Most of the mining hashrate supported memory-hardness hard-fork it seems, and in regards with mining there are also voices in the community (mining and trading channels in discord) in regards with non-outsourceability (aka pool-resistance).
The motivation for pool-resistance was to avoid dominance and possible censorship by big pools. However, with pool-resistance the mining seems to be dominated by large farms and maybe private pools.
On improving non-outsourceability, proposal by @mhs_sam :
" According to Ergo’s WP, memory-harness (fpga and asic resistance) and non-outsourceability (pool resistance) are two core properties of ergo. There are lots of statements in WP and other papers, blog posts, tweets bolding these two features. These are core features and makes ergo different and interesting. I know that, some economics which could not be predicted before launch is against these two features, but we cannot predict the future. Giving up at this early stage makes other claims in doubt too. I am a fan of ergo since the first time I heard about it because of its novel and revolutionary features, it was a new and different chain. It was the most distributed and decentralized protocol I’ve ever seen. I mean it, even the development of ergo is not centralized, consider me, rsmmt, Bigevil, york as a few examples. I saw in the discussions that it is possible to fork only of security features. I see these two features beyond security. These are the philosophy of ergo, the nature of it. Without these features ergo is like many other blockchains with less and more features. Ergo claimed something and these claims are broken somehow, so what is the action? I see the only option is to insist and fix them. These core features and maybe others may have flaws in the future again and should be fixed again and again.
Regarding technical discussions, I think it is not hard to maintain memory hardness with some modifications in the prehash computation and its parameters. Also, the bigevil’s magic works due to predictable output of the signature function, It could be avoided easily by replacing the signature<d with hash(signature)<d."