1.4.21.1 was released today. It’s a critical patch for a small subset of nodes which have iterated far faster than expected.
Fix: Invalid proofs and softlocking
The reverse difficulty bomb of 1.4.19/20/21 has a lower bound on how far the difficulty should go, otherwise two things will happen at the extremes beyond iteration 700,000:
Proofs produced past iteration 700,000 would be invalid (even if the math checks out, those proofs do not validate at the protocol level)
At iteration 800,004, the arithmetic underflows and difficulty spikes to an impossible to clear value, essentially soft locking the node
This patch unwinds the invalid proofs if present, and ensures moving forward the difficulty bar stays at a safe value.
We remain on track for the expected emissions rate when 2.0 launches, placing us somewhere in the 700-800MM total tokens range.
If you are not near iteration 700,000 this patch will change nothing for you, but if you’re near it, you should probably pick up the update.
As always, be sure you are backing up your store.
An updated ETA for 2.0 will be announced later today.
Some dishonest miners rapidly accumulate increments to 700k using m3 max and then transfer the config to multi-core regular machines to quickly obtain tokens. Will there be any punishment for this behavior?
The protocol was developed to reward nodes with faster compute as well as provide the option to upgrade hardware. Anyone who implemented that strategy operated within the bounds set by the protocol.
700k was only feasible to achieve this week and 2.0 is coming very soon, so the window to exploit such strategy is very small and unlikely to have a significant upside over letting the m3 nodes continue to run.
What somebody who achieved 700k increment on an Apple Silicion M3 can do is shut down the node on the M3, and move the node’s peer id and store to a cheaper hardware which will produce proofs slower than the M3, but still faster than if they’ve ran it from scratch on the cheaper hardware. But now they have to decide what to do with the M3: they can either start from scratch on the M3 or put the M3 out of commission. Neither option makes too much sense when we have a few days left until 2.0 when the reward formula changes, since an M3 starting from scratch will likely produce proofs slower than the 700k increment cheaper hardware, and I don’t think it’s possible to rent an M3 by the hour so they either have to pay for it until the end of the month or they’ve outright bought it.
