>2. If ASICBOOST was actually used, we'd see ample evidence on the blockchain.
The covert form of ASICBOOST (where they don't use the version field, the form that was only recently publicly discovered) would only show up as a higher than usual number of empty blocks or blocks with reordered or missing transactions (depending on how the attacker implemented it; it's not necessarily both).
>3. ... As you can also examine for yourself, the transactions are ordered essentially by fee-per-byte, which is not what we would see if they were shuffled to create collisions.
I find it extremely disingenuous that the author only focuses on looking for reordered transactions and ignores the often-noticed unusual number of empty blocks that Antpool mines[0], and the article reaches absurdity when the author claims that the lack of reordered transactions means there's absolutely nothing to worry about.
>4. Building more efficient mining chips is what miners are supposed to do, as it secures the blockchain. Framing an optimization as an attack is disingenuous. Going from 28nm to 16nm was not an attack on the network. Better mining algorithms have never been considered an attack.
If for example a mining "optimization" encouraged every miner to mine only empty blocks, then calling it an attack doesn't sound out of place. An optimization that only often encourages miners to make empty blocks isn't categorically different.
>7. Segwit is being offered as a solution to ASICBOOST. Yet Segwit has at best a tenuous connection as a response to ASICBOOST mining. There are countless other, non-controversial solutions that can disable ASICBOOST, should one choose (wrongly) to do so.
Greg Maxwell's very post announcing the issue introduces a fix for the covert form of ASICBOOST that does not involve implementing SegWit!
[0] https://bitcoin.stackexchange.com/questions/50184/why-does-a... http://www.livebitcoinnews.com/antpool-generating-large-amou...