Hacker News Reader

Comment by jeromeof | Hacker News Reader

The most expensive implementation of Conway's Game of Life ever - over $2,000 per step! (Probably the slowest too!)

Kind of amazing that a super slow, super expensive computing architecture like this has actually turned out to be very useful for some things.

Then try to calculate the energy / processing power spent per step.

Assuming some market efficiency, those fees should be fairly similar to the cost to process that transaction into a block, so, 2,000 usd

This assumes that there's no inflation I believe. But on Ethereum there is inflation.

Until EIP 1559 goes into effect in July, then it might vary between deflationary and inflationary at times.

The electricity isn't just paying for the one transaction it's paying to secure all historical transactions and all current balances.

When counting how expensive a visa transaction is, you don't also consider the price of all the bank vaults in the world and the staff protecting those vaults, because protecting existing ownership is orthogonal to enabling new transactions.

There is a point to be made, but you've made it all wrong.

You could argue that the $2000 transaction is actually validating all previous transactions... which sure. I will grant you that.

But by that reasoning, you need to amortize all future purchases as paying for part of the Game of Life transaction! If you do that then the cost of the transaction approaches INFINITY DOLLARS as time goes to infinity!

Obviously, this way of looking at the cost of the transaction is useless to the point of being meaningless. So let's stick with $2000, which has a clear meaning in reality.

The security isn't there to protect historic transactions, it's there to protect the holdings of current holders. If I had a bitcoin at one point and no longer have that bitcoin, I'm completely indifferent about the security of the transactions involving that bitcoin because it no longer impacts me.

Also, the cost per transaction - even if viewed through that lens - does not approach infinity as time goes to infinity because there would be an infinite number of transactions as well.

I honestly can’t tell which model you’re arguing in favor of.

The proof-of-work mechanism for blockchains is broken into two elements. There's the inflation element, which is a reward for miners that is independent of transaction volume, and then there's the transaction fee element.

The inflation element exists to secure the blockchain, and is proportional to the total amount of value being held on the blockchain. As the coin price goes up (independent of any transaction activity), the amount of reward for PoW mining also goes up. It doesn't make sense to count this part of the PoW reward a cost of transacting because the purpose is unrelated to transacting and also if transacting completely halts the expense does not go away.

> protecting existing ownership is orthogonal to enabling new transactions.

But for blockchain, those are the same thing. You cannot add new transactions to the blockchain without doing the work to secure it.

Yes you can. If the token price falls substantially and the fee pressure disappears you can continue transacting without adding work to secure the blockchain.

This is actually one of the largest open problems for Bitcoin. In a few years, the inflation will drop below a point where it is enough to secure the blockchain, meaning the security of Bitcoin is entirely dependent on revenue from transaction fees, which only exist if there are more transactions trying to get into blocks than there is block space to hold the transactions.

If Bitcoin doesn't have enough transaction demand, you will be able to transact on bitcoin for nearly free without contributing to the security at all, which is a very bad thing for Bitcoin as a whole.

How much would it cost on Tezos?

Reply on news.ycombinator.com