

  • The trains in San Francisco's Muni Metro light railway, for example, won't start up in the morning until someone sticks a floppy disk into the computer that loads DOS software on the railway's Automatic Train Control System (ATCS).

    I bet they could replace that with a floppy emulator, that'd load disk images from a flash drive instead.

    • by gavron ( 1300111 )

      $19 http://www.insight.com/en_US/... [insight.com]
      $30 http://www.amazon.com/GoTEK-S... [amazon.com]
      $35 http://www.amazon.com/Emulato... [amazon.com]
      Many others: http://www.shopfloorautomatio... [shopfloorautomations.com]

      Seriously spending $200M+ because you're stuck on floppies when a $18-$35 device will make those floppies obsolete in 10 seconds is absurd.

      I know, the naysayers and downvoters are saying "OH, but you ALSO need a USB stick!" You can get 100x for $106 at
      http://www.amazon.com/Wholesa... [amazon.com]

      So multiply how many floppy readers you have by $20 and that's yourr

      • by darkain ( 749283 )

        That first one at least, USB floppy drives? Are absolute DOGHSIT for quality. I would NOT want any production system relying on them. Hell, not even my home systems as a hobby. I purchased one a while back, and the "drive CD" was actually shipped and stuck inside the floppy drive, requiring me to open up the drive just to get the CD out. But besides this, this reading/writing reliability of the head was absolute trash. "Oh, maybe USB isn't powerful enough" - well, turns out it actually is! I built a greasew

      • It's not the cost of the replacement software. It's the cost of proving that replacement hardware will not introduce bugs that cause a failure.

      • They have definitely heard of these, and they have definitely been tried.

        Those emulators all work only if the floppy drive is used in a common way used by popular operating systems.
        They do not work if the original software used proprietary techniques for speed, copy-protection, unique operating requirements or other purposes.There were a lot more ways to control some old hardware to do odd and unconventional processes than are accounted for on modern emulators. Emulators just don't handle techniques like in

    • by ArchieBunker ( 132337 ) on Sunday May 18, 2025 @06:59PM (#65385809)

      What you say is 100% correct. Now with that out of the way, do you feel comfortable with your signature on the document saying the emulators won’t crash any trains? Liability is what this comes down to.

      • by dbialac ( 320955 )

        What you say is 100% correct. Now with that out of the way, do you feel comfortable with your signature on the document saying the emulators won’t crash any trains? Liability is what this comes down to.

        Who says the hardware from their running from the 90s even has USB ports? I'd encountered them previously on a Sun, but I'd never seen the ports previously nor anything that could connect to them. They weren't very common until the late 90s. If I recall right, Mac started the trend in the consumer world with the iMac. It beat Windows to a degree of widespread availability and it wasn't until when Windows 98 came out that you started to generally see USB ports on the hardware and USB devices. I don't know if

        • Look at the Gotek emulator. It acts like a floppy drive to the host system and stores floppy images on sd or usb drives. It connects through a typical ribbon cable.

    • by PCM2 ( 4486 )

      San Francisco resident here. As far as I can tell from all the various things I've read, the news surrounding this issue is badly reported based on interviews with uninformed and technically un-savvy public spokespeople.

      I mean, some of the stories talk about needing to find programmers proficient in "90s-era programming languages." Which ones might those be? I'm unaware of languages that were widely used in 1998 and are completely obsolete today. I'm betting the real challenge is finding programmers profici

    • Virtualization is great way to freeze a development environment at a point in time, quite useful as a lazy sysop, no library version mismatch problems... I wonder if you could take a disk image and go from there ... ?
  • by azander ( 786903 ) on Sunday May 18, 2025 @05:45PM (#65385691) Homepage

    We have an old Newspaper Laber called Prism. REQUIRES Windows XP SP2 due to 3 very old Propriety SCSI, and Serial cards to make it work. I'm looking for motherboards to replace the one we have before it dies. New hardware is upwards of $500,000 USD to replace it. Needs to have 3 PCI (1) slots. We can't get drivers for them either. Nobody makes them and the Manufacturer removed them.

    I keep old hardware around to read old files. Have an old laptop that rents out for $500/day to local operations for their CNC machines. They need it maybe once every 5-6 years, for a single day. It has paid for it's old circa 2000 price 4 times over, and my time, since I started doing that. The business are VERY HAPPY to have it even for a day to upload new plans instead of having to purchase a new multi-million USD device to replace it instead.

    THAT is the cost of closed source hardware.

    Az

    • Have an old laptop that rents out for $500/day to local operations for their CNC machines. They need it maybe once every 5-6 years, for a single day. It has paid for it’s old circa 2000 price 4 times over, and my time, since I started doing that. The business are VERY HAPPY to have it even for a day to upload new plans instead of having to purchase a new multi-million USD device to replace it instead.

      THAT is the cost of closed source hardware.

      I'm not entirely sure anyone would be convinced to make sure their next multi-million dollar hardware purchase is open source hardware to avoid $100/yr consulting costs. Is that really it? For the years that hardware has vendor support they'd be paying a healthy fraction of the upfront cost in support and maintenance. After that, extended support, maybe through a different company, again big bucks. A long tail after that of $100/year is ... nothing. I'm assuming they have a support contract of some kind wit

  • Is Apple hardware or software used much to run other equipment compared to Windows and *ix? It is probably a tiny fraction. It is much easier for Apple to deprecate their systems when the majority of their systems are used as desktop computers. Sure maybe there is some occasional older software with compatibility issues but the domain of issues is much smaller since they are pretty much only in the desktop space.

    • Yes, this story is about old Windows because old other operating systems were scarcely used for anything by comparison because they were scarcely used by comparison in general... except DOS of course.

      Most of us know of a few examples of other kinds of systems being used for special purposes, often Amigas because they had cheap genlock capability. The Prevue Guide is the obvious example. But as many of those systems as they were, they were nonexistent compared to CNC machines with only DOS drivers.

      • by Monoman ( 8745 )

        Yes it is and I should have made it clear I was commenting about the "Apple is ""really aggressive about deprecating old products," " part. There are reasons most companies chose x86, Windows, DOS, and *ix over Apple.

  • When a public service splashes out on something that works, or at least they can tolerate, they're loathe to spend any more money on it. I bet these ticker boards just work so they'll bodge whatever is needed to squeeze a few more years out of them.
  • This bit in the article about printers that use Win2k jumped out at me: "A while back we looked into upgrading one of the computers to Windows Vista."

    Why would you do that? And not just in a picking-on-Vista-for-sucking way, in general: Win2k is very out of support, Vista is very out of support; so either way you are going to need to have the whole setup dedicated to keeping those systems away from potential trouble. There are some incremental improvements(vista would mean being able to ditch SMBv1 and p
    • Lots of systems can only be upgraded one major release at a time. In cases like this jumping 3-4 major versions at once isn't an upgrade, it's a rebuild. They're likely comparing the two options.

  • The thing is, particularly for a large organisation, replacing an old system is very expensive, and as a "maintenance" task doesn't have a very appealing business case (when up against revenue generating projects that compete for funding). During that period in the year, when the budgets for the following year are set, and people start to cut items from their wish list to fit into the ever tightening budget, those sorts of maintenance projects inevitably drop off the slate.

    You might have a old Windows syste

  • Who's to blame for the these geriatric systems? My guess is that it's not Microsoft but rather the companies that make these systems. They choose stability rather than progress or efficiency, very much in the same way that space systems avoid new hardware and software and instead prefer hardware and software that has both been certified and shown to be stable via long use. These systems tend to be financial systems, safety critical systems, or non-maintainable systems (like spacecraft).

    • by Bert64 ( 520050 )

      Not just the companies that make the systems, but those who buy the systems.
      If you're buying a system you intend to use for 20 years, you shouldn't build it using components where the manufacturer is telling you support will end after 5. You need components that will be supported for the lifetime of the system, or you need to have a workable upgrade plan.

  • None of this stuff is bad, let alone "ancient" in any real sense of the term. It tends to be simple, fast, stable, and maintainable. DOS 3.3 is positively modern compared to some of the rock solid mainframe stuff written in languages like FORTRAN and COBOL in the 1960s. By comparison Windows has been going downhill for almost two decades now. Windows 7 was the high point of the Windows line and every version since then has been worse overall, sometimes much worse. The only problem with Windows 7 is Micr

    • by Shades72 ( 6355170 ) on Monday May 19, 2025 @12:36AM (#65386229)

      To my understanding, Windows XP/Server 2003 were the last versions of Windows that implemented the BSD network stack into Windows. From Vista on, Microsoft made their own networking stack, as the BSD one didn't comply with Microsoft's vision on how to implement a network stack into an OS. Maybe they thought they could do better, perhaps the ideas from MS about the subject were overreaching, most likely both.

      The network stack in Vista/Win7/Server 2008 wasn't too radically different from the BSD one. With Win8/Server 2012 the notion of telemetry came about and the MS network stack was being prepared for that purpose. From 2015 to around 2019 Win10/Server 2016 had a network stack much more optimized for sending telemetry without affecting the end user internet experience too much. Later versions of Win10/Server 2022 had much more telemetry and Microsoft did not care too much anymore if the user experience was affected or not. The network stack in Windows 11/Server 2025...well, with kernel 24H2 Microsoft decided that adverts are now part of what the stack needs to deal with too. Till now, they also decided that the end user experience should not deviate too much from the (later) Windows 10 user experience.

      With 25H2 and or Windows 12/Server 2028, I see the trend of deteriorating user experience continue, not necessary with more telemetry or adverts, but with Recall, which will sent your own data to MS for analyzation. And while MS has your data anyway...might as well train their AI with it. And don't worry, you as the end user will "happily" pay for bigger/faster connections from your ISP. And Microsoft will be more than happy to take your subscription money to sell you your own data back to you via their Co-Pilot scheme.

      I'm hosting my own AI/LLM solutions, am subscribed to ChatGPT and Claude, use the free version of Gemini...all of those provide me (somewhat) useful responses. Far from being a master of AI/LLMs, adept would be the best description of my skillset. Tried CoPilot again last week, tried it 6 months ago. What a disappointment it was back then. And last week's tryout showed very little to no improvement.

      Maybe CoPilot works for a bit if you are "hook, line and sinker" into Microsoft eco-system. But I'm not. And I don't see a real benefit anymore in doing that either. Anyway, expect Microsoft to "optimize" their network stack again for their new Recall functionality that they will push onto Windows users in 25H2 or sooner.

      • by butlerm ( 3112 )

        Thank you for that incredibly informative response. I was not aware of the history of their changes there, not that I can understand how things went so very very wrong. I means something like DECnet on VMS is reliable from what I understand and doesn't exactly implement a BSD networking stack either, except on Ultrix (which does using something like Berkeley sockets for its TCP/IP implementation) - DECnet is something quite different, and generally speaking highly regarded. And of course the NT kernel w

  • I am pretty most of the embedded systems I use -- like bank ATMs -- are running MS-DOS. Even as we speak. (My old girlfriend used to program them.)

    I myself did some embedded programming for credit card auth terminals (also cashier's check dispensing machines) for the major company that provides those to merchants. It was in Z80 assembler and some C on a home-grown just-barely-an operating system we wrote. That was less than 25 years ago. (Merchant credit card terminals were not yet on the network in the ear

    • by Bert64 ( 520050 )

      Most ATMs are running windows 10 these days, older ones running 7 or xp have largely been phased out at least by the major banks.

  • You wonder how much actual benefit the operating system upgrades have actually produced. If all you used your computer for was as a glorified typewriter NT was fine. If you have a mission critical system that works, what benefit is there? Particularly compared to the cost and risk that the new system won't work.
    • by Bert64 ( 520050 )

      Very little, but the problem is that the old systems are no longer produced - no software support, no spare parts, no security fixes. The market moves on.
      These are absolutely not traits you want on a mission critical system.

      • I don't know why those are really issues. Outside software support isn't something you need if you have a system that is already proven and working. Security for most such legacy systems is mostly a lock on the door to the computer room. They aren't attached to the internet and no one is writing new malware forl egacy computers. Parts are probably not an issue either since there are thousands of old computers out there to use for spare parts. Its not like floppy drives aren't available if you need them. In
        • by Bert64 ( 520050 )

          New old stock parts often can't be relied upon either, it depends on the nature of the failure.
          For something that's failed due to wear and tear sure, a spare part that's sat unused for 20 years is usually fine. But if the problem is due to something like old capacitors leaking then that part that's been sitting on a shelf for 20 years could be in even worse shape.
          You could also have moving parts that have seized up due to non use.
          Media also degrades over time even if not used and carefully stored.

          Once spare

  • None of this surprises me. In an enterprise setting you don't usually screw with something that works and has been working. I guess that's why we still have COBOL around; it's why I worked on a refrigerator-sized unit of a computer which ran our missile system when I served with the U.S. Navy (the Sperry/Univac Mk 152; I believe that same system was being run by some major-city public transit system - I forget which one). Constantly upgrading to the latest and greatest has a cost which goes beyond the initi
  • by kackle ( 910159 ) on Monday May 19, 2025 @11:19AM (#65387465)
    We are a multi-department, multi-disciplined manufacturer. When the plasma steel cutter in the shop failed, the younger IT staff didn't know what to do about the failing DOS computer that ran the ~ $150,000 machine. So they called upon this gray-hair to leave his programmer's desk and walk back there with his voltmeter. I saw in seconds that there was a bloated capacitor on the motherboard, next to the video IC that was the likely culprit of the failing video output. I ordered the $1 capacitor which fixed the issue.
  • "The challenge with upgrading these machines lies in the high costs associated with hardware compatibility, regulatory compliance and the need to rewrite proprietary ATM software," he says.

    A responsible organization accepted that every important piece of computation and machinery will need replacement. Every year, you put a little away for every machine's replacement. Of course, protecting that replacement reserve is hard. Hands find their way in for this project and that. Eventually, you're at the point where you NEED to replacement the machine, it's a recession, and the money just isn't there.

    So you go to the press to talk about how expensive it is to replace everything so that people are l

  • by abulafia ( 7826 ) on Monday May 19, 2025 @11:22AM (#65387477)
    I make things out of metal as a hobby. I have a modest home machine shop - I'm limited by what I can reasonably run in a (large, but) urban apartment.

    A few of the machines are CnC, and I went out of pocket for commercial software for modeling and to run them back when perpetual licenses were a thing. That was roughly 20 years ago.

    About 15 years ago, I saw the writing on the wall, virtualized that machine and removed its networking, and treat a copy of that image that hasn't booted since then like my birth certificate. It would cost about $3k/year to license modern software to do the things that does.

    It is effectively feature-complete for me, too. I don't collaborate, so I don't care about all that stuff, I obviously don't use the cloud, and also don't care about all the integrations, because this is just my hobby.

"I've seen it. It's rubbish." -- Marvin the Paranoid Android

Working...