This post will connect a couple posts from yesterday and explore storing data in images.
Connections
There’s a connection between two blog posts that I wrote yesterday that I only realized today.
The first post was about the probability of sending money to a wrong Bitcoin address by mistyping. Checksums make it extremely unlikely that a typo would result in an invalid address. And if somehow you got past the checksum, you would very likely send money to an unused address rather than an unintended used address.
The second post was about QR codes that embed images. I give a couple examples in the post. The first is a QR code that looks like the symbol for a chess king that links to my Kingwood Data Privacy site. The second is a QR code that looks like the icon for @CompSciFact and links to that account’s page on X.
Someone responded to the first post saying that nobody types in Bitcoin addresses; they scan QR codes. The second post suggests QR codes are very robust. I wrote the post just having fun with the novelty of using QR codes in an unintended way, but there is a pragmatic point as well. The fact that you can play with the pixel sizes enough to create images means that there is a lot of safety margin in QR codes.
If you send Bitcoin by scanning a QR code, it’s very likely that you will scan the code correctly. The individual pixels are robust, and QR codes have checksums built in. If somehow you managed to incorrectly scan a QR code for an address, bypassing QR code checksums, then the protections mentioned in the typo post come into play. The checksums in a Bitcoin address apply whether mistype or mis-scan an address.
Storing data on paper
QR codes are a way to store a relatively small amount of data in an image. How much data could you store in an image the size of a sheet of paper?
Ondřej Čertík did an experiment to find out. He wanted to see how much data he could store on a piece of paper using a regular printer and reliably retrieve the data using a phone camera. You can find his results here.
QR code internals
I’ve thought about blogging about how the error detection and correction in QR codes works, but it’s too complicated. I like writing fairly short articles, and QR error correction is too complicated to describe in say 300 words. Other people have written long articles on how it works, and you can find these articles if you’re interested.
It might be possible to write a short article about why it takes a long article to describe QR code error correction. That is, it might be possible to describe briefly the reasons why the algorithm is so complicated. I imagine part of the complexity is just historical contingency, but I also imagine the problem has some intrinsic complexity that isn’t obvious at first glance.
Ondřej used Golay codes for error correction, and it worked fairly well. It would be interesting to compare the robustness of QR codes to a design that simply used Golay codes.
From Bitcoin to Jupiter
James Burke hosted a BBC series called Connections in which he would trace unexpected lines of connections. For example, I recall watching one episode in which he traced the connection between a touchstone and nuclear weapons.
In this post we’ve connected Jupiter to Bitcoin. How? Bitcoin addresses lead to a discussion of QR codes, which lead to discussing Ondřej’s experiment, which led to Golay codes, which were used on a Voyager probe to Jupiter.
The post Storing data in images first appeared on John D. Cook.