Even if scaled up to global usage, I very much doubt Bitcoin mining would be worse.
As computers get better, traditional applications, like databases, perform better and you can either serve more customers, use cheaper (and often more efficient hardware), or do nothing and use your outdated hardware.
With bitcoin, your enemy is not the performance of the raw work you are doing (you can hash better), but the fact that everyone else also has these increases. Because of that, the limiting factor is not performance, but cost, and so over time the average cost of the bitcoin network will not go down like the cost of centralized systems do.
Let's do a thought experiment: Let's imagine that instantly all computers get 100 times faster but it costs 10 times the energy to run these faster machines (not actually that unrealistic) and the cost becomes 10 times cheaper.
Traditional centralized banks can simply use a 10th of their existing hardware to have no energy-cost increase but a 10 times increase in customers they can serve. Sure, it's a bureaucracy, so it'll roll out slowly, but they're saving money too so maybe a little faster.
Bitcoin, everyone will simply buy 10 times as much hardware since it's cheaper and they want to get a bigger bit of the pie (mining reward) before everyone else manages to buy all their hardware. You now have an equilibrium in terms of equipment cost reached again, but suddenly the hash-rate is 1000 times more and the energy consumption is 100 times more.
The above might sound a little too theoretical, but if you look at the increase in hash rate and estimates for power usage of cpus, then gpus, and then asics, you'll see that in general power consumption for bitcoin constantly increases.
You're arguing that the existing cost of banks / the existing system is high because it's legacy. Your argument doesn't say "and it's constantly, and provably, going to waste more energy as hardware gets more cost-efficient". Even if you want to claim that banks are currently less efficient (which I doubt they are), I hope the above convinces you that bitcoin over time will constantly use more energy thus resulting in an obviously more wasteful result, while banks and other centralized systems can take advantage of improvements and thus become more cost and power efficient over time.
Not really. Eventually electricity will dominate mining costs, not hardware, and people will only invest in more efficient hardware, not cheaper hardware. The total cost of the electricity used in mining will never exceed the total value of the block rewards plus transaction fees, since miners would rather stop mining than mine unprofitably.
Whether the total electricity used increases or decreases over time depends on whether the increase in price of bitcoins outpaces the block reward halving every 4 years, or transaction fees become a significant source of miner income (https://en.bitcoin.it/wiki/Funding_network_security). A single bitcoin would need to be worth $1 million in 50 years to keep pace, for example.
But I think it's more likely we'll transition to better (cheaper, more secure) mining system before then anyway, which is entirely possible to do without having to redistribute bitcoins (it would essentially be a fork that the "economic majority" agrees with: https://en.bitcoin.it/wiki/Economic_majority)
If you have 300 BTC and you suspect someone else is almost able to mount a 51% attack and thus devalue all your BTC, it makes logical sense for you to put more money into mining, even more than you can gain from the combination of block rewards and electricity, to secure your existing investment. As you do that, others will be forced to as well to avoid you having too much power and to protect their BTC and so on....
That does not tell the whole story though, power usage per gigahash is a more accurate measure, because the higher the GH the more secure the network. From CPU, to GPU to ASIC the power use per gighash has fallen significantly.
If you really want to compare it to the banking sector then you need to include the full and real costs of that banking system, power alone is not the only cost. One bail out, as seen in the last few years, dwarfs the power cost of bitcoin. That is without considering the knock on effects of the banking meltdown on the overall economy.
And bitcoin can take advantage of improvements to reduce power consumption of ASICs (as there is no next step tech that will provide an increase in gigahash as ASICs did, so now it goes to refining the power usage. Additionally larger farms are looking at renewable alternatives and uses for the excess heat being produced. It is still a very young sector so there is a lot of working out to do, but it is being done.
There is no fiat energy.
On the converse the cost of the bailout is pegged to the price of money. Fiat lowers that cost via inflation. The surplus currency created by the bailout is reflected in low interest rates.
The key is that money is more fungible than energy. Buying energy and converting it into bitcoin is likely to be inefficient when the market reaches equilibrium. As a value store, nobody wants rubles when they try to cash out.
There's some efficient amount of money to spend on security, and markets will generally aim for that amount, even if they don't necessarily hit it exactly. In the version of Bitcoin where the mining reward is negligible and miners are mostly paid with transaction fees, it should happen there as well. But the mining reward is currently substantial, which greatly over-allocates resources to security.
Add that in to the fact that Bitcoin mining has huge externalities in many parts of the world, due to the fact that electricity is often produced by burning coal, and coal externalities are poorly accounted for, and you have a big problem. Or rather, you would have a big problem if Bitcoin were big enough to matter.
I would say the upper bound on the amount of energy someone will spend on BTC is roughly the cost of (prestige of mounting a 51% attack) + (BTC market cap of one individual). If people allocate resources to BTC efficiently then one rogue person will just mount an attack and take over the network for the combination of prestige and fortune. Any person with large BTC wealth will want to have more security to protect his investment appropriately, and your security against said attacks is buying more hashing power. The fact that over time hardware becomes cheaper just means it's a constant losing battle.
Other markets that allocate money to security I think can reach an efficient amount because the system they're protecting does not, as a core part of it, require you to have more processing power than your user's / people you transact with. The two security models are so drastically different I don't think a comparison can be made. Perhaps the closest comparison is password hashing where it's good practice to increase difficulty over time to slow down attacks, but with 2FA and lockouts after N attempts, you actually don't need to allocate significant resources there.
And changing over a course of 10 years (there's an incentive to constantly upgrade hardware anyway) is probably still way faster than the banking system would change on a global scale if it was running into a problem.
While this is a different story because hardware and the PoW algorithm is involved, Bitcoin has a story of "breaking" changes. There will be more with the increase of the block size, etc. So I doubt this would be aniticipated way before it made it's way into the protocol, so the affected parties could prepare as well.
Your thought experiment misses an important point: the cost of energy. AFAIK that's already dominating hardware costs for miners.
That means: since the cost of bitcoin mining will tend to scale up (or down) linearly with the total value of bitcoin, so will energy usage (barring changes in energy price). It will not keep going up when the bitcoin value has reached a plateau (for whatever reason).
Dollar price of electricity doesn't matter. You can't magically print out more coal by signing a document, nor you can clear your "kWh used" counter by going to war with someone else. We have something like 50 years to cut down on our energy use and switch to better sources if we want to have a technological civilization alive on this planet; setting up our economy to run on an exponentially-growing resource waster might not be the smartest idea now.
Physical security also scales well. The volume of secure space you have increases to the 3rd power with only an approximately 2nd order increase in the costs of creating it (your basic area/space tradeoff working for you in this case).
The argument that payroll doesn't scale is also absurd: who are all these Bitcoin entrepreneurs trying to run exchanges, if not payrolled staff?
Finally and most importantly: the difficulty factor of Bitcoin does not reduce the computational power required to mine it. It is the exact mechanism of the provably increasing cost: the difficulty goes up, so more and more power is required to preserve the status quo.
Any changes in energy cost are thus ignorable, since in a Bitcoin-only world they're constantly eaten up by changes in difficulty factor, whereas in the non-Bitcoin world they become real-wealth for everyone.
No. the difficulty goes up so more hashing power is required, whether this means more actual power usage will depend on the machines being used and their efficiency.
Except... The difficulty also goes down.
But don't let facts get in the way of your flaming argument...
A decline in hashing power can only happen from people looking at real energy consumption and saying "to hell with this". At which point the network compensates to keep hashing difficulty the same, but does not reduce energy consumption for any of the remaining participants.
But correspondingly, the mean difficulty does go up continuously - that's the fixed number of coins. So even when the block to block difficulty is adjusted, the net trend is less coins per unit hash, which means the marginal energy cost of the coins approaches infinity.
Miners optimise for USD$/kWh, not for coins/hash. The latter is entirely virtual and largely unrelated to the network power consumption.
which means the marginal energy cost of the coins approaches infinity
You seem to ignore that Miners, too, have to pay for their energy.
So, in your model, either the market price of the coins must also approach infinity, or the miners will just stop mining.
One of the first things you realize when you try to setup payment processing is that middlemen would not actually exist in the payment processing industry if it was an efficient market. But these companies do exist because the entire industry structure has developed to pass risk from credit card companies to intermediary businesses whose sole function is basically to manage risk and enforce policy compliance. It is a bit like if Google didn't give email addresses to individuals because it couldn't be bothered policing spam, so you got your email through a third party.
Anyway, even if we ignore the deadweight losses imposed on the economic system by all of the businesses which cannot get credit accounts (merchant accounts), the costs of supporting this credit-centric system are enormous. PCI compliance costs alone are staggering, and serve no purpose except to lower the risk held by credit card companies. And even with this, the inefficiency of central risk management is so great that VISA doesn't even bother to investigate fraudulent purchases of less than 500 USD. And why should they when costs can simply be pushed back to merchants and turned into a profit center by issuing chargebacks.
And this is where bitcoin gets interesting -- by inverting the risk structure and letting anyone accept payments without the need for a payment processor to issue credit, Bitcoin eliminates entire horizontal swathes of the payments industry, while placing competitive pressure on legacy payment providers. If the technology scales to the point you are concerned about, it will decimate the number of people required to be employed managing risk in everyday transactions.
Put more succinctly, you need to shift from thinking about the energy costs of bitcoin as a wasted ongoing expense and start considering them the price of eliminating an otherwise intractable trust problem in finance. It is like paying for a standing firefighting force. Expensive in one way, but a lot cheaper than the alternative.
Centralized systems could be more efficient if their incentives were in line with the incentives of the people they serve. But their incentives are not and they are not fair.
You are theorizing about something based on how it feels without really looking into the reality of it. Bitcoin mining is so many orders of magnitude smaller than even fragments of current financial infrastructure it would laughable to talk about, even if electricity usage was any sort of a major issue, which it is not.
... and exponentially worse as time goes on (by design).
Only for people who don't understand the design.
Google "bitcoin difficulty".
> Only for people who don't understand the design.
Personal attacks all the way because your pet golden hammer was attacked. Always the same result.
Neither is Bitcoin my pet golden hammer (I don't even know that metaphor), nor is my statement a personal attack.
It's merely pointing out that your claim is so fundamentally wrong that you can't have a good understanding of the design.
If you don't want to learn the protocol then maybe think about it this way: Energy consumption is not an exponential factor in bitcoin simply because miners have to pay for their energy, too. And the number of miners does not grow exponentially.
Your premise is wrong. Bitcoin has a mechanism called "difficulty". It's very much at the core of the protocol, google it and ponder how it affects your argument.
Furthermore, the current economics of mining have a fixed expiration date set to 2140, that's roughly when the last coin would be mined.
The protocol will need to be changed fundamentally long before that date. Since we don't know what that change will look like, we also don't know how it will affect the network's power consumption.
Any "thought experiments" that naively extrapolate from the current power consumption are as pointless as trying to guess your car's mileage based on its fuel consumption during ignition.
The protocol is still in the bootstrap phase. If it prevails then the bitcoin in 10 years will be very different from the bitcoin of today.
Bitcoin mining will always be a race on who can spend the most energy.
No. It is a race on who can spend their energy the most efficiently. (in terms of hashes per kWh).
That's why old mining methods (CPU miner, GPU miner...) are obsoleted by new ones (ASICs).
You can pump a Gigawatt into an old GPU mining rig and it will only cost you a lot of money.
Pump the same Gigawatt into a modern ASIC rig and it might make you money.
the hash rate doesn't grow on trees.
why the hell would you doubt it would be worse? it does 7 transactions per second.
http://www.forbes.com/sites/timworstall/2013/12/03/fascinati...
here's on estimate: Bitcoin Mining Uses $15 Million's Worth Of Electricity Every Day
Those stats assumed CPU/GPU and FPGA, not ASIC, miners.
For instance, the 5-to-500-gigahash-per-second rigs would have to be running at over 300k watts of electricity using those statistics. :)
So in reality, the network is much more efficient -- the high-end miners are using more than 100x less electricity (which would still be 3000 watts for high-end rigs per those numbers), and probably closer to 1000x less, than those estimates.
GH/s/w: 4.6 // http://www.coindesk.com/a-look-inside-kncminer/
watts to provite current hash rate: 300,000,000/4.6 = 65,217,391.30 w
kw to provide 1s: 65,217.39
an hour of that: 65,217.39*3600
234,782,604 kwh
10c per kwh => $23,478,260.4 per hour
24 hours => $563,478,249.60
my own figures are even more wacky so I would welcome some better statistics.
65,217kW is the energy usage. Providing that for an hour takes 65,217kWh - not multiplied by anything.
So that's $6,521 per hour, $156,521 per day.
ps it wasn't obvious at all, that's why i wrote "even more wacky" as mark of how solid i thought the figures were.
3600*7 = 25200 transactions for $6521.
they'd really want to see about increasing the block size.
Cost per transaction would go down drastically with increased usage.