Author Topic: Channels makes POS cheaper than POW  (Read 1729 times)

zack

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 384
    • View Profile
Channels makes POS cheaper than POW
« on: September 22, 2015, 12:07:00 pm »
I agree POW is a good way to distribute coins. I think funding the answering of valuable questions is another good way to distribute coins. Selling coins in exchange for food to feed developers of the blockchain is another good way to distribute.

Besides creating coins, we also need to destroy some value to maintain consensus. If we don't destroy value, then a Sybil attack could be used to exclude some transactions from the history.

The consensus used in Flying Fox is a cheaper way to secure consensus than proof of work, but you have to understand channels, which allow off-chain transactions, to understand why. Putting your money into a channel reduces network liquidity, which destroys value. The value that gets destroyed for maintaining channels can simultaneously be used to secure blockchain consensus, so we don't have to destroy value twice. We can kill 2 birds with 1 stone.

If truthcoin is going to scale, eventually you have to put it into channels. Your current blockchain consensus cost estimates make POW look cheap because you aren't including the cost of the channels. The fiat banking system needs actors who behave like channels. When you spend money internationally, you are usually giving money to someone who professionally owns piles of money around the world, and earns a profit from moving value for customers. Moving small payments off-chain is how payment networks scale.

==Why proof of stake is more expensive than pow==
Usually, taking part in the consensus process for proof of stake means putting up a bond. The bond-holders work together to make the next block. A person with 10% of bonds has 10% control over what the next block will look like.

This is expensive because the money that is in bonds cannot be used for anything else. Value is destroyed based on the interest rate.

==Why delegated proof of stake is more expensive than pow==
When you vote for someone based on how much money you have, your money is still spendable. So money isn't destroyed base on the interest rate like in POS, but look at presidential elections to see how much money can be wasted on advertising candidates. The candidates don't have to have any investment in the blockchain to hold their position.  Candidates can pay for votes. Attackers can pay candidates to attack.

==About channels==

Channels transactions have some major limitation in comparison to normal transactions.
1) There is a finite amount of money in each channel. spending money in the channel changes what percentage of the money each of the 2 participants controls. The money on the customer's side of the channel is very liquid, he can spend it almost anywhere. The money on the validator's side can only spend to one person. It is nearly unusable, so value is destroyed based on the interest rate.
2) If your partner disappears, you have to wait a delay until you can get the money out.
3) If your partner closes the channel at the wrong point in history, then you have delay amount of time to provide counter-evidence and stop your money getting stolen. If there isn't enough space in any of the blocks for you to provide your evidence, then you lose.

Usually, one of the channel participants is very rich and very well connected. He is hired for his ability to cheaply move your money to many places. He is also hired to give you extra liquidity in your channel so that you can receive payments. This person has their node running 24/7 to process payments.

The other channel participants is not well connected. He only logs on occasionally to spend money.

==Something cheaper than pow==
If we combine the channel relationship with the delegation vote, then we can be sure that the validator is well invested in the system.
The only way to receive a lot of votes is by providing lots of liquidity.

So for this POS, instead of making bonds, you make a channel. For each channel, one of the pair of participants is the validator. If the channel has twice as much money, then the validator participants has twice as much control over adding blocks to the blockchain.

This is similar to DPOS in that you can still spend your money. So, like DPOS, we aren't wasting very much value due to interest rate.

It is dis-similar from DPOS in that if you vote for someone, and they don't give you any liquidity in the channel, then you can't receive payments. So the candidate would have to pay more in order to buy votes in Flying Fox compared with DPOS.

There will be a market rate for the balance between liquidity and validating power. For example, say the rate is 1/6
If I as a customer wanted to put 1000 coins into a channel with a liquidity provider, then the liquidity provider would be willing to put 5000 coins on the other side of the channel, and neither of us pays the other.
If I needed the ability to receive more than 5000 coins, then I would have to pay the liquidity provider to provide a bigger limit.
If I didn't need to receive money, and was only planning on spending, then I would set the limit to 0, and they would pay me for giving them more validating power.

So long as the validators own more than X of the coins, then the market rate between liquidity and validating power should stay below 1/X.
« Last Edit: September 22, 2015, 04:29:08 pm by zack »

psztorc

  • Administrator
  • Sr. Member
  • *****
  • Posts: 464
    • View Profile
Re: Channels makes POS cheaper than POW
« Reply #1 on: September 23, 2015, 03:13:48 am »
I think it is a good idea, but I don't think it will be cheaper than PoW.

The (very clever) overlapping benefits of storing money (as stake "deposit" and as payment channel facilitator), won't overlap in the MC=MR sense. People will find the process of running channels to be a really great deal: channel fees + transaction fees + coinbase rewards! So huge sums of money will be poured into channel-creation, and locked there for a long period of time. One thing is that, since we don't yet have LN channels, we can't observe the replacement of (larger) transaction fees by (smaller) channel fees. My guess is that the LN effect will be independent of how the LN might be double-used for consensus.

(And of course you'll still have the Tendermint long-range Ask-a-Friend problem).

If you only intend to deploy this idea after the PoW has created the initial distribution of coins, then, I expect this to resemble what will happen anyway (LNs collecting many transaction fees).
Nullius In Verba