The sooner PoS goes live the better, but only when it is checked and tested and audited many times over. The idea of simply rushing it to live deployment without completing all auditing first as the primary backup plan in case of malicious actions related to ASICs or Bitmain seems incredibly reckless, IMO. I think that while we don't know enough to say that the ASICs are a significant threat in the interim before Casper launches, we also don't know enough to say they do not pose such a threat. The fact that Monero forking revealed 70%+ of their hashrate may have been due to hidden ASICs should be a stern warning to the Ethereum community. ASICs may account for the majority of hashpower added since early December, and could well hit 51% of the total before Casper is ready. While I don't think an immediate emergency hard fork is necessarily warranted, I do think it would be prudent for the Constantinople fork to be scheduled for as soon as is reasonable (in the next ~month or so) and include in it a small change to the hashing algorithm that is likely to break existing ASICs. Maybe include another issuance reduction at the same time as a mitigating factor just in case the change fails to kill off ASICs, but hopefully it would buy enough time either way to finish Casper properly and get it out before new ASICs can be made to overwhelm and centralize the network.