Per @nathan-at-least :
Hello all, back from vacation. While I support the design requirements both @zookozcashand I laid out above for a Zcash upgrade, I’m going to advocate within our company that we postpone Harmony Mining from NU2-Blossom (with NU3 being the next candidate slot).
There are four logistical/technical (not strategic) reasons for doing so:
- We are behind schedule for Blossom / NU2, according to The Network Upgrade Pipeline, Upgrade B 2019. According to that plan we should be in the midst of third party security auditors analyzing our specification while we work on implementation.
- There is no clear leading candidate for block selection design (including difficulty-adjustment), and I believe the “rewards ramp-up” + time-locks are inseparable and thus inter-dependent on the block selection design.
- @daira’s simulation is precise for two designs, and shows in both cases a reduction in rollback protection. Even though @zookozcash clarified that a regression is acceptable, I want a stronger understanding of how much of a regression a given design introduces (and then to weigh that against advantages we’re trading off for).
- I haven’t looked at existing designs yet, and I’m not aware of any company engineers doing so, yet, and our best effort should verify scrutinize similar designs, their security models, and our different requirements and design choices. (Much love to @zawy12 for linking directly into Grin DAA code in this comment! That’s a good start, though I’d want to start by looking for security models / design rationale.)
We’ll continue R&D for these design goals on the NU3 Upgrade A 2020 slot.