Is it possible for BCH to change to an ASIC resistant algorithm? With Eth changing to POS ??

I think most of the people using the word "decentralized" incorrectly. I think people are confusing it with "distributed". See Vitalik's post on the subject.

There are two levels of centralization that matter:

1)33% of the hashrate controlled by a single entity for long periods of time. This enables selfish mining.

2)51% of the hashrate controlled by a single entity for even a short period of time. This enables the 51% attack and all of its consequences -- censorship, monopolization of mining revenue, rewriting of history, etc.

Decentralization means that the hashrate is split between enough entities such that no single entity or coalition of entities is likely to reach either of those two thresholds.

Aluminum smelting is an example of a decentralized industry. No, it's not something people can do at home, since people need a plant that costs around $100M. But there are enough plants distributed around the world and owned by different entities to ensure that nobody can corner or monopolize the market.

The thing people are asking for from ASIC-resistant mining is distributed hashrate, not decentralized hashrate. But as Satoshi recognized, there's no need for that. Distributed hashrate does not actually address any of the issues with Bitcoin's threat model that aren't addressed by decentralized hashrate.

In fact, the use of general-purpose hardware for mining makes a cryptocurrency less secure. With ASICs, miners have a significant investment whose value is determined exclusively by the value of the currency they are mining. If a Bitcoin miner successfully attacks Bitcoin, the value of their hardware investment drops. However, with general-purpose hardware like GPUs and CPUs, that isn't the case: if someone successfully 51%-attacks Ethereum Classic, the value of their GPUs stays the same. This makes it feasible for someone to short ETC, then attack it with a 51% double-spend attack, and profit both from the double-spend and from the short.

The vast majority of cryptocurrencies that have been successfully 51% attacked have been mined by general-purpose hardware:

Feathercoin -- 80 blocks orphaned in 2013. Feathercoin uses scrypt, which was GPU-mined in 2013.

Bitcoin Gold -- attacked repeatedly, beginning in 2018. BTG uses Equihash, which was designed to be an ASIC- and GPU-resistant algorithm, but which ended up being GPU-mined until 2019.

Vertcoin -- attacked with Nicehash GPU rentals.

Ethereum Classic -- repeatedly attacked with Nicehash GPU rentals

One of the only two instances I can think of in which an ASIC-mined blockchain was successfully 51% attacked was Bitcoin ABC. But that's an example that proves the point: the miners who attacked ABC were not financially motivated to profit off of the attack; instead, they were attacking out of retribution, because they were angry with Amaury Sechet and Bitcoin ABC's behavior prior to the fork. As Bitcoin ABC was threatening Bitcoin Cash with a soft fork (i.e a developer-sanctioned 51% attack), miners thought it fair to respond by 51%-attacking ABC. By attacking Bitcoin ABC, these miners thought they were defending Bitcoin Cash.

The other instance is the attacks against Bitcoin SV, which may have been similarly motivated: CSW repeatedly threatened to 51% attack Bitcoin Cash prior to the BCH-BSV fork, which may have motivated miners to attack BSV once BSV shrank enough for the attack to be feasible