I read someone’s estimate — rough, I’ll grant you — that mining bitcoin (just bitcoin) consumes as much total electrical energy as an entire medium-sized nation. Divide network hashrate into how much power the most efficient ASICs use, and I don’t think it’s an unfair estimate. You can’t dismiss the argument that maybe we can do better as “naysaying,” nor that there is something about cryptocurrency that the people making the argument do not understand. How many joules go into making a bitcoin block? (Really, I’m curious.)

It doesn’t take the entire output of a coal-fired generator to handle trillions in trade volume on traditional exchanges. Minting coins doesn’t require an entire nuclear plant dedicated to it. Even for newfangled cryptocurrency, the human race can handle a hundredish billion dollar market cap without using that much energy. It’s incredibly wasteful and destructive to our planet. How do you think the Chinese are throwing cheap electricity at all that mining gear? Provinces covered in solar panels? (For that matter, how do you think the Americans are?)

Literally the entire concept of mining is brute force through the SHA2 problem space for nothing (to create the currency of value; nobody is saying bitcoin is worthless). Imagine if, instead, you dumped that energy into storage or even useful computation (again: proof of work, not the currency, is useless computation), and what you could do with it. There needs to be acknowledgment that brute force cryptographic hashing is a pretty serious design flaw. We just happen to be working around it now that bitcoin is a Big Deal.

To drive this home, Bitfury is building out 30+ MW mining facilities that require immersion cooling in 3M Novec, because state-of-the-art datacenter CRACs aren’t capable of rejecting waste heat from that kind of power density. We are talking 150+ kW cabinet-equivalents, where anything over 15-20 kW per cabinet with CRAC is currently considered noteworthy. AIUI, they have to underclock their gear when relying on datacenter air. Two-phase immersion is a cool, efficient technology that will probably reimagine the datacenter and should be developed further, but machine learning and science workloads are barely starting to crack that nut. Bitcoin ran directly to the front of the line of “incredibly power-intensive workloads that we haven’t encountered in datacenter design before.” That’s good and bad.

straw man. I was responding to the statement that the computation is doing nothing, and not arguing one way or another that the energy used is reasonable for the benefits gained.

Try responding to the most charitable interpretation of others' comments. It's clear that nerdponx is referring to the same thing that jsmthrowaway went into greater detail about.

Try this as a habit: If in response to something someone else has said, you find yourself thinking, "That would be ridiculous! Who would think that?", then follow up by answering your own question: "That would be ridiculous, so the person I'm responding to probably doesn't think that."

Pondering how ridiculous it would be for you to choose to believe that someone else actually believed the first ridiculous thing is an exercise you can pursue at your own option.

While I grant that you may be correct in this case, this isn't a tenable position to take in general. An overwhelming amount of people do believe patently ridiculous things, and frequently talks about them as if they make sense.

I would agree with you if this conversation were taking place anywhere other than HN. This community isn't infallible, but I argue that the level of discourse here is far above the mean.

No, I actually agree with the person. Cryptocurrency is horrendously wasteful. But that person ignored what I said because that's not what I was discussing. That's the straw man argument I mentioned.

How am I building a strawman? I didn’t ignore anything you said; I’m responding directly to it. The computation does nothing. Until the brute force finds a hash that meets difficulty, it’s computing literally nothing of value. That’s computation and energy we do not get back. You’re not folding proteins, you’re not rendering climatological models, you’re not training machine learning. You are hashing, then throwing it away. It’s a “nope, not it, next.” There is no measurable value until by chance you find one, then the energy you threw away getting there gains value in that you threw it away. Millions and millions of watts of absolutely zero computation of benefit to humanity.

Yes, the end result is a secured ledger. The path to that secured ledger is entirely wasted computing. Bitcoin is wasteful solely based on this fact. How can you consider Bitcoin wasteful as I do and also claim people who point this flaw out are naysayers who don’t get it? Your argument makes no sense.

> . Until the brute force finds a hash that meets difficulty, it’s computing literally nothing of value

It's not doing nothing. It's searching - searching for a hash that can be used to protect the ledger from attacks.

Saying it's doing nothing is like saying Dijkstra's algorithm is doing nothing until it finally finds the shortest path. That's just plain wrong. Sorry.

> brute force through the SHA2 problem space for nothing

wrong, bitcoin network brute forces through sha2 to find the one that secures the distributed ledger and transaction platform on top of it. the rest is irrelevant since it's based on wrong assumption.

OK. Of what utility to mankind is SHA256(SHA256(candidate)) after it is discarded by the miner? What do we get from that?

isn't that obvious? it is of utility of verifying that candidate can't secure the ledger.

this really comes down to question whether you think the distributed, writable by anybody ledger is useful to humanity. personally i think it absolutely is. do you?

Not quite. It most certainly doesn’t come down to that, and it’s a reduction of what I’m saying. I’m drawing a distinction between the specific way bitcoin verifies its ledger versus cryptocurrency and distributed ledgers themselves. I took pains to enunciate that repeatedly, so let me try to illustrate with an example:

Say I have one watt of power. I apply it to a computer for however long and it spends that energy folding a protein, simulating a gravitational system, or modeling a hurricane. I apply it to another computer and it spends that watt for however long computing a double SHA256 and throwing away the result. If, at the end of both calculations, I have worked toward proof of work on a secure ledger, which energy was better spent for humanity? That’s the difference: while the secure blockchain is a noble goal, what the computer is doing during that time is something we badly need to rethink. “The double-hash of these bytes is <foo>” isn’t even useful if we collected it. It’s meaningless computing. Yes, we ascribe meaning when blocks are made, but it’s a very irresponsible technical decision to waste energy in order to prove that work was done.

This would be fine if it were a low amount of energy, but it isn’t. We are talking gigawatts that are potentially being spent on this, which is part of the point I was trying to make. Electricity isn’t free, and it’s important to consider our planet. Bitcoin essentially puts computers in a busy loop, and it’s a very hamfisted means of achieving its goal. We have already seen alternative concepts, including this one, that illustrate there is potentially another way to achieve what we want without throwing a small star’s worth of energy at the computing infrastructure behind it.

Coming after people and assuming they’re criticizing cryptocurrency as a concept whenever a specific piece is questioned is tough to discuss against. I think cryptocurrency can do better, and there are undoubtedly better ways to come up with to secure a distributed ledger. I’m alarmed that the adoption of bitcoin has exploded our utilization of resources and that people feel that worry about this is an indictment of the whole concept; this polarization makes it difficult for all of us to improve.

So, no, it isn’t obvious what value I should see out of a computer spinning itself through SHA2 for no benefit aside from securing a ledger. In any other domain, we’d consider such a design ridiculous. There has to be another way, one where that same (intensive) calculation can benefit humanity for the power that we are throwing at it, and we still end up with a secure blockchain. I’m not smart enough to invent it. I hope we can support the people who are, and perhaps admit that bitcoin’s proof of work scheme is a mistake in the larger landscape.

I worry that given how deep we are into ASIC development, arguments against what I’m saying are only going to get louder as time goes on. We are probably long past the point of no return with bitcoin already.

There is no other way to secure a distributed ledger at the moment other than using large amounts of power. That leaves two choices:

1. Admit that you think that crypto currency is not worth the energy cost. Even if you believe that, it's just wasted breath on your part because they are absolutely not going to go away.

2. Propose another solution. No one else on earth has found a better solution than proof of work. Do you have something better?

It’s not wasted breath to say it’s not worth it. The point is that distributed ledgers will not be the future of currency and banking - they’re not worth it. They’ll surely exist as a niche technology or religious artifact for the libertarian set.

> which energy was better spent for humanity?

Wrong question. It implies energy can only be used exclusively for single purpose. The right question is given X watts, what is the ideal distribution for Y purposes and the answer is that we dont know but we tend to value some things more than others and that will be reflected in the distribution

> no benefit aside from securing the ledger

This is the crucial point that many who raise the same objection dont get: proof of work has to be futile by design. In fact it is misnomer and the correct name should be Proof of energy spent, as the energy is the ultimate resource we’re paying for security of the ledger. If we accept that ledger security must be tied to solving some more “important” problem, and there indeed was a non futile problem space, some f(candidate) that yields two results: (candidate-ok?, problem-solved?), the incentive to search that space drops to zero because it is no longer tied to solving the “important” problem. And suddenly the ledger is useless.

If you know how to resolve this conflict - publish a paper and you will be the next Satoshi.

> Wrong question. It implies energy can only be used exclusively for single purpose.

No. Both computers did hard work, and earned credit toward proof of work in my example. I’m rather repeatedly saying energy can be used for two purposes (securing a ledger with a difficult computation that also means something), and I have no idea how else to explain this.

I’m not sure why your incentive would drop to zero. Say I swap in a universal simulation from the Big Bang instead of SHA2. A miner simulates a portion it chooses, which involves quintillions upon quintillions of gravitational fields. Eventually it simulates a rare event, like two black holes colliding or something. It somehow notates this as proof that it worked on the problem for a while, hashes that proof combined with the transactions, and boop, mined a block. Meanwhile, all along, we have learned science. I could think of other problems that work: SETI@home and the other BOINC apps come to mind, as does LHC analysis. Large search space, demanding algorithm, rare thing to find (SETI/Higgs candidates), and valuable science along the way. Some issues I could see are unpredictable nature of block time, verifiability being expensive, and problem spaces that move forward in time, like my first example. But I don’t see those as insurmountable.

Why would incentive drop to zero to search the space? I’m not following that claim, nor why you’re scare quoting important.

> earned credit toward proof of work in my example

they both did work, they didn't yield any proof of it.

> Say I swap in a universal simulation from the Big Bang instead of SHA2

if you have an idea - express it in pseudocode. what is your `f(candidate)` here? what is the return value? how does return value ties in "is candidate valid" and "simulation data" in a way that prohibits modifying source code to only do "is candidate valid" part?

you're trying to handwave your way through some of the most complex questions in computing. "what if we just swap sha2 for blah, where blah is something that solves humanity's problems" - well sure, let's do this, show me the code.

> Why would incentive drop to zero to search the space?

because the axiom we're operating with is: though shalt not run ledger security code without "useful" code. as soon as the problem is solved - nobody secures the ledger anymore.

> why you’re scare quoting important

because you're forcing upon me what i should consider important problems humanity has to solve. your opinion is that distributed ledger isn't one of them. your opinion is wrong.

> your opinion is that distributed ledger isn't one of them. your opinion is wrong.

How many times do I have to say in a single thread that this is not, and will never be, my opinion?

3 options here, pick one:

- acknowledge the fact that ledger has value and therefore corresponding amount of resources are OK to be assigned to that valuable problem

- admit that no resources should be assigned to ledger problem because there are more important ones

- present your formulation of compound "important" and ledger problem such that it satisfies the conditions i mentioned above

am i missing anything?

And that hard to reproduce hash represents computational trust, which is a new type of resource that just happens to be very expensive in energy cost using our current technology.

It would be expensive by design no matter what the technology.