Before miners get angry, consider that whatever the community does will attempt to preserve the efforts you have made to make Bitcoin a success.  Paragraph five, below, includes a provision to protect you, so please don't write me off.

The competition is essential to protecting the data in the blockchain.  I worry that any time (eg all time so far) the rules of that competition remain static, a group of people willing to develop and optimize hardware that performs the work will form, and their products will find a following - the miners.  These miners and hardware producers will advance the technology until the cost of competing in the space is so high that mining bitcoin is, for all practical purposes, centralized.

I propose that the proof of work algorithm be scheduled to change periodically.  The current definition is pretty simple and there's no reason not to continue using simple definitions.  We could develop a list of hash algorithms which could be indexed so that the first few bits of the difficulty-change block's hash (or even every block's hash) could be used to select one to be used for the next block.  The principle is to discourage specialization of hardware designed for what we must admit is an arbitrary computing exercise, intended only to enable competition.

Of course chip designers could start working on hardware that can handle all the algorithms defined, but the protocol can also warn them that from time to time, the community will alter the content of the list of hash algorithms, specifically to ensure that general purpose computing machines (ie, what the average tech aficionado will have) is the best device for mining bitcoin.

If such a variable PoW were to be used, I recommend that most of the elements in the initial list of algorithms be the current PoW algorithm so that most blocks can be solved by the existing mining community, and only one every now and then will be available to everyone else.  Over time, that ratio would fall, giving the miners time to convert their expertise into more productive activities.

I refer you to the stories at the beginning of each chapter of Douglas Hofstadter's
Gödel, Escher, Bach: an Eternal Golden Braid, a few of which describe a competition between a record-player-making tortoise and Achilles, who works on making records that break the record players.  It offers some through provoking musings that can easily be related to this thing Satoshi made.

notplato