Topic: . (Read 1660 times)

Can there ever be enough.

2 should be the minimum. Greater than 42 would be even better.

500 nodes even if in the same datacenter can serve the network better than 1 node. If the datacenter shuts down, it's likely that at least 2 of those 500 will be set up elsewhere. And 2 > 1.

what also matters, and hasn't been taken into account with those numbers, is the geographical distribution of the nodes. I.e., I'm sure X% of the DASH masternodes are hosted in data centers, as well as n% of litecoin and y% of bitcoin. Tomorrow the datacenter with 500 instances of dash masternodes decides, for whatever reason, that its not going to allow people to run the service. So those 500 nodes were in effect 1 node. Etc etc.

so, a node is NOT a node. A node in the same datacenter, and especially on the same *silicon*, is not another node. Each node is a point of failure. The closer the probability that set(n) of circumstances can destroy a set(x) of nodes means that set(x) of nodes are one point of failure, and thus one node.

"If you ignore the paid nodes and consider just the volunteer nodes"

It's kind of ridiculous to do so.  A node is a node....pretending they aren't there is like pretending incentivized miners aren't there.   I am guessing it might suit your purposes to do so, but in reality, Dash of course has the second strongest full node network, trying to spin it otherwise and say Dash only has in effect 200 nodes, is false.

That would be incorrect becase you don't give up your 1000. If you run a masternode for a short time, get one reward, and shut it down, you have more coins than what you started with.

The people running a full node on BTC earn zero income from the node itself, no different from a node on, say, Quark. The fact that roughly 6000 people are doing it anyway (compared to 32 on QRK) says something about the strength and appeal of the coin.

I'm also pretty sure that it costs quite a bit more to run a Bitcoin node than most (all?) other coins. For example, when I was mining on p2pool a while back I had to upgrade my bandwidth allowance (at extra cost) just to keep the Bitcoin node going, and the bandwidth is higher now. No other coin that I can think of has that level of bandwidth requirement.

I wasn't saying anything bad about DASH btw. If you ignore the paid nodes and consider just the volunteer nodes (about 200) that puts DASH behind only BTC, LTC and DOGE, which is close to where it is in the market cap rankings (DASH is a bit higher than DOGE there).  The only other two that are close to DASH in terms of volunteer node count are PPC (proof-of-stake) and NMC (anomaly as I mentioned).

PoS rewards are generally too low to incentivize people enough to stake / run full node.

makes sense, I hope Dash does go in that direction. 

He says high tps activity has to be done off chain (with period settlement to the chain which keeps things honest). I'm pretty sure Dash is going to go in that direction too. Chains are good for some things, not good for everything. At least that is one view. There is another view that says trust in Moore's (Kryder's, etc.) Law and it will all work out. It's kind of impossible to know which will turn out to be right really.

Thanks for the link I will have a listen (https://soundcloud.com/epicenterbitcoin/eb-095)

Not sure personally if it's possible to scale a currency to 10,000's transactions per second like you would need for everyday use, without users having to use high spec hardware / connections, although if you keep ~7 transactions per second then I guess it's not an issue.

I edited the post above with a link where you can listen to the interview where he explicitly says that renting nodes in a datacenter and not doing transactions is "quite useless". You can agree or disagree with him, but that's what he says.

As for 20MB blocks, he's against that, so his opinions are consistent at least.

"It’s the amount of economic interest that is relying on full-nodes and has direct trust and control of those full-nodes"

With this statement I interpret it that Adam is saying you need to tie-in the economic interest of the full node operators to ensure the security.  Using your own full node is one way to achieve that, but operating a masternode for other users is another way because the operator doesn't get rewarded if they don't provide a secure service in the interests of the end users.

About not renting nodes in data centers, if Bitcoin increased to 20MB blocks, on current broadband etc, I don't see how 20x the throughput can work in the foreseeable future on consumer hardware so there is no choice but to scale-up the node's hardware and use datacentres.  

It's how you keep a datacenter-based full node network decentralized is the big question, and Bitcoin full node network is already becoming centralized - again I think Dash is showing the solution because virtually all full nodes are on server hardware and decentralized across a lot of hosting companies around the world, because a user anywhere in the world can get paid to operate a decent node.  

You are mistaken. That is exactly Adam Back's point (well, that, and if you run a node yourself and validate your own transactions it makes everyone more secure). Read what he wrote again.

"it’s not just running a full-node, you have to actually use it for transactions"

What do you think he means by that?

He has also warned against renting nodes in data centers, which is exactly what you get when you pay people to do it.


Also https://soundcloud.com/epicenterbitcoin/eb-095 (54:16)

"It's not just running a full node. If you just run a full node in a data center and you don't do any transactions, that's actually quite useless. So it's not the number of nodes, it is the amount of economic interest that is relying on those nodes and has sort of direct trust or control of them"

Interestingly I just noticed that the quote you copied above appears to be a serious misquote since it drops the part from the middle of his statement about just running a node in a data center being useless! Not your fault, coinjournal did that.

the trust comes from the number of nodes and how decentralized they are though, running one node yourself won't make your transactions more secure I think.  My point is, you don't get a high number of decentralized nodes with a busy network like Bitcoin, unless you incentivize them, Dash is an example solution to this in action.

but it's ok to agree to disagree on this

Yes he did:


That's the key. As a transactional participant you have to be running your own node to get the security benefits he's talking about. So that is individual merchants, investors, exchanges, etc. Setting up a system to pay people to run nodes independently is exactly not what he is suggesting.

Anyway, this is independent of my point that you can't use paid nodes as an indicator of interest and support for the coin (across coins).

We can agree to disagree about whether nodes-for-nodes-sake is a desirable goal. I will agree that if you think it is then incentivizing them is appropriate.

But size of the full node network is a bottleneck in the amount of transactions per second the currency can handle as full nodes process the transactions, Bitcoin it's only ~7 per second and hit capacity several times already with their fullnode network.  

Adam back talked recently about how the number of nodes as a measure of security / decentralization:

“If you boil it down going down from the requirements about what Bitcoin is and why decentralization and permissionless innovation [are important], you can translate that into what are the mechanisms that make bitcoin secure, and the full node auditors — it’s not just running a full-node, you have to actually use it for transactions. It’s the amount of economic interest that is relying on full-nodes and has direct trust and control of those full-nodes. This is what holds the system to a higher level.”

...and how nodes are key in the blocksize debate: http://coinjournal.net/adam-back-on-the-overlooked-importance-of-full-nodes-in-bitcoin/

And there are groups trying to incentivize Bitcoin full nodes e.g. Bitnodes https://getaddr.bitnodes.io/nodes/incentive/ and http://www.coindesk.com/adopt-node-project-aims-bolster-bitcoin-network-security/

If we take Dash's working incentivized full-node network and transpose it to Bitcoin as a rough example, at 100x the market size, Bitcoin could have 300,000 servers for it's full node network instead of 5,000, or 30,000 at 10x the spec depending on how it was incentivized - wouldn't it be a different debate on block size in Bitcoin for example if that was the case?

Respectfully I disagree. Nodes in and of themselves don't actually do much that is useful other than seeding blocks, and you really only need a few nodes to do that.

The reason I exclude paid nodes is not that one kind of node is better than the other, it is that I'm looking at volunteer nodes as an indicator of participation. If you pay people to do something, sure they'll do it, but it loses value as an indicator of anything other than the rate of payment, or at least it can't be viewed as an apples-to-apples comparison from one coin to another.

You make a valid point that the cost of running a node is function of the amount of activity on the network in terms of bandwidth, storage, etc. so maybe that should be considered as part of the indicator along with whether people run nodes or not. If I had to guess that would probably increase the scores of DOGE and BTC and decrease the others.

I think "Volunteer nodes" are part of the problem not the solution.

Bitcoin incentivizes miners, but not full nodes - the result is a declining and centralizing full node network that can't scale as usage grows.

You need to incentivize all the users or you end up with infrastructure not strong enough to support large transaction size / frequency / storage requirements, for example Bitcoin's block size debate, and node incentivization problems e.g. http://www.coindesk.com/adopt-node-project-aims-bolster-bitcoin-network-security/.

Dash solved the problem with the protocol paying full nodes a % of the block reward same as miners, resulting in a very strong infrastructure, ~3,000 nodes running on servers, compared to ~5,000 in Bitcoin, at < 1% of it's market cap.

https://www.youtube.com/watch?v=FY1mciGGhO4

First I would break it down by those that pay for nodes and those that don't. The ones that don't pay for nodes but still have a lot of nodes are pretty strong. That's BTC, LTC, and DOGE. The PoS coins are hard to assess, other than Dash (since the paid nodes are an identifiable subset). Dash has 3259 paid nodes. Even subtracting that, it still has around 200 volunteer nodes, which is decent.

NMC is a bit surprising to me. Would not really expect that to have 200 nodes.


Seems to (incorrectly) show 0 for many coins that are not BTC forks.

you could go to https://bittrex.com/Status
you will find there the connections for each coins.
you would have to take into consideration things like if the coin is new or not and if it's PoS or not.