Topic: Flaws in LN (Lightning Network). - page 6. (Read 2054 times)

Cryptography. Use Google:

https://bitcoin.stackexchange.com/questions/63073/lightning-network-are-decrementing-timelocks-used-in-bi-directional-channels

So come again: who/what decide (i.e. judge), is cheating A who claims his last transaction to B is N+1, or B who claims the last transaction he received from A is N?

It's actually quite clever game theory, you don't need to be honest, but if you try to cheat you will lose money.
So it's in your best interest to be honest.

It's rediculous to require to be trustfull in trustless environment, isn't it?

That is no longer true. It is sufficiently safe to play around it with the suggested amounts (up to $40) on mainnet. Heck, some people almost insist that you pay them only via LN for any smaller purchases now (even though on-chain fees play no role currently).

Aren't we already there though? Given how many people use: a) SPV wallets (which are not connected to their own node). b) Online wallets. c) Exchanges and other services as their wallets; I don't see this as a real argument here. However, it is indeed quite inconvenient that you can't use LN without an active node. Watchtowers should help.

You and whoever you are transacting with are both the referees.  You keep each other honest.  It's explained here.


In essence, they can't trick you into spending from an old channel state because you both have to agree on what the current state is.

Also, there's apparently going to be less reliance on this penalty method as Lightning matures.  People are already talking about eltoo and achieving the same result without needing penalties.

I have already asked this question.

There are 'penalty' in LN for those who're trying to cheat. But who is The Referee to decide who is a cheater and who is setting up a counterparty as a cheater?
In other word, who or what prove/certify the last LN transaction from A to B was N and not N+1, which B has just dropped in order to set up A as a cheater and therefore steal coins of A as 'penalty'?

Yeah, this is an issue now, but it this will likely become less and less of a problem as more channels with higher capacity are added to the network. In my experience lnd works better the smaller the transaction is. For sending extremely small transactions it works great.

I believe there is also plans for splitting transactions and sending several small transactions through different routes at the same time.

That's ok because the "vision" for the Lightning Network is to accomodate your day to day micro-transactions, which is not practical to do on-chain, with "unfairly" cheap fees.

But if you want to send $1,000,000 of Bitcoin, then doing it on-chain should be the only way in my opinion.

The most common problems i have observed in LN are

1. Compatibility issue as theymos stated earlier, when we have an up gradation there are always some compatibility issues, although it can be eliminated with time but still it is one of the major issues.



2. Limitation of funds transfer through channels, unfortunately LN will not support big/bulky transactions, you have to rely on your active channels and their capacity.

Yeah, that is some good analogies you make there. Same as with exchanges, most users use coinbase and other online wallet/exchanges to store and use their bitcoin. How many actually download bitcoin core and run a full node? A very small percentage I think. On your phone you can't really run a full node either, has anyone a full bitcoin client with 200 GB of blocks stored on their android phone? no. But do people say bitcoin is centralized because of this? of course not.

Even with most users using a "light wallet", the network itself will be decentralized and yes, maybe we will have a thousand "big" nodes that have much more channels and much higher balance than the average users, but it will still be extremely decentralized.

One of the major reasons why LN might lead to centralization is:
This could lead to a situation where most users will actually prefer an online-wallet that also handles their respective LN-channel(s), just to spare them the hassle of running something themselves.
Of course, that's the overall trend with any online software (e.g. email being used over gmail etc.), but LN kind of encourages that kind of delegation from the user's side to service providers.
On the other hand, users are free to run their own Bitcoin and LN nodes, just the same as with running their own SMTP-sever, so it's effectively more a kind of "concentration" rather than real "centralization".

I find all this talk about centralization very strange, if you actually run a lightning node you'll quickly end up with 5-50 payment channels. Both incoming and outgoing. I've had a node running for a couple of weeks and already have 25 payment channels, most of them created by others connecting to me. This is extremely decentralized. Look at the network topology and how it's evolving, most nodes has a lot of channels in both directions and it's growing rapidly.

I think that LN will eventually be an important part of the overall ecosystem, but there are some downsides:

First, the biggest downside of LN is that the recipient has to have their LN node online and listening at the time of the transaction in order for the transaction to occur. This is fine for payments to web stores etc., but I think that it will largely preclude usage of LN for more peer-to-peer transactions (eg. forum trades). It would not be good if people were expected to use only LN, since that would result in almost all individuals using trusted-third-party wallets in order to accept payments.

Second, LN often does not integrate easily with existing BTC payment systems. For example, I would have to almost completely rewrite the bitcointalk.org payments system in order to make it work with LN. (Someday I'll do it, but not soon.)

Third, it's possible that the network will end up excessively centralized. If 90% of LN channel-value goes through a small handful of nodes, then that would be a real problem. That sort of centralization could lead to: 1) a lot of people getting their funds locked up for a long time; 2) possibly a slippery-slope to further centralization, eg. "gatekeepers"; and 3) possibly even losses if too many unilateral channel-closes are necessary at one time network-wide. However, the degree of centralization in your channels is completely controllable at your end. If you want to avoid channels which go through a certain highly-popular node, you can do so (possibly at higher cost), and then you will be immune to problems related to that node. You can never be forced to accept centralization with LN. So ultimately this ends up being a software problem of properly informing end-users of centralization and risk. Depending on how the network ends up looking, this might or might not be a difficult problem to solve.

In a few years, the majority of transactions will probably go through LN, but I think that it's currently a bit over-hyped. LN can only be part of the overall picture, and it'll take quite some time to figure out how to get it to fit smoothly and safely into everyone's lives.

Can it also be done with least fee or even fee-less? I mean, I've got to heard and learn about LN that we don't really need to pay anything unless and until we close a channel we do our transactions in, so is it possible? As well, doesn't the network charge a fee and is it user-defined? Can LN technology ever be implemented in exchanges once the testnet is done with their tests?

Diligence is a virtue.

As with basically anything else, technological progress, i.e. the development of Wallet Software that hides this complicated functionality under a user-friendly interface, will mend the problem.

In short: give it some time, and using LN will be just as simple as using Bitcoin itself.

The major "flaw" of LN is that it scales transaction numbers, but actually reduces the number of possible users of the network.
That, too, will most likely be overcome with technological progress, but for the moment it renders the LN basically useless for its intended purpose of taking Bitcoin to mainstream usage without the need for centralization.

There's also a game-theoretical, macroeconomic effect that might lead to Bitcoins "in" the LN being less valuable than Bitcoins on the mainchain. But that's hard to prove and calculate. It, too, will most likely be overcome with technological progress, but it might always remain true to a certain extent.

There is actually no flaw in the scenario described and no weakness.

After the attacker sends 0.15 btc to his me2 channel through you and closes it, you still have 0.3 btc in channel 1.  What this scenario suggests is that the attacker then convinces someone else on the lightning network to set up a channel/channels with you and send him 0.3 btc through you.  That's fine and that will then deplete channel 1 to a balance of you having 0 in it, and him closing it out.  But two other things had to happen, and the most important of those is that you now have 0.3 btc in a channel or multiple channels with other participants of LN.  You are out nothing and have lost nothing.  The second thing of course is that the attacker had to get someone to send him 0.3 btc.

The accounting behind LN channels balance routing is very easy.  You only transfer some money to another person in a channel you are in if you receive the same amount in another channel (and you can even charge a fee for doing so). 

I'm not too much familiar with the LN, but I have heard predictions that it will be fully operational by the end of next year. As far as I understand it, one of the main problem is routing. There are estimates that without improvements, using the current protocol, the LN can scale to millions of users. I believe there is also an implementation on the mainnet right now.

https://p2sh.info/dashboard/db/lightning-network?orgId=1

https://1ml.com/statistics

We shouldn't rush it, because when it comes to money, it's better to be safe than sorry.

Whoever tells you to just use LN without warning you that LN is in testing mode, that it's risky to use LN and that you should really use LN for now only on testnet is incorrect (to say it nice).
I was also overconfident at some point and somebody shocked me telling that LN needs years to become commercially useful. Now I know that LN still needs tests and improvements. Badly.

On the other side, indeed, Franky is Bcash supporter and doesn't waste any opportunity to tell people how bad Bitcoin is. And while omitting that LN is early-beta, he will tell you (pretty correctly documented) all its flaws.

Before you take franky1's side, you should be aware that he is a Bitcoin Core, the developers, hater, and he might be a closet Bitcoin Cash supporter. It might be something that he would deny, but most of his debates are aligned to the ideas of people like Roger Ver.

Nothing wrong with it, and it's his right. But sometimes he uses gaslighting, like Roger Ver, to confuse the newbies.

https://en.m.wikipedia.org/wiki/Gaslighting