This is not at all surprising. A lot of the machines on the Bitcoin network are ASICs (Application Specific Integrated Chip), which are extremely good at doing one thing: calculating SHA256(SHA256(x)). You can't use them for anything else, so it's a little unfair to compare them to personal computers.
Nitpicks aside, though, the Bitcoin network's hashrate is mindblowing.
In my top-level comment, I wasn't disagreeing with anything you've said but was bemoaning the use of the words "computing power" since those ASICs are doing computations as well.
ASIC's consume very little power compared to a PC. One watt per gigahash is a rough value, although most promise quite a bit less. You'd need about 100 PC's, at 300W each, to produce a gigahash, using the assumptions in this article.
Ah but that's the unfortunate twist. If you can get a gigahash for 30000W with PCs, you would have no incentive to go down to consuming only 1W with an ASIC. You would consume 30000W with ASICs and get 30000 gigahashes.
[EDIT: Corrected the numbers to match what you wrote. It is irrelevant, though -- if you get a gigahash for X watts with PCs, you'll use X watts to get X gigahashes with a 1W/gigahash ASIC.]
All you are saying is that the transition to 30kW of ASICs will not happen overnight. If you stopped at 1W, you would quickly find that you are making less money than you were before -- your competitors will also switch to ASICs, and will consume more than 1W because they want to make more money.
To put it another way, not going to 30kW of ASICs is equivalent to taking 30kW of PC mining hardware and turning some of it off (in a pre-ASIC world where PCs are the most efficient mining hardware available).
How much energy, resources, etc are required to sustain cash and credit card transactions? Think of all those banks, skyscrapers, ATMs, computers, money printing machines. All of those things exist to power the security of the modern financial industry.
If bitcoin powers the post-modern financial industry, the resource savings will be enormous.
And yet the actual functionality of Bitcoin - a simple ledger - could be implemented using much less resources than are used to maintain the Bitcoin network. It's curious, isn't it?
Mining is what makes the bitcoin network secure. The higher the hashrate, the more secure the bitcoin network. As I mentioned in another comment, if you compare the resources required to secure the bitcoin network vs the resources required to secure any other financial networks, it's miniscule.
First of all, there's no such thing as a perfectly secure system, second of all, you tar people who are interested in decentralization at your own peril. The type of power that is being consolidated in multinational corporations is greater than anything that's ever existed in the history of the world. If we don't want to wake up as serfs one day we should be taking decentralization and other technological democratic aids seriously while we have the chance.
As opposed to the power Bitcoin exchanges, Bitcoin mining pools, and Bitcoin developers have over Bitcoin? The supposed decentralization in Bitcoin is theoretical at best; in practice, power is concentrated in a few key places.
So really, we could make a better system, one that uses far less power, by just accepting that some things cannot be decentralized and designing a system accordingly.
Right, so because there are emergent power structures of some sort, it's all equivalent and there's nothing actually decentralized about it. I mean, Linux might be "open source", but really Linus is calling the shots, so it's really no different than Windows at the end of the day.
You had said that we should criticize decentralization at our own peril because power in the current finance system is being concentrated on a small set of corporations. The only difference between that and Bitcoin is the scale -- the amount of money involved with Bitcoin is nowhere close to the amount of money that multinational banks are dealing with.
The difference is that with Bitcoin, we are also pouring otherwise useful energy down the drain as part of the security model. The is purely because of a fantasy about an ideal kind of money that requires no central authority. If we were to drop that fantasy and accept the existence of authorities in the system, we could create a digital cash system that:
* Has the same kind of payment security properties as Bitcoin
* Actually allows for anonymous payments
* Supports offline transactions
* Requires many orders of magnitude less power than Bitcoin
Of course, as soon as you say "central authority," the Bitcoin crowd dismisses the idea entirely, regardless of its merits and regardless of the existence of concentrated authority in Bitcoin.
Wrong. You can have a central authority that issues money and still have a censorship resistant system -- this is what paper cash is, and this is what is possible with the digital cash systems academic cryptographers have been studying since the 1980s.
In other words, you can set things up so that no single authority can fail in a way that would cause the entire system to fail. You can have N authorities such that a minimum of K are needed to keep the system working.
They aren't perfectly secured. They seem secure for two reasons:
Try to hack a bank, or exploit credit cards, or even break in a vault, and you'll see what kind of external measures governments and private agencies put into protecting the system (hint: you'll get locked forever, or shot, or worse)
Now imagine you succeed, and somehow don't leave any exploitable trail, do you think a story will pop somewhere? I don't.
Centralized money systems are the reason today, what you make in a month working 16 hours a day in some country will buy you a drink in another. This allows rich countries to straight up buyout any poor country's production, and for trivial amounts. It leaves them in a state where people are starving, even though they produce enough to feed 10x their population.
Imagine a world where a bag of rice costs the same in every country. Yeah, currencies are awesome.
Edit: just to be clear, I'm not saying rich countries are evil(er), because this would not work without the cooperation of poor countries leadership.
Well, take a rice plantation. You need people to cultivate, maintenance, supplies, management. What would those people make on average in the US? 20k?
20k buys you 40t of rice a year on the current marker. How many people do you need on an exploitation to produce 40t/year? 3? 5? 10?
There's your number. With a year's work (doing the exact same thing), you can buy several people's yearly production. The reason for that are currencies.
Without them things would equalize over time (it's happening in EU right now, though only internally). It's a long and bumpy process, but eventually it removes the ability from a country to drain another for next to nothing.
Edit: deleted post said there was evidence a uniform supply pricing was actually bad for poor countries
first of all perfectly secure systems do not exist. but lets say for a minute you mean visa or mastercard are perfectly secure, do you think that the electricity use of the bitcoin network exceeds the electricity use of VISA all over the world? They run their own always on computer network and offices, which need people and computers that require electricity to be able to function properly. Then factor in that this is repeated in almost every country they operate in, sometimes more than once depending on the size of the country. So the electricity use of bitcoin is just a drop int he ocean compared to what the current banking industry uses.
"do you think that the electricity use of the bitcoin network exceeds the electricity use of VISA all over the world?"
That is the wrong comparison. The right comparison is this: does Bitcoin require more electrical energy per transaction than VISA?
If the answer is not "yes" today, it will eventually be "yes." Here is a thought experiment:
Suppose that a new technology is developed that halves the energy required to process a VISA transaction and check for fraud. VISA has every incentive to adopt that technology, and thus reduce its power consumption.
Now suppose that a new technology halves the energy required to process Bitcoin transactions i.e. to compute hashes. Since miners profit by computing more blocks, their incentive is to adopt the new technology, compute hashes at twice the rate, and their net power consumption remains constant.
However Visa has no incentive to search formore efficient technologies where are you in bitcoin my name is constantly being developed to become more efficient and it isn't necessarily the number ofhashes you are getting but thenumber of hashes you get per watt of electricity used. so the incentive is there for minors always, this is not the case with visa
I would add to that that the only part that requires this proof of work when decentralized is the timestamping. We can have a system where everything is decentralized with a timestamping subsystem (that can also be composed of many computers for a balance of power) that blindly timestamps transaction hashes without even knowing what it is timestamping so that it cannot censor it. This would be a perfectly reasonable system for me and one that would spare that huge cost of electricity burned on proof of work.
> The higher the hashrate, the more secure the bitcoin network.
wat
edit: clarification, the total hash rate has zero bearing on the overall security. The question is about how hard it is to gain majority, in other words, how that hashing power is distributed.
At the time of writing, the two largest pools have 50% of the hashing power: https://blockchain.info/pools. If they decide to work together and perhaps DDOS a smaller third pool, they can have 51% pretty easily.
Inb4 'they wouldn't do that', the largest pool (ghash.io) is known to have attempted double spends before. Besides, surely one wouldn't argue that Bitcoin's security is a matter of confidence in a few pool operators being friendly?
Seeing as lots of that hashing power probably comes from ASICs and FPGAs, if anything it's more of a waste of Silicon that computing power, as without Bitcoin there probably isn't much use for all that hashing power...
This isn't the completely worthless observation it appears (like, for example, the snarky retort that Babbage's difference engine outpowers all the world's Bitcoin ASICs put together, on a decimal arithmetic benchmark). The Bitcoin ASIC network is a modest amount of equipment: in dollars, watts, or transistors, about the size of a one supercomputer from the TOP500. So this really underscores that general-purpose PCs are terrible at very simple & very parallel problems.
Adding a FPGA to the CPU/GPU mix could narrow the gap:
What a waste of time. I haven't given too much weight to criticism of Bitcoin's use of computing resources--there's enough low-hanging fruit--but this is actually somehow reprehensible.
Hardly a waste of time. You haven't seen this for what it truly is; a waste of time is your cash sitting in a clearinghouse for days before going overseas. a waste of time is any value transaction that takes more than minutes on a global scale. A waste of time is when the authorities freeze your accounts and cancel your cards. Cryptocurrency solves these problems and more.
Breaking news : a single caterpillar excavator has more excavating power than all the toy shovels in the world together.
I still find the approach of the article interesting, but what would be relevant is to compare the power consumption of the bitcoins miners compared to the power consumption of all the personal computers in the world.
I was thinking the other day, let's say someone operating the bitcoin network wanted to crack a specific AES256 code - couldn't they try to use the network to insert that into different blocks with different difficulties to try to get the network to crack it for them?
Or it doesn't work quite that way?
Would be interesting fiction (ala Cory Doctorow style) if say the NSA created a crypto-currency to make everyone else crack messages for them.
Doesn't work that way, bitcoin proof of work uses SHA256 twice, which is of limited value. It gives you a near collision to a double hash. It's burning energy for work.
One could make a case that the use of sha256 would drive down the cost of comodity ASICs, enabling a well resourced attacker to use mining hardware 'off-label' to find sha256 hash collisions. Same goes for scrypt, with the adoption of its use as a password store.
Well ... except by definition bitcoin is hashing at that rate AND computers are being used for other tasks, so they can't be using 100%. There are also billions of embedded systems that have computing power but can't be used for bitcoin mining. What about CPUs in phones and tablets?
If you define "world's computers" as "desktop PCs" and you exclude GPUs (which my son seems to use to heat his bedroom), then maybe this statement makes some sense ... but it's really a pretty weak milestone.
i almost broke out laughing after reading this :D THE ALIENS ARE MINING BITCOIN!
ok, now i see the ASIC story in the comments, that kinda helps this have some sense, provided we skip over the part of the text that says "if we estimate"...
63 comments
[ 4.6 ms ] story [ 113 ms ] threadNitpicks aside, though, the Bitcoin network's hashrate is mindblowing.
[EDIT: Corrected the numbers to match what you wrote. It is irrelevant, though -- if you get a gigahash for X watts with PCs, you'll use X watts to get X gigahashes with a 1W/gigahash ASIC.]
To put it another way, not going to 30kW of ASICs is equivalent to taking 30kW of PC mining hardware and turning some of it off (in a pre-ASIC world where PCs are the most efficient mining hardware available).
I don't think we'll get to a point where BTC has a significant footprint anytime soon.
If bitcoin powers the post-modern financial industry, the resource savings will be enormous.
They will all still be there, there will just also be lots of machines calculating SHA256 hashes
Perfectly secure systems exist that don't require this lunacy.
So really, we could make a better system, one that uses far less power, by just accepting that some things cannot be decentralized and designing a system accordingly.
The difference is that with Bitcoin, we are also pouring otherwise useful energy down the drain as part of the security model. The is purely because of a fantasy about an ideal kind of money that requires no central authority. If we were to drop that fantasy and accept the existence of authorities in the system, we could create a digital cash system that:
* Has the same kind of payment security properties as Bitcoin
* Actually allows for anonymous payments
* Supports offline transactions
* Requires many orders of magnitude less power than Bitcoin
Of course, as soon as you say "central authority," the Bitcoin crowd dismisses the idea entirely, regardless of its merits and regardless of the existence of concentrated authority in Bitcoin.
The main point stands though - without thw decentralisation fetish, this thrashing would be unnecessary.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.105...
In other words, you can set things up so that no single authority can fail in a way that would cause the entire system to fail. You can have N authorities such that a minimum of K are needed to keep the system working.
Try to hack a bank, or exploit credit cards, or even break in a vault, and you'll see what kind of external measures governments and private agencies put into protecting the system (hint: you'll get locked forever, or shot, or worse)
Now imagine you succeed, and somehow don't leave any exploitable trail, do you think a story will pop somewhere? I don't.
Centralized money systems are the reason today, what you make in a month working 16 hours a day in some country will buy you a drink in another. This allows rich countries to straight up buyout any poor country's production, and for trivial amounts. It leaves them in a state where people are starving, even though they produce enough to feed 10x their population.
Imagine a world where a bag of rice costs the same in every country. Yeah, currencies are awesome.
Edit: just to be clear, I'm not saying rich countries are evil(er), because this would not work without the cooperation of poor countries leadership.
20k buys you 40t of rice a year on the current marker. How many people do you need on an exploitation to produce 40t/year? 3? 5? 10?
There's your number. With a year's work (doing the exact same thing), you can buy several people's yearly production. The reason for that are currencies.
Without them things would equalize over time (it's happening in EU right now, though only internally). It's a long and bumpy process, but eventually it removes the ability from a country to drain another for next to nothing.
Edit: deleted post said there was evidence a uniform supply pricing was actually bad for poor countries
That is the wrong comparison. The right comparison is this: does Bitcoin require more electrical energy per transaction than VISA?
If the answer is not "yes" today, it will eventually be "yes." Here is a thought experiment:
Suppose that a new technology is developed that halves the energy required to process a VISA transaction and check for fraud. VISA has every incentive to adopt that technology, and thus reduce its power consumption.
Now suppose that a new technology halves the energy required to process Bitcoin transactions i.e. to compute hashes. Since miners profit by computing more blocks, their incentive is to adopt the new technology, compute hashes at twice the rate, and their net power consumption remains constant.
However Visa has no incentive to search formore efficient technologies where are you in bitcoin my name is constantly being developed to become more efficient and it isn't necessarily the number ofhashes you are getting but thenumber of hashes you get per watt of electricity used. so the incentive is there for minors always, this is not the case with visa
apologies for paul wording Ion phone
wat
edit: clarification, the total hash rate has zero bearing on the overall security. The question is about how hard it is to gain majority, in other words, how that hashing power is distributed.
At the time of writing, the two largest pools have 50% of the hashing power: https://blockchain.info/pools. If they decide to work together and perhaps DDOS a smaller third pool, they can have 51% pretty easily.
Inb4 'they wouldn't do that', the largest pool (ghash.io) is known to have attempted double spends before. Besides, surely one wouldn't argue that Bitcoin's security is a matter of confidence in a few pool operators being friendly?
Adding a FPGA to the CPU/GPU mix could narrow the gap:
https://news.ycombinator.com/item?id=1933192
Just... don't bother with me. I've been watching this since the first hype and you aren't anywhere near the first breathless comment I've read.
but this is ridiculous. Personal computers are 3 ghz * 2-4 cores * a billion devices.[1] It is huge!
That's why we can work at layers that are 10 layers above what we're actually doing. Such as a style sheet interpreted by a browser.
Try dropping down to C sometime. These machines are not toys.
[1] random lazy link: http://www.ask.com/question/how-many-computers-are-there-in-...
Or it doesn't work quite that way?
Would be interesting fiction (ala Cory Doctorow style) if say the NSA created a crypto-currency to make everyone else crack messages for them.
No, the work done is very specific. It's hard to imagine a usecase where it would be useful to take advantage of it.
If you define "world's computers" as "desktop PCs" and you exclude GPUs (which my son seems to use to heat his bedroom), then maybe this statement makes some sense ... but it's really a pretty weak milestone.
ok, now i see the ASIC story in the comments, that kinda helps this have some sense, provided we skip over the part of the text that says "if we estimate"...
but, i guess since i clicked, i loose.
how very scientific