Bitcoin Forum

Bitcoin currently only have around 5k full nodes, this is less than back in 2013, when we were hovering around 7k full nodes.

As the chain is getting huge, quickly. It's no longer trivial to host such big blockchain, the disk space and bandwidth expense is quite high.

We will need to start discuss rewards for hosting a full node that stores the full blockchain, they add a lot of value to Bitcoin and it's unfair that they have to do it for free.

I would believe that the actual number is somewhat more. It is impossible for Bitnodes to locate every nodes since some nodes may have already hit their limits and Bitnodes is unable to connect to them. Bitnodes is also likely not be able to record nodes who do not enable incoming connections.

Back to the nodes question. Nodes are fairly important, I agree. They are the backbone of Bitcoin and they help to relay transactions and blocks and enforce the network rules. That being said, nodes should be as diverse as possible. Having a lot of nodes on the same ISP or datacenter would cause them to be centralised and wouldn't really help the network.

Implementing a reward system at protocol level would require a lot of work, a third party reward system such as the one 21.bitnodes.io uses to have would work better.

Btw, I can probably run a full node on my old computer for $10 per month in electrical costs.

the question is how to implement a reward system that is both good and is impossible to abuse,
and then who is going to pay that reward? an additional fee to transaction?
and then there is the problem of reaching a consensus.

if you want to contribute to bitcoin network not just use it as a wallet with full verification on your own it is good to know this though:

Segwit actually kind of does this in a way by allowing nodes to become lightning nodes as well. But good luck getting miners on board with giving up some of their fees.

incentivizing nodes just makes TX fee's rise. adding barriers of entry.
incentivizing nodes just creates sybil attacks (one person running several nodes to increase earnings)

it doesnt cost alot to run a computer. you just turn it on and leave it on.

the only people it does cost are the childish mo-fo's that run it from a cloud service because they dont want their parents finding out.

if you dont have a basic computer at home. then dont run a full node. we should not be incentivizing people to run nodes on cloud services like amazon. otherwise it begins to defeat the point of being a distributed decentralized network if the majority are all running on amazon

To be honest, when I first saw the statement, I was rather confused over the other statement. Ordinary folks can run full nodes even though they aren't really contributing to the network to that kind of extent. Full nodes without port 8333 open can still relay transactions, blocks and enforce the network rule.

Running a node without port 8333 open STILL helps the network since it helps by storing the blockchain and your peers can still request blocks from you. That statement isn't fully accurate and users should still try running a full node.

10 EUR / month can get you a VPS capable of running a full node 24/7 at 99.9% uptime with >100 connections.


feel free to donate -> http://188.68.53.44/

The problem you have to solve first is how to distinguish a proper full node from a pruned (?) or fake one.


Id guess a homerun node is even cheaper, considering it could run on an old laptop. CPU/RAM requirements are pretty low if its running constantly.

---

it is not bullshit the quote alone is out of context, you should read the reddit post. I've added a line above that quote which i think helps clarify things a bit.
the argument was about helping bitcoin network or in other words contribute, and for that purpose the quote is right, you won't be contributing to anything if ....[connection and bandwidth are such and such]

Quote properly. The idea of "leeching" is bullshit and comes IMHO from p2p torrents which have next to nothing to do with bitcoin.


It does not. What is "contributing to the network" according to you that requires inbound connections?

As I said, even if you are not opening port 8333, you are still relaying and verifying transactions and blocks for the network.

Connection polarity doesn't matter, you will still be relaying nodes/transactions to your peers. The network will still benefit for that. You would just be connected to 8 peers to help to relay information to them. The reddit post is 3 years ago where the nodes are still plentiful. With the decreasing number of nodes, running a full node would help even if you do not open your port.

Exactly, nodes accepting inbound connection are available to SPV nodes and contribute more bandwith resources. All other nodes, even pruned nodes help the network though. Quotes like the above (esp. without context) might have even contributed to a decressing number of full nodes. If you tell people running nodes the only way they can (e.g. due to ISP restrictions) is bad for the network they are inclined to stop.

but declaring an spv node and a pruned node as being the same as a full node makes people not want to be a full node because they are lulled into the delusion that theres no point being a full node.

much like core lulling people into the false belief they dont need to upgrade when segwit is activated.

full nodes are full nodes for a reason.

The node reward doesn't have to be big or a game changer, but even a small fee would give more people a reason to run nodes..

If the rewards were too big it would even risk centralization due to full node farms.

I'm basically repeating what franky said, rewarding nodes in a noticeable way just makes everything more expensive and it doesn't really benefit everyone. Incentivizing more businesses to run nodes (such as Coinbase or whatever, I don't know and that's just a name) would be a better choice as we know some of their profits go to maintaining the network and they feel the effects less overall.

and how would they get paid..

suddenly a 1in 2 out becomes a 1in 5400 out transaction?!?. where 5398 outs are paying nodes a few sats each.. um no thanks

im all for people trying to make a buck with LN. because that is just a side option much like using blockchain.info or coinbase.com... but trying to turn bitcoins onchain mainnet into something commercial will cause issues for real utility of the mainnet.. so again no thanks

all i can see happening is a re-ignition of what happened in the mining pool emergence. instead of 1 node 1cpu.. we start seeing people running multiple nodes and multiple computers. and becoming node farms

This will be a very complicated process to differenciate between people running a full node, and those who run full nodes and not opening a single inbound connection and those people running pruned nodes. There were even some people who spoofed nodes on Azure, so if they manage to fool the system, then they would be getting paid for "fake" nodes on cloud based services. This will just be a extra pain in the ass to manage and police this. ^hmmmmm^

the real foolish thing is. even if there was a way to eliminate all fake nodes.

we will see a fight over how much they can/should earn.
we will start to see nodes rejecting and refusing to relay peoples transactions unless they get some income.
then it will be a fight over how much income.

some will want ~$15/hr (0.02btc per hour for ~2500tx) which works out as 0.00000133/tx
but thats just for one nodes income. lets say there were 5400 nodes

suddenly that 0.0072btc just to pay the nodes to relay the transactions, before even thinking about it reaching a mining pools mempool and the mining pools tx fee for miners

id say let LN go commercial and be the paypal2.0 with its CLTV/CSV payment held(maturity)/chargeback mechanisms and fee's. as a side option

and leave bitcoins mainnet as far away from economic game theory as possible.
bring the mainnet back to using code rules for priority and slow down the fee war to work over decades not months

I didnt do that. I just said full nodes that are not accepting inbound connections are still contributing to the network. Less so then nodes that accepting inbound connections or nodes that can help you sync, yes, but still contributing. Whereas the gavin quote suggest that unless you can hold 100 connections you are bad for the network.


They fully verify data, not fully fill the default 125 connection slots.

im not talking about connection slots. im talking about being a full part of the network. more so im on about validation and access to full data.

---

but to answer your opinion when it comes to connection counts

the less connections you have the more hops(relays) data has to do to reach everyone

EG if 74 nodes had 74 connections each. the data will reach all nodes in one hop.. 74*74=5476
anything below 74 wont be enough to reach all the 5400 in one hop. some will require a couple hops
however there is a big leap with less nodes.
18 nodes connected to 18 nodes connected to 18 nodes can reach everyone in 2 hops
9 nodes connected to 9 nodes connected to 9 nodes connected to 9 nodes can reach everyone in 3 hops
6 nodes connected to 6 nodes connected to 6 nodes connected to 6 nodes connected to 6 nodes can reach everyone in 4 hops

so athough having 6 connections, and moving it to 9 connections (if they all done it) gets everyone having the data in 3 instead of 4 hops

trying to get everyone having the data within the next hop requires 74 connections+

so many deem this as being a bottleneck if some of them 74 nodes are not then passing it onto 74 nodes themselves

i personally dont see a reason to need 125 connections as a default but if everyone had just 6 connections.. they are just causing data to need to bounce around a bit longer then whats deemed most efficient.

---

though to counter that. only 75 nodes need to connect to 75 nodes each... meaning the other 5300 nodes RECEIVING it can do whatever the hell they like because the data would have already been received by the time the other nodes wanna play funny business about how many they should connect to

and this is where i feel core are re-inventing the 'supernode' concept. using things like their 'fibre' brand.
lets say they have 80 specific nodes with 80 connections each . everyone gets the data in one hop.

leaving everyone else free to loop it through however many/little they want

though good for data migration, it does then make 'fibre' the gate keepers of data if they are the centre of the distribution network

I am running 2 full nodes at this point. I personally don't need/want any reward for doing so as I just want to help the network, but it might be a good incentive for people to run full nodes. It time on time again surprises me how low the number of running nodes is, and yes, it has been in decline for years. Another thing is that the number of nodes should be places all around the world, and not hundreds of nodes being hosted from just one data center. That would not really help to achieve some level of decentralization. And then you have the problem of which nodes you must reward. Other problems are how much would they get paid, and what will the pay schemes look like, who will pay for it, etc.

may i ask why you are running 2 nodes. are they on 2 PC's in two locations.. or one of them on a datacentre one on pc.. or both data centre

Doing business is the incentive to run bitcoin nodes, if there are not enough businesses to keep the network running then bitcoin is a failed experiment and should die.

You can run full nodes and charge people for using them, there's your incentive.

Beside the blockchain getting bigger, one of the main reasons that bitcoin nodes went down overtime is a pretty obvious one: in the begining if you wanted to use bitcoin you had to run a node yes or yes since bitcoin qt was the only wallet in existence. As more wallets with no waiting time to sync like electrum were released naturally the nodes went down. This is dangerous, so we should indeed think about ways to incentive people to run nodes, but its not as easy as it seems. Incentivizing people to run certain nodes with money is also a very tricky scenario... it should be on people's best interest to run nodes if they want their investment to survive overtime.

By placing rewards, it's gonna start a whole new economy from it. Soon mining companies would start monopolizing this too.

I agree with you and wouldn't it be especially helpful if a lot of people were doing that just from home? Most people have the option of leaving an older computer running all day and if they would get some kind of incentive to do so, we'd have a lot more full nodes on the network.

I agree, running full nodes is becoming a serious issue due to the size of blockchain and access speed and people who are willing to take this issue on themselves deserve some reward. But aren't they getting BTC for it, already? I thought if you ran a full node, you would profit from the network fee.

I know its been 4 days, but that was exactly my point. A full node does verify all data whether inbound connections are accepted or not.


Thats an overly simplistic network structure and I would be very surprised if 4 hops have a significant impact on orphan rates (I suspect thats where you are going with this).


Again, the argument I am against is not that better is not possible. It is. I refuse to believe and have yet to see an argument that a node that is not accepting inbound connections is harmfull to the network. The above was a nice though experiment, but it didnt actually lead to any harm towards the network. Maybe I just missed and you can ELI5.


IIRC the relay network was created to circumvent problems with the great firewall. Thats a very specific problem and has little to do with the average user running a full node.

---

Same someone also running 2 noes, one is in a datacentre running 24/7 without wallet, the other is a home node with a wallet.

---

No, thats miners.

---

my issue is that if everyone runs only bitcoin core, (no code diversity) and all run it from amazonaws, (no location diversity) where all nodes are pruned and unable to allow new nodes to sync.

bitcoin become limited to being just a relay network that begins to crumble when new nodes cant join because existing nodes dont allow full syncing or have inbound connections closed.

i see no point in people paying $20 a month to some data centre when out of 7billion people.. or even just 2 million that are using/have bitcoin. can easily run a full node from home.

yea theres a few hundred or a couple thousand that say they cant. but that doesnt stop the majority that can. its like saying lets not let activision or blizzard release any new online games because some people cant play the game

if core want to be the centre/core of bitcoin then they should concentrate on being a full node and stop all the wishy washy pruned/litenode stuff.. leave that for electrum/multibit to play with.
pretending that not upgrading, or running pruned mode, or not having inbound connections is 'fine' is putting people into a false sense of security and wasting their time.

people who want to be full nodes need to know the darn assed truth about what's involved.
teaching people that anything below say 18-74 connections means that some nodes wont be getting the block data in the next hop, so that the recipient from you might be having to pass it around because you have not passed it around as much yourself.
yes its minimal disruption.
but a healthy network is about being a strong network where its all uptodate and verified efficiently. and should one supernode go offline there are enough other supernodes to cover everyone getting the data efficiently.

i understand core want to dominate and be the centre so that users can just be crappy litenode relayers.. but thats stupidly centralizing the network where litenodes do not have independent network involvement but are just a shadow/false pretence illusions of decentralisation.

i dont see any reason to incentivise nodes if the only reason people want to run a node is for a pay day
i dont see any reason to incentivise nodes if people dont understand what a FULL node intels
i dont see any reason to incentivise nodes if core devs want to be the gate keeper main supernode and then have weak shadow nodes pretending to be full nodes even though they have not upgraded or have upgraded but then decided to turn off certain settings because they feel someone else can do the work for them

I agree.


So why advocate that they may only do so when opening their ports otherwise they are harming the network? (I havent seen you do this actually, I just really want to hammer down my bullshit point). Its probably enough for 1 out of 10 do this. Home run nodes have specific problems though, which is why I pay for a datacentre one. For one they are usually not online 24/7 and can cause issues with other services like VoIP and streaming. On the other hand you have ISP related issues, like quickly changing IPs (e.g. 24 hour DSL disconnects) that rendes it irrelevant whether or not you open the port as you are no longer found on the old IP. Whatever the issue may be, let as many people run a full node of their choosing. I dont care if its core, unlimited or any other of the countless full nodes.


A pruned not does not require you to trust the electrum server(s). Maintaining a trustless wallet is a good reason to run a node even if you disk and available data volume is small.


Me neither, as you and countless others have said, even if we could solve the "prove you are a proper full node" problem, it would introduce another cost.


I disagree, let everyone that wants run a full node and let them learn as they go. Dont make this a boring licence shit, where you first need to pass a test in order to run a full node. Not that you can anyway. Let people fall into the rabbit hole and figure shit out. Some might stick around.


Software that defines a protocol is a problem, yes. It requires every other implementation to follow all quirks the reference implementation has. Replacing it with another software is not a solution to this problem though.

I am not sure which is worse:

1) people running their nodes on AWS (https://aws.amazon.com/about-aws/global-infrastructure/), Azure (https://azure.microsoft.com/en-us/regions/)
2) people running their nodes on Comcast, Optimum, FIOS

Both are geographically decentralized but under the centralized control of a few companies. Considering that each home internet connection in a particular region from a particular home ISP all connect to the internet backbone at the same choke point, there is probably no difference in decentralization between all FIOS home nodes in NYC vs. all AWS nodes in US East.

I tend to think that 2) would have more restrictive ToS than 1) as far as running servers is concerned.

It is fine to discuss rewarding nodes, but I doubt an agreeable solution can be found. PoS is definitely out of the question (I remember the OP suggesting that Bitcoin be converted to PoS a couple of years ago).