[zmonx]

This is a very good summary of the situation, thank you for posting this.

However, one critical issue is omitted in the above: The suggested approach (SegWit), albeit probably more by coincidence than by intention, also corrects a shortcoming of the existing transaction protocol that lets you reuse parts of the computation when trying to find a suitable block, leading to a net gain of ~20% hash power over other miners if you exploit it.

Critically, said approach ("AsicBoost") is patented! Therefore, not all miners can use the approach, leading to a centralization of hash power in contrast to the original intention of Bitcoin.

Miners who currently benefit from this oversight in the protocol have no short-term interest in removing it via a change of protocol headers that SegWit entails, in addition to the other reason you mention, i.e., higher transaction fees if there is a backlog due to small blacks.

[kbaker]

I didn't want to mention it since it is still not proven that this is happening, only strongly suggested.

Of course I support any efforts to get rid of shortcuts in the POW function ASAP, and by any means necessary. The whole reason behind a POW function is to ensure a fair competition.