asanso

New paper with Giuseppe Vitto:
"Top Gun: Degree Annihilation Attacks on Poseidon". We study algebraic attacks on the CICO problem for Poseidon and Poseidon2, introducing a degree annihilation framework that reduces the complexity of the underlying polynomial systems. Concrete analyses over KoalaBear and other ZK-friendly fields. Details on ePrint 👇👇👇👇

Unsolicited advice to the AI-native generation.
I use AI a lot. Probably more than I should. For coding, writing, thinking, exploring ideas, occasionally asking stupid questions I would be embarrassed to ask another human and, yes, writing this post :D
It is an extraordinary tool, and I have absolutely no intention of going back.
Lately I keep thinking about alpinists climbing peaks above 8,000 meters.
Some of them train to climb both with and without supplemental oxygen. Oxygen is wonderful at that altitude and nobody skips it to prove a point. They train without it because equipment fails, and discovering that you cannot function without it when you are already in the death zone is probably not the best moment for that particular revelation.
Learning to climb without oxygen doesn’t make the oxygen any less useful, it just makes you a better climber when you do have it.
I think about AI in much the same way.
I use it constantly, and it makes me faster and sometimes even smarter (or at least gives me that pleasant illusion). I also know I can still sit down with a difficult problem, think through it, write something, or figure out why a piece of code doesn’t work without asking a machine for help. I want to keep it that way.
So my advice to younger people is not to avoid AI. Quite the opposite: use it, abuse it. Just turn it off every now and then. Write the paragraph yourself, or find the bug by hand, and let yourself get stuck for an afternoon.
There is nothing morally superior about doing things the hard way. You just want to know you can still climb when the oxygen stops flowing. More prosaically, it is hard to notice when the machine is wrong if you have never done the job without it.

But hey, I am an old fart, so do not listen to me.

Quote of the day: Sometimes you have to take two steps back to take ten forward

Autumn

Big PQTS call today 👀

It’s not exactly a mystery anymore: we’re moving toward hash-based cryptography.

Today we get into the weeds and pick the default leanSPHINCS parameters.

14:00 UTC. Come argue about parameters with us :) link ⬇️