San Francisco's train control system still runs on floppy disks, risking a "catastrophic failure"

midian182

Posts: 9,748   +121
Staff member
WTF?! Joining the list of organizations still using archaic technology is the San Francisco Municipal Transportation Agency (SFMTA), which has a train control system still reliant on floppy disks. It's said to have been cutting-edge tech when it was first installed – back in 1998 – but the city fears that if there isn't an upgrade carried out soon, there will at some point be a "catastrophic failure."

It's somewhat ironic that in San Francisco, home of Silicon Valley, the train control system is running off floppy disks, an entry in our Once-Iconic Tech Products That Are Now a Fading Memory feature.

Mariana Maguire, SFMTA Train Control Project, told ABC7 News that the SFMTA was the first agency in the US to adopt the floppy-based automatic train control system in 1998. That was the same year that the original StarCraft was the best-selling PC game, and even by that point, CD-ROMs were making floppies obsolete.

Reporter Luz Pena shows off a 3.5-inch floppy in the video of the report, but the article mentions that "5-inch" floppy disks are used in the system. It's probably 3.5-inch disks as 5.25-inch variants were even less likely to be part of "cutting-edge" systems in 1998.

SFMTA director Jeffrey Tumlin said the train management system was meant to be in place for 20 to 25 years. It is now entering its 26th year of service.

When asked about the urgency of upgrading the system, Tumlin said that although it is working fine, every year brings an increasing risk of the disks suffering data degradation. He added that at some point there will be "a catastrophic failure."

Also read: Once-Iconic Tech Products That Are Now a Fading Memory

The problem is that upgrading to a modern system will not only take around a decade, but also cost hundreds of millions of dollars. Tumlin said he hopes the majority of the money will come from state and federal grants, while the rest will come from "Muni's rapidly declining internal capital resources."

Floppy disks have clung to life decades after most people assumed they were long gone. It was only in 2019 when the US military dropped 8-inch floppies for SSDs in its nuclear weapons systems. We also heard that Boeing, Chuck E. Cheese, and many small businesses that couldn't afford to upgrade were still using floppies as recently as last year.

Permalink to story:

 
While floppies themselves are kind of unreliable under the wrong circumstances, 'Old' doesn't automatically means insufficient or unsafe: More than half of all of the banks around the world still run on Cobol simply because there's no good reason to change: Trust me if anybody could afford the absolute bleeding edge of technology it's the people collecting interests and fees for all of society.

Second and this is the big one, I flat out reject citing the terrible high cost to upgrade systems costing 'hundreds of millions of dollars' Just take ONE as in a single car manufacturer's revenue from a single year and tax it 10% and you would effectively have hundreds of millions of dollars given how their revenue in 2023 alone for all of the top 10 is measured above 100 billion each.

Consider it a fair taxation for the untold damage car culture has created as we near the point of no return in a climate catastrophe we're already experiencing in several parts of the world so 10% extra taxes a single year, given how like all corporations they pay 0 taxes anyway, would seems like we could not only modernize the train control system if we wanted to but we could build entire new rail systems all over even 'unfeasible large' territories like the US (Hint: it's not really unfeasible if the US already had an extensive rail network built 2 centuries ago) and actually solve some of the traffic issues, costs of living inside metro areas, etc.
 
While floppies themselves are kind of unreliable under the wrong circumstances, 'Old' doesn't automatically means insufficient or unsafe: More than half of all of the banks around the world still run on Cobol simply because there's no good reason to change: Trust me if anybody could afford the absolute bleeding edge of technology it's the people collecting interests and fees for all of society.

Second and this is the big one, I flat out reject citing the terrible high cost to upgrade systems costing 'hundreds of millions of dollars' Just take ONE as in a single car manufacturer's revenue from a single year and tax it 10% and you would effectively have hundreds of millions of dollars given how their revenue in 2023 alone for all of the top 10 is measured above 100 billion each.

Consider it a fair taxation for the untold damage car culture has created as we near the point of no return in a climate catastrophe we're already experiencing in several parts of the world so 10% extra taxes a single year, given how like all corporations they pay 0 taxes anyway, would seems like we could not only modernize the train control system if we wanted to but we could build entire new rail systems all over even 'unfeasible large' territories like the US (Hint: it's not really unfeasible if the US already had an extensive rail network built 2 centuries ago) and actually solve some of the traffic issues, costs of living inside metro areas, etc.
California is already one of the highest taxed states in the Union, yet they dont have enough money. Hmmm....... But that's always the answer isnt it? Want more rails? Raise taxes! Climate change? Raise taxes? new floppy drives? Raise taxes!

Maybe that isnt the answer?

If you want to fight the "climate catastrophe" then feel free to stop using electricity, running water, and modern infrastructure. All those conveniences are bad for the environment. Funny, I dont see people gunning to live like the Amish.....
 
Can't they just use a drive replacement, I use a Gotek in my Amiga for exactly this purpose... What would be the problem with them doing something similar ?
Well, you see, that would make sense and be cheap. Not enough money for the higher ups to skim off of with their contractor buddies. Everything has to cost extreme amounts and require total replacement at the taxpayer's expense.
 
Well, you see, that would make sense and be cheap. Not enough money for the higher ups to skim off of with their contractor buddies. Everything has to cost extreme amounts and require total replacement at the taxpayer's expense.
In my experience, large 'all singing and all dancing' systems are promised and users end up with a half baked solution that never gets properly fixed or updated. All too often system analysts under estimate the complexity and wrinkles within new systems.
 
In my experience, large 'all singing and all dancing' systems are promised and users end up with a half baked solution that never gets properly fixed or updated. All too often system analysts under estimate the complexity and wrinkles within new systems.

The problem tends to be unrealistic requirements, misunderstanding of the use case, or simply not understanding how complex the system actually is.

Yes, I live this life. And too often the answer is "The Requirements Changed", "How the system was designed doesn't work for us", and "Well, we only budged this much, so just get it working ASAP".
 
The problem tends to be unrealistic requirements, misunderstanding of the use case, or simply not understanding how complex the system actually is.

Yes, I live this life. And too often the answer is "The Requirements Changed", "How the system was designed doesn't work for us", and "Well, we only budged this much, so just get it working ASAP".
My response to how long and how much was always to point to the last few similar projects and say “about this long and this much.” Everyone would bellow “that’s too much and too long!” It was never about requirements as it was about disbelief and denial even when looking at our (or industry’s) actual track record.
 
Floppy disks are probably a "little" bit safer to the same degree a manual vehicle transmission is
safer. A lot of crooks probably never heard of a floppy disk and a lot of car thieves don't know how
to "drive a stick". ;)
 
My response to how long and how much was always to point to the last few similar projects and say “about this long and this much.” Everyone would bellow “that’s too much and too long!” It was never about requirements as it was about disbelief and denial even when looking at our (or industry’s) actual track record.
"But we've learned our lessons since last time."

Disregarding different jobs, requirements, and personnel involved.
 
If the problem is "floppy disks", use a different storage medium and emulate the floppy drives if necessary. If the problem is more than the floppies themselves, say so. I don't get this "hundreds of millions of dollars" thing.
 
If the problem is "floppy disks", use a different storage medium and emulate the floppy drives if necessary. If the problem is more than the floppies themselves, say so. I don't get this "hundreds of millions of dollars" thing.

The "hundreds of millions of dollars" is for bribes & graft reasons; as you said simply using a floppy drive emulator on a flash drive or three would only cost a pittance; said emulators are pretty well developed thanks to those interested in retro gaming. - Edit, aside from retro gaming, after looking into it a bit more there are some very good floppy emulating hardware intended for older industrial equipment, like CNC machines. Those sorts of devices would in all likely hood work just fine for this train system.
 
Last edited:
I recently recovered a dozen more 3.5" floppy disks for my retro hardware and restoration projects. Some even were under rain and sun. After a good cleaning (which includes disassembling and reassembling) and running them through a software for re-magnetize them, almost all of them worked fine. Anyway , I put those in the low quality section of my collection (I classify them by quality/state)

A friend also gave me some additional 5.25" ones, so I have more. Although I destroyed one while restoring a floppy drive. A shame because it was of very good quality, I didn't realize it when I chose it for the first test of the unit, should had used a low quality or bad one.

and about data degradation. Well, I have some floppy that still have the files as they were written in 1994, and they can be read without any problem. However, there is software that helps keep floppy disks in acceptable condition, including the data.
 
Last edited:
Our hospitals (and police stations) here still do that too, lol. Altho, as of last year.. they slowly started to switch to disks and windows 10. No more Windows XP I guess. ;p
 
While floppies themselves are kind of unreliable under the wrong circumstances, 'Old' doesn't automatically means insufficient or unsafe: More than half of all of the banks around the world still run on Cobol simply because there's no good reason to change: Trust me if anybody could afford the absolute bleeding edge of technology it's the people collecting interests and fees for all of society.

I agree there is no need to change an older technology if it works properly. But 2 points: floppy disks have to be swapped to some other reading devise simply because reliability and availability of floppies are currently very bad - prone to data corruption, new disks - if available - are very expensive, and those have limited life span.

But for the Cobol part... those legacy software blobs are not there because they are good. They are there because 'old masters' created them with limited to no documentation, provided hundreds patches with little to zero documentation, and there are no Cobol engineers with required knowledge to translate this to anything modern. The cost of update would include cost of reverse engineering, full process analysis for this part of organization, identifying hidden dependencies and software tricks people used at that time to get something working... and handling issues after transition because something was still omitted. Cobol doesn't work well in financial institutes - it is simply a portal to old, dangerous world...
Cobol is a living bomb, badly optimized, poor security, not scalable tech, where people for 30+ years were throwing another software knots and functions to achieve some obscured results. The only real way to fix that is to get some specialized model and AI but there is a risk this AI after going through that crap will become self-conscious and take over the world...
 
Consider it a fair taxation for the untold damage car culture has created....
So you walk everywhere? Or do own your own donkey?

...as we near the point of no return in a climate catastrophe
Stuff and nonsense. Even ignoring the wider issue of climate alarmism, the fact remains that the entire transport sector accounts for less than 18% of GHG emissions -- and personal autos less than 10%.
 
Back