The more I looked at @Dusk ’s validator selection, the more interesting one contradiction became to me: the result needs to be reproducible, but future participants shouldn’t be predictable too early.

That’s where the seed matters. Dusk updates the seed from the current block generator’s signature over the previous seed, then uses that evolving value in the deterministic selection process. Once the required information exists, everyone can independently reproduce the result. But beforehand, future selections shouldn’t be easy to calculate.

I like that distinction because deterministic doesn’t necessarily mean predictable.

Still, it makes me focus on the seed itself. If future committee selection depends on an unpredictable input, then the assumptions around how that input is produced become part of the security model.

So I’m wondering: does Dusk’s seeded sortition strike the right balance between reproducibility and unpredictability, or does the seed become a security boundary that deserves just as much scrutiny as the selection algorithm itself?
$DUSK
$BTR
$BTW
@Dusk #dusk
🧐 What’s the key challenge in Dusk’s seeded sortition?