This USB flash drive can only store 8KB of data, but will last you 200 years
tomshardware.com
tomshardware.com
SO - unless the data you're storing is worth a major effort by future computer historians, just find an old daisy-wheel printer, and save your 8K to archival quality paper. That'll be good for a thousand years or so.
Vs. reading data on this device, 200+ years out?
As long as it hasn't burned, or crumbled, or molded, or...
Still depends on humans in 200 years having use of computers and electricity. Civilisational collapse would likely mean humans will lose those skills (like we forgot how to make cement/concrete after the fall of the Roman empire and it took 1000 years to rediscover).
This. The best long term archival media that can outlive generations has proven to be non-electronic ones (paper manuscripts, animal hides, cave paintings, rock engravings, etc).
Even if you can still find USB readers 200+ years from now when we'll all use quantum computers and wear AR headsets or neuralink implants (provided we don't extinct ourselves through war and climate change till then), there's no guarantee the systems of the future will still be able to read file systems and data formats encoded in the standards of today, but to read stuff off paper/stone all you need is your eyes, no technology.
For example, some retro computing youtubers I follow are struggling to decode 5.25" floppy data from 50 years ago because it was encoded in some niche proprietary non-standard format that modern SW has no idea about, so they have to low-level scan the magnetic flux off the things and then trial and error various encodings hoping something will stick. Meanwhile, a random small church in my country just found 500 year old books in its attic that are perfectly preserved and readable, and they have survived a church fire and over a dozen wars.
Not saying such USB drives aren't useful for a specific niche, I'm sure they are, but if you're looking for 200+ year endurance you'll definitely want to diversify your data storage to non-electronic mediums as well and not bank it all on a USB drive with new unproven technology.
If modern society survives continuously for 200 years, I don't see any reason we couldn't read current formats if they're actually standard and/or well-documented; we can, for example, still read original unix filesystems, if nothing else by running actual original unix in (open)simh, and that's 45-50 years old already.
Also, a lot can happen in 200 years that can have unforeseen effect on the FRAM of that USB stick (EMPs from nuclear explosions, solar storms, etc)
Even old USB 1.0 devices sometimes struggle on modern HW.
Despite the claimed 10,000 year lifespan, I doubt this will satisfy the kind of hackernews who prefixes four-digit years with an '0' like 02024. We just don't know whether we will have or be able to make the machinery to read it back millennia from now. It will make a nice archival solution for businesses and governments who want a data preservation solution over a time horizon of decades to centuries, though.
I doubt it. Even governments and businesses suck at preservation: just look at game studios who loose source code of valuable IP they themselves developed 20+ years ago. People working there just don't care about something so long term when their own skin isn't in the game and they're not getting paid enough for it to care much about what happens when they're gone or what their predecessors did.
This isn't the famous Voyager space probe where people hang around well past their retirement date to keep something alive due to emotional attachment, in most cases digital data just is left to rot and die out as people leave companies/institutions and new people come in with different agendas and career goals.
Microsoft's magic Tesseract data storage cube is cool, but good luck finding the tech to read it 10k years from now when Microsoft is long gone. Written paper and cave drawings will outlive us all.
I'm definitely on board with the analog approach, but this is interesting. What's the reason for recommending a daisy-wheel printer compared to other printing technologies? Is it just that the deformation caused by the impact of the type serves as additional redundancy or is the ink (or its adhesion to the paper) somehow superior to other printing processes in this regard?
The USB itself is thousands of lines of code for USB driver in computer and special USB physical interface as well which might not be even manufactured in 200 years.
It would be like today looking on some crazy parallel interface from 70s using non standard bit sizes (i.e 12 bit sizes bus) with complicated timing patterns and 5V TTL on physical layer if you are lucky.
Unless a dog eats it
If you want to assure durability there are better methods. Good paper, parchment, fired clay, or high tech ideas like storing data in glass. But in terms of easy and cheap methods for one or two human lifespans this occupies a niche (together with archival-quality CDs and the like)
Unless we hit another Dark Ages, I think its safe to assume the USB standard will survive.
200k on paper. 30k if you don't want to require an actual sheet scanner to read it.
To read it requires software, but it's open source software not tied to any particular hardware or os platform, and you use any kind of imaging hardware you want of the time. It doesn't matter how much cameras or os's change in the future.
Also FRAM reading is destructive. FRAM internally destroys and re-creates the data every time it's read. MRAM does not. idk offhand how the two compare on things like data loss from environmental stresses like extended periods at low or high temperatures, or thermal cycling etc.
For 600 dpi printer you will need a scanner with at least 900 dpi physical (let me emphasize, physical, not interpolated) resolution.
The Tom's article spins it as a pointless storage device and pushes the key feature of it including a built-in serial terminal text editor down under several paragraphs of snark.
I'm not telling you to buy one and I don't know who needs one, but it's a bit unfair to judge it as an archival storage device when it doesn't even seem to be trying to be one.
One extra pedantic point: assuming one could reasonably expect 200 years out of not only the FRAM but the rest of it as well... you don't really want your really good data to survive 199 years before a catastrophic slightly early failure so maybe leave instructions to access and/or duplicate the data before, say, half-time (100 years).
[0] https://www.cnx-software.com/2024/05/15/blaustahl-usb-storag...
Or go with the 64kb version for 6 dollars.
[0] but ignoring obvious simple answers like "depends on the size of the stone"
But the long term storage angle doesn't make much sense to me, you can get 8kb in a couple of QR codes.
Much better would be to market the number of read-write cycles as the USP over flash - so you could use it as some kind of constantly updated but persistent, encryption key or parity code or whatever.
I'm going to make a safe bet that there's no way any of the capacitors involved survive anywhere near that long. Now of course, technically the data is still there, but I doubt it in 200 years it's going to be all that easy to replace the microcontroller while retaining that data. A competitor in this race would be the connector:
> The Blaustahl includes 4MB of NOR flash for firmware and a USB Type-A male port that requires no additional drivers on most operating systems.
Something tells me that USB type-A is unlikely to survive the next 200 years.
Hey, what happened to usb-b?
You can find USB-B ports on e.g. laser printers (where size doesn't matter) or on some older USB drives (where you'd want USB 3 speeds).
I think i even had a sata to usb box using the huge usb 3 version.
Those were fine-ish, but their micro versions can't die in a nuclear explosion fast enough...
The pico microcontroller, on the other hand, has normal flash and won't last very long. Here's an interesting article:
https://www.ni.com/en/support/documentation/supplemental/12/...
Maybe keep it in the freezer?
Otherwise I would expect tin whiskers or some other process to dominate reliability at long periods.
More generally, all electronic components can fail [2] and there's a lot of such components in this device. This is applying component-level thinking to a systems level problem with no real way to verify their claim (they don't even claim to have done any aging test to try to simulate how long the device could last) and why this should be called out as nothing more than a marketing gimmick.
[1] https://www.edn.com/class-2-ceramic-capacitors-can-you-trust...
[2] https://en.wikipedia.org/wiki/Failure_of_electronic_componen...
Why? All capacitors on the flash drive are MLCCs, which unlike their electrolytic counterparts aren't exactly known for failing - let alone simply due to age. When push comes to shove you can just remove them without meaningfully impacting functionality anyways.
[1] https://page.venkel.com/hubfs/1_Images/Resource%20Center/Cro...
[2] https://www.edn.com/class-2-ceramic-capacitors-can-you-trust...
- ESD / supply voltage spikes. This device has basically zero protection against them. The RP2040 is well-known to be quite sensitive to them, so I'd expect it to kill the controller relatively quickly. The FRAM chip itself isn't as exposed, but could still be damaged.
- USB-A becoming obsolete. I fully expect USB-C to replace it on new machines within a decade - and we'll probably switch to something completely novel within the next 50 years or so. You'd have to get a PC out of a museum to access it.
- Flash failure. The data is stored on an FRAM chip, but the controller's firmware isn't. That chip is rated for 20 years (but I do expect it to last quite a bit longer).
- Crystal oscillator failure. It's literally a crystal, which is vibrating. They can fail.
- LDO fail-closed, exposing the MCU, flash, and FRAM chips to an overvoltage.
- Tin whiskers. The device is almost certainly using lead-free solder, which means it is susceptible to the solder growing tiny "spikes", which can lead to shorts. We've gotten quite good at managing these over the years, but all bets are off when it comes to a 200-year scale.
Most of those can probably be recovered from, as you'd still be able to attach debugging tools directly to the chip and bypass the controller. The biggest risk is probably losing the documentation, which makes accessing it an awful lot harder.
I also vaguely recall that Sonic 3 used this storage type? Like it was a weird outlier for cartridge-based saves IIRC.
Edit: Yeah, it did: https://hackaday.io/project/13425-sonic-3-feram-adapter
... which is certainly not guaranteed for 200 yrs.
Contradiction in terms.
I mean, maybe i'm asking too much, but a reliability training for journalists, will be nice. /s