Hacker News Reader

Comment by ivanhoe | Hacker News Reader

The huge difference is that digital data doesn't ever "wear out" - it might get lost or it might be in format that you don't know how to open on your current device - but it will never become partially unreadable or fragile like the physical media. Both of these problem are relatively easy to solve. It does cost money to make backups and do the format upgrades (or backup the old tools), but it generally costs a lot less than storing physical books over the same period of time and worrying of humidity, acidity of paper, insects, mold, etc. And not to mention how super-expensive the paper-books restoration is, if (or better once when) they get damaged by time.

View on HN

Digital data doesn't wear out, but it also doesn't degrade gracefully. Usually if a printed page starts to degrade there is at least the opportunity for someone to notice and take corrective action before the content is lost completely (scanning, processing and re-printing, perhaps).

Digital data tends to go from "fine" to "gone" with no middle ground.

That's what backups and RAID arrays are for. We certainly know by now how to build a redundancy that will guaranty the data safety. For most of the libraries it's really a budget problem - they've been given huge expensive buildings with controlled air conditions for storing paper books and that's sort of taken for granted, but then governments take cuts on providing them with proper IT infrastructure and knowledgeable staff for maintaining the digital data copies.

> That's what backups and RAID arrays are for.

the point is exactly that: if you have to rely on very complex setups to keep the data fresh, it means that the data is not safe.

can you imagine an hard disk surviving from the ancient Egypt to us?

Wasn't the point of the original comment that the setup to keep paper safe is also very complex?

Another question I wonder about: How hard is noticing the failures?

The cold hard truth here is that curating information is actually a very difficult problem - but paper records will survive longer than digital records if no-one's actively looking after them.

It's 100% true for a book forgotten on your attic, but if discussing organized effort to preserve information I argue it's an illusion - the libraries have to be actively looked after too. There is a fire protection, regular deratization and disinsection, if roof starts leaking it has to be fixed. If you just leave a book in the middle of a field unprotected it won't last a year. When libraries are left without funds their books get destroyed fairly quickly, I've seen that first hand in my home country.

Digital data requires some maintenance, but with cloud services that's offloaded to service providers. I haven't logged in into Glacier or the related AWS account in years, and yet it still backups my hdd regularly. Now, if a comet strikes the Earth and destroys all the civilization, yes, paper books will be a lot more useful than HDDs or dead cloud servers, but in more realistic scenarios data copies give you more flexibility. Just look at Ukraine, they moved all their official digital archives out of the country at the very beginning of the war. You couldn't ever pull it off with paper archives, it would take months and cost a fortune.

yes, it's hard, but papyrus kept inside a very strong, possibly waterproof, possibly not affected by rust (alluminium, stainless steel etc), metal box will last much longer than hard drives kept inside the same strong metal box.

true, but we could always print encoded digital data to that papyrus and call it a data-storage method :)

And now you need a parser and a machine :)

Bit rot: bonjour

Well the thing about digital data is that it's all represented in an uneven compressed way. Take your average piece of paper with text on it, it's essentially a 2D array of white and black dots. Remove half of all black dots and you can still mostly see what the text was.

Same is true with raw digital representations. Take on the other hand a jpeg which is a bunch of coefficients for sine waves, where every tiny change ripples out to ruin the entire thing. Even the basic 3 byte RGB per pixel representation is already compressed with only 8 bits being used to save 256 values. Flip the MSB and your value is already completely screwed.

An equal encoding to real paper would be using 256 bits to represent each colour, summing up the bits to get integer values. That way you can flip 20 bits and you'll still easily tell what the colour was, without any extra processing.

This of course comes at an exponential increase in storage requirements, plus we can be smarter than just using brute force. Taking a compressed representation and then adding on a few layers of error correction would likely be far more reliable while preserving most of the spatial efficiency. Of course then you need the software for unpacking and correcting to survive as well...

This completely ignores https://en.wikipedia.org/wiki/Error_correction_code Those have been known for a long time and they are used in practice for both transmission and long-term storage.

Big storage arrays come with health metrics and auto-rebalancing around broken drives.

The idea that digital storage media degrades without "opportunity for someone to notice" is outright false.

Devices themselves can fail too, not just media.

Sure? We were talking about data though.

> The huge difference is that digital data doesn't ever "wear out"

it depends.

once a broken copy is out for whatever reason, you have a broken copy.

also most of the formats used today to archive digital media are lossy (mpeg4, h264, jpeg etc)

> but it generally costs a lot less than storing physical books

if we discount the electric bill.

> Both of these problem are relatively easy to solve

not really.

I used to own a huge collection of DVDs

they're as digital data as it gets, but no one owns a DVD player nowadays and many of them degraded to the point that even an high quality reader cannot read them properly.

problem is you don't get a blurry image like with analog media, you just get broken and/or unreadable data, so basically that data is lost forever.

> also most of the formats used today to archive digital media are lossy (mpeg4, h264, jpeg etc)

"Lossy" doesn't mean it wears out. If you are a digital media librarian, you will use MPEG and put it in a container with error-correction codes and store multiple copies.

"Lossy" refers to the encoding between the full data you got from your scanner/digital encoder and the final file. It doesn't mean that meaningful information is lost from the original analog artifact and it doesn't mean it's not great for long-term preservation.

I wonder how this difference plays out if we are looking at a timespan of centuries instead of decades. Multiple generations of computers or various kinds of computing and reading technology would have to be stored and maintained along with the digital books that depend on them. And those are physical devices that degrade in many ways and require a power source to be used.

However, this might be a non-issue if we find ways to emulate the functionality of those devices authentically and integrate it into ever-evolving hardware, even if it turns out to be drastically different from what it was before. It is important to consider that the difference in appearance of even a single character between the real device and its emulation could significantly change a book in form and even in content.

xkcd as relevant as ever: https://xkcd.com/1683/

Reply on news.ycombinator.com