So Greg finally admits there is a discount for witness data (relative to now) in Segwit, but of course he has an argument for why it is better. The argument seems to be based on judgments like "UTXO bloat being a bigger problem than witness bloat." If those arguments are true, that could be a good reason for changing the economics (other than, you know, changing the economics), and then the massive benefit to Blockstream would be purely coincidental: he would have found a win-win result for the community and Blockstream. The question is whether the reasoning for changing the discount ratio is sound and - if so - whether it makes it worth changing the economics of transactions.
Whatever the case, we have 76 million reasons for thinking it isn't purely coincidental, prima facie, so his argument had better be way more than handwaving and judgment calls (especially circular reasoning that starts from the premise that small blocks are good).