Hacker News new | past | comments | ask | show | jobs | submit login
New hard disk write head analytical technology can increase hard disk capacities (tohoku.ac.jp)
94 points by XzetaU8 9 days ago | hide | past | favorite | 51 comments

Some days I dream of a distant future in which an average being (human or otherwise) comes in contact with a hard disk drive containing humanity's knowledge. I think about the vast number of layers that have been combined to create such a marvel of technology, simply to store data in an efficient way. Sure, you might be able to figure out the physics of magnetic storage, or even the basic electronics, but what about the motor controller? The architecture and firmware on the storage controller? The electrical interfaces to the storage controller?

Let's assume you even get that far.

What about the physical layout of data on the drive? The filesystem? The file formats?

That's not even all of it, and there is just so, so much that would be incomprehensible to a single individual, however intelligent.

It's truly incredible how we've been able to combine years and years of knowledge into something most humans would just take for granted.

“Any sufficiently advanced technology is indistinguishable from magic.” -Arthur C. Clarke

I've thought about the same thing, but I suspect the main hurdle to overcome would be recognizing it as a data storage device in the first place. If one had an inkling of that use, then some future archeologist wouldn't need to know much about the electronics or controller. They might image the magnetic fields of the platters at 1000x the resolution and sensitivity of the drives original read/write electronics, then treat it like past archeologists might have treated a tablet with cuneiform.

At that point, you might record storage media and technology information at lower resolution to make it easy to read and easily identifiable. Like a text on how this thing works engraved on it, with a magnifying glass. We partially do that on product labels that say you should open it only in high vacuum or laminar chamber. (Simplified to "do not open" for normal people.)

The LongNow foundation has thought about that, too. They start of with “how would people be able to understand that text?”. See https://blog.longnow.org/02012/10/12/decoding-long-term-data....

I remember reading about a project that started with instructions on how to build a stage 1 instrument (a microscope, IIRC) that allows you to read the instructions to build a stage 2 instrument that allows you to read the instructions to build a stage 3 instrument, etc. but cannot find it.

Sounds kind of like this paper from 2015[1], which has been a source of inspiration for me for the last 5 years since I read it.

[1] https://archive.org/details/tr2015004_cuneiform/

Thanks! It’s not what I remember, but interesting, nevertheless.

Is this the one you are referring to?


No. That doesn’t describe technology needed to read it, and doesn’t require different tech (only better and better eyes). There was a way more convoluted one. Maybe it was a SF story.

Ive had this idea kicking around for over a decade to make ceramic or granite slabs of sufficient dimensions as demonstrated by artifacts found from archeological digs so as to withstand elements and time and etch the surfaces of sufficient depth so as to maintain their features and then have an analog projection apparatus that uses sunlight so as to magnify the feature, also constructed of similar means. Design the components all open source buy plots in cemeteries around the world and any where else that wont be subducted via techtonics in any time soon and just bury wikipedia or whatever else. The Long now meets that scene in the time machine basically. All in the hopes of confronting this exact scenario you’ve outlined.

The Long Now foundation isn’t only building a clock. It also is thinking about long-term conservation of data. See https://rosettaproject.org/)

Reading about their clock and my subsequent research into how one would "communicate" through design that a particular site (say, nuclear waste dump) was harmful and not something to be messed with or explored was the original impetus for this thought experiment.

It's fun to dream!

Off-topic question: has anyone used M-Disk BD disks to make long-term backups? I got recently interested in it after reading that US Navy tested it in Arizona and found it to be the only medium without read errors after the tests. But that was apparently more than 5 years ago and I stopped hearing about disk drives around then. Is it an unreasonable idea to buy a USB writer drive and give it a go?

I use them. Burn each copy twice. Par2 files on separate disks. So far no read errors (verified yearly)

Well, even normal BD is very resilient, apart from delamination. But that's a gradual problem you can preempt with multiple parity discs for each bundle of data discs, along with periodic checking of the discs (delamination is optically quite obvious).

I really want to. I just wish that the BDXL discs were, like, half as expensive as they are.

It's not M-Discs fault of course, I just have too much data...

yes! i've used Verbatim BDXL M-Disc 100 GB discs for backup. i just store the data in a Vercacrypt container for security. there are some low quality drives and discs out there so you need to do your research but I've been using it for couple of months and they've worked great. you can get 5 discs for $60 on amazon.

Add video codecs (and the operating systems they run on) on top and you know why the library of congress decided lasering movies onto film (3 black and white rolls, one for each color channel) is the best storage format for something as simple as moving images and sound.

A cool project that responds to or takes account of this (trying to make a storage artifact that someone plausibly could learn to use without most of our technological knowledge) is https://rosettaproject.org/. It's a somewhat high-density archival storage medium that tries to give a clear hint of what it contains and how to read it, just requiring optical magnification.

Of course, that means its overall capacity is many orders of magnitude less than the hard drive you describe.

This is a fascinating thought experiment.

I think equally as interesting an experiment would be to imagine if an alien race found this drive, instead of a future human.

I imagine, to them, it would be like a modern day auto mechanic admiring all of the analog wonders of an antique car from the 1950's. Even though they may have no experience with the technology they are probably able to recognize the roots of the technology and perhaps are able to quickly deduce how it works with simple observation.

I’ve wondered how extremely bright ancestors figured out the basics. Carrying water. The benefits of cooking food. How rocks or logs rolling could become a wheel. How you actually kill a large animal the first time. How to try berries in a new area. How did they figure bread out?

Bread is actually not that difficult to figure out. Grains are easier to digest if you grind them up. Somebody didn't want to use their teeth to do so and invented a form of dough. Yeasts are everywhere, so it's just a matter of bringing the dough sufficiently close to your fire to get something that resembles proto-bread.

Sure but each bit isn’t obvious before you’ve seen it. Lots of people you know would just chew the grains after boiling because that’s what we’ve always done. Every so often someone stops and thinks about trying to crush it.

How many remote tribes have bread? Obviously generally requires precursors like farming.

You just oversimplified picking grains at the right time, grinding them to a powder, mixing it raw with water, but not too much, adding a third weird ingredient, and cooking it. At every stage you have a product that’s not really edible. Someone had to experiment.

Minimal bread requires only grain and water. At first this would have been unleavened bread, but because the yeast is already present on the grain, leavened bread can be accidentally invented just by forgetting about it for a while.

This is basically a creationist argument - life is complex so someone must have pulled the strings. But given enough time all of that can randomly go exactly right.

Obviously it did, I was just arguing against “is actually not that difficult to figure out.” Everything is easy after you know how it’s done, but before it you only have grain berries and maybe a soup with grains.

it's important to understand that evolutionary processes are not random. People experimenting with proto-bread would tend to prefer the best results, even if some steps are discovered by accident.

> Some days I dream of a distant future in which an average being (human or otherwise) comes in contact with [today's technology].

> “Any sufficiently advanced technology is indistinguishable from magic.” -Arthur C. Clarke

That is precisely the premise of the "science & sorcery" genre or hyper-sci-fi as I like to call it, where relatively primitive people regularly stumble upon ancient technology that they can't hope to comprehend but use it for its incidental effects.

For example a smartphone may be good as a compact light source but not much else, or a hard drive for its shiny sharp discs and magnets.

See the Numenera RPG, set a billion (!) years into Earth's future after several civilizations have risen and collapsed: http://numenera.com

I desperately want there to be a series of books written in this universe.

There’s a CRPG called Tides of Numenera and some tabletop adventures of course.

Would a hard drive even last that long.

I hope that we have something, something like the seed bank which saves the most important data of the past. Would kind of suck if no one had a SATA in 500 years though

Laser engraving tiny text on a hard surface maybe? https://nanorosetta.com/

Even if HDD disappear, the sheer depth of research that went into it will remain impressive for very long.

That's true, although the technological literacy to make them might disappear. Their disappearance could also have other knock-on effects in other industries, like shrinking the market for high-precision air bearing spindles, magnetic testers, lapping machines, and so on.

and atomic thin metallic coating (IIUC disk platters have one atom thick platinum layer sometimes... crazy)

The HDD market will it disappear for a very long time - decades away. Remarkably, with energy assist technologies, write density improvements will actually accelerate over the coming decade.

It's the same with internal combustion engines. And on a smaller scale, when that bit of custom code you wrote is obsolete because someone else wrote a better library.

I find it a bit sad and hard to let go of such things. I'm sure I'm not the only one.

There's a lot of science in them but it's a bit less extreme (in my newb eyes) than hard drives.. I mean the head control involves air flow. And it's so minute.

Modern internal combustion engines (think formula 1) are probably way more complicated than you realise. I mean, why would they not be using technology as cutting edge as hard disks? They've been around for a lot longer so have a lot more knowledge sunk into them.

yeah ok, there are ICE that are extreme levels of sophistication too. Even some mainstream class cars have fancy engines. But it's not as extreme as the average hdd any user had in his computer without realizing the level of physical depth involved.

Alternatively, I wonder when we will start using MEMS for storage?

Say you have a MEMs device which actually physically flips an atomic sized bit, on or off, like a light switch.

A little gearing machine is what will flip the atomic switches. Or read back its binary data.

Hard drives can be demagnetized, and its controller board will eventually fail. This is its Achilles Heel.

Flash drives can get electrocuted and fried.

Tape is still magnetic, but it separates the reader and writer device, from the actual medium. But it’s too large and cumbersome.

Something like a MEMS storage device, will be fully solid state. Fully enclosed. Electrically neutral. Magnetically neutral. And can last for millions of years. It’s the ultimate time capsule.

MEMS based storage is a thing, though the most common architecture appears to be basically a magnetic system with a very large number of read/write heads (I think: the summary papers I read were extremely vague on the actual read/write/storage mechanism). The issue mostly seems to be the storage density isn't very good: the actual size of a bit is still quite large and the read/write heads also take up a fairly significant amount of space, but throughput is potentially huge because of the many heads operated in parallel. MEMS tends not to operate at the atomic scale, features of the mechanisms are usually a fair bit larger (and mechanisms which make break/make contact are generally difficult to make reliable and long-living in MEMS: surfaces will tend to stick together when they touch and rolling/sliding interfaces will wear away quickly. Almost all industrial MEMs use involves structures which only move by flexing, not the pretty looking gear systems which tend to get media coverage).

This sounds like an opportunity for a PhD dissertation.

Discover a new ability to coat micromachines, to make it resilient to wear and tear.

Or to find a micro lubricant that can be used on MEMS micromachines.

Btw, nice response.

Excellent news. The pattern of the last five to eight years of the prices being effectively the same when I go to buy a new hard drive is incredibly depressing compared to the yearly halving of my youth.

I just bought a 14TB drive for 250 canadian - equivalent to $0.0141 USD per gb. According to a backblaze blog from 3 years ago [1], they were at $0.024 per gb four years ago.

So they certainly havent halved, but the prices are definitely still dropping.

[1] https://www.backblaze.com/blog/hard-drive-cost-per-gigabyte/

Which drive did you pick?

Was a good sale on WD Easystore external drive from Bestbuy. Back up to MSRP now.

Thanks! I'll keep an eye out.

Careful, or you might end up with a shingled drive with radically worse write performance than you might expect.

Note that shingled drives are totally fine for many use cases. I bought this 14TB drive to act as a secondary backup for non-critical data, I expect it'll be a write-once-read-rarely kind of situation. Even if it is SMR, it'll be fine for me.

Don't use large SMR drives as OS or software drives, or drives that do a lot of changes when you're above 50% fill capacity on the drive

Won’t that kill any chance of recovery though? If you need this under normal conditions then surely if there is damage then it’s game over

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact