Hacker News Reader

Then there are other societal costs. Like people losing power. https://www.independent.co.uk/news/world/iran-bitcoin-electr... Even rising cost of electricity and added pressure on ASIC/GPU production due to bitcoin is an externality we all pay for. Ask anyone who has been trying to buy a GPU for the last 6 months.

You're lamenting the short-term downside, but you ignore the long-term upside. If it hadn't been for the crypto boom, GPUs would not be where they are today.

I find that really unlikely. Machine learning, rendering, and other massively parallel workloads were plenty to keep gpus advancing. All the crypto boom did is create shortages and raise prices. At best we got some additional chip fabrication capacity, but crypto continues to eat that.

> At best we got some additional chip fabrication capacity, but crypto continues to eat that.

That’s the funny thing about this genre of arguments (Bitcoin improves GPU tech; Bitcoin subsidizes clean energy) — even if true, the gains will only be realized if there’s a crypto price crash, since otherwise mining absorbs the gains that it is responsible for by definition.

Not at all, the refinement in semiconductor production is a permanent benefit into the future for every computing application. GPU mining has only caused temporary shortages three times in the past 10 years. Prices don't need to crash either, due to difficulty adjustment the profit will trend towards marginal cost, which can only be reduced with better technology.

Improvements in semiconductor manufacturing would happen either way eventually; to the extent that Bitcoin accelerates them it is (by definition) by creating the demand that necessitates them.

It's not zero-sum. These improvements are externalities, they are permanent and apply to all other uses of computing.

Similarly, one can point out all the advances that the Apollo program brought that are unrelated to space travel. COVID brought in lots of scientific insights that will bear fruit regardless of whether we have to deal with any other coronaviruses in the future.

Sure, and smoking is responsible for advances in lung cancer research. It's the broken-window fallacy, though. The resources that were reallocated to lung cancer research due to smoking could have been used for other things.

> It's the broken-window fallacy, though.

The broken window fallacy does not consider the benefits from improvements of technology that may result from destruction. Suppose that smashing a bunch of windows would lead to permanent improvements in glassmaking efficiency. In that case, the benefits in the long run would far outweigh the short term costs. Of course, that's not a given with any form of destruction, but I am arguing that it is the case with cryptocurrency.

The fallacy does account for that. In a counterfactual where the windows were not broken, the effort spent improving the efficiency of glassmaking would have been directed (by the invisible hand) to improving the efficiency of something else.

It doesn't account for that. If it did, it would assume improving the efficiency of one thing is equally valuable to improving the efficiency of any other thing, which is absurd.

Like I said, not all destruction is destined to create excess value, but some destruction is, and I am arguing that this is the case here specifically. A counterfactual where this money would've been spent on something even more beneficial doesn't really matter. It wasn't spent that way and it could also have been spent in numerous other ways that are even less beneficial.

> Machine learning, rendering, and other massively parallel workloads were plenty to keep gpus advancing.

I think the opposite is true. Now that we have cheap GPU accelerators, the machine learning revolution is taking place. But where does that cheap GPU accelerator come from? Clearly, the more frivolous "gaming demand" was paving the way for machine learning, not the other way around. It's similar for mining.

> All the crypto boom did is create shortages and raise prices.

Rising prices are a signal to the market that production needs to be increased, which lowers prices in the long term. Shortages are short term.

> (most mining occurs on the grid)

The post you were replying to is disagreeing with this. Power made available for extremely low cost next door to a hydroelectric dam etc. might be technically "on the grid" in that there is some connection between that town and the rest of the grid, but if the long-distance transmission lines are already at maximum capacity (which is usually the case, otherwise power wouldn't be any cheaper in those towns than anywhere else) then that surplus power is de facto not available to the rest of the grid.

Are you suggesting that the vast majority of mining operations are running on direct/dedicated links to local power generation facilities? I can understand the argument that some significant fraction are, but there's still a huge amount of mining that would by necessity have to happen on the grid. If consumers or small-time business owners get involved in crypto they have no choice but to use the grid. The costs involved in building out dedicated lines like that are not trivial, so I'm kind of skeptical that it would be widespread enough to make on-grid mining a rounding error. Do you just mean that in terms of revenue, all the big-earning miners are hooked directly up to dams?

> Are you suggesting that the vast majority of mining operations are running on direct/dedicated links to local power generation facilities?

Not necessarily in the sense of having their own physical wire, but if the path from the power generation facility to where the mining is taking place doesn't take up any long-distance transmission capacity (which is why power in those towns is cheap) then it's de facto the same thing.

> there's still a huge amount of mining that would by necessity have to happen on the grid. If consumers or small-time business owners get involved in crypto they have no choice but to use the grid.

If they want to participate as full nodes, sure. Maybe they shouldn't. And even if they are, they're going to be an insignificant fraction of mining volume.

> The post addresses this question in detail. In brief summary, extracting "green" energy from the grid (most mining occurs on the grid) means that the energy available on the grid is more likely to be from CO2-generating sources.

This is a hypothesis, but where is the evidence, i.e. actually increased CO2 emissions? My argument is that most of this mining is using electricity that would otherwise not be used or produced. That is why it is so cheap in the first place.

We have concrete evidence from the recent storms in Texas, where CO2-producing power generation was spun up in order to respond to demand once the price caps were lifted so it could be profitable. It's up to you to decide whether Texas is an outlier.

If the power was literally not being used the cost would be negative - this is a real thing that happens. If it costs you more than $0/kWH you are creating demand and potentially increasing CO2 emissions.

> It's up to you to decide whether Texas is an outlier.

That isn't really related. Like I said, it's possible the temporary spikes in prices can make mining with fossil fuels profitable, but that is in the short term.

> If the power was literally not being used the cost would be negative - this is a real thing that happens.

Price only goes negative when there is too much energy produced, but downregulating production would be even more expensive. This is mostly a problem with wind energy combined with coal or nuclear plants. Otherwise you would downregulate production so you don't have to pay the penalty. Consuming this excess energy in a profitable manner would keep prices above zero.