It can't look like a floppy but actually be a keyboard or trojan spyware device or whatnot. It doesn't get direct access to my bus. It generally doesn't have conversations with my OS when I'm not looking, and I can see and hear when it's doing stuff. IT never felt the need to glue the slot shut. It's straighforward in the way paper is straightforward.
Perhaps the worst thing about the floppies still in use is that they're not even that floppy. You can't bend them without breaking them!
If we're going dedicated disk storage, I'd prefer something better. We had eSata for a while, but nobody used that. With today's SSDs, I can see a market where small eSata SSDs (or a variant thereof, carry some kind of NVME connector) take stuff to and from devices.
Alternarively there's good old SD-cards. They work with phones, with laptops, with desktops, they're dirt cheap and they're available in huge sizes. They're also easy to lose, often made from the worst flash memory produced, and can be WiFi cards and other weird expansions, but the attack vector is probably much smaller than USB.
In terms of security, the only real benefit floppies probably have is the terrible speed and the tiny capacity. Insert a USB drive with an exploit and a device is compromised within milliseconds, insert a floppy and you're spending several seconds listening to the disk being read. If something funky is going on, you have some time to yank the disk out. It's not much, to be honest, but it's something.
One time, this was when I was 8, a friend was immensely frustrated with a game, just extremely pissed off, and grabbed a C-64 game and ripped the internal, softer floppy out of the shell (8" pic, but same idea):
He did this by grabbing it and ripping it out via the center hole. The external shell was untouched/damaged. He then crushed it into a ball, like crumpling up a piece of paper.
Of course, I was pissed. But amusingly I then took it, and flattened it out, and then shoved it with all its creases .. back inside the floppy shell after opening it by its top flap.
I love the older actual floppies for their floppiness. Today we're back to bending things like screens and phones but bendable tech has been with us for decades!
If you have IOMMU enabled (like any sane platform does), no external devices have direct access to your bus. I know the DMA attacks you’re referring do and they don’t work in 2022.
Yes, the sound! It's a very important security factor, allowing you to spot suspicious activity. I still use HDDs for important data for that reason too.
I'm surprised nobody ever manufactured the floppy-disc equivalent of those audio-to-casette-player adapters. Making one would be a fun project...
But the real, practical answer is that there are floppy drive emulators that speak (whatever protocol floppy drives speak) to the host, but which let you plug a USB Mass Storage device into the front:
Just rip out the machine's floppy drive and put one of those in its place. Button on the front pages through virtual "floppies" stored as files on your USB stick. It's sort of like using a game-console flash cart.
They don’t actually emulate a floppy diskette, though. They have a signaling protocol that works through the floppy mechanism but they won’t just work in a floppy drive with the stock BIOS floppy calls. They need a driver to implement the host side of the protocol.
Sure, yeah. A real "cassette adapter" for floppies would need not just a regular one-head magnetic transducer inside the diskette, but some mechanism to determine which track the read-head is trying to read. Maybe another, much more sensitive magnetic transducer, to determine how far away the read-head is from it. Maybe a strain gauge, if the read-head actually makes contact with the medium. Or maybe just a bunch of metallic pads that close different voltages on an ADC ladder circuit — presuming that doesn't disturb the flux pattern being read out by the head.
Though I suppose you don't really need to know where the head is. You could just have an array of very small magnetic transducers, with good magnetic isolation between each one, where each one is looping over a different track; and whatever track the read-head angles onto, it just ends up "plugged into" one of the transducers that was already there waiting for it. Kind of like how a "cassette adapter" for 8-track players would have to work, if one of those was ever made.
Bonus thought: try to imagine how a "cassette adapter" for VHS would have to work. Tape with conductive wire run through it in on a matching slanted raster, etc. There's a reason they never made them :)
No, the reason is that there are no situations where you'd want one. If you have a VCR connected to your TV, this means it has a video input, so why not just plug your video thing straight into that input?
There are TV/VCR all-in-ones with no external inputs other than antenna-in. A VHS “coupler” would be the only way to route past the lossy tuner-to-baseband phase of one of these (unless you’re willing to open the TV’s chassis to bodge something in.)
For those interested in the details of the protocol, there's some information here (apparently from an attempt to create a Linux driver): https://schou.dk/flashpath/
That's not really what I meant; the device you linked is an example of the second kind of thing, a "floppy disk emulator."
A floppy version of a "cassette adapter" would be a floppy disk on one end, and a USB cable on the other. You'd feed the adapter disk into a floppy drive; the cable would dangle out of the front of the drive; and then you'd plug the cable into a USB mass-storage device.
As with an audio cassette adapter, this would be a way to feed data into a device through the magnetic read head — making the drive believe for analogue RF reasons that there's a flux-recorded medium in there; one that reads off as whatever data you're playing down the line.
> The only way to get anything off off it is with a floppy or an IDE HDD reader.
Another way is to dump the data in hex to the serial port, and write a serial port reader for your modern machine. That's how I transferred files from my old PDP-11 to the IBM PC in the old daze.
The company I worked for (Data I/O) initially told me not to write the program. I did it anyway as a rogue project. It turned out to be ideal for Data I/O's customers to connect Data I/O's LogicPak to a PC, and the salesmen started making copies on their own to give to their customers. It increased sales, and wasn't a big klunky program with a zillion unneeded features like Kermit.
Eventually, Data I/O made it an official part of the LogicPak.
It was a VT-100 terminal emulator for the PC, with a side feature of capturing the received data to a file, or writing the contents of a file to the serial port.
The systems are running RTX setups. Realtime motion planning and commands. I am sure it can be done in VM's (with some heavy engineering time) but its really beyond it's planned or expected life.
I've worked on legacy DOS systems, and virtualising them is generally the right approach. You can use just about any modern hardware, and the VM will nicely abstract it away. Plumbing hardware through to the VM can be a challenge, but not impossible to overcome.
Can you not find an ISA Ethernet card and copy the files over the network? There are plenty of TCP stacks out there -- I was even able to use ssh when I tried it years ago (though I had to use ssh v1 and increase the handshake timeout since my 386SX was struggling to keep up).
An NE2000 in a box without a bunch of other ISA cards competing for IRQs isn’t too bad. A DOS packet driver is pretty plug-and-play and will open up various DOS tools for common protocols (FTP, HTTP being the most useful for file transfer).
Yeah, but anything still running and important these days (and not already virtualized) likely has one or more custom ISA cards doing undocumented things.
You could get it to work, probably, but do you want to explain to the boss why the mass spectrometer is dead because you didn’t like using the floppy?
Got any recommendations for a good IDE HDD reader? I've seen some cheap USB ones on amazon, but all the reviews are full of people who either got non-working hardware or lost data using them. Seems like a fairly simple device, but it's like you either get garbage under $5 or something that sells for hundreds with little in between.
i wouldn't say the only way, though not fast, if you don't want to implement ISA 16-bit 10BaseT ethernet, you could use a rs232 serial port and something like 57600 8N1 connection and serial cable to a nearby linux system. at least then there would be no need to physically move media between systems.
I'm far from an expert in these things, but I've seen SD cards with WiFi servers that allow remote browsing photos on them straight from the camera itself (not using the weird WiFi-over-SD extensions to give cameras themselves WiFi, which the SD standard allows for).
Surely someone could take one or those IDE/CF adaptors and add small (100MB) write only and read only partitions controlled by a chip like an ESP32. You'd get two drive letters, one you can use to push data to the WiFi server and one you can use to pull data from there. That way you don't even need to write much of a driver for the DOS system at all because there's no locking and filesystem corruption to deal with. The parts are worth a dollar at most, all you'd need to figure out is the interfacing with the file system. You can even do Linux style device control by exposing fake files to DOS that the system can try to write to for things like configuration.
Maybe someone in the legacy gaming space can work out an open source product for this! I think many people would love to transfer stuff through easier means and selling such devices to enterprises stuck with old hardware can be a significant business opportunity.
I worked on an old but perfectly functioning device that scanned medical samples. The only way to get its data out was by way of floppy. Upgrading the device would cost far more than the time savings of cutting out the floppy transfer.
Do you not worry that it's obviously going to stop working at some point, and it'd be better to get onto a newer system while you have the time now and you aren't doing it when it's actually broken?
In grad school (07-13) at a top 10 engineering school world wide some of the fanciest equipment in the department (microscopes?) ran connected to windows 3.1 machine. The facilities manager kept spare parts in the closet.
The manufacturer had long abandoned the equipment and no alternative software was available.
The equipment would cost millions to replace and what they had apparently worked great. So the 386 chugged along.
I don’t buy this. Worst case scenario, the microscope talks to the PC using a proprietary hardware interface (say, an ISA card). It would cost far less than millions to hire a competent electronics engineer to reverse engineer the ISA card and the driver that talks to the ISA card, and fabricate a modern interface. Best case scenario, the microscope talks to the PC via a standard interface, and would be even cheaper to reverse, since you’d just need to worry about the software layer.
On the other hand, it would truly cost millions when the aforementioned ancient interface inevitably breaks and becomes impossible to reverse engineer, necessitating purchase of a brand new microscope.
>what they had apparently worked great
This is the real reason: because the scope works fine right now, nobody thought it prudent to invest the tens of thousands it would take to future-proof it. Exactly like companies that prioritize short-term profits over long-term success.
Lets say it costs 100k to reverse engineer a card, drivers, and software (I doubt it will stay on budget, I see how the software industry operates. The Uni will probably get hooked on SaS contract)
So now the department is down 100k for a hack. Does the hack even work? Probably not. The "engineer" probable over sold himself and got hired anyway because of the talent shortage.
So the uni is down $100k, the system chugs along with Win3.1 just fine, and every research group understand that when it finally breaks they'll all have to pitch in for a new one.
Btw, as to your comment on this department's myopia, the university is top 10. The department is top 5 world wide.
>So now the department is down 100k for a hack. Does the hack even work? Probably not. The "engineer" probable over sold himself and got hired anyway because of the talent shortage.
It seems disingenuous to assume the absolute worst case scenario is what will inevitably happen. The best case scenario is equally likely IMO: a reverse engineering genius undergrad will create a robust modern interface as part of a summer research project in the lab.
>Btw, as to your comment on this department's myopia, the university is top 10. The department is top 5 world wide.
That’s an appeal to authority fallacy. Just because you’re top 10 or top 5 or number one doesn’t make you immune to mistakes. Meta and Google are both at the top of their respective industries but make boneheaded decisions all the time.
If expensive essential apparatus you don’t have is available off the shelf, you buy it or go home.
If expensive essential apparatus you already have needs floppy disks or serial ports or twinax or a green screen terminal and a dot matrix printer or Indigo2 Maximum Impact with a black box hanging off the scsi port or whatever, you have a tech keep that shit running. If it breaks and would require serious PI time to manage reverse engineering and replacing components, you buy new apparatus if it exists to be bought!!! That’s what you do, or you get steamrolled by other labs that are doing research instead of running an equipment nostalgia club.
And the sensible way of keeping mission-critical legacy hardware running is to emulate everything that can be emulated using modern, interchangeable parts. An SD-card-based floppy drive emulator will be a lot more reliable than a decades-old combination of physical drive + floppies. An entire DOS or Windows 3.x computer can likely be replaced by modern industrial systems (also popular in the retro community) that reimplement i586 in SoC form.
"because the scope works fine right now, nobody thought it prudent to invest the tens of thousands it would take to future-proof it."
My "appeal to authority" was in response to this comment - this department has proven its ability to manage its affairs for far longer than the lifespan of the software running their equipment.
Windows 3.1 came out in the early 90s (late 80s?). How often have the driver APIs, the electrical buses, etc changed since? Should the department have bought new equipment at every change to be "up to date"?
This department did the right thing. They invested in other equipment and kept a couple thousand dollars of spares on hand.
Same story here: We work with a custom build machine that is controlled by an C64 mainboard... it works perfectly since forever and C64 parts will be available for the foreseeable future, so why change?
Given its popularity in the retro community spares may continue to be available for a long time. You can already get entirely new boards and drop in replacements are appearing for the various custom and out of production chips.
You’re often better running the old one into the ground - the new one will be just as expensive when it comes time but you’ll be getting the current model at the time.
It’s also the story of why I don’t care anymore if some ‘engineer’ comes crying his XT that he needs 10 minutes ago doesn’t boot anymore when ‘it was working perfectly fine yesterday’. Yes it costs money to stay up-to-date and if you don’t want to pay and it breaks, you get to keep the pieces.
And the cycle of companies who only do R&D for project based work. This is the real issue. I would say it's a problem with middle and high level management, but I am still not sure if that is fair.
That's a little short-sighted isn't it? If you define "works" as "worked the last time we tried it", that's not really something you'd want to have to depend on for anything reasonably important.
I'd say if if you cannot replicate the system it's already time for some investment. But even before that people should have the foresight to keep systems up-to-date or at least bridged to contemporary systems.
Or replace the floppy drive with a floppy drive emulator which uses an SD card as a storage medium. They're popular in the retrocomputing community. No other hardware or software changes required, and you replace a whole bunch of moving parts that are bound to fail with something simple and solid-state. Same deal for hard drives in old PCs: image them, and then pop in a faster and moving-parts-free drive emulator.
Then you have to make sure that there are no IRQ, DMA, or IO Port conflicts on what is probably an industrial control device device with data collection boards boards using all kinds of non-standard resource addresses.
And then, god knows if the device driver for the Ethernet card can coexist with all the other drivers and TSRs, or even if it's running a mainstream version of DOS. Maybe it has a DOS Extender that makes something fail in a non-obvious way.
Adding hardware to Old PCs isn't what you're accustomed to.
They often have a serial or parallel port, especially the industrial ones. Was the only way of communicating with them. That port might be used for the actual job the computer does though.
Yes, exactly. On these old systems you often run into issues with IRQ conflicts for PCI cards. It is common to disable USB and use a PS/2 port for a mouse and keybord.
The US Department of Defense moved away from using 8" floppy disks for its nuclear missile command in 2019. They were replaced by a "highly secure solid state digital storage solution" [0] which sounds like code for a floppy to SD card adapter.
I think this article's mention about MiniDisc is wrong -- they are probably talking about MO disc instead. (MiniDisc wasn't widely used as data media as it required a less available unit for writing those. MO, in other hand saw big adaptation in Japan, at least as much as ZIP disk did elsewhere.)
At least for MO, it looks like they are nixing them earlier than floppy disks, as manufacturers are no longer making drives and discs for it.
MiniDisc in an MO media. And I'm not sure about Japan, but judging from what came over to North America and the UK, writing to MD was SUPER popular. Though that might be because the major use case in export markets was as a more compact CD player/writer.
MiniDisc, at least in Japan, was mostly used for writing musical data (and it required proprietary software to write in special ways, and wasn't mountable as a file system. (e.g. you couldn't just copy over .mp3 file and listen to them, so it had to be converted/encrypted/recorded onto MD in a way it is not readable in ordinary means.)
There was a specification called MD DATA for this purpose, which could be used to store data in more ordinary way, however such discs could not be read by musical device, and had to be one of handful device specifically designed for this purpose -- while it was also used storing pictures from digital camera, as far as I know, it only showed limited use compared to MO as data medium. (Those MD DATA devices could also read musical media, but couldn't record them.)
Later in MD lifecycle, specification like Hi-MD did allow data featured storage class support (which you could write like 300MB of data to standard MD discs, and 1GB on Hi-MD media) this was very late in the MD lifecycle.
MiniDisc used as MO technology at its base, but it was definitely not used interchangeably, and MiniDisc was really specific media (mostly) used to music.
As a sort of related piece, I recently read the book Rising Sun by Michael Crichton (1993). In the book he describes a future where the Japanese effectively become global hegemon, dividing American industry into pieces that are easily bought by Japanese conglomerates, all of whom are years ahead of the Western world in regards to technology.
It's a fun piece of media that projects a future that didn't happen, and even more interestingly, was really the opposite of what happened.
The book played on fears of American inferiority to the rise of eastern nations, and is a great reminder that no matter what we may think of as true now may not be the case in the future.
Hearing about how far Japan has fallen behind in the technology race really brings to mind the fact that at one point, everyone was afraid that Japan would take over the world with its high tech industries.
Worth a read just for how wrong Crichton really got it.
The article only makes a quick mention of one case of data insecurity, but I feel like it's the real driver for this change. There have been several data loss incidents just this year, and data storage and portability like the fax machine in Japan are anachronistic reminders that data privacy and security only move forward after damage is done.
I know that there must exist adults working in the software industry that have never seen a floppy disk. Because I'm old, and it's been probably fifteen years (to be conservative) since I've held a floppy disk in my hands. I can understand a resistance to change, as we all know how well moving it "online" can go versus a working system, but where in the hell were users even supposed to find these disks even, say, five years ago?
As an off-topic peeve, why is "Nix" capitalized in the title when nothing else is?
Indeed, and it'll be interesting to see if the little floppy icon that has become the universal pictogram for 'save' in applications survives into a time when even the hoariest devs and designers have never seen such a thing in the flesh. I suspect that it will, as the symbology has taken on a life of its own, divorced from its origins.
> As an off-topic peeve, why is "Nix" capitalized in the title when nothing else is?
Good question. The entire title is original, even though the article title (“Japan declares war on floppy disks for government use”) is clear and concise.
No, they are not manufactured. The last manufacturer (Sony) ceased in 2011 or so. Some entrepreneur bought a ton of stock and is still selling it, there was an article around about this which I read recently but I can’t find it at the moment. A lot of customers who buy floppies are buying them for various government equipment, or shops with CNC machines programmed by floppy, etc. If you imagine that a CNC machine is designed to last decades and costs tens or hundreds of thousands of dollars, then buying floppy disks makes a lot of sense. If your government ministry requires floppy submissions because nobody bothered to update the rules then it makes less sense.
My life experience: be in another country, have electronic documents on flash pendrive, bring to government facility, have recepitonist read it, take back home, discover sophisticated trojans on the drive. After that, I only bring DVDs.
One of the problems of floppy disks IMAO is that they need to use a lot of plastics for storing tiny amounts of information. We don't live in that cheap one-use plastic utopia anymore
Can be useful for security, but is not a wise way to use the hardware resources.
I lived and worked in Japan and I was shocked how outdated the country was in many areas.
Opening bank accounts felt like I would imagine opening bank accounts was in the 1980s here in Germany. FAXes are still a huge thing. People buy CDs and DVDs. The architecture looks like it stopped developing somewhere in the 70s or 80s. Work culture is archaic, from absurd hierachies, to protocols (who sits where, who speaks first, who is allowed to speak louder, who must speak quite), to outdated offices (even for the most modern companies that i visited), tower computers, black suits with ties... puh, it was a nightmare. I dont want to start on the society and its outdated values (compared to where we are in the West): how they understand democracy, how they treat minorities, how they deal with their nazi-past.
I wonder every single time when media is reporting on the war against China and that Japan and Korea are our friends because we have the same values. Neither Korea nor Japan share most of values with us, in my opinion. Japanese and Korean friends agree.
There were other things that felt modern, but those were mainly driven by those state-sponsored and protected engineering companies that were founded during the post WW2 times.
Overall I would say: I did enjoy my time there, but I hated the racism, the work-culture and the bureaucracy.
Who to blame? I blame Raegan (Trump's big hero) who became president prior being a TV star and Raegan's slogan "Let's make America great again" and his campaign against Japan as he believed that he must destroy the Japanese threat by launching a trade war against them and by fighting the currency. Basically 1:1 what Trump copied 1:1 - just against China.
> I wonder every single time when media is reporting on the war against China and that Japan and Korea are our friends because we have the same values. Neither Korea nor Japan share most of values with us, in my opinion. Japanese and Korean friends agree.
In this context, it just says that Japan and Korea are both in western democracy camp.
Literally read this while loading an old Famicom disk game into a Nintendo disk drive. Thirty years and it not only still works, it was more capable by its design. Nothing wrong with magnetic media, but good luck on your modernization effort.
Floppy disks are not manufactured today. Maybe there is nothing “wrong with magnetic media”, but there is definitely something fishy about running your business on equipment that can only be replaced with used stock & the various supplies of NOS floppies.
Floppy disks are also extremely difficult to repair, and realigning a misaligned head requires equipment and media that is difficult to find and expensive to acquire. If you never had to adjust the head on a floppy disk—well, let me tell you—you have to get a special floppy disk to realign it and use an oscilloscope to read the signal test pattern from the special disk while you realign it.
“Not manufactured for over a decade” and “requires specialized equipment and training to repair” is a bad combination.
To be clear, there are a lot of scenarios where floppies make sense today. You may have an old CNC machine or an old chemical sample analyzer that takes floppies, and it may cost five or six figures to upgrade.
The thing that's wrong with magnetic media (specifically the floppy disc and audio tape) is that the head is in contact with the medium, which slowly wears it out. It only has a certain number of uses before it dies (also, the plastics will probably disintegrate at some point). Hard disks are much more reliable for longevity. Not sure about flash (or backup tapes - I assume they've been designed with longevity in mind)...
Sure, that's why it failed, but it's bizarre to me to focus on something metaphysical like that. If something can go from horrible to something else without changing a single atom, just Sony publishing some things under a license? What was actually horrible about it at that point, how did you let the circumstances overshadow the object?
That tells me something about you, but nothing about MiniDisk. Which was a cool little futuristic cartridge thingy.
USB isn’t just storage, it’s way more than that. If you plug a USB stick into your computer you have no way of knowing beforehand what the device actually is or what it does. Is the device a normal USB flash drive? Or is it actually something far more malicious?
There are USB sticks, in the wild, which impersonate a keyboard. When connected to a computer, at some point, they will send keyboard commands to the computer in an attempt to fire up PowerShell and run commands that install malware.
In general, you should not use a USB stick unless you know its provenance. Consider them highly dangerous. Be especially suspicious of USB sticks that you find “lost” near your business, since somebody may be targeting your business—this is a common technique both for pentesters and for actual criminals.
USBs usually have a CPU inside that sits between the storage and PC. This way the PC doesn't need to know a thing about the storage and the CPU can do whatever it needs to implement the USB protocol. However the CPU can be malicious and modify files
I guess this is the answer... but honestly would hate having something that barely fits in my pocket around... I suppose sdcards are the new floppys but they are much too small, not to mention that they could stop being writable at any moment, but I guess its better than both...
The last Visual Studio that I had to install from floppies needed to be transported in a couple of sturdy shopping bags - something definitely had to give. Maybe p-code would have got a second wind to reduce the amount of space software binaries needed, but as it was CDs rode in to rescue us (and bloat!)
But trying to fit things on them was fun. Was a fun way to learn dd and how boot sectors work. And what new technology can double capacity with a hole puncher? None I tell ya.
It can't look like a floppy but actually be a keyboard or trojan spyware device or whatnot. It doesn't get direct access to my bus. It generally doesn't have conversations with my OS when I'm not looking, and I can see and hear when it's doing stuff. IT never felt the need to glue the slot shut. It's straighforward in the way paper is straightforward.