PoW doesn't run on "power" in general, but on electricity. It can be trivial to convert some forms of power into electricity, but not all. One particular form of power which could seen as "wasteful" is that of burning fuels to create heat. A mining chip is a 100% efficient electricity to heat converter. If the cost of useful mining chips could become small enough, it would never make sense to use a traditional burner for heating a space, but using mining equipment would be preferable because you could recover some of the cost of the energy used to heat the space in potential mining rewards.
This is the goal we should aim for: As we're approaching the upper limit of Moore's Law, mining equipment will have a much longer lifetime and focus on reducing the cost of production could turn households and other buildings into data furnaces. It may not even be necessary for mining to be profitable - as long as there is there is a large enough ROI for users of such data furnace to cover its initial cost and eventually reduce their heating bill.
> 0: and it is a definite article, like the internet. There by definition can't be more than one at any given time.
Technically multiple chains can and do exist at any time because there is no "given time" - time is relative. Two miners at two ends of the earth may both produce a valid block at a given time (say, in UTC), but the nodes in proximity to them will receive their blocks at different times, due to distance and the fundamental speed limit of information transmission. The multiple chain conflict lasts until the next block is produced. Such conflicts could last for multiple blocks in a row, but with a probability which rapidly declines with number of blocks.
> Two miners at two ends of the earth may both produce a valid block at a given time
Er, no, I mean you can't have more than one (distinct) live blockchain standard; eg if you have Bitcoin and Litecoin, one of them must be using less than 50% of the available hashing capacity (because otherwise it would add up to >100%), and therefore not be secure[0].
Good point about some power use having beneficial side effects (heating) in addition to the actual work though.
0: because if it ever actually needed the security - was more valuable to attack than the majority chain - then the miners on the majority chain would have a economic incentive to attack the minority chain and gain more from attacking than they lost from undefended attacks on the majority chain.
Obviously it's possible to have two blockchains in practice, just like it's possible to have two internets in practice, but there's a constant pressure to drain applications from the minority (quasi-)singleton into the majority singleton until the minority goes defunct from lack of use.
This seems empirically false as for example Bitcoin Cash is gaining more applications and usage compared to Bitcoin to a larger degree than it's 3% hashrate would suggest.
There have even been reorg attempts that have been defended by miners that support the minority chain, making it a bit more difficult to determine how secure a chain really is.
This is the goal we should aim for: As we're approaching the upper limit of Moore's Law, mining equipment will have a much longer lifetime and focus on reducing the cost of production could turn households and other buildings into data furnaces. It may not even be necessary for mining to be profitable - as long as there is there is a large enough ROI for users of such data furnace to cover its initial cost and eventually reduce their heating bill.
> 0: and it is a definite article, like the internet. There by definition can't be more than one at any given time.
Technically multiple chains can and do exist at any time because there is no "given time" - time is relative. Two miners at two ends of the earth may both produce a valid block at a given time (say, in UTC), but the nodes in proximity to them will receive their blocks at different times, due to distance and the fundamental speed limit of information transmission. The multiple chain conflict lasts until the next block is produced. Such conflicts could last for multiple blocks in a row, but with a probability which rapidly declines with number of blocks.