Talk:Paper data storage

Twibright Optar
This project is interesting, but isn't worthy of inclusion in this article due to it's immature state. I spent a number of hours tinkering with it. It's only available as source code, and the Makefile has a mistake in it where the math library isn't linked properly due to misplaced -lm tag. Once the Makefile is corrected, I got a program that I could not achieve reproducable data from it's own digital output.

The encoder produced a PGM file, the decoder only reads PNG files. Converted the PGM to PNG, but it couldn't interpret the data correctly. Returns the error, "Cannot find uppper left corner." 184.156.59.81 (talk) 08:19, 2 December 2014 (UTC)

Missing details and other technologies
The barcodes (and now QR Codes and other symbols) are a common way to hold data on paper. They need a mention in this article in my opinion. And another common solution is OCR, and this can be extended even on binary files by coding them as base64.

Someone mentioned Optar in the talk, I tried it without success too but there is another one on Windows : Paperback. I stored a 117kb file on a sheet of paper with it's default settings (200 dpi) and scanned it at 600 dpi. The file was successfully recovered with no errors. []

The article mentions at it's end how greyscale pictures can hold megabytes or even gigabytes of informations. This may refer to this article : Techworld but it is only theory. They have proven that no device is accurate enough today to recognize such number of colors and no printer can reproduce such a range of colors or even greyscale accurately. Some technicians compared this as the search of perpetual movement for data. — Preceding unsigned comment added by 145.129.41.75 (talk) 13:24, 14 December 2014 (UTC)

Trying to do it with greyscale is a bit silly, to my mind. Almost all scanners are colour, and colour printers with pretty accurate alignment exist (and the new gel-based inks can greatly reduce wicking and smearing in inkjets, which tend to be better aligned than lasers in the first place). You can try messing about and attempting to accurately print and then determine between even just 8 or 16 levels of greyscale at high resolutions ... or just use multicoloured dots. Red vs yellow vs green vs cyan vs blue vs magenta vs brownish vs white gives you triple the bit-depth at the same resolution vs monochrome without much more difficulty in reading the information back in. With a small-droplet neutral black added in, that's 4-bit density with basically no real change in accuracy (as each signal is either "off" or "on" with no fuzziness in between - though to be sure, you may want to tune the ink pigments to the scanner light frequencies...). Then you can work on whatever methods may be needed to extend that to 8-bit if you really want to, but it's a law of diminishing returns even at this level.

Even so, if we assume that you can somehow record 1GB per page, and print on both sides without any issues arising from that, you'd need basically an entire standard ream of paper to represent a VHS-tape-sized 1TB hard drive, or a whole box for a 5TB one - so even at our current level of development it's already starting to look a bit impractical. You'd also need some way of bootstrapping up to an operable system that can read the pages, after whatever cataclysm it is that wipes out all other forms of digital storage (magnetic, optical, etc) but manages to leave printed pages and the devices for scanning them intact. Not to mention the time it would take both to print it and to scan it in; if we assume one minute per page in both cases (a practical maximum for the resolution involved plus all the general physical and computational overheads), that's roughly equal to a fairly slow USB2-based external HDD, gets kerbstomped by even fast USB2 devices let alone USB3 ones, and isn't really so much faster than DVDRs read by a good quality drive (which themselves offer rather higher data density) that it can justify the hassle, storage space, cost (a box of paper isn't super cheap, and the ink and electricity then needed to print it all would push the cost quite close to that of an actual 5TB drive), use of resources, and inherent fragility of the medium (paper can get liquid spilled on it, wrinkled, torn, burnt, light- or heat-faded, dirtied, acid-affected, etc far more easily than an optical disc, or well-shielded hard drive). 193.63.174.115 (talk) 11:37, 15 October 2015 (UTC)


 * You have some good points, but I'm going to disagree on the "inherent fragility" of paper. My understanding is that many documents printed with ink on paper in the first 50 years of printing (incunable) are still readable today, centuries later. I don't know of any hard drive that is still readable even 10 years after stuff was written to it. Ink-on-paper may be fragile, but the evidence suggests that hard drives are even more fragile. --DavidCary (talk) 03:30, 17 May 2016 (UTC)