Funding of network security with infinite block sizes

[Mike Hearn]

Oh goodie, more Google conspiracy theories. Actually I never had to ask for approval to use 20% time on Bitcoin. That's the whole point of the policy - as long as there's some justifiable connection to the business, you can do more or less whatever you want with it and managers can't tell you not to unless it's clearly abusive. That's how we ensure it's usable for radical (i.e. unexpected) innovation.

But even if I was being paid to work on Bitcoin full time by Google, the idea that I'd want Bitcoin to grow and scale up as part of some diabolical corporate master plan is stupid. Occam's Razor, people! The simplest explanation for why I have worked so hard on Bitcoin scalability is that I want it to succeed, according to the original vision laid down by Satoshi. Which did not include arbitrary and pointless limits on its traffic levels.

The idea that Bitcoin can be a store of value with a 1mb block size limit seems like nonsense to me. That's reversing cause and effect. Bitcoin gained value because it was useful, it didn't gain use because it had value - that can't be the case because it started out with a value of zero. So if Bitcoin is deliberately crippled so most people can't use it, it will also cease to have much (if any) value. You can't have one without the other. The best way to ensure Bitcoin is a solid store of value is to ensure it's widely accepted and used on an every day basis.

If Bitcoin was banned in a country then I think it's obvious its value would be close to zero. This is one of the most widely held misconceptions about Bitcoin, that it's somehow immune to state action. A currency is a classic example of network effects, the more people that use it, the more useful it becomes but it goes without saying that you have to actually know other people are using it to be able to use it yourself. If there was immediate and swift punishment of anyone who advertised acceptance of coins or interacted with an exchange, you would find it very hard to trade and coins would be useless/valueless in that jurisdiction.

The reason I'm getting tired of these debates is that I've come to agree with Gavin - there's an agenda at work and the arguments are a result of people working backwards from the conclusion they want to try and find rationales to support it.

Every single serious point made has been dealt with by now. Let's recap:

• Scalability leads to "centralization". It's impossible to engage in meaningful debate with people like Peter on this because they refuse to get concrete and talk specific numbers for what they'd deem acceptable. But we now know that with simple optimisations that have been prototyped or implemented today, Bitcoin nodes can handle far more traffic than the worlds largest card networks on one single computer, what's more, a computer so ordinary that our very own gmaxwell has several of them in his house. This is amazing - all kinds of individuals can, on their own, afford to run full nodes without any kind of business subsidisation at all, including bandwidth. And it'll be even cheaper tomorrow.
• Mining can't be anonymous if blocks are large. Firstly, as I already pointed out, if mining is illegal in one place then it'll just migrate to other parts of the world, and if it's illegal everywhere then it's game over and Bitcoin is valueless anyway, so at that point nobody cares anymore. But secondly, this argument is again impossible to really grapple with because it's based on an unsupported axiom: that onion networks can't scale. Nobody has shown this. Nobody has even attempted to show this. Once again, it's an argument reached by working backwards from a desired conclusion.
• Mining is a public good and without artificial scarcity it won't get funded. This is a good argument but I've shown how alternative funding can be arranged via assurance contracts, with a concrete proposal and examples in the real world of public goods that get funded this way. It'll be years before we get to try this out (unless the value of Bitcoin falls a lot), but so far I haven't seen any serious rebuttals to this argument. The only ones that exist are of the form, "we don't have absolute certainty this will work, so let's not try". But it's not a good point because we have no certainty the proposed alternatives will work either, so they aren't better than what I've proposed.

Are there any others? The amount of time spent addressing all these arguments has been astronomical and at some point, it's got to be enough. If you want to continue to argue for artificial scaling limits, you need to get concrete and provide real numbers and real calculations supporting that position. Otherwise you're just peddling vague fears, uncertainties and doubts.

[1714937962]

1714937962

[Peter Todd]

Accepting this as true, for which you present a strong case in your thread on trusted banks, then this is indeed a complementary service which may well attract a significant user base in the future. It may even succeed in handling 90% of the transaction volume which would otherwise hit the main blockchain. Is that your optimistic scenario?

However, and correct me if I am wrong, but such a trusted banking service does not exist yet. Not even in a prototype form, let alone one that can rapidly substitute for blockchain transactions. Acceptance of new services like this will take some time, at least a few years, surely. The people on this forum are ahead of the masses on bitcoin usage, yet they universally appreciate that their holding is stored on thousands of nodes worldwide. How many of us would quickly and permanently move our bitcoin holding to one single service instead of having it stored directly on the main chain?

We already have off-chain transactions, for instance Mt. Gox, and AurumXchange's code systems. Similarly web wallets allow for transfers from one user to another directly; MPOE reports that quite large amounts of BTC are exchanged between users every day. The Silk Road is another example of off-chain transactions - deposits go into a big shared wallet and transfers within the system are totally off-chain. In that case there are significant privacy advantages, not to mention how it makes implementing escrow easy.

There isn't going to be a single service that does this, that's my whole point: if you achieve scalability by just raising the blocksize, you wind up with all your trust is in a tiny number of validating nodes and mining pools. If you achieve scalability through off-chain transactions, you will have many choices and options. I'm sure the users of the Silk Road have very different ideas about who they can trust than users of BitPay...

You use the word "trust", but it takes time to earn it. It has taken Bitcoin four years to earn the trust that is fueling its success today.
I argue that there is not enough time left for that level of trust to be earned by complementary services before the 1MB arbitrary constant becomes as effective as any ddos attack in the history of bitcoin.

Please consider this chart and let us know, in your considered opinion, whether trusted banks will be fully ready, with a proven track record, before its blue line reaches 345,600.

https://blockchain.info/charts/n-transactions?showDataPoints=false&show_header=true&daysAverageString=7&timespan=&scale=1&address=

Off-chain transactions aren't something that will be implemented in one big go. Like I said above, the markets can naturally adjust bit by bit gradually moving uses of Bitcoin from on-chain to off-chain as the fees gradually increase. At worst growth in the least important, lowest value, parts of the Bitcoin economy is slowed while off-chain tx solutions catch up.

On the other hand, if the blocksize is raised and it leads to centralization, Bitcoin as a decentralized currency will be destroyed forever.

It might not be sexy and exciting, but like it or not leaving the 1MB limit in place for the foreseeable future is the sane, sober and conservative approach. Exactly what I want from the developers of the brand-new and still poorly understood technology underpinning Bitcoin's $1.5billion market cap. For that matter, I personally have a good chunk of my wealth tied up in Bitcoins and want them to be valuable in the long run.

You know, I work in engineering at a company filled with ex-aerospace guys, guys who are careful and try not to get people killed by things they don't understand. They all find my work on Bitcoin interesting, and a big part of that is because of how crazy high-stakes it is. If anything Bitcoin is kinda like being given alien technology that happens to be able to make a plane that never has to land, and unfortunately doesn't even even seem to be able to land. It's brilliant at moving passengers around, although the current plane can only fit 1000 of them. It's also the only one we have so if you want to change anything you have to do a bit of wing walking and be careful not to get any body parts near the props. Oh, and not to mention, it's currently raining cats and dogs, well, actually mostly dead puppies...

Now, under those circumstances do you think I'd go up to my boss and say "Gee, I dunno, how about I climb out that access hatch at the back and add some sheet metal until the plane is long enough to shove 200,000 passengers in there?" Fuck no. For one thing this plane of ours damn near fell out of the sky a month ago, and it only took 700 passengers to make that happen. We also don't really understand why it flies at all; we've all got our own theories and some of the theories are probably even right, but we've never flown this plane in serious turbulence, let alone been attacked by those monocled guys in airships off in the distance. Heck, we got worried enough when one of them sent us a nasty telegram the other day about how shabby our records were or something.

No, instead I'd be told to work on some prototypes and write some papers and see if maybe I could make some small nimble plane that could fly along-side our hulking monolith. Preferably a plane that can land for maintenance.

[caveden]

Great post Mike! You should create a topic with it in the OP, as it deserves more than to be buried here in the 7th page of this topic.

If Bitcoin was banned in a country then I think it's obvious its value would be close to zero.

Hum, are you sure about that? I'd like to know what happened to the market price of gold in US during the ban of the 30s...
If Bitcoin is aggressively banned everywhere, then I'd agree with you. But if it's still allowed in a significant number of places, I don't think so.

[Nubarius]

[...]
The idea that Bitcoin can be a store of value with a 1mb block size limit seems like nonsense to me. That's reversing cause and effect. Bitcoin gained value because it was useful, it didn't gain use because it had value - that can't be the case because it started out with a value of zero. So if Bitcoin is deliberately crippled so most people can't use it, it will also cease to have much (if any) value. You can't have one without the other. The best way to ensure Bitcoin is a solid store of value is to ensure it's widely accepted and used on an every day basis.
[...]

I fully agree with this.

[...]
On the other hand, if the blocksize is raised and it leads to centralization, Bitcoin as a decentralized currency will be destroyed forever.
[...]

Peter, I've struggled to try to understand you concerns but I still find your arguments convoluted and hard to follow. You want it to be possible to run a Bitcoin node on average hardware, but you don't want the people with average hardware to be able to use the network. How can that work? It's obvious that a blockchain recording all transactions is bound to be resource-intensive, but as far as I can see that's the very nature of a proof-of-work blockchain. Satoshi's idea was that such a resource-intensive system could be viable in the 21st century and he always presented Bitcoin as an accessible payment system, not as some sort of infrastructure for large financial services. And that idea is also what practically all of Bitcoin enthusiasts have been promoting during the last three years. I think I'm not the only one who's been telling his friends about how easy it is to get a Bitcoin wallet, and make payments to anyone around the world, and how this system could eventually be used to pay for a restaurant bill or to buy a book online. This is probably the idea that first attracted most of us to Bitcoin. If you think that such a thing is not possible and that Bitcoin can only survive as a low-volume network where payments are outrageosly expensive, well then that's like saying that you don't really believe in the original Bitcoin idea. You could be right, who knows, but keeping the block size limit is what effectively kills the Bitcoin dream, either by turning it into a specialist service used by a minority or, more likely, condemning it to irrelevance and oblivion. We need to find a compromise, and I think the sensible thing would be to agree to a higher block size limit. Those, like myself, who would remove the limit altogether will find a much higher limit more reasonable whereas those who are adamant that the limit is necessary could accept a value like 50 MB, and avoid crippling the system during the next few years.

[Mike Hearn]

I don't agree with the idea that keeping the 1mb limit is conservative. It's actually a highly radical position. The original vision for Bitcoin was that it supports everyone who wants to use it, that's why the very first discussion of it ever was about scalability and Satoshi answered back with calculations based on VISA traffic levels. The only reason the limit is there at all was to avoid anyone mining huge blocks "before the community was ready for it", in his words, so it was only meant to be temporary.

Peter, your usual response to this is that Satoshi made mistakes, etc, that he was wrong and that you know better than him. Which may or may not be true. However, regardless of whether he was wrong or right, that is how the project was envisioned and trying to fundamentally change it now is a controversial position - every bit as radical as wanting to change the inflation formula.

It's actually me that's the conservative one, because I advocate "staying the course".

[fornit]

i agree with mike that staying with the 1mb limit is far more radical - and unreasonable - than increasing block sizes. however, i think there is little need to find a final solution at this point. a conservative increase of the max block size - for example to 5mb - is a very simple and predictable short term solution. 5mb would work on computers and bandwidth a lot of people already have so you dont even have to argue with moores law or anything to show that you wont run into trouble.
and when the exponential growth hits that new limit hard and fast and there are still lots and lots of full nodes around, the 1mb-extremists will be very very quite and it will be much easier to get a consent on something more final  

there might be more elegant solutions, but pushing for anything that might even remotely look like it wont run on an already existing, affordable consumer computer is a very hard sell. at least for now.

ps: mike, i think you have trouble grasping how alien exponential growth is for pretty much everyone. it eludes the natural human capability to make good estimates and seems overwhelming and scary. psychologically, predictability is extremly important right now and thats one of the main reasons i advocate for a fixed increase in block size for the time being. i think its worth limiting the best case scenario for short term bitcoin growth for that.

[justusranvier]

ps: mike, i think you have trouble grasping how alien exponential growth is for pretty much everyone. it eludes the natural human capability to make good estimates and seems overwhelming and scary. psychologically, predictability is extremly important right now and thats one of the main reasons i advocate for a fixed increase in block size for the time being. i think its worth limiting the best case scenario for short term bitcoin growth for that.

I think you might even have succumbed to this with your 5 MB recommendation.

With no further increases in the adoption rate, average block size will reach 1 MB by the end of 2013 and 10 MB by the end of 2014. An increase to 5 MB only buys a few months, and if something happens between now and then to increase the rate of Bitcoin adoption all bets are off.

[fornit]

I think you might even have succumbed to this with your 5 MB recommendation.

With no further increases in the adoption rate, average block size will reach 1 MB by the end of 2013 and 10 MB by the end of 2014. An increase to 5 MB only buys a few months, and if something happens between now and then to increase the rate of Bitcoin adoption all bets are off.

i am well aware of that. as i said it limits the best case scenario. but if we really do hit 5mb in say summer or fall 2014 that also means bitcoin is twenty times as big as it is now. imho, at that point its likely too big to fail - pun intended.
so i would like to intentionally sacrifice the best case to avoid the worst case. short term at least.

[acoindr]

Well, one thing seems clear to me by now. No amount of continued arguments will change entrenched views here. So, on to the next question. What do we actually do?

In negotiations where parties are far apart and demonstrably unwilling to move the only solution AFAIK is one where nobody gets what they want entirely, but instead uses something all can live with.

In my opinion that would be a change that appears most "safe". That seems like raising the limit by some safe appearing amount and seeing how things go. I estimate that would be raising the limit to something like 5-10MB.

[marcus_of_augustus]

Okay, here's some numbers so we are all on same page and not throwing around "vagaries"

The maximum size of the blockchain growth at present 1 MB block limit is 52.5 GByte per year.
At an increased block limit of 5 MByte => 262.8 GByte per year.
Similarly for 10Mbyte, max. growth => 525.6 GByte per year.

When someone says "they want to increase the block limit so that the network can grow", they can mean only one thing, they want the number of transactions on the network to grow since clearly the number of new people willing to run a full node with increased storage decreases (above some limit) as the size of the blockchain increases. So just to be clear here, they do not mean they want the network to grow but that they want the usage of the network to grow, which is not the same thing.

So as long as everybody is clear that increasing the size of the blocks is limiting the number of full nodes then that is okay, but what you are using as the metric for the "size" of the network is important here.

Is the size of the network the number of nodes or the number of transactions? What is the stated goal here, maximizing transactions or maximizing network nodes?

[acoindr]

Okay, here's some numbers so we are all on same page and not throwing around "vagaries"

The maximum size of the blockchain growth at present 1 MB block limit is 52.5 GByte per year.
At an increased block limit of 5 MByte => 262.8 GByte per year.
Similarly for 10Mbyte, max. growth => 525.6 GByte per year.

When someone says "they want to increase the block limit so that the network can grow", they can mean only one thing, they want the number of transactions on the network to grow since clearly the number of new people willing to run a full node with increased storage decreases (above some limit) as the size of the blockchain increases. So just to be clear here, they do not mean they want the network to grow but that they want the usage of the network to grow, which is not the same thing.

So as long as everybody is clear that increasing the size of the blocks is limiting the number of full nodes then that is okay, but what you are using as the metric for the "size" of the network is important here.

Is the size of the network the number of nodes or the number of transactions? What is the stated goal here, maximizing transactions or maximizing network nodes?

I don't think storage is a big problem, even storing the full blockchain with 10MB figures. Memory has kept a good pace at becoming cheaper, more so than higher bandwidth options. Even with today's costs you could handle the 10MB figures for about $65 per year in storage costs, which will only get cheaper.

Dell Price      $129.99 - 1TB Hard Drive

http://accessories.us.dell.com/sna/productdetail.aspx?c=us&l=en&s=bsd&cs=04&sku=A4489204&dgc=ST&cid=248711&lid=4318543&acd=12309152537461010

[justusranvier]

When someone says "they want to increase the block limit so that the network can grow", they can mean only one thing, they want the number of transactions on the network to grow since clearly the number of new people willing to run a full node with increased storage decreases (above some limit) as the size of the blockchain increases.

This is not at all clear.

Increased storage requirements caused by increased transaction demand may reduce the number of home users willing to run full nodes but at the same time this growth can only happen if the number of people using bitcoins is increasing. Perhaps the fraction of home users willing to run a full node is decreasing by a certain percentage, but it's not at all obvious this percentage will be larger than the percentage increase of the user base as a whole.

In addition, if the  transaction rate is growing because of increased adoption bitcoin will be spreading into entirely different sectors than just home users; there will be more bitcoin-based businesses. It's not a given that the decrease in the number of home users running full nodes will exceed the growth in the number of businesses running full nodes.

It would be silly to assume that the composition of users during the extreme early adopter phase will be in any way representative of the composition of users in the future as Bitcoin moves higher up on the adoption curve.

[acoindr]

...

In addition, if the  transaction rate is growing because of increased adoption bitcoin will be spreading into entirely different sectors than just home users; there will be more bitcoin-based businesses. It's not a given that the decrease in the number of home users running full nodes will exceed the growth in the number of businesses running full nodes.

It would be silly to assume that the composition of users during the extreme early adopter phase will be in any way representative of the composition of users in the future as Bitcoin moves higher up on the adoption curve.

I agree.

[marcus_of_augustus]

When someone says "they want to increase the block limit so that the network can grow", they can mean only one thing, they want the number of transactions on the network to grow since clearly the number of new people willing to run a full node with increased storage decreases (above some limit) as the size of the blockchain increases.

This is not at all clear.

Ok, maybe not crystal clear down to the exact numbers but qualitatively it is correct. This can proved easily by considering the upper bound, take the total installed 'potential nodes' on the planet with more than max_block_size*144*365 available storage space and there is your total available number of nodes. Make some assumptions about how many of those available nodes will actually run bitcoin and you get your maximum practical network size, measured in number of full nodes.

Increasing the block size limits the number of nodes unless there is added sufficient incentive for new nodes to bring online the new storage/relaying capacity. Whether that is a good or bad thing is a separate issue.

[jmw74]

Well, one thing seems clear to me by now. No amount of continued arguments will change entrenched views here. So, on to the next question. What do we actually do?

In negotiations where parties are far apart and demonstrably unwilling to move the only solution AFAIK is one where nobody gets what they want entirely, but instead uses something all can live with.

In my opinion that would be a change that appears most "safe". That seems like raising the limit by some safe appearing amount and seeing how things go. I estimate that would be raising the limit to something like 5-10MB.

"So you say that Porsche you have for sale is worth $40k, I say it's worth two bucks.  Why don't we meet in the middle and call it $20k, that's fair right?"

People can't come along with a radically different vision for what bitcoin should be, and expect everyone else to meet them in the middle.

I think it's pretty clear who would win this battle if it came down to a fork.  The one with cheap payments.

[caveden]

Is the size of the network the number of nodes or the number of transactions? What is the stated goal here, maximizing transactions or maximizing network nodes?

Repeating what was already said once again: you can safely transact without being a full node. On the other hand, what's the point in being a full node if you can't even transact since there's no more room in the blockchain for your transactions?

[marcus_of_augustus]

Is the size of the network the number of nodes or the number of transactions? What is the stated goal here, maximizing transactions or maximizing network nodes?

Repeating what was already said once again: you can safely transact without being a full node. On the other hand, what's the point in being a full node if you can't even transact since there's no more room in the blockchain for your transactions?

Well that's disingenuous, there is always room in the blockchain (up to 1MByte per block at present), it is just the price to get in the blockchain that is at issue here. As I've tried to make clear on numerous occasions, above in this thread also, you cannot simply divorce the discussions on block size limits from fees as simply as you are wont to do here.

The whole discussion is about who is going to be paying for the N*max_block_size*365*144 Mbytes annual global storage requirement for the blockchain (N - number of full network nodes), trying to  block one's ears to discussions of fees is ignoring half the argument. Shall we have some quantification of optimal size of N for those who seem to be saying it is a number that can be discounted?

[Frozenlock]

I think it's pretty clear who would win this battle if it came down to a fork.  The one with cheap payments.

I'm not so convinced... What if someone introduced inflation instead
of fees? Transactions would be "free", but I'm pretty darn sure the
resulting cryptocurrency would be immediately discarded.

Rather, the one which would win the battle is the one that could
preserve the value. So yeah, I hear you Mike when you say that Bitcoin
has value because it's useful... but it also has value because it has
the potential to maintain it. Bitcoin became money because it had all
the required characteristics to be so, not just 'some' of them.

The biggest selling point of Bitcoin for everyone is the capped 21
millions. If that's not telling something, I don't know what will. The
only ones for which the BTC price is irrelevant are those who
immediately transfer to another asset, like... oh I don't know... a
payment processing service? *Wink to bit-pay*

Anyhow...
I'm still trying to wrap my head about the assurance contracts, but I
don't really see it. Need more IQ points perhaps?

Assuming there's no need to limit the block size or the bandwidth, I
don't really see why/how one could pay for THash. You pay to be
included in a block, miners compete to get the fees, a hashing war
follows... and that's it.

Isn't it always more profitable for a potential attacker to just mine
with the rest, instead of waiting in the shadows with an idle
supercomputer to occasionally reverse a transaction 1-2 blocks ago?

[Peter Todd]

Well, one thing seems clear to me by now. No amount of continued arguments will change entrenched views here. So, on to the next question. What do we actually do?

In negotiations where parties are far apart and demonstrably unwilling to move the only solution AFAIK is one where nobody gets what they want entirely, but instead uses something all can live with.

In my opinion that would be a change that appears most "safe". That seems like raising the limit by some safe appearing amount and seeing how things go. I estimate that would be raising the limit to something like 5-10MB.

Raising the limit as the first response to scalability problems sets a precedent that the limit will be simply raised again and again as Bitcoin grows. Why spend the effort solving the problem, when you can simply accept less security and punt the issue another year into the future? Fast growing internet startups aren't exactly known for their long-term planning.

I think it's pretty clear who would win this battle if it came down to a fork.  The one with cheap payments.

Bitcoin as a payment system is interesting in that as it becomes easier and faster to complete the fiat->Bitcoin->fiat loop required to make a payment, the economic influence of that use becomes less and less important, all things being equal. The reason is simple: the faster you can complete that loop, the fewer Bitcoins are tied up making payments, and thus the demand for Bitcoins for that application goes down. Similarly those users care less about what the value of Bitcoin is at any given moment.

Conversely your investors, the people holding Bitcoins who believe they will maintain their value in the long run, perform far fewer transactions, yet constitute the economic majority and for now are the people paying for the security of Bitcoin. (via the still large inflation subsidy) This group has every reason to oppose changes that will sacrifice the security of their investment just so people can make cheap transactions.

[TierNolan]

Increased storage requirements caused by increased transaction demand may reduce the number of home users willing to run full nodes but at the same time this growth can only happen if the number of people using bitcoins is increasing. Perhaps the fraction of home users willing to run a full node is decreasing by a certain percentage, but it's not at all obvious this percentage will be larger than the percentage increase of the user base as a whole.

Some kind of distributed verification of the block chain would be a potential way to get around the size problems.  When you connect a node, you could say how much you are willing to verify and how much hard disk space you will allocate to bitcoin.  You would then only check some of the information.

This requires that the protocol be modified slightly so that a node can broadcast proof that a block is invalid and then all nodes that receive that proof will discard the block.

There could also be distribution of the storage.  Making sure no info is lost is a potential weakness, but as long as there is enough overlap that should be unlikely.  Also, there would still likely be full nodes which would store everything.

Also, proving a block is invalid sometimes can't be done if info is withheld.  You can't prove a block is invalid, if you don't have some of the transactions referenced by the block.  There would also need to be some system for broadcasting something like "tx with hash = <some-hash> does not exist".  This would be proof that the block in question is not valid.  It isn't clear how to prevent spamming of such messages.