I agree that measures should be taken to prepare to switch the proof of work function, but don’t think an algorithmic switching procedure like the one you propose is needed currently.

The danger to users of Bitcoin from a ‘classic’ hard fork is that if a majority of miners judge that their interests lie with that fork, then they are incentivised to attack the original Bitcoin blockchain with a 51% attack in order to ‘encourage’ migration to the new fork. This must be prevented. A change of PoW would do so effectively. Further, even the prospect of such a PoW change factors in the thinking of professional miners, as it stops them contemplating ‘experimenting’ with classic, knowing that they could just move back to mining Bitcoin if classic fails to be profitable.

The value of a PoW change, therefore, lies as much in the threat as in the execution. However, there is not much need to plan to change the PoW more than once, or algorithmically. This is because in the event of a PoW change that does its job — which is to say, destroys the capital investment of miners — it will be common knowledge that miner defection to a hard-fork unwanted by owners of bitcoins will cost miners their incomes and the value of their investments. Thus, while the new PoW function will inevitably be implemented in ASICs, the owners of that ASIC hardware will have no incentive to repeat their ‘experiment’ of trying to enforce a hard-fork without regard to Bitcoin owners.

This also means that a PoW change is not something that should be taken lightly in the absence of bad behaviour from miners. Miners are currently delivering a product that Bitcoin holders want. Their belief that it will continue to be in demand is what justifies their investment in increasing their mining capacity, producing a better, more secure Bitcoin. This happy state of affairs should be left well alone, until and unless miners cease co-operation.