Hacker News Reader

Comment by gruez | Hacker News Reader

>That's not entirely accurate. His solution was to increase the block limit [1], and use faster computers [2].

This doesn't address the storage requirement problem. 15M transactions per day * 250 bytes per transaction = 1.369 TB per year. After a year most home users wouldn't be able to run a full node without shelling out for extra hardware. After 10 years you'll need to spend hundreds on hard drives just to get started. Sure, it's still decentralized in the sense that you can still run a node, but it'd be closer to usenet (users connecting to independent server farms) than what we have now.

View on HN

Yes, and we’re already seeing exactly what happens with bigger blocks from Ethereum. The entire thing is being run on Consensys’s centralized Infura servers on AWS. It’s not even decentralized.

This is an unavoidable physical constraint given Bitcoin’s architecture. You can either prioritize decentralization (and with it, censorship resistance), or you can prioritize TPS or in Ethereum’s case heavy-weigh smart contracts (at the cost of decentralization).

Everyone having a copy of the whole blockchain is not needed. What you need is sufficiently distributed hashpower that no actor can perform a 51% attack. That's all.

Small-time users can use thin wallets; mega-power users can download the whole blockchain and spend whatever (in perspective, not super significant) amount of money they need for sufficient storage. Users in between can carry a reduced form of the blockchain that doesn't care so much about a full transaction log but accurately describes the state of every wallet's balance (thereby functioning the same from a "verify that they're not bullshitting me" perspective).

You just described a way to decrease the security of Bitcoin. Users already decided collectively that no amount of security should be exchanged for other features inside Bitcoin. Even taproot, which is a relatively small change takes years to get to a point that it's accepted.

It's like Kosher food: if a small percentage of people want food to be kosher, it's often easer to make all food Kosher.

As Michael Saylor says, Bitcoin is good enough to store hundreds of trillions of dollars. Just don't f*ck it up.

While I agree with you on everything, and I'm all in on BTC as a store of value, I love some of the contracts that are built on top of Ethereum. Uniswap is an example system that should be implemented in Bitcoin as well over time, as it brings transparency to trading and asset ownership. It doesn't need turing completeness (so gas is not required), though as it needs state management, I think it's important to be able to account for the cost of UTXO created.

Yes Ethereum is useful and interesting as an experimental platform to see what's possible with this new tech, and what better-designed L1 protocols, programming languages, and APIs may be able to support.

But I'm not comfortable with their marketing it as a "world computer" among other things, implying it's more production-ready and robust than it really is.

> After 10 years you'll need to spend hundreds on hard drives just to get started.

Do you think that 10 years from now, a 15 TB hard drive will cost at least $200?

Maybe? Hard drive growth speeds have slowed down in the past decade or so. While there's no doubt that storage capacity growth will outpace transaction volume growth, and that eventually it would be feasible for a typical home user to run their full node (at visa's TPS). It remains to be seen whether that's 10 years away or 50+ years away.

Reply on news.ycombinator.com