As an engineer, it saddens me to see the huge inefficiencies inherent to cryptocurrency mining these days. Surely, when we're spending in the billions on wasteful hashing, we could dedicate a small fraction of that to engineering better consensus algorithms.
Unfortunately, we have an ecosystem where everyone is incentivized to ignore the problem because they're too invested in maintaining the price of Bitcoin and other PoW cryptocurrencies. They often don't pay for the externalities — environmental costs. The specialization of mining also severely hurts the distributed and decentralized nature of these crypto-currencies.
I'm a huge fan of crypto-networks, and have made a career in the space. But, I want Bitcoin to die off and leave room for newer generation technology.
This is a one-sided comment. There’s a lot of research going into different variations of proof-of-stake. 4 out of 10 largest crypto (by market cap) are PoS. There’s a lot of work Ethereum community (Casper), Cardano folks, Tendermint etc are doing to create better consensus models.
Proof-of-Stake statistically equates to "Give the rich more money".
Consider Ethereum's stats (and other similar PoS minting methods):
Presale ICO / Premine ( max cost $0.50 USD per ETH )
= 72,009,990 ETH
Total Supply today (Feb 23rd 2018)
= 97,800,000 ETH
Source:
https://etherscan.io/stat/supply
Now imagine a financial system where all the wealthy have to do simply own money (spawned from software or premine) to get more money. Or where voting is done by merely by controlling a large sum of money/tokens.
Except only the top few largest "stakers" are paid, others are either disqualified entirely for being too poor to stake, or will not recieve block rewards for being statistically irrelevant.
PoS is a lottery with reusable tickets. Winners get more tickets.
With PoW you use your money to buy mining equipment and you will get more money, and voting power. With PoS you buy more of the currency with that money instead.
Right, but the rest of us also have to suffer due to the economic repercussions (or being unable to get an affordable graphics card without waiting forever)
Just pay more for the card and you can have one whenever you want. Isn't that what HN thinks employers should "just" do with tech talent because "there is no real shortage"?
Nano uses dPOS and does not mint any new coins with PoS nor does not charge any tx fees. Instead of "Give the rich more money" it's more like "preserve the value of my money by fulfilling the promise of this coin".
Nano/Raiblocks were created from software, a reproducable distributed database protocol. Except in the case of Nano, the vast majority of the tokens are in control of the owners with a few being given away to the public for clicking captchas.
What evidence is there 90% of Nano distribution isn't owned by a single person?
This is a major aspect of transparency in decenteralized p2p protocols, and the issue with Nano is there's no paper trail or proof that the devs didn't take all the supply.
We do know 13% of all Nano in circulation was stolen in the Bitgrail hack.
PoS is not going to be very popular as it removes the literal "money out of thin air" that was happening with GPUs in the past few years (coming to end in medium term). PoS strongly favors already well capitalized owners, instead of allowing new ones.
Only for the largest account holders, whom statistically will receive most of the newly generated money (and even more as they continue to accumulate newly printed PoS tokens).
PoW distributes block rewards proportionally to hash power and these proceeds can be used to buy more hash power to achieve a compounding effect. From what I understand all of the viable PoS systems offer a fixed rate of return on the stake. This is like PoW forcing everybody to use the exact same hardware; a benefit to the little guy at the expense of the large account holders.
Transitioning from PoW to PoS seems like a possible solution. Or shortcut the PoW phase by seeding your coin with snapshots of one or more existing coins that had a “fair” distribution phase.
The ugly truth is that proof-of-stake systems, at least current ones, give up decentralization properties. Users with the most funds (stakes) can effectively validate and rewrite the blockchain. Or in the case of a network split, the 2 chains can no longer be merged (because they implement "safety" mechanism such as preventing nodes from rewriting more than X blocks.)
I'm more concerned that it takes something "wasteful" like mining to make people wake up. What about DRM techniques like HDCP? Even with hardware acceleration we've been wasting power for at least a decade to encrypt every single video stream so that we can stop piracy. Piracy still exists. Mining most definitely uses more power than video encryption but only one of them has really produced anything of value.
Efficient? Maybe, you can still have inefficient hardware. The hardware dedicated to HDCP encryption is single purpose, just like a Bitcoin miner. I'd expect the Bitcoin hardware is far more tuned for efficiency than most hardware HDCP circuits since power usage is the major cost of owning a miner in the longterm.
I'm not aware of any actual metrics and like I originally said I'm pretty sure bitcoin mining still uses more energy.
Bitcoin (well, its descendants now) makes my international money transfers dirt cheap while DRM prevents me from using media that I paid for on my favorite devices.
Have a sense of proportion. Think about how much electricity, paper, waste and money is spent to keep up the banking industry, anti-fraud departments, AML/KYC compliance teams, clerks and all that show. Bitcoin is actually an improvement.
It's funny you mention anti-fraud departments and AML/KYC compliance teams since Bitcoin does not provide those and thus their cost exists in addition to the cost of mining.
But Bitcoin doesn’t provide the services that the banking industry does. Bitcoin is more comparable to physical gold than to a modern bank account. Banking was built on top of gold (and later fiat) because it provides valuable services. If bitcoin becomes a widely-used currency then the same banking system will be built on top of Bitcoin.
Proof-of-Work consensus algorithms (so far) work by wasting more and more energy.
Max efficiency peaks in the network with a very basic CPU
PoW algorithm merely sets a lottery number granting write access for a single computer
Every additional increase in processing power reduces efficiency by increasing the size of the number to be guessed
All PoW does is increase the capital cost for any user who joins after x-amount of time has passed.
I like that analogy: “setting a lottery number” - that feels more accurate than the typical way the lay media explains mining, usually a variation of “solving lots of puzzles”.
“Puzzle” implies some cognition is required to come to a solution - not that effort is being wasted by brute-forcing something.
It's a problem in the realm of distributed computing.
PoW attempts to create a mechanism to verify accuracy of transactions in the database among a group of users who would want to cheat, suppress, or manipulate the records.
So PoW uses the lottery to create a competition among all users for high computational power (capital sacrifice), but unfortunately the computational "work" is nothing more than a random number generator.
Ideally, the computational work used in PoW would not be worthless random numbers but could instead be computations that provide value to scientific research. Distributing computing projects like Berkeley Open Infrastructure for Network Computing (BOINC), Folding@Home, and Seti@Home, are all very similar distributed computing projects that predate Bitcoin.
The main problem is PoW for transactional verification needs to be a problem that's hard to known in advance, yet very easy to verify once known, and there must be a consistency to the solving of transactions. This can be done with a centralized validator, but presumably the ideal solution would need to be distributed/decentralized as was the point of Satoshi's p2p cash system.
The problem is it's very hard to have consistency of successful problems solved when you have complex useful computational problems being solved, whereas PoW simply asks "guess a number, now guess a higher number if the network hashrate is faster".
If anyone solved a method to process useful valuable computational work, instead of random numbers, and applied it algorithmically into a PoW transactional protocol - this would be a holy grail solution as it wouldn't be considered a race to the bottom of inefficient computational waste.
Golem seems to be attempting a distributed computational work for digital tokens model, but I have doubts their design will be genuine, effective, or sustainable.
Thanks for the comprehensive response. It does seem like there should be a way to accomplish something more meaningful for society as part of PoW. That link was an interesting read, I was unfamiliar with "Notary digest chains." I have now some new reading. Cheers.
Bitcoin mining seems inefficient now, but when the network is carrying 100x more or 10,000x more transactions, won't it be extremely efficient? (On a per-transaction basis)
O(n) total world energy expenditure on Bitcoin scales to O(1) transactions, so it currently only gets worse as more energy is poured into the system for the same transaction output.
There have now been about a half-dozen various suggestions to get the network to carry more transactions, all of which are mostly failures, except for Lightning Network, which is now in the process of becoming a failure.
I have been wondering if an algorithm of Proof of Spending could somehow be in the mix. People should be rewarded for spending their money to enable a healthy economy. Of course something like this couldn't be the main part of an algorithm, but maybe somehow be considered. Not sure if there has been any research in that area.
Switch, no. The ASIC miners would never vote for something that would destroy half of their investment. A fork could totally happen. That's what Bitcoin Gold did by forking the blockchain to using the Equihash algorithm.
95% of people wouldn't be interested in crypto if it didn't make them money. You're asking people to give up the money to improve the system for what? So they can't make as money?
Crypto isn't charity. If you want better tech, you have to prove why it is worth it.
This 4 billion dollars and all the money wasting on power and ASIC chips is a direct tax on the bitcoin ecosystem. This is the incentive you and your currency/consensus model have to beat bitcoin. In addition to the engineering problem, the number of people worldwide who prioritize environmental preservation mean that a currency that can preserve or even improve the environment are going to be preferred. What if we changed proof of work to be proof of not-cutting-down rainforests? 100 secure micro-satellites circle the globe taking pictures of areas prone to deforestation and distribute currency based on the areas not being logged?
This means you're an "engineer" but not yet an economist.
Try to study economics if you're really serious about your career in cryptocurrency area, you'll find yourself thinking completely differently. At least you'll get to a point where you don't make a wholesale criticism like this but can say "I know it would be better to have a tech that consumes less energy, but I also understand why it's not a trivial issue, because physics and economics."
One of the biggest fallacies in the history of economics that repeats itself over and over and over again is that people make decisions based on what they can see but they neglect what they can't see that happens as a result of what they can see.
As much as I would like to see PoS work, it may just be the case is that at the end of the day, PoS and PoW are the same in the grand scheme of things, not because of some greedy people with financial interest but because of physics. It's hard to make sense of this unless you study economics, no matter how great of an engineer you are.
[Edit] I hope my comment works as a wakeup call for those of you who are open minded enough to learn a piece of the puzzle that's obviously missing from your mindset (I shared it because once I learned it, I was so regretful for not having studied even a little bit of economics in my life until now), but looking at all the downvotes, looks like most people don't like to hear what they don't like to hear. I don't really mind, suit yourself. But just remember "Cryptocurrency" is "Crypto" + "Currency", which means it's not just CS, but also economics.
As I am reading your comments more, it seems like you just don't understand this trifecta of software engineering, security, and economics. All blockchains that run code on untrusted nodes have the extremely difficult problem of securely verifying other nodes transactions which usually devolve into solutions that solve O(constant) problems on O(N) nodes with N/2 majority consensus governing the network. That is literally the worst solution you can come up with in computer science because it does not scale whatsoever.
Contrast this with running code on nodes you trust which scales linearly and you should be able to see why everyone says blockchains are a solution in search of a problem.
> "I know it would be better to have a tech that consumes less energy, but I also understand why it's not a trivial issue, because physics and economics."
So thus I ask, what "physics and economic" problems do blockchains solve better than current solutions bar illegal markets?
Proof of Work and Proof of Stake both waste tons of energy, and of course, the entire economic system both of them are trying to replace also waste tons of energy.
The problem is, Proof of Work is very objective and measurable, which is why most people criticize Bitcoin for its waste of energy since it's so easy to measure.
But proof of stake and the legacy economic system both lack such objective measures, which is why it's hard to compare them. The term "stake" itself is subjective and doesn't take into account the economic ripple effect it can trigger as a result.
And these ripple effects DO exist, it's just that most people are not aware of them and just take for granted. Which is why I think it's crucial that you study the history of money before you can confidently say you understand "how money works".
This is not to say proof of stake won't work. It's just saying that there's a high chance that there's hidden cost to proof of stake which in the long run can map to the amount of energy wasted by proof of stake, which is backed by newton's 3rd law. You can't create energy. Everything that happens on earth is simply a transfer of limited amount of energy from the sun.
If this sounds all woo-woo to you, that's why I suggest studying economics.
I personally think the only way to overcome energy consumption is to come up with innovation that's as different from newtonian physics to quantum physics.
> The problem is, Proof of Work is very objective and measurable, which is why most people criticize Bitcoin for its waste of energy since it's so easy to measure.
> But proof of stake and the legacy economic system both lack such objective measures, which is why it's hard to compare them.
Bitcoin is estimated to cost $2.5 Billion dollars per year [0] to validate about 77 million transactions per year.
Visa processed about 100 billion transactions in 2015 at the cost of 12 billion dollars (revenue - net income [1]).
That makes Visa 270 times more economically efficient for cost per transaction [2]. Also, I get many other benefits using Visa like fraud detection whereas bitcoins 1 way transactions make them extremely susceptible to hacks.
Visa may well spend only $0.10 on a $1000 transaction but given that the total cost to me is $30 (either directly, in the case of a foreign ATM withdrawal, or indirectly through merchant fees), there are many use cases where crypto is currently more economical from the consumer's point of view.
If banks and payment processors were more competent (like maybe processing transactions on Sunday) and less greedy, crypto would be a lot more useless.
There needs to be a HN bot that just copy and pastes the same old stuff to comments like mine so people don't have to waste energy doing it over and over again.
The fact that you're comparing Visa with Bitcoin is exactly my point: Humans make decisions based on the things they can immediately see and neglect things they can't see. That's what you're doing when you're comparing visa to bitcoin.
I can't really explain this in a single HN comment, which is why I recommended studying economics. If this was so easy that I could explain it in a page nobody would need to read books or take classes.
> The fact that you're comparing Visa with Bitcoin is exactly my point: Humans make decisions based on the things they can immediately see and neglect things they can't see. That's what you're doing when you're comparing visa to bitcoin.
Then can you explain the economic utility that bitcoin provides because it is more innificient way to store and exchange "value" than what banks provide?
> I can't really explain this in a single HN comment, which is why I recommended studying economics. If this was so easy that I could explain it in a page nobody would need to read books or take classes.
I have taken many econ classes and I have read dozens of econ textbooks. You are handwaving about "ripple effects" and that if I knew econ it would all make sense.
> Then can you explain the economic utility that bitcoin provides because it is more innificient way to store and exchange "value" than what banks provide?
Let me ask you back. Can you prove with 100% certainty that banks are more "efficient" than bitcoin? What does "efficient" even mean? Did you take into account all the jobs required to run the banks? Did you take into account all the energy these bank employees consume? Did you take into account all the side industries that arise around banking system? Did you take into account all the corruption that took place over and over and over throughout history that eventually took everything down to zero? This question is really meaningless when you consider all these factors.
People tend to arrogantly think that they understand the world when they don't. All they do is try their best to make sense of things, and even with these scientific theories, nothing is absolute. Note that I'm not calling YOU arrogant. I'm calling humans as a whole.
I am not saying PoW is better. I'm saying people who think PoS is absolutely better than PoW are leaving a lot of things out of the picture when they draw that conclusion. Nobody knows what's better or what's more efficient. Maybe banking IS more efficient, but maybe in the long term (as in 100 years) it may be more inefficient. (History shows that there's A LOT of invisible inefficiency in the banking system, manifested as economic depressions)
> Let me ask you back. Can you prove with 100% certainty that banks are more "efficient" than bitcoin? What does "efficient" even mean? Did you take into account all the jobs required to run the banks? Did you take into account all the energy these bank employees consume? Did you take into account all the side industries that arise around banking system?
What? I thought you said you were well versed in economics. This is literally one of the core concepts for the benefits of economics because you can evaluate things on their "efficiency" (costs) versus the value they provide (revenue). Every "job", "energy cost", and "side industries that arise around banking system" are evaluated in the cost to run the business and if they were greater than the value they provided, they would not be in bunessines because revenue would be less than sum of those costs.
> Did you take into account all the corruption that took place over and over and over throughout history that eventually took everything down to zero?
Obviously corruption is not good for our systems neither are other economic problems such as negative externalities [0]. However, I don't see how cryptocurrencies solve them in a way that makes them better than current solutions as a whole.
> This question is really meaningless when you consider all these factors.
I'm not asking you to write a phd. Just give me one good economic use case outside of illegal markets for cryptocurrencies that is better than current technological solutions.
> Just give me one good economic use case outside of illegal markets for cryptocurrencies that is better than current technological solutions.
Transfer $1000 from your Coinbase (or other electronic) account in the US to cash in Indonesia, in less than 20 minutes, at a better rate than spot. I did it yesterday with ETH and even made $20 on the spread after all the fees.
The cheapest conventional alternative would have been a Visa or Plus ATM card which would have cost around $14-$47 depending on how good your card is.
I used limit orders to add liquidity in both countries and so paid no trading fees. There was a 1% “cash withdrawal fee” at the Indonesian exchange which is all I paid in total.
I don’t know how temporary it will be; I’ve been using crypto to transfer money internationally since 2013 when BTC was $800 and it’s always been the cheapest method (although I sometimes had to write bots and wait a few days to get a good deal). I imagine the spreads will continue to trend towards zero so I’ll be paying 0-1.5% total instead of earning 2% which is still much better than banks for three-figure to low four-figure amounts.
PoW requires brute force. PoS requires tied up capital. Do you really think the energy usage is the same...physically? Sounds like nonsense with no backing.
Also, I didn't say they ARE the same. It COULD be the same, or maybe PoS could come up with a tradeoff scheme that's just good enough to work but essentially conserves energy much more than PoW. But the thing about energy is when it is destroyed somewhere, it gets created elsewhere. That's how it simply works.
The most important fallacy in economics is that people can only observe what they can observe, not because they're idiots, but because it is spacially or temporally impossible. For example a huge global inflation can happen as a result of a very small stupid decision made by a government, but this may happen 10 years after the original decision.
This is why humans make bad decisions because they make decisions based on what is possible to observe, but historically these ALWAYS came back to haunt them. 100% of the time.
It MAY be the case that "this time it's different", but the more you study economics the more you become really humble. Trust me. Try it. Just need to read a couple of books.
"need to read a couple of books." No. Just no. You are wrong about energy. Power plants do not just burn a steady state amount of fuel. You are also wrong about capital if that is the "energy" you misnamed - capital is created all the time, not debt but actual prosperity through innovation. Please refrain from your "read a book to be like me" humdrum. I am a mathematician - I know enough economics to be quizzically annoyed by the falsities spoken
The human races net work output might be fixed - but through the spawned creations and mechanics - the work one does and thus the spawn they create, can be vastly consequential. Energy consumption is not fixed. Human work is fixed but when augmented by technokogy it is wildly divergent. There is a difference between using millions of watt hours for mining versus using it to create things that have a far longer shelf life..things like crops or PCs which will continue to provide further value. Bitcoin mining is short sighted.
Those companies at at the top all still have competition who need to buy the same supplies, so they aren't monopsonies, but they're close enough that this is an opportunity to use the word 'monopsony'.
Are these things true or equivalent? These areas are not necessarily booming. They are stable with few consumers of the goods.
At this point, every scheming hustler is in the market for a bit coin miner. Same in a gold rush -- any fool can dig with a shovel. It's not until boredom or consolidation happens that the shovel maker margins start getting thinner.
Better is to think of it as making something a lot of people want. Lots of miners = lots of people who want shovels/ASICs. Whereas mature industries like those end up as monopsonies i.e. a single buyer with lots of suppliers.
In most of those cases the guys higher up the food chain have appreciating assets such as land, consumer brand value, mining rights, developed oil fields and mines, etc. The guys lower down the food chain make depreciating assets and have to get up every morning and make new stuff to sell every day or they go out of business.
In this case, the shovel seller is believed to use the shovels a bit before selling them. In that way, they extract all the value from the shovels themselves, before a purchaser ever digs.
If it’s BTC then a) the real figure is much lower because they can’t sell $4B of BTC without drastically moving the market, b) what portion of that is due to their BTC holdings increasing in value by more than 1000% in the past year?
I’m guessing Bitmain can’t pay for electricity with BTC.
Are there estimates of how much (fiat) money flowed into cryptocurrencies in the last year? It would have to be much more than $4B for Bitmain to be able to sell $4B worth of BTC and still see the 1000% gain.
The daily volume at the beginning of 2017 was $200M USD.
The daily volume at the end was $8T USD. $4B over 365 market days is somewhere between 1-4% of the daily market.
If liquidating their bitcoin tanked the market in 2017, it wasn't too noticable.
A bunch of that is probably going to be in BCH, since they were requiring that for mining hardware purchases for a while, and that's an even shallower market.
Right, that’s my question: are they selling it as it comes in, thus have actually made $4B in USD (despite the downward pressure of selling), or have they been hodling BTC and now have $4B of BTC at the current market price (which is not actually worth $4B)?
> The increased demand for special hardware like GPU’s and ASIC based systems has created a black market where hoarders buy the hardware in the market and sell for exorbitant prices on websites like E-Bay.
Since when is ebay considered the black market?
I always thought "buy low, sell high" was the ebay motto...
Bitcoin mining is a real world version of Keynes' quote about labour productivity[1].
The important difference is that the money doesn't belong to any country, and is transformed purely into personal wealth. This will lead to wealth inequality the likes of which we have never seen. Governments should charge miners a lot more for just existing, and using all the infrastructure the public has funded to spin money out of thin air which will never help pay for that infrastructure or contribute to any of the economic activity which keeps a country going.
Why we let people dealing in financial instruments acquire so much wealth and power continues to baffle me.
This is also a good argument for the wealthy to pay their taxes. The roads people use to get to the store cost real money. But that doesn't really happen either.
Why not create a coin awarded for proof of AI-modeling and open-up the network to researchers? Seems like it would be a lot more productive if all that processing power went to neural networks.
“ASIC-resistance” just means “needs lots of RAM”. Eventually someone will figure out how to make oodles of RAM sufficient for a many-core custom processor. Heck, modern SSDs (e.g. Intel Optane) might be fast enough to replace DRAM for these kinds of applications.
Another kind of ASIC-resistance might be using dynamically-generated algorithms employing self-modifying code. You’d have to own a Xeon Phi to get good figures for that, methinks.
It’s a shame PoW is based on arbitrary hashing instead of, say, “Proof-of-Folding@Home” - that would do some good at least, methinks.
> It’s a shame PoW is based on arbitrary hashing instead of, say, “Proof-of-Folding@Home” - that would do some good at least, methinks.
It already exists, it is called Gridcoin and is based on Proof-of-Research. Coins are awarded based on contributions to the Berkeley Open Infrastructure for Network Computing (BOINC). There are many scientific computing projects you can contribute processing power to that utilize the network.
> Eventually someone will figure out how to make oodles of RAM sufficient for a many-core custom processor.
It's already been figured out. The issue is that ASIC manufacturers don't see the need to switch from BTC, when they are already making money. Hint, it's the type of ram.
Unfortunately, we have an ecosystem where everyone is incentivized to ignore the problem because they're too invested in maintaining the price of Bitcoin and other PoW cryptocurrencies. They often don't pay for the externalities — environmental costs. The specialization of mining also severely hurts the distributed and decentralized nature of these crypto-currencies.
I'm a huge fan of crypto-networks, and have made a career in the space. But, I want Bitcoin to die off and leave room for newer generation technology.