⭠ Return to thread

Thank you, Mr. Green. May I ask again about comparables? How much energy does fiat currency "use" to achieve the same ends? Are there other energy-using tasks that fiat currency performs or relies upon that Bitcoin does not? (I'm pretty sure the answer is yes.) We would need much more information to make an informed fiat vs. Bitcoin energy comparison.

I'd also note that many other immense computational processes exist which no one seems to be complaining about, energy-wise. So it's interesting that Bitcoin is being singled out on this metric.

Like I said, I'm neither here nor there on Bitcoin. I just don't find the energy argument particularly compelling.

Expand full comment

Helen, the answer I gave you is the best proxy. Economic activity is measured in GDP terms (you can debate whether this correctly captures all activity). The immense computation processes outside of bitcoin contribute to economic activity, e.g. Google data centers drive economic activity worth $180B per year. Google uses ~12 terawatt hours (1/10th as much as Bitcoin). Bitcoin produces ~$6B in economic activity (of dubious value imho).

Expand full comment

Is the energy consumption driven by mining or accounting and trading? I have read that mining is energy intensive. But, isn’t it almost all mined? I’d think it that energy consumption for accounting and trading would be small.

Expand full comment

Thank you - I think I understand better now.

There are other energy hogs out there besides Bitcoin - AI, for example. Blockchain/AI will be driving more and more economic activity as they mature.

And again, traditional banking/fiat currency uses more energy than Bitcoin - but the energy use is less obvious because much more distributed and integrated into other sectors.

Expand full comment

Can you clarify what you mean by AI is an energy hog? As I understand it, training a deep neural network using machine learning algorithms is energy intensive, but that executing the resulting trained network is relatively cheap. For example, Apple trained a facial recognition substrate using images of 100 million faces, which no doubt took considerable power, but billions of people can unlock their phones with FaceID in less than a second without having any noticeable affect on the phone battery. Similarly, Waymo has no doubt used massive amounts of computing power training their cars in image recognition and self-driving, but Waymo cars themselves drive around all day using roughly the same amount of energy as regular cars. I have also worked at several tech companies that use ML in their business in ways that are net profitable. If AI were really an energy hog, this shouldn’t be possible due to having to pay for the compute resources. To me, the energy usage of ML seems reasonable. Can you point to instances where it is excessive?

Expand full comment

This is what I read, but I don't have any personal knowledge to contribute. I don't have a problem with energy consumption, but the author of this article does (with Bitcoin). https://www.wired.com/story/ai-great-things-burn-planet/

Expand full comment