12 Feb 2022 c.e.
Who Gets What and Why, Liquidity Ads Edition

Alvin Toth's book on market design, "Who Gets What.. and Why", outlines a few aspects of market-making that I thought it would be interesting to apply to liquidity markets on lightning.

Toth outlines five things that every market design should take into consideration:

  • Market Thickness. Is everyone who wants to participate online at the same time? In the same venue?
  • Congestion: Do participants have ample opportunity to evaluate the available matches?
  • Evaluation: Of what's available, what has value to me?
  • Game Theory/Strategy: Do I need to 'game the system' to get what I want or can I express it directly?
  • Safe Participation: Am I going to be taken advantage of?

Before we can apply these, it'd be useful to have an understanding of how liquidity ads work. Let's see what we're working with.

Liquidity Advertisements

In lightning, liquidity ads are posted by a node who wants to offer liquidity. The ad contains the rate that they're charging for it, how much they expect to be reimbursed for on-chain fees, and their commitment to a cap on channel fees on the liquidity they do provide.

The ads are placed on a network-wide bulletin board system, effectively. Every node gets a copy of the ad that every other node is handing out.

To take up a node on their advertised offer, any node may attempt to open a channel with them, telling the node the advertised rates they expect to lease liquidity for and the number of sats that they'd like put into a channel on their behalf.

If the node that advertised has the funds available, they'll accept the offer and the channel will open. The funds are locked into the channel for 4032 blocks, or about 1 month. The fees for leasing the channel are placed on the advertiser's side of the channel, effectively transferring the balance from the opener to the other side. This is cool though, because the fee you pay is more inbound liquidity!

(You can read more about inbound liquidity and its importance in another post I wrote).

Now that you've got a bit of an overview of how liquidity ads work, let's see how liquidity ads stack up using Alvin Toth's five marketplace designs.

Market Thickness

Toth uses the example of a farmer's market to explain thickness. Vendors are often not allowed to begin selling until the market opens. This is to prevent buyers from arriving earlier and earlier in an attempt to get the best head of lettuce or the last carton of yogurt.

Instead, the market establishes start and end times. If you want to participate, you need to be there when it's open.

Of course, the earliest arrivals will most likely get the best head of lettuce that day. But everyone who wants to participate knows exactly what the start time is.

Liquidity ads aren't an amazing design for guaranteeing that everyone who wants to buy liquidity shows up at the same time. They do, however, provide (relatively) equal access to information about available liquidity. Any lightning node that comes online and is set up to receive gossip will receive the same information -- the marketplace itself is the lightning network. Which is to say you don't have to register with a separate service or download any extra software in order to see and take advantage of the available liquidity marketplace.

On the other hand, the information is sent out by the gossip network, which means that the first nodes to find out about new liquidity are the ones that are directly connected to the node that's advertising it. The next nodes to get the new info, are the ones that are connected to them, etc. So timing of the arrival of new offers of liquidity is a bit dependent on who's making the advertisement and where they are in the network relative to you.

Congestion

Congestion is a problem when the marketplace is too thick, or when there's too many people trying to make deals all at the same time. The issue is that, depending on how the market is designed, you may not have time to evaluate all of the available 'matches' before they've had to pair off with another. He makes a distinction between two different 'types' of good markets: commodity and matching.

In a commodity market, you offer a good for a price, and all buyers can see the offered prices. Liquidity ads works this way, as the contracts for how long of a liquidity lease you're buying has been standardized -- every lease is exactly 4032 blocks long. Sellers list their prices and channel fee caps in their advertisements -- any buyer has a rigid set of criteria they can apply to all of the provided offers.

In a matching market, you need to consider applicants for fit before making an offer. Dating for a partner or placing medical students at residency programs is a good example of a matching market. There's a time period for evaluation that's necessary before figuring out exactly which available offers is a good one to take.

Liquidity ads have a slight matching component to them as well -- as a buyer you may need to evaluate the node that's providing the liquidity ad before you can determine its value to you. For example, you might want to figure out how much existing inbound capacity it currently has, or its connection to other nodes in the graph via channels. It might take some time to figure this out -- hopefully in the future there will be 3rd party services that can help with quickly identifying good and valuable matches.

LNRouter.app is one such service that offers both stats about nodes as well as a view of the ads for liquidity that they're offering.

Evaluation of Options

This matching component of liqudity ads gets at exactly the third criteria, evaluating your options. Since it's a bulletin board system, it's fairly straightforward to look through all the available ads and then use whatever evaluation criteria you're using for inbound liquidity that day to decide what's usable for you.

This evaluation criteria however is ad hoc currently. Every buyer of liquidity needs to have a general idea of what they're looking for in a liquidity seller -- maybe you care about uptime? Or the historical liveness of the candidate node's existing channels? Or how long they've been around? It can be difficult to get this information right now, but there's a few third party services that have popped up to help fill the gap. I'm excited about more of them coming up.

Liquidity ads have a small disadvantage here compared to auction markets. A seller of liquidity doesn't have the opportunity to discover for the value of their current liquidity to the buyer -- there's no public orderbook that they can inspect to see prior liquidity buys from other nodes. A seller can see current bulletin board prices, however this doesn't tell you how big the market of buyers is (or what rate they're buying at).

This lack of an orderbook has further implications. Sellers don't have much good signal for the thickness of the market for their liquidity. Currently sellers receive offers to buy whenever the buyer makes them. They have no way of knowing if selling 1M sats to the current buyer will prevent them from having enough liquidity to sell out 5M sats to the next buyer. Designing truly decentralized marketplaces is hard for this reason -- there's no centralized sharing of previous order history which can help inform decisions about pricing and demand.

(Note that this is also a problem faced by JoinMarket. Decentralized markets are difficult to coordinate orderbooks for.)

Finally, it's hard for buyers to know which sellers still have funds available -- it's possible for an ad to be posted for a node that has no currently available liquidity, or whose available liquidity would satisfy a buyer's inbound needs. In the current market, a buyer must ask to buy liquidity in order to discover the available capacity of a seller.

Game Theory/Strategy

The fourth aspect of market design is whether expressing preferences directly is conducive to have your desires met.

In his book, Toth uses the example of the Boston public school matching systems where certain rules about grandfathering and siblings made it difficult to just list what schools you wanted and expect to get a preferred choice -- instead parents needed to know the intricacies of who else was picking certain schools to have an idea of how likely they'd be able to get into a school before listing it. This required parents to do informal polls -- whoever had the best information usually was the most successful at getting their children into the school the wanted.

Liquidity ads, by comparison, are pretty straight forward. A buyer of liquidity expresses their preference by directly petitioning the seller. Given that some sellers may have limited funds, there's probably an advantage (depending on how high of demand that liquidity is then) to being directly connected to a node that's sending out a liquidity ad for the first time, as you'll receive it sooner than other nodes in the network.

However, it's impossible to predict which nodes will be emitting them first and most lightning nodes will only exchange gossip with a limited number of peers.

One way this could be improved is for a centralized service to accept and publish liquidity ad information -- this would give nodes at far reaches or who aren't up to date on their gossip downloads an opportunity to see new advertisements as they were published, at the cost of having reliance on a centralized service (which then gets the information edge for all nodes, for itself).

Safe Participation

Finally, by participating in the liquidity ad market, what avenues are there for me to get cheated? If you offer up liquidity, are you going to be taken advantage of by the opening node? If you buy liquidity from a node, how do you know that they will leave the channel open so that you have an opportunity to use the inbound liquidity?

At the protocol level, we've tried to keep leasing liquidity straightforward by aligning a liquidity seller's incentives with the buyers. This means it's in both the buyer and seller's best interest to have the liquidity that was sold be used by the buyer (or kept available to use for the duration of the lease).

We do this by locking the output of the funds to the seller for the duration of the contract -- if the channel closes before the lease has expired, the seller won't be able to re-use the leased funds until the lease duration ends. This incentives the seller to keep the channel open since closing doesn't give them their funds back. Additionally, as long as the channel is open they'll be able to earn routing fees with the leased capital.

When opening a channel to any new node, whether as a buyer of liquidity or just a normal lightning node, it's always advisable to do some due diligence on your channel partner before committing funds to the channel. While the funds are locked into bitcoin contracts that you have the ability to withdraw at any time, there are certain classes of attacks where your channel peer can manipulate fees etc cetera to send your funds to miners. Liquidity ads expands the opportunities for deploying capital, and for nodes to request capital deployment from you (for the right price).

If you're selling or buying liquidity, it's probably a good idea to use smaller amounts for first time leases with new nodes, and to scale up your capital commitment over time as you learn more about your peer. There's some third party services[1] that allow you to see node uptime and ratings. Liquidity ads provide you with the information you need, however, to know exactly with who you're transacting before committing to the new channel.

In Exitus

Liquidity ads create a decentralized liquidity marketplace for lightning. They provide a cheap and anonymous way to find and source inbound liquidity for your node, and provide an accessible venue for any seller to list their liquidity for sale.

There's definitely room for some improvement.

It'd be great if there was a method for sellers to post executed sales to a centralized dashboard or data provider who could then provide better pricing and demand data to all market participants.

There's also room for better analytics around nodes offering liquidity and maybe a nicer way for them to prove their total available liquidity via the advertisement or as part of the node connection handshake, so that prospective buyers have a more granular way to sort available ads by amount.

Thanks to Anon1 for their help with reviewing this!

[1] LNRouter and Amboss are both great for this.

#lightning #liquidity-ads #market design
>>