The economics of Bitcoin mining are a bit weird in that it impossible to make it more energy efficient.

The system auto adjusts the computational complexity of mining bitcoin so that it always costs a little less than one bitcoin to mine a bitcoin, and at scale the only variable expense is electricity so as the price of bitcoin goes up, so does the amount of money that must be spent on electricity.

Current 6.25 Bitcoin are mined every 10 minutes. So globally about $2 million must be spent on electricity every hour.

In a little over 2 months the block reward cuts in half to only 3.125 bitcoin every 10 minutes. That will have the side effect of reducing the money spent on electricity for mining bitcoin so long as the price of bitcoin remains the same.

“The System” is not really that intelligent. The statement that “It will always cost a little less than one Bitcoin to mine a Bitcoin” is only correct because the incentives in the system steer everyone toward that. There’s no direct link between the two. Bitcoin Miners are intently aware of how much energy they consume, and if the price of Bitcoin dips below what they are paying for electricity, they likely will shut down their rigs, because no one wants to mine at a loss.

The real issue with Bitcoin is that the algorithm used to find more Bitcoins is kind of basic in terms of its difficulty mechanism. It was the first one ever used for cryptocurrency. It was originally envisioned that owners could mine more bitcoin with spare cycles on their CPU, but since it was first designed, people have come up with custom mining chips that can mine faster and much more power efficiently. But paradoxically, this has made things worse, because the bitcoin mining difficulty simply scaled up to account for all that. So now the only way to mine Bitcoin is to have this custom hardware – it’s too hard to do any other way – and you need so much of it that you are just as power hungry as before.

There are other algorithms that don’t have these same problems. They have been designed to use other computing resources (like gobs of memory) that are much harder to concentrate on custom chips, making it much more expensive (spatially as well as computationally) to simply spam more of them. Ethereum uses a totally different model now that doesn’t rely directly on power consumption at all.

OG Bitcoiners seem to think that the massive power consumption is a net benefit, because it is spent in making the overall network more secure, and less likely to be attacked. So they will never try to change their block algorithm, even though other projects are more secure with less power consumed. And if that opinion holds, the only way to eliminate this source of power consumption would be to crash the price, and cause the Bitcoin miners to have to mine at a huge loss to continue.

the massive power consumption is a net benefit, because it is spent in making the overall network more secure

I really have trouble understanding this argument. Joining a mining pool secures nothing.

The whole point of mining is to arrange transactions into blocks, and then generate a cryptographic hash of the block that meets some difficulty criteria. It costs some small amount of computing to do that. But an astonishingly large number of hashes won’t meet that difficulty criteria, which is why miners have to try a gazillion times to find one that works.

However, once a block has a valid hash, it is added to the chain. Then, the hash of that valid block must be used in the next block, which will be equally hard to find.

By “security”, what is really meant is “How can I be sure that a transaction can’t be undone once it is committed”? And it’s because all these blocks are stacked on top of each other, and cryptographically related. Once a transaction appears in a block, and a few blocks get mined on top of it, it becomes prohibitively difficult to un-do it, because someone would have to put in the computing power to re-authenticate a string of blocks, all while the rest of the network is adding blocks to the valid chain at a faster rate.

The security of this whole arrangement has so far been working good as well.

In order for someone to try and perform a 51% attack, they’ll need to either compromise a large swathe of existing miners (e.g if the government seized control) or create/acquire hardware totaling more than 100% of the existing network today plus growth while you attempt to build more than 100% and then maintain growth over the rest of the network.

As the network grows that becomes exceedingly more difficult to perform.

I have really high hopes for something like proof of work, but it’s not without it’s own problems either, and with Ethereum, it’s the first massive scale test, so it’s not as battle tested as proof of work yet, although it’s been used in smaller projects so there has been some testing. With more money on the line though, comes more will to try and break it, or use an exploit you may have held back beforehand.

One interesting difference with POW/POS is that if a miner/entity does somehow perform an attack, they keep the hardware and can continue to try. With POS, they should get slashed in which case the money is gone. But with POW you have the barrier of actually acquiring the correct amount of hardware, meanwhile in POS, you just need the money so there’s no manufacturing/lead time.