Bitcoin Forum

This is from a Google Hangout with Daniel Larimer. He starts talking about the issues with Bitcoin mining and PoW at 1:01:20 and goes until 1:03:44. Only a few minutes long.

https://www.youtube.com/watch?v=tBcJ5rsAGaI

Seems to me like he's bringing up some pretty serious flaws here. He later goes on to say that PoW is dead and Bitcoin will have to change at some point. Any thoughts?

Thanks!

The blockchain maintains the list of all the transactions in bitcoin. It must be the same to everyone and yet not created by a single person or entity. Proof of Work is what enforces both of these features. The original vision was that anybody could be a miner and participate in securing the blockchain. However, as bitcoin becomes more popular and valuable, the mining rewards lead to the creation of mining facilities. Similarly to the gold rush, at the beginning there were pioneers with pickaxes but soon after they were replaced by conglomerates with heavy machinery. Now that mining pools have mostly taken over, he questions the decentralization of bitcoin mining and therefore bitcoin security is in the hands of a few. Furthermore, what is too costly for an individual may be acceptable for governmental entity. In conclusion, he prefers to have a group of trusted organizations in charge of the blockchain. Mining would be rendered useless and the cost of maintaining the blockchain greatly reduced.

I agree to most of what he said except for his conclusion and solution. I think finding a group of trusted entities is very hard and is the reason why previous digital currencies failed. POW mining is quite wasteful in computational resources but if it is what it takes to have an independent currency then I accept it.
There are other suggestions out there (POS, POS + POW). They may work better or not. I'm not convinced that anyone knows. In a sense this reminds me of physics. There are many theories out there but we can't verify any of them because we are nowhere close to that energy level. Today, there are a ton of different options but no one has put enough effort into breaking BTC. At this point, you'd need 100 PHash/s to attempt it. But who knows, maybe someone will build the super bitcoin collider.

I think the easy explanation is that he doesn't know as much about it as he thinks he does.

I think these are not major flaws or blunders. Even administration would be aware of them. So not an issue to worry about.

Daniel Larimer is a nobody.
He does not know what he is talking about.
Big "conglomerates" doing the bitcoin mining does NOT effect the security of the block chain.
Lets pretend only ONE person in the world is mining bitcoins.
Big deal.
Who cares?
There is nothing that one person can do to hurt bitcoin.
Everyone else using bitcoin has the block chain.
Easily over 1000 people have the block chain downloaded.  No I didn't bother to look that up, it is a safe guess.
This one remaining miner cannot change the record, or the records of future transactions.

Perhaps I am missing something, but why does anyone care if everyone STOPS mining bitcoin?
The Bitcoin QT software will still behave as a wallet and peer to peer transaction data will still be shared among everyone elses wallet, correct?

Or are miners acting as a control node, in addition to mining?

?? I was under the impression that mining confirms transactions and brings security to the network...unless I'm completely wrong lol?

And I wouldn't just say Daniel Larimer is a "nobody" he pretty much created one of the first DACs with Bitshares, and it's been doing pretty well lately. Wouldn't be surprised if it passed Litecoin at some point.

i would listen more to bitcoin developers rather than this nobody
bitcoin developers raises couple concerns about bitcoin, in short it needs more developers thats all

No, Bitcoin completely depends on mining to operate.  A miner in control could completely destroy Bitcoin if they wanted to.  if they had a chain that took more work they could release it any time and wipe out all the transactions and miner rewards during the period they mining in secret.

Correct! A miner that constantly as more than 50% of the hasing power can reverse transactions at will reaching back to the moment when he got that 50% + hashing power.

I feel that many ppl in this thread express their believes without really knowing much about D. Larimer and POW. Doesn't seem like a productive discussion.

On the issue: POW is costly, slow, tends to centralize (because of ASICSs and Pools) and requires trust (in the few pool operators).   

Here is more info on the system Daniel L. proposed http://wiki.bitshares.org/index.php/DPOS http://bitshares.org/delegated-proof-of-stake/
It is operating already in BitSharesX.

Please consider that the only miners who could generate a longer chain than the rest of the network would themselves need to invest nearly 3 quarters of 1 billion dollars just to have enough hardware to 'overstep consensus' that is provided from the current pools (currently - this figure will grow larger as more entrants turn on their mining hardware for the first time in efforts to mine for bitcoin).  Saying that miners could destroy it if they had a longer chain is not true - they would simply have their own blocks confirmed and would receive a larger portion of block rewards (rightfully so I might add - for they too are contributing mining power to the network)  If an ill-willed miner with a half billion dollars and a chip on his shoulder thinks its better to hijack and effectively freeze Bitcoin transactions - he'll be noticed quick and blacklisted by the legitimate mining traffic even quicker. 

I don't think this is true. If someone were to have 51% of the hashrate and somehow change the protocol then the rest of the network could just ignore that blockchain.

You only need to trust pool operators for a very short period of time because if the pool were to do something nefarious the miners could simply leave and join another pool.

In your first paragraph you mixed up a few things: The amount of hashing power a miner or a group of miners have has noting to do with changing the protocol. A 51% attack does not require changing the protocol. The other 49% can not ignore it if it is the longest chain unless they made a "social" agreement to ignore it -> Then is is not much different anymore from (D)POS, read about it (see the links above)!  

You are right with your second paragraph. But what I said was that someone who "constantly has more than 50% of the hashing power can reverse transactions at will" (corrected quite a few spelling mistakes form my original post :) ). If this is the case it doesnt matter what the other 49% of the network do. But this is basically the same with POW as with DPOS or any consensus network security model.

I didn't say they would destroy Bitcoin, I said they could.  How much they would have to spend would depend on a number of circumstances.   Most of the calculations you see are bogus because they assume the entity would buy miners from vendors.  At this point a 51% attacker would develop their own chips and produce their own hardware.  They could mine in secret and release the chain after a long period of time with a bunch of empty blocks and wipe out all the transactions and block rewards over that period.   

mining in secret = orphaned blocks = NO longer chain, I'm just sayin  :-\

Are you denying the possibility of a 51% attack? :)

That equation is not true. When someone mines a longer chain in secret and publishes it later the until then longest public chain including all its block will get orphaned because it is not the longest chain anymore.

But that all was not the point of the OP...

Let's also not forget that ASICs can easily be taxed by governments which guarantees law enforcement gets their cut of the mining profits. Special purpose chips can only do one thing so taxing it doesn't disrupt the economy any more than taxing tobacco.
CPU mining was general purpose so politicians couldn't put a tax on Intel or AMD chips.

How can ASICs easily be taxed? I think it would actually be quite difficult without raising taxes on computer equipment. The only way that I can see ASICs being taxed is via sales tax, which is generally not outrageous.

ASICs aren't general purpose computer equipment. It's highly specialized single purpose equipment. The only thing a Bitcoin ASIC can be used for is to mine Bitcoins. So who would complain if it were taxed? And how will you mine if your government puts a 10% tax?

You'll pay the sales tax.

Still haven't seen a decent response to this problem...

There's very rarely anything without flaws, everything is more or less in a constant state of ebb and flow, sometimes getting better, sometimes getting worse.

It may be completely unfounded but for a variety of reasons I am quite confident bitcoin can grow from strength to strength as there are seemingly enough evolutionary drivers associated with the cryptocurrency.

Just sayin'.

 :)

The problem is not stated completely correctly since mining pools do not have as much control as he states.  They depend on the pool members who can switch instantly so I think that part of the problem is overstated.  Beyond that he makes some good points but the system is still experimental  and it is unclear to me how it will pan out.  Several people have claimed the current system is unsustainable over the long term.  The problem is that the solutions proposed are mostly centralized models and that eliminates the main point of Bitcoin.  Right now we are in the gold rush era since the block rewards are so large.  In just over 100,000 blocks the reward will be halved again and there is pressure to keep the transaction fees low so I think that will create a big upheaval in the mining ecosystem.  I don't think the current rate of increase can be sustained past the next halving.

Note that Dan Kamisky went through this same thing before and claimed POW would be over by the end of last year.  His issue was that Avalon asics had a huge percentage of the hashing power because they designed the first working chips.  Of course they had no incentive to disrupt the Bitcoin ecosystem so they sold the miners instead of taking over the system.

The bottom line is that it is experimental and people who claim to know what will happen or claim something "dead" is hyperbole.

One thing Daniel Larimer mentions which should be of concern to everyone who supports Bitcoin:  The average cost per transaction.

This obviously needs to be low because the intent is to be below the current trusted 3rd party financial institute's fees.  I base this on the following from Nakamoto's paper:

"Completely non-reversible transactions are not really possible, since financial institutions cannot avoid mediating disputes. The cost of mediation increases transaction costs, limiting the minimum practical transaction size and cutting off the possibility for small casual transactions, and there is a broader cost in the loss of ability to make non-reversible payments for nonreversible services."

Larimer suggests that eventually the transaction fees will be around the $8 mark. This got me thinking along the lines of what the cost to process a transaction is today.

So what is the cost of adding a transaction into the blockchain?  NOTE: I am not talking transaction fees, I am talking the actual cost of processing the transaction based on the cost of computational power. Since the transactions are processed by miners, I would suggest the answer to is the cost of the power to mine divided by the number of transactions as a starting point.  Note, this is a starting point, it does not include hardware required, services, bandwidth, etc. which would need to be added in to get the total cost before dividing the number of transactions (ie: only making it worse). Disclaimer:  Please understand, I am not an economist and neither do I have a degree in finance.

Total number of transactions for a year (from Blockchain.info): 45,795,557
Estimated cost to power Bitcoin mining (from Haas McCook): $780,000,000

I get $17 per transaction.  This is a seriously shocking figure.  Even if McCook's estimate of power used in Bitcoin mining is twice as much as he suggests it is, the cost is still $8 to process a transaction.  Miner revenue for this same period is $834,649,130.70

With an increasing hash rate, though thankfully more power efficient miners, this is likely to get worse.  With mining equipment having a ROI life of say 6 to 9 months, mining gear could be purchased up to twice a year in order to maintain revenue.  This should really be the concern of the Proof-of-Work mechanism used.  Competition is increasing the cost versus decreasing it like you would expect in other industries.  Food for thought.

Please find McCook's paper here:
https://cdn.panteracapital.com/wp-content/uploads/The-Relative-Sustainability-of-the-Bitcoin-Network.pdf

Blockchain.info URL:
https://blockchain.info/charts/n-transactions-total?timespan=1year&showDataPoints=false&daysAverageString=1&show_header=true&scale=0&address=

CK

Donations towards my thesis welcome:  1ADEzUG7HVQkq3mu5g85KkupAAbzRTppyd

Here is the rate of transactions which has not gone up significantly in 2 years
https://blockchain.info/charts/n-transactions?timespan=2year&showDataPoints=false&daysAverageString=1&show_header=true&scale=0&address=

It is reaching a record of staying above 50,000 per day for 2 months continuously but it is not some huge increase.

Agree 100%. I calculated around 30$ per transaction at the current level of POW security. https://bitcointalk.org/index.php?topic=395761.msg6852307#msg6852307
With BitsharesX it is $1 cent per transaction while providing more security per time passed since the tx was broadcasted.
...Someone has to pay the costs, either the users ($8-$40 USD per tx; ~ $40 was calculated by Tim Swanson) or the coin holders (as it is today, ~10% dilution = 10 % increase in money supply per year with Bitcoin atm).
Those are bare facts.

Block #318997
Summary
Number Of Transactions    118
Output Total   2,390.60927025 BTC
Estimated Transaction Volume   1,315.50637339 BTC
Transaction Fees   0.01761036 BTC

Why talk about the transaction cost without considering the transaction fees vs the block reward?
Today miners get 25 BTC from the reward and 0.01761036 from the fees. Basically, transaction costs are negligible. The idea is that as the rewards go lower, the total fees offset the loss. Obviously, it is not with 120 x 0.0001 btc that you get there. Very roughly, we need 250 000 transactions per block.

So the main question is - How do we get bitcoin to support this type of volume and more importantly how do we get bitcoin to these kind of volumes?

my calculation is based on the assumption that nothing else changes.

Why should tx volume / mass adoption catch on just because the block rewards is halving? :) 

I disagree with your conclusion. Today the majority of the revenue the miners received is from block subsidies. These block subsidies allow for miners to spend more money on electricity then they earn in TX fees alone. Block subsidies are in place to give miners an incentive to keep the network secure while bitcoin grows and does not have enough transaction volume to support the cost of securing the network. The theory is that over time, the number of TX per second will grow to an amount high enough that TX fees are able to pay to keep the network secure.

You are welcome to disagree and I encourage debate. Just from your answer I am not sure which point of my conclusion you are disagreeing with.

I conclude that the average cost to protect the Bitcoin network is nothing lower than $17 and this is going to get worse as more POW is added to the network. The average increase since inception has been 18% per transition.  Competition is meant to make things cheaper not more expensive but this is not what we see in POW.

Cost to power Bitcoin mining / number of transactions that are being processed by Bitcoin mining = cost to protect Bitcoin Network

And again, not including full nodes, mining hardware, operational costs of miners other than power, the backbone Matt Corallo has implemented, or any fast payment services available.

Subsidy fees to miners are there for 2 reasons - compensate miners, which you pointed out and introduce/mint new coins which you didn't mention.

Doing this calculation today ignores the award. Miners work to get it, they don't care about our transactions. You say 18$ because you divide the mining cost by the transaction count but they would be doing the same even if the blocks were empty.
Would it mean that the transaction cost is infinite?
We'll have a clearer picture when all the bitcoin are created. Then, we'll know the transaction cost. If it's greater than 5 cents, bitcoin is not viable for microtransactions. Until then, it's an upperbound estimation with a significant error margin.

You are 100% correct, the calculation does ignore the reward. Reward has nothing to do with calculating cost. I did mention the reward total in my post, though it is not relevant in calculating the cost, but it is the incentive that drives the cost up, and one of the major points in my comment.  POW model increases the costs of securing the transactions because it increases competition for the reward.

You are correct, they would be doing the same if it was empty.  The computing power that powers mining is incredibly high, with very few transactions in relation to the computing power to process/safeguard the transaction at hand. This can only be attributed to the reward for solving a block as the transaction fees are negligible today, again pointing to the reward being the reason.

If there are no transactions, then there can be no cost to secure the transactions, so then my formula would be:

Cost to power Bitcoin mining / number of transactions that are being processed by Bitcoin mining (where number of transactions are greater or equal to one) = cost to protect Bitcoin transactions

In reality, there are transactions, so one is able to do the calculation.

With regards to the "until then, it's an upperbound estimation with a significant error margin."  Can you suggest another method of calculating what the cost of securing a transaction is today?  This was my question I attempted to answer. I appreciate you taking the time to reply with a critisism but you don't offer an alternative other than wait and see.  Again, the question is today, not the unknown future.

I do understand that once the incentive is gone, miners will only be claiming transaction fees, and that these fees would need to cover the cost of the miners or there would be no incentive to mine, it clearly states this on page 4 of Nakamoto's paper, and indeed we will be able to calculate the cost of processing a transaction, but that will also have nothing to do with income, rather just the cost of the "CPU time and electricity" as Nakamoto puts it.

Using the same method I used previously, using AntMiners, so this includes the cost of hardware and power consumption and cooling, I get about $7 per transaction.

Cost of power (including cooling)+ Cost of Hardware (S3+ and PSU)/ Transaction volume (must be 1 or greater) = Cost to secure each transaction.

This assumes:
everyone is running efficient hardware at $0.08 per kWh.
extra 1/3 of power cost is added to account for cooling
conversion rates are as of the time of this post


That is mining at its most economical but still accessible to the average person (subjective)


This also answers whether there is money to be made in mining.  If total income of miners is $834,649,130.70 (using yesterdays posted amount), and the expenses are calculated as per the above ($317,834,173.10).  There is a huge profit in mining.  The question is who is getting it?

People who generate their own electricity, who don't need to pay for cooling and who have invested in and run efficient hardware.

I really don't agree with the stated method as a measure of "transaction cost", since the "cost" to miners must be less than 0 in order to incentivize mining. If it were not, there would be only hobbyist miners, and we would not be seeing network hash rate growth currently. Maybe there's some cost for the solo miner using Antminers and spending more in electricity than the block reward, but they are mining with unprofitable hardware, inflating this measure arbitrarily. Those people don't increase the costs of transacting in Bitcoin at all. In fact, we could likely run the entire block chain on a single node today, if we could only find a trusted central authority.... /sarc. Anyway, "mining operating expenses" / # of network transactions is a poor measure of the cost to transact in bitcoin.

The true "cost" of transactions for comparison should be simply the transaction fees / amount of BTC transacted.

None of this even matters until the mining subsidy is removed in 100 or so years; at that point if transaction fee value is greater than the cost of operating a mining facility, mining will continue to grow until the cost of mining is no longer below the transaction fee reward (or more likely until an overshoot and correction), at which point the network will eventually reach equilibrium. The cost per transaction will remain roughly the same, since the network will self-adjust based on profitability of mining. This scenario would likely require a much higher purchasing power for BTC as well as a higher volume of transactions on the network.

But even if the subsidy were removed today, what would likely happen is people switch off hardware, increasing transaction fee payouts to the remaining players until equilibrium is reached in terms of transaction fees = costs. Either way, equilibrium is roughly reached, at which point the amount of BTC gained from transaction fees would have to be very close to the overhead cost of maintaining the network. Since transaction fees ultimately = cost, my suggested method above is a much more accurate way of stating the cost of transacting in bitcoin

TLDR - cost per transaction is actually almost nothing, still, and the $30, $17, etc. numbers are based on misleading definitions and incorrect assumptions.

Right. It's quite simple actually. The price is determined by the equilibrium between suppliers of the service and the people paying for it. Today, everybody is happy to settle on 0.0001 BTC per transaction (excluding large transactions). It's as if there is a massive subsidy on bitcoin transactions. So much that it hides the real cost. When it goes away, we'll settle at a different point. I can't tell you where. We can look at where the reward comes from though. If 25 BTC are created, they dilute the value of the previous coins. But also there, people factored that in. They already consider the final number of coins in their valuation of bitcoin.
So basically, the transaction cost is the transaction fee because we are paying for the rest by accepting the scheduled inflation.

You have got to be kidding me. You lost me on this part.

...but that would decrease the security of the network as POW security depends on the amount of hashing power available to the network.

What you say makes total sense.  The cost is subsidised by the reward.

I do appreciate this reply, it gave me something to think about.

there is no problem. its the same uninformed people making videos and the noobies get scared lol. just stop worrying.

That's right. It's why I said that Bitcoin needs to solve the transaction limit problem (1 MB) and get massive adoption so that while fees remain low, their total value is enough incentive to continue mining. If fees are too high, people will not use it. If total value is too low, miner will quit. We need both for a secure system. Fortunately, smart people saw that it takes time and built the schedule over a hundred years.

Bitcoin is perfect, his argument is henceforth invalid.