Bitcoin Forum

When running a node becomes expensive enough that people can't do it for free you'll still be able to find full nodes willing to accept incoming connections. You'll pay for that service in a variety of ways:

1) Transaction fees: You connect directly to a miner who lets you do so because they want your transaction fees. They may require some # of transactions per unit time, and part of the agreement may be that you only send transactions to them. (easily verified) In return they'll run your bloom filter against incoming blocks, although don't be surprised if they force you to give them a bloom filter specific enough to identify exactly what addresses are in your wallet as part of the deal.

2) Pay-for-service: You pay for the service directly. In return they resend your transactions to miners to get them mined, possibly with preferential deals (kickbacks) that may or may not be public knowledge. They also run your bloom filter against the blockchain, and again, they may or may let you do so in a non-specific manner. Given AML regulations I wouldn't be surprised if the services that operate out in the open only allow you to tell them what addresses you are interested rather than a bloom filter obscuring that information. (AML rules apply to case #1 too)

3) Datamining: Google and other search engines already provide a lot of services purely in return for the data they can gather. The blockchain itself is a rich source of transaction data, made richer by figuring out the real identities behind the pseudonymous addresses on it. Just like #1 and #2 if you can determine who is sending what transactions and owns what addresses you can integrate that into a rich dataset to do things like get real-world information on what vendors are actually popular, which in turn can feed search engine results and other services.

It'll be interesting to see how AML regulations apply to all these services in the future. I suspect they'll eventually be subject to the same know-your-customer rules as any other financial service provider to help authorities link identities to Bitcoin addresses. This doesn't have to be very intrusive: in case #3 that might be as simple as using your Google login to authenticate with Google's Bitcoin servers.

Excellent post! Unfortunately that everyone is sticking their heads in the sand about this.


This really makes you wonder why Google is so happy to pay Mike Hearn to work on Bitcoin.

I don't think Google pay Mike Hearn to work on Bitcoin, he moonlights on his own.

That's incorrect. He does it as 20% time, which is a google program where they let you devote 20% of your paid time to a project of your choosing, but approved by management. I have spoken to google employees among my friends and acquaintances and they say coming up with a good 20% time project is actually really difficult because finding one with the right balance of innovative, novel, and still valuable to google is quite hard. Your 20% project idea can be far out, like google's self-driving cars, but it still has to be something that google could plausibly profit from in the future.

For what it's worth, Pieter Wuille also works for Google as a Site Reliability Engineer, but unlike Mike he apparently does work on Bitcoin purely in his spare time.

Yes.  Mike's efforts are valuable, such as recently getting Bitcoin extension in HTML5.

Some are. He also wrote bitcoinj, which is quite clearly Copyright Google, and gets used in all sorts of things. (like my smartphone wallet)

But he also sees anonymity as a bad thing and sees no problems at all with a Bitcoin where only a select few can participate and pushes quite hard to make that happen. Time will tell, but I think his overall contribution to Bitcoin will be extremely negative.

Interesting information to glean in this thread.  People are always weak links in great ideas.

Paying full nodes for transaction data and to relay transactions is exactly what Bitcoin needs, to prevent non-mining full nodes from dropping out and the network relying on only a small number of mining pools for all transaction verification.

With Bitcoin it's perfect too, because the currency the Bitcoin network establishes can be used to pay for the work the network nodes do.


I think it would be tough to regulate, because full nodes can be any where in the world.

Thanks for the information! :)

The way to keep non-mining nodes in the network is to simply ensure that the blockchain grows slower than hard drive capacity does.  People with lots of bitcoins have an incentive to run a node, simply to keep their money secure.

Any other solution results in a loss of confidence in the currency and the emergence of a rival blockchain, and most likely that rival will simply be a fork.  Except that it won't be the fork, it will be the original that 99% stick with.

I personally believe that a large company such as google, which has tons of data centers and the such will come in to bitcoins, see the money to be made (especially the $1b+ in bitcoins to be released over the next few years at the current price) and basically make a huge Asic center designed to mine bitcoins all day that controls say 70% of the network hashing power.

It's bandwidth that's the biggest cost of running a full node, not hard drive space.

Bitcoin has to grow at a rate of something like 1,000% a year for the next 10 years to become a major global currency. Groundbreaking new technologies have astronomical adoption rates. You're going to tell businesses to not adopt bitcoin to prevent transaction data from increasing the bandwidth requirements of nodes too quickly?

The only viable option is to allow the blockchain and the number of transactions to grow as fast as it needs to, and allow specialists to take over full node operation when the requirements become too much for average users.

Thin nodes making micropayments to full nodes (the two parties can use "rapidly adjusted off-the-chain micro-payments" using lock time (https://bitcointalk.org/index.php?topic=25786.0) to avoid unnecessary transaction fees and blockchain bloat) for transaction data would help disperse and increase the number of full nodes running.


The biggest risk to BTC from a rival blockchain is if protocol changes are made to BTC to artificially constrain its own growth to reduce full node bandwidth/hard drive capacity requirements. That would give a rival network space to grow. We need to setup BTC so it can grow rapidly as a network. The larger the BTC network gets, the more useful it is to potential users.

I have thought about this as well. We are relying too heavily on the fact that bandwidth is essentially free for most people with a monthly data plan.

But I do not see this as a permanent solution.

I do think the best way forward for the Internet is to have a sort of pay to play type of setup where you pay for your bandwidth usage from remote site to your computer. I believe bitcoin is the answer for how to pay for this, but am not sure of the best method yet.

In such a scenario, each node in the network would get paid based upon the amount of traffic that passes through their node. So people would want all of the Bitcoin traffic to pass through their nodes so they can get paid for it. Also, locations where a lot of traffic is generated will have a lot of people around it helping to pass the traffic.

Umm, i would expect mtgox to pay me to have my bitcoins with them, well, bitcoin.

Normal users will access their online banking account and will have one account with bitcoins
 or use electrum or some sort of lite client that doesn't download the whole blockchain?

good post

Why would Google get into a commodity market where players have little differentiation?

Google is about markets with the largest first-mover-advantage possible. Markets where you can only go big or go home. Bitcoin hashing with ASIC's is a tough business with profit margins that go down as you operate on a larger scale, not up. The one exception to that rule is your validating node, and every Bitcoiner should recognize that it is imperative that we ensure the ability to run a validating node and mining operation continues to be within reach of average people, and possibly anonymously.

If Google got into mining it would be by running a mining pool, not by operating hashing hardware itself.

Yes, but blockchain transaction volume doesn't have to grow.  Even now most activity happens off of the block chain.  For example MtGox does a lot of business every day, with day traders constantly making trades.  Those trades are not part of the blockchain.

In the future we're going to see a world where many bitcoin users never interact with the blockchain.  Instead they'll keep their bitcoins in bank accounts and spend them with debit cards.  This has to happen if bitcoin is to succeed, because the alternative (regular people safely using raw bitcoins and not having them stolen by hackers) will never happen.


At this point you aren't talking about bitcoin.  You're talking about a currency controlled by governments and central banks.  Also I wasn't saying that average users need to be able to run nodes.  I'm saying that average nerds need to be able to run nodes.  Not only is the idea of an average user running a node absurd, but the idea of an average user using a bitcoin client is absurd.  Raw bitcoins will never catch on.  They are simply too hard for regular people.  They are going to need to use them through debit cards, small denomination paper wallets, etc...

But limiting the growth in the size of the blockchain would mean massively increasing the reliance on trusted third party payment processors, which goes against every thing that BTC was created to avoid.

BTC-only transactions require no trust and no third party intermediaries. That's something BTC needs to maintain if it's going to have a real impact.


No, because running a node is a purely technical process. There is no political involvement in it. A node can be any where in the world and accessed over the internet.

Relying on a network of large 'trusted' third party payment processors to handle almost all transactions is what would hand control of the network to governments and their regulatory agencies.


Nakamoto's original vision was for 100GB blocks and a network that can handle global-scale activity. If you want small networks relying on third party payment processors, there are more than a dozen BTC-variants that you can use for that. BTC should follow Nakamoto's vision.

Correct. Anyone who wants small block sizes forever should dump their BTC and load up with YA Coin, Royal Coin and China Coin. Problem solved.

Do you really think Google is that selfish?

Since running a node will not be hard for someone with infrastructure, small companies (like mine) could run one and just give it out to customers few or no strings attached.  Even if it cost $100 a month it would be good advertising and a customer perk that spread out over many customers really costs almost nothing. 

Enthusiasts could also just run a node for family and friends.  I have seen people spend thousands of dollars on gaming PC's, spending $1000 for good hardware for a full node will not be a problem for people into it. 

Yep. Very sensible indeed, but you might as well talk to retep's hand, because the face doesn't understand.

retep is 100% convinced that scaling Bitcoin will lead to centralization as small hobbyist and home nodes drop off due to bandwidth and storage requirements.

However, his dream solution is a nightmare worse than the problem.
He would have the blockchain reserved for big banks, millionaires and "important" users willing to pay $20 per transaction. Everyone else being forced to go through 3rd parties for their Bitcoin transfers.

He ignores that in his scenario all the thousands of Bitcoin hobbyists with nodes at home are going to switch off their Bitcoin software if they are effectively banned from the using the blockchain because of sky-high fees. It will actually create the centralization which he detests.

The solution is scaling Bitcoin by encouraging small businesses such as yours (Littleshop) to pick up the slack, adding higher capacity nodes and maintaining decentralization, and a blockchain available to all users with fees of a few cents. This is the future we need.

If "Moore's Law (http://www.mooreslaw.org/)" is still valid, the technology will keep up with bitcoin. We first used CPUs, then GPUs.. now we have come to ASICS. It is possible that something else is around the corner.  ???

AML makes no sense. One can provide this service with offshore hosting, or even as tor hidden service.

I did some quick calculations a little while ago and calculated that at the current rate of growth of the blockchain, it'll be at 60 GB in five years.  (I'm running a full node on my computer.)  The key phrase there is, of course, current rate of growth.  But assuming it continues as it is, I don't see hard drive space being a problem. 

Bandwidth?  That's another question entirely.  I'm running basic DSL right now and it's not really impacting my internet speed negatively.  So long as that doesn't change in the future, the bandwidth affect it has is fairly negible.  Again, that assumes that the bandwidth requirements of a full node will be the same with a 60 GB blockchain size as they are with a 8.91 GB blockchain size.  (Current blockchain size as of today.)  That is of course probably not a valid conclusion. 

If the webhost I'm planning on using when I host my business offered unlimited storage, I'd happily host a bitcoin node there.  Unfortunately, it "only" offers 100 GB and I'd rather not see my entire web hosting space eaten up by the bitcoin blockchain in several years.

Now, if anyone wants to persuade Google to roll out their Google Fiber nationwide, I don't think the bandwidth constraints of the blockchain would ever be an issue.  I'd happily pay the $70 / month they're asking for a gigabit internet connection.