4 ms·
I wish the articles would use bytes as an information storage measure.
by quickthrower2 3y ago
I wish the articles would use bytes as an information storage measure.
- clnq 3y ago> For example a 20 x 20 mm nickel Nanofiche sheet can hold up to 8,000 pages of text rendered at 150 dpi. All this depends on the font used, and what kind of information is stored. But using lorem ipsum, and Verdana, and only uppercase characters, spaces, and periods, I found that you can fit about 49,500 characters in a letter format page at "150 dpi" [0] This is 396 million characters in a 2 x 2 cm sheet. The Latin ASCII alphabet has 26 letters, but in my experiment, there were commas and periods. Though at the resolution I used, commas and periods would be difficult do differentiate if there was almost any degradation or printing error. So let's say it's a 27 character set. Maybe we can stick a few more numbers or symbols in there to get to 32, although some numbers like 1, 0, and 5 look like letters, so not all would be usable at this resolution. If each character can represent 32 states, that is equivalent to 5 bits per character. 5 x 396 million is 1.98 short billion bits. That is 247.5 million bytes. Or 247.5 megabytes. This is in 4 square centimeters, so the data density is about 61.9 megabytes per square centimeter. If we give up human readability and the text would be regular language (according to Zipf's Law), we could expect up to 0.2 compression ratios. Or 310 megabytes per square centimeter. On the other hand, we could then use quite dense data storage methods like barcodes. But these calculations are a bit pointless because this is meant to be read without a computer. A DVD has a readable surface area of about 89.2 square centimeters on each side (some DVDs are multi-sided) and 4.7 to 8.5 gigabytes of storage in that space. So it stores between 54 and 97.6 megabytes per square centimeter. For storing plain text, it might be on the same order of magnitude in terms of space efficiency as a DVD, but it is human-readable. If the data is not human readable, then it could be at least 5 times more efficient than a DVD. Which is closer to a Blu-ray disk. Back-of-the-napkin, might contain errors, many assumptions made. The goal is to understand the order of magnitude with some errors. In short - it looks like it's not 100x more or less efficient than writing a plain text file to a DVD. P.S. I chose Verdana as it was a quite legible sans-serif font. I then created a texture of random black and white pixels, and applied it at 25% opacity to simulate a significant degree of printing error at the supposed resolution they can print at. The font at the chosen size remained just about legible with that kind of interference [1]. The exception is periods and commas, which are difficult to tell apart. That is why I considered them one character before. Making the font even 10% smaller would make it really difficult to read. This is all not grounded in any technique or proper estimation method, I just wanted to show why it seems like going smaller might be difficult. [0] https://i.imgur.com/Pi8acSk.png https://i.imgur.com/Pi8acSk.png [1] https://i.imgur.com/stFkXQy.png https://i.imgur.com/stFkXQy.png