sounds like maybe an asicboost kinda hax optimization. literally just learned about how the blake2b fork has dramatically increased the amount of data in the block headers so idk, maybe has something to do with that also.

ok, my clanker found more details: dug it up, and the "no point making miners" line is about existing hardware, not about difficulty collapsing. the header is 164 bytes now, not 80. activation was block 961,640 on 30 august, and the layout carries a much larger work field plus "multiple ASIC profiles" (knots pr #359, commit 658cc68, 22 august). that is why electrum and sparrow go blind: they fetch 80-byte headers and verify the pow themselves. the reason nobody should build a miner: blake2b-sia is a strict subset of blake2b, so the existing sia asic fleet mines it with stock firmware. goldshell's sc line is sold as "blake2b-sia", tens of terahashes, and it has been mining siacoin for years. the pr says outright it "aims to be compatible with pow hardware produced by several (6 iirc) vendors", and connor's lab ran stock goldshell boxes against the regtest and testnet4 builds and they mined accepted blocks. so the market is already served by a mature, second-hand fleet. a new design competes with that, not with sha-256 miners. difficulty did the opposite of collapse. first retarget fell 41.3 percent because the sha miners left, then it went 4x, 4x, 4x as sia hashpower piled in, then 2.99x, 1.353x, 1.10x, 1.17x. it is climbing toward the blake2b fleet's size, not falling. asicboost is not the mechanism here. it was a sha-256 midstate trick; the analogue in this change is the asic-profile work in the header, which lets different vendors' midstate conventions fit the same format. the part that matters for your swap is not the header: it is the 45-day coinbase lock (6,480 blocks, active since 973,440) and the 800 kwu block weight limit while rdts is on. freshly mined coins cannot move for six weeks, so any swap funded by mining is on a different clock than one funded by ordinary utxos. sources: blake2bminer.com (status page, kept current from their own node), knots pr #359, and the goldshell sc line. the site is refreshingly honest about what is measured versus assumed.

and this is concerning to me because sia miners will now flip between sia and blake2b whenever their relative value ratio flips in favor of $ per hash

dive bombing attacks (short very large rise in hash power) and because of how extreme this can be, double spends can be achieved.

i did not even think about how "dive bombing attacks" would enable reorging the chain to rug. i hope someone is going to consider cancelling this update lol.

i'm especially experienced in the topic of sha256 mined coins with fluctuating mined value that led to the dilation of the average block time from 5 minutes to 12 minutes after about 6 years of nicehash. that's why i designed this

limenka

which was my design for how a no-spam bitcoin fork should look: - difficulty adjustment is a proportional/integral (no derivative) adjustment that has a strength bias to increase difficulty slower than it decreases it so the chain recovers faster from big hashpower mining blocks on it - every block must have a Verifiable Delay Function proof which is one based on hashes that doesn't require a ceremony and a leakable secret. this VDF prevents most reorgs and stops a big miner from stomping on the chain's difficulty to mint coinbases (such as empty blocks, a denial attack) - the blocks are no longer the clock, the bytes/time becomes the clock, and the time bounds allowance for new blocks has the restriction to monotonic (1 second minimum advance of timestamp) and nodes reject a timestamp more than 1 minute in the future (which is the VDF time, roughly). which means that there is a secondary incentive after a big miner clobbers the difficulty, miners are rewarded because they can mine a bigger block that pays more for the time of the delay. lots of simulations in that codebase too, keeps the sha256d hash function and doesn't change the damn header. i'm literally right now wasting (maybe) tokens accounting for this block header bloating that just got inflicted on blake2b chain. the implications of this change are pretty brutal. competing with the main chain of siacoin is a really really bad idea.