I love the idea of literally building hardware transparent enough to be inspected by eye… but if you build a board where you can see all the traces, could you also add a transparent conductive material such as indium tin oxide to add other traces that you couldn’t see?
Hardware security expert, RobertT, pointed out long ago on Schneier’s blog this exact risk when I talked about using older nodes for visual inspection. He said you can’t un-invent technology. The nanoscopic CPU’s, radios, etc will still exist regardless of what node I’m using. My opponents, maybe at the fab itself, might slip something like that into my designs. So, they always have to be considered in how one secures production of devices even with old or transparent technology.
“We rely on logic placement randomization to mitigate the threat of fixed silicon backdoors”
And that just made no sense to me given how FPGA backdoors would likely work if aiming to hit lots of people. It’s a nice step that even I would’ve done. It’s not enough, though. You definitely want a gateway and/or secure tunnel in front of it so they can’t activate their built-in triggers or artifices easily.
I would find it very funny if we decided to randomize board layouts to get more secure hardware. Handling the different lag between instances alone might make this a massively losing proposition, especially in a world where so much research is done regarding delivering a clock signal everywhere at the exact same time.
Is this relevant, I wonder? https://libresilicon.com/
I find these types of posts are pretty philosophical.
If you are in camp where you believe context is important: absolutely you can!
If you are in the camp where you believe it must satisfy all contexts: absolutely you cannot!
What is even trust?