Belowis a complete rebuttal of Rick’s Youtube video on the Lightningnetwork. Please note that I have edited the names he gives Bitcoinand Bcash so as to not confuse noobs with labels used by amicroscopic economic minority of the market.
“Hey guys, thisis Rick. So a lot of people have asked me to comment on Lightning,the Lightning network. I don't know where to start. Yeah, I'm goingto tell you exactly what I think about the Lightning network, andit's not going to be pretty I'm afraid.
There's a lot ofBitcoin supporters that say that Lightning is going to wipe the floorwith everything and everybody else must be so scared they're s******gtheir pants or what not. I'm not making this up, people are actuallywriting this.
That's not thecase. I don't see the Lightning network going anywhere. And with theprevious week's presentation of how Blockstream failed by trying tobe a horizontal and a vertical actor in the marketplace at the sametime.
Let's take a lookat what the Lightning Network is, with that in fresh memory.”
Ifail see what analysis of the Blockstream company’sbusiness endeavours has anything to do with the features ofLightning.
“So, let's takea look at the famous Lightning Network. It's built, or not built onperhaps, but marketed with 2 bad assumptions, 2 wrong assumptions.
The 1st is: thatBitcoin can't scale on chain. That's a wrong assumption, and we'll becoming back to that.”
Bitcoincan scale on chain. You just sacrifice decentralisation when you doso (Through the elimination of long tail small miners, enablingcoercion of the network). Since decentralisation enabling selfsovereignty is the most important feature of the network, increasingthe blocksize on-chain is called ‘down-scaling’ or ‘reversescaling’.
Bitcoinmaintaining the current limits of block size is called ‘scaling’through the passage of time. Meaning, that as time goes on andthe same limits are in effect, prices for equipment, data usage andelectricity prices continue to lower, thereby increasing thedecentralisation of the network. This is scaling.
“The 2nd isthat Bitcoin use must have high fees in order to be long termsustainable. Which is also wrong.”
Rickshows ignorance of basic economics with this statement. Low or nofees on the distributed/broadcast layer enable market participants towaste as much space as they want on a shared market space. Marketparticipants see no consequence on throwing everything and thekitchen sink down a limited pipe (Tragedy of the Commons). Ifthe market then attempts to address this by repeatedly enlarging thepipe, 2 negative consequences will be induced: 1. Small long tailminers will be unable to remain competitive and will go out ofbusiness 2. Market participants will remain numb to pricesignals/warnings further compounding the 1st point. Oncethe small long tail miners have been put out of business, the marketcan be coerced.
Put another way, a resource being in greater supply than thedemands of the market participants (In this scenario, block space)results in the participants not economising or valuing that resource.However, whenever a resource is in higher demand than its supply, themarket participants begin economising (Principles of Economics, Chapter 3).
And both of theseare kind of a precondition for the Lightning network to have a usecase in the first place.
TheLightning Network has more use cases than just extremely low fees. Itadditionally enables instant transactions and onion routed privacy.
“But eventhough these are wrong assumptions for the mere existence of thenetwork, let's take a look at what predated this.
This leads toproblems with Bitcoin. Had been observed for some time, and just ayear ago the Blockstream people in particular were saying that Segwit(Segregated Witness) will solve all of this. ‘If everybody's ready,if we can just have Segwit all of these problems are going away.’
It didn't quitehappen that way, did it? So, now the story has changed. The problemsremain, for the most part, but the solution is different.”
Ican’t speak for Rick’s strawman ‘The Blockstream people’.Segregated Witness enabled transaction TXIDs to become non-malleable.This is both a major security fix, and a feature.
Theresultant security fix aspect fixed a possibility of a maliciousactor changing the TXID of a transaction to another TXID. If amalicious actor executed this, a transaction could be made invalid(Censorship of transactions).
Secondly,because of this security flaw, TXIDs were previously consideredunsafe to rely on for any features to build on. Now that this isfixed, TXIDs are reliable and previously impossible features arepossible.
There are 4 majorproblems and a conceptial problem with the Lightning network as itstands today. All of which are showstoppers. So in order tounderstand what these problems are let's first look at what theLightning network is.
It is currentlyworking on Bitcoin's main network. But, it's important to rememberthat it's working among people who are all friends with each other.There are no adversaries in this setting, and we'll be coming back tothat because that is a key concept here.
Thereare now a number of adversaries on the Lightning network, and attackshave been seen. Unsurprising, as there has been enormous growth inthe network. However, the network lives on, and continues to grow.
So what is it?It's being marketed as the Internet of Money. Just like how you cansend a video stream or a message from somewhere on the internet tosomewhere else on the Internet, you're supposed to be able to send amessage containing money. This is done by having a network ofbalances where I have a balance to 1 hop, the next hop also has acertain balance in it and so on.
So if I want to pay Charlie something but I don't have a channeldirectly to Charlie. Maybe we have some neutral friends where we bothhave channels open. And so the idea is that this payment can hopbetween Charlie and I in multiple hops and then eventually make itall the way by adjusting balances in the network as a whole.
So in this wayyour payment is routed. Routed, this is a key concept that we'll comeback to. Through other people's balances and the entire network ofbalances keeps updating itself as people want to pay each other.
The proponents ofthis network have illustrated it like this: That in the middle of thenetwork .... "the middle of the network". Just that wordimplies that we have a heavily centralised solution at hand here.
In the middle ofthe network you have really heavy duty data centres and they wereillustrated as banks, probably the NSA, a couple of other people.Most of them benevolent, some of them adversarial.
Outside of thesedata centres you have a number of highly connected nodes which wouldtypically be big corporations. Highly redundant, always online and soon, and so forth.
Althoughsome centralisation will occur, there are limitation factors whichinhibit significant centralisation which I will highlight below:
A large organisation may be enticed to place a large amount of Bitcoin on the Lightning network to gain fees from large single transactions passing through. But if they do so, that money would be in a hot wallet and would be a highly attractive target for hackers. This threat discourages placing large amounts of money on one node in the Lightning network
Users don’t need to pay in 1 transaction for a single purchase. For a small purchase (eg. Buying a beer) it might be most efficient for it to be 1 transaction. However, if you were to buy something expensive (Like a new car) this could be done in a hundred, or a thousand transactions on the lightning network using a multitude of routes. These transactions could additionally be split over a time period that would only appear instantaneous to human observation. This means that it is possible for wallets to subvert significant centralisation by dividing large sums by instances of time, and by different paths. This can all be done as a back-end technology of the wallet, not involving user interaction.
As I’ve said before, if Bcash were to gain significant popularity (Which it hasn’t), this would kick out the small long tail miners (The little guy) and enable centralisation of the network (Due to the increasing network throughput requirement costs and other increasing hardware costs). Lightning Network nodes are not at risk of this. Even if you have a cheap Raspberry Pie setup, you could still route payments.
In other words, extensive use of Bcash kicks out the little guy from mining thereby enabling coercion of the network. Extensive centralisation of Lightning nodes doesn’t stop a little operator from running a node. Therefore, if a centralised Lightning node is corrupted enough to attempt to censor a transaction, the payment would just be routed around through the smaller nodes. Additionally, lightning payments can be onion routed, so intermediate nodes won’t know who the sender and receiver even are. But with Bcash on-chain and centralised in the above hypothetical scenario, techniques like analytical analysis and port scanning against a public ledger can enable a centralised entity to target and censor.
Unlike other routed network examples (Like the Internet), Simple buying and selling using only Bitcoin on Lightning involves no Gateways. This means that there are no mandatory choke points that a user must converse with to send payments (Unless the receiver is stupid enough to only open a single channel, which is their own choice). The only gateways in Lightning are only required if you want to do either a submarine swap or a cross chain atomic swap (Meaning, transacting to outside the Bitcoin Lightning network). But since any node operator can choose to offer these services, the user isn’t forced to go through a mandatory route / potential choke point.
And so here[pointing to outer node] we would have our channel, we would open achannel and the idea being that we could send money to somebody onthe other end of the network as a whole. Creating hops as we gothrough the network. In this way its supposed to look pretty muchlike the Internet. You would be here in Berlin, for example, send apacket to the nearest node, and it would be routed to somebody youwanted to pay in San Francisco, for example.
There are anumber of flaws with this model, and they're not fixable. The firstobviously is that this is at an experimental stage right now, itdoesn't really work. People have lost money trying to experiment withthis on the BTC fork of the Bitcoin main network. And these peopleare still euphoric using it because it seems to solve all of theseproblems that have been plaguing Bitcoin BTC as a whole. But it isimportant to remember it is experimental and there are noadversaries.
The fourproblems: You must be online to receive funds. This is a huge stepbackward from where Bitcoin was, from where Bitcoin is. The wholeidea is that somebody should be able to send you money and you shouldnot have to worry about that address being maintained, it ismaintained by the Bitcoin network, in the network case. Here, ifyou're not online for any reason at the very moment somebody istrying to update their balance to you, you're not getting the money.
Merchants can'thave this. They can't lose out on funds just because they need toreboot a server. They can't lose out on funds just because somebodyput a shovel into a glass fibre somewhere. This is going to be areally hard sell. And it follows that no cold storage is evenpossible because all storage in the Lightning network needs to be ononline machines which have the private keys. This has "hack medude" written all over it.
If you get anoutage, you're not getting any funds. This is such a major problem,it's a show stopper for merchants. And as we saw last week, merchantsare the key group for cryptocurrency adoption. If you don't have asolution that's good for merchants, go home, that's basically it.
Submarineswaps solve this issue. Submarine swaps enable a 3rd partyto live swap say a Bitcoin transaction with a Bitcoin Lightningtransaction without any trust. If for example, a customer usingLightning wanted to spend their Bitcoin with a merchant but thatmerchant wasn’t on the Lightning network, a “submarineswapping” node holding Bitcoin both on and off the Lightningnetwork could sign a transaction sending LN-BTC to the submarinenode, and BTC from the submarine node to the recipient. This can allhappen in the background (Back-end technology) without userinteraction.
Second problem,routing is unsolved. The Lightning whitepaper is huge, it does notsolve the routing problem. And obviously if I want to send money toCharlie and Charlie is over there and we don't have a directconnection between us, then solving the routing is key to making thiswork in the first place. But it is not solved.
So 'how' you ask?Doesn't the Internet work? Obviously if the Internet works then thisis going to work. It's just a matter of copying how the Internetworks isn't it? Well that's the key here. If we can't find a way fromBerlin to San Francisco, we can't send money from Berlin to SanFrancisco. But the Internet can find packages from Berlin to SanFrancisco. So obviously there is a way, isn't it? Hence, this 'TheInternet of Money' selling point. It kind of evokes the idea thatsince the Internet works, the Lightning Network must work. Thiscompletely ignores how the Internet actually works, so I'm going toexplain that.
Mesh routing isan unsolved problem in Computer Science. In particular, mesh routingwith adversaries present in the network is an unsolved problem. Andobviously in any financial network you're going to have adversaries.You're going to have people who want to rob other people of theirmoney. We can see this in the cryptocurrency sphere, which is thickwith scammers, and frankly criminals.
Tocall mesh routing ‘unsolved’ is opinion disconnected fromreality. Ironically, Rick points to the Internet as an example of amesh network (Which clearly works), but then says it is an‘unsolved problem’. The fact that there are bad actors onthe network, does not in itself mean that the problem is unsolved.
Mesh routing inan adversarial environment is an unsolved computer science problem.So how does this work with the Internet, if this is an unsolvedproblem? How does a package get from Berlin to San Francisco. Well,frankly the Internet has not solved this problem. The way this doeswork is something called Border Gateway Protocol (BGP). It is anagreement between Internet Service Providers to all be nice to eachother and agree which packets are going where on the network.
When you'resigning up at the ISP level to connect to the Internet, you'reagreeing to be a good guy. There are no adversaries at this level onthe Internet. And as hard as it can be to wrap your head around thatidea, that's really the case, there are no bad people at this levelon the Internet. There are no bad people at the routing level, andthat's the only reason routing on the Internet works.
Sometimes peopletry bad things at the routing level of the Internet. And that's whenyou get stories like this: A Russian controlled telecom decided theywanted all the packages that went to Visa, Mastercard and Symantec togo to them. So they just published that 'Hey guys, these packages aresupposed to come to us'. And so the entire Internet started to routepackages for Visa, Mastercard and Symantec to this Russian Telecom.
This is how theproblem is not solved on the Internet. That everybody is friends witheach other. And when somebody does it wrong, which might not havebeen wrong from their perspective, but when somebody behaves in anadversarial manner, they're called out on it and disconnected fromthe network. This is not solving the routing problem. This iseverybody cooperating to carefully maintain a very delicate globalrouting table at the ISP level. This is done manually, for the mostpart, setting up routing tables.
Obviously, if youhave a financial network where people are supposed to be able to comeand ago out of their own volition, this must handle an adversarialenvironment. The routing problem is not solved. This is not animplementation detail, this is fundamental to the entire concept, andit's just being glossed over.
Ironically,Rick points to an example of a bad actor on the BGP level of theInternet (Due to their adversarial actions), but haspreviously stated that ‘there are no bad people at this level onthe Internet’. The BGP market is highly exclusive and involvesstate actors and large ISP organisations. Just because theseorganisations are large, or command great authority, does not meanthat they are benevolent without adversarial potential.
Tobelieve that all these large actors would always act in a way definedwithout adversarial intent or leanings, is naive. To understand thatit would be almost always against their best interest to behave inthis manner, is to understand the intents and natural behaviour ofthe many actors and how they interact in this system.
Theactors on this level are not necessarily friends with each other, butthey are also enemies, and sometimes even both friends and enemies atthe same time. To simply state that everyone is friends with eachother, and therefore in turn always act in friendly ways, is topresent a scenario disconnected from reality.
Hereis a video of a network map recording from when thePakistani Government attempted to block Youtube on the BGP level ofthe Internet in 2008, and screwed up. Instead ofblocking Youtube throughout Pakistan, they redirectedall planetary traffic attempting to get to Youtube into the Pakistanistate owned ISP. This in effect caused Youtube to be inaccessible tothe entire planet for a number of hours. Justbecause the Pakistani Government is stupid, doesn’t mean that theiractions weren’t adversarial.
Third: A legalliability for nodes, which is not even being discussed. Because oncewe've established that there's going to be a relatively small amountof central high availability nodes doing this routing. Kind of likethe number of ISPs in use today, if this is supposed to be working inany way, shape or form. Then you're going to see financial regulatorsjump on this. Because, remember we're depositing money into achannel, in terms of the Lightning network.
When I'm openinga channel to somebody, I'm depositing money into that channel so thatbalances can flow in both directions in order for these balances toflow to whatever endpoint in the network I decide to send money to orreceive money from.
We're opening ourchannel here, and the entire network needs to update along a selectedroute to our endpoint. This means that I'm depositing money,effectively, with this actor.
Incorrect.When you open a channel you and the other party broadcast how muchmoney you have committed to the channel. But you still have commandof your own money you put into the channel. At no point do you givethe other party custody of your money without consent.
I'm effectivelymaking my own money unavailable since it's going to be locked up inthis channel. And from a regulatory standpoint that means that I'mdepositing money with this account here. They become custodian ofthose funds.
Ahah, key word.Key regulatory word. Because once you're custodian of other people'sfunds, then all sorts of regulations kick in. In particular KYC andAML requirements. Yes, that means that those high availability nodeswill need to request identity papers from everybody who wants to opena channel with them. This cannot happen in today's regulatoryenvironment without every channel being preceded by somebody showingidentity papers to a regulated actor. That's a showstopper, that's acomplete and utter showstopper.
Aspointed out above, when you open a channel this is not ‘depositing’.The other party does not have custody of your funds. Therefore it isnot a custodial system.
Fourth, thereactive security model. You have a network where if somebody doessomething bad, the entire security model depends on other peoplebeing able to shut out the adversarial act, as in people claimingmore money than what they were entitled to and closing down thechannel. Reactive security. Reactive security in a network topologymodel where you have high availability centralised servers in a meshin between themselves and users on the edges connecting to thoseservers.
We've seen thissecurity model before. We've seen this for quite some time actuallyin a network that is still being heavily used but was created in the1990s. Its been in use since. And frankly, a lot of the Internetattacks such as Denial of Service were initially targeting thisparticular network. So we learnt a lot about security from theexistence of this network.
That network isIRC (Internet Relay Chat). Where you have a number of servers andpeople in the network can connect to any of these servers in thebigger network. Open channels with each other and forward messages tothe channel, or to the end user, being routed through this network.
This has been inexistence since 1990, for almost 30 years, and it still doesn't work.It still doesn't work in terms of the security model. It has notsolved the security problems inherent with reactive security whenloosely connected users connect to a centralised mesh of servers.
And remember now,IRC is the breeding ground for the most introverted security expertson the planet. There are no better security experts than those whohang out on IRC, and they have not solved these problems. We stillsee net splits, we see channel takeovers, we see kick bans of channelowners, and so on and so forth. 30 years into the existence of IRC.And this is the security model that's being proposed for theLightning Network. You need to solve these problems first, you reallydo, before you can claim that this is any secure, solid way ofhandling money.
Andyet, IRC still works and is still used today. Rick is correct thatboth IRC and the Lightning network are reactive security models.However, IRC networks are a server/client type network (Typical mesh) whileLightning is a P2P network (True mesh).
WithLightning, each node opens up multiple channels to adjacentnodes. These adjacent nodes in turn have channels with other nodes,and so on. Because data (payments) isrouted and not broadcast to everybody, bringing down 1 node wouldresult in payments being routed around the incapacitated node. Foran attacker attacking a Lightning node (With onion routing)they have no idea whose payments are being routed through that node.At best, they can see who has direct channels to that node. Anygiven node is essentially blind to the service that they areproviding more than 1 hop away.
Anattack on an IRC service could result in termination ofthe IRC service or other malicious attacks on users of that service.A successful attack on a Lightning node only results in the neighbournodes being terminated from the attacked node. Allother nodes would simply route their payments around theincapacitated node. Even the neighbour nodes could have multiplechannels open to other non-impacted nodes and the paymentswould simply take another route.
And then there'sa bunch of additional minor problems. We mentioned the cold storage.There's the obvious problem of needing to open a channel before youcan even start using it, which is an enormous hassle when you need aconfirmed Bitcoin transaction just to start using it. You can't justclick a button and start using it. It's not production ready, and soon and so forth.
Althoughyou need to open a channel to start using it, which takes time, thischannel can be setup to be opened for an extremely long time. Tore-fill the channel, you could do a submarine swap instead of closingand re-opening the channel.
And even afterthese four major problems, which are all showstoppers. There's onefinal observation: and that is that the market will always favoursimplicity over complex solutions to problems. And the Lightningnetwork is an extremely complex solution just trying to solve scalingof the Bitcoin network and reduction of fees on the Bitcoin network.
Therational customer prefers simplicity over complexity with regards tohow they operate and interact with the product. But the rationalcustomer does not care about the level of complexity under the hood.For example, a car is simple to operate (Steer with hands, pushpedals, etc), but a car under the hood is an extremely complexfeat of engineering. That complexity enables greater transportationcapabilities not previously possible.
The Lightningnetwork might be a brilliant solution to a very narrow problem. Butit is not a simple solution to this set of problems, which isscalability on-chain and the fee problem.
There are a lotof simpler solutions already out there to these 2 problems. Some ofwhich we need to remember. Some of which are actually Central Banksolutions. Central Banks and Commercial Banks dependent on CentralBanks are in many cases far more favourable to a merchant solutionthan what we've seen here. And for that reason, for those reasons,I'm considering the Lightning network a dead end. It is not going toget built on, it is not going to gain adoption. It is going to remaina toy that's being tinkered with and eventually left by the way side.
We saw in lastweek's video how important it is that other technologies built on topof your horizontal technology. I don't see this being built on. Andif its not being built on, it's a dead end.
Inconclusion, I find Rick’s analysis of the potential of theLightning network to be highly lacking in analysis of playerincentives (Adversarial or otherwise) within the subjectenvironment. Additionally, his network topology examples compare tonetworks with mandatory choke points (The Internet with gatewaysand IRC with centralised servers), which Lightning lacks(Lightning on Bitcoin only would be P2P single plane / true mesh). Ibelieve this lack of analysis may be responsible for his other broadassumptions made on prior YouTube videos.
Forexample, In this YouTube video he made the assumption last year thatthe miners and large businesses could simply ‘fork off’and the market would follow them.
Other Bcash supporters believedthis roughly around the time of the fork, including Craig Wright(Ironically, Craig still appears to believe this with his upcomingBcash fork).
Thisassumptive behaviour lies in contrast to my own principles andactions of the time. In contrast to either Craig or Rick, I recalltelling many people prior to the fork that ‘I did not know whatwas going to happen’. Simply because something like this hadnot happened before on Bitcoin. Now that it has, we’ve discoveredthat forks are an excellent method for market participants to go outand try their own thing. However, the market participants determinewhat product is desired.
