The problem with a lot of cryptocurrencies is that the mining is difficult and wasteful; you're doing a lot of computation that only makes economic sense if a lot of other people are also doing it, and it's plain that many of those people are trying to do it using someone else's equipment, which is a form of theft (specifically, conversion).

I like DogeCoin (much shibe) but this is still a Bad Thing. It would be much better if the problem being solved to create a blockchain were also useful in some other context, eg protein folding. Of course, some such problems are handled better than others by computers; protein folding is something that humans seem to do well but for which we haven't managed to develop great algorithms yet.

But if part of the work can be done in the human brain rather than in a CPU/GPU/ASIC, is that bad? The current computation model basically means that mining rewards go to whoever invests the most capital, ie has money to throw at the problem. This is inherently inequitable, because it means people with more money or a willingness to appropriate the equipment of others will mine/dig/discover more coin. Of course, the argument is that since mining is rather random, in theory you can start mining on your small CPU tomorrow and hit some coin, or get a random amount of coin for a fixed amount of mining. But random rewards devalue expertise and experience in favor of an unnaturally even probability distribution. I really think the computational abstraction is a barrier to cyptocurrency adoption, because ultimately your economic incentive is only as good as people's adoption of a sunk cost fallacy.

Here's a back of the envelope calculation. In theory (by supply-demand idea), you would make a negligible profit mining bitcoins. Each bitcoin is worth about $600. If our supply-demand assumption is true, the $2,160,000 dollars of Bitcoins that come into existence every day, that's also about $2,160,000 of electricity used. Or about 18,000,000 kWh per day. Or about 9,000,000 kg of C02 per day (9000 metric tonnes). That's about 3,285 thousand metric tonnes per year, or about the entire annual CO2 output of a country such as Papua New Guinea.

Most of the numbers floating around the internet and nearly every mainstream news piece about bitcoin's energy consumption are wrong. They cite an outdated statistic maintained by blockchain.info that assumed everyone was using a GPU even when nearly all of the network was ASICs.

I would estimate the current network (25Ph/s) averages around 2-5W per Gh/s, leading to 50-125MW total power use. bbosh in this thread is off by an order of magnitude.