My time tracing rats nests of network cables across four floors has prepared me for this. #gardening
Remote
Thomas Sibley
@trs@metasocial.com
mastodon 4.7.3Often riding a bicycle or tending the garden. Computing generalist. Papa to two. He/him. Cars ruin cities.
244 Followers
243 Following
25 Posts
Joined November 08, 2022
Website:
Papa posts:
lol I just discovered I was doing gravel laps around Dave Matthew's house this morning without realizing it.
Replaced a rivet! #BikeTooter #SEAbikes #repair
Replying to
TIL about the Jim Creek Naval Radio Station not far outside of Seattle. Big antenna, low frequency. Looks funny on aerial imagery.
Got a "you're absolutely right" in an email from support and now I'm glaring at my computer. (Like most days.)
Replying to
@itamarst@hachyderm.io @glyph@mastodon.social What I often saw over and over again was that too many layers would try to opportunistically/transparently multithread and end up creating CPU contention.
Boosted by @octothorpe@mastodon.online
Replying to on metasocial.com
One takeaway is that disrupting/breaking QRs with a minimum of edits (and with only a sharpie) is a _lot_ easier than adversarially modifying one to data under your control.
The crucial format bits which specify the code's error correction level and mask pattern is only 15 bits (5 data + 10 ECC). It appears in duplicate in the code, but many of those 30 bits will be dark already and you can black out the rest. As the layout of the bits is linear, it's easier to black out multiple bits at once.
Massive Attack / Tom Waits - Boots on the Ground
Replying to
@rf @virtuous_sloth https://www.thonky.com/qr-code-tutorial/error-correction-table has good info here.
For a 2-L code (most common I see), there are 272 data bits + 80 ECC bits. That's ~23% of total capacity!
In an example code I've been playing with, if I flip a single data bit (e.g. B/1000010 → C/1000011), then 43 ECC bits change (~53%).
Replying to
@jernej__s@infosec.exchange @benjistokman@mast.benstokman.me @virtuous_sloth@cosocial.ca my understanding of those is that they tend to rely on the higher error correction levels, while most codes without a logo use the lowest level (L), and thus are maybe more easily tweaked.
Replying to
@eeeps@front-end.social Great blog post! I learned a ton. One question. You wrote:
An implication of this is that there are many times more photons shooting out of the blue side of that gradient, than there are from the green side, even though every swatch on the gradient has the same apparent lightness. Weird!
Isn't it rather that our eyes have non-uniform perception of the same number of photons at different wavelengths? Or am I missing something?
Replying to
In my very limited testing, this breaks the code (and isn't super obvious at a glance).
Blacking out the timing patterns (alternating dark/light squares between finder square patterns) doesn't break it in my testing. Nor does blacking out the alignment pattern.
Replying to
@virtuous_sloth@cosocial.ca Nod. My hope is to better understand just how challenging/not simple it might be and how much changes can be intentionally minimized. For example, it seems there's opportunity to help a tool minimize changes by marking up which bytes in the encoded data it can have free reign over (vs must keep the same). Imagine a QR code that points to a short URL and you can create your own short URL that's one character different.