The only people incentivized to run a node are businesses which leads to a hub and spoke model of major nodes ran by businesses like bitfinex, bitrefill, wallet of satoshi, muun, etc.
Person run small nodes don't have the liquidity and the people running them don't have the expertise to properly manage channels, rebalance, etc. Which is a huge hassle for someone who just wants to buy a coffee.
Bitcoiners need to be honest about this, instead of sticking their heads in the sand and shouting down anyone who points this out. For a normal user who won't even go through the trouble of doing a coinjoin to acheive a basic level of privacy, thinking they will run their own lightning node is complete fantasy land.
This perspective (specifically the first paragraph) was extremely common as an argument against LN around the time it was first being introduced (2017-18). Here's what I think is a fundamental error embedded in this way of thinking:
A common complaint, from outsiders, about Bitcoin itself is 'well I can't verify the code! So it's effectively controlled by a tiny group of Core devs'. The nuance is this: even if only a few thousand people can fully vet and audit the code, the numbers are high enough and distributed enough that someone is definitely going to catch malicious changes. This is the dynamic of open source: anyone can contribute and review and test, even if of course most people don't.
It's linked to bitcoin's own architecture: nodes can leave and join at will. The dynamic created by this openness is what counts, not the specific set of contributors/nodes/coders that exist today.
By the same token, in Lightning, of course hubs will spring up, there will be pockets of centralization here and there - but new nodes can spring up any time to route around those hubs. And this argument is critically dependent on how easy it is to spin up Lightning nodes for at least a subset of users - and indeed, it really is easy, and it only gets easier every year. Even people on phones have very easy to use non-custodial options (example: Phoenix) so the situation is even better than the 'start-a-node-optionality' that I describe.
So in that sense LN is even better than 'fully distributed/decentralized' - it's instead completely open - you cannot stop any set of people choosing a pure hub and spoke model for their dealings, and you cannot stop the exact opposite, either.
I agree. Node centralization is not a problem, because small nodes won't ever die, and their existence will always counter-act the censorship of hubs. For end users, auto-pilot will work just fine for managing channels. 'Running a node' should stand for 'running a profitable routing node' - that's not trivial (but easy enough for an average developer, whom there are millions), and end users - shouldn't route. They should use non-custodial wallet with a proper auto-pilot.
We are at the very beginning of LN, and so much progress is ahead, it makes not much sense to assume that current limitations are forever.
reply