this post was submitted on 18 Nov 2023
1 points (100.0% liked)
Ethereum
5 readers
1 users here now
Resources
- Website & Blog
- White Paper & Yellow Paper
- Documentation & Stack Exchange
- Learn Solidity
- Source Code on Github
- Bounty program
- Chat on Gitter
- Network Status & Gas Price Market
- List of DApps
- Meetups
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
It poses an interesting question: What happens if everyone starts using Ethereum as a data archive service?
One interesting aspect of this, and this applies to most blockchains, is that you pay for the data a single time upfront but you expect it to remain there forever.
Does this make any sense to you at scale? Does this at all sound sustainable? Because it isn't. It's a tragedy of the commons problem. It can even get worse if a naive pricing mechanism is added for rent, where you rely on hardcoded parameters to gauge the price of storage in a fluctuating market.
You only solve this issue with a market mechanism, something fundamental. Automatic Transaction Rebroadcasting coupled with a finite ledger is a good solution: basically, all transactions have a set lifespan and at the end they must be rebroadcast, paying the average fee/byte new txs paid over their lifespan.
If you fund it well enough (or if it's on Saito and your staking rewards offset your rent) you can have data that lasts forever on a chain with a sustainable economic model.
It might seem like a downgrade because you don't get to pay once and live forever - but consider that paying once and living forever is a straight road to economic ruin and is simply pushing costs onto future users, thus continually removing users from the chain until it dies.
If you actually can pay for storage at a market rate, the chance that your data survives hundreds of years is far greater.
You're forgetting that the cost of storing data has always decreased. By having the revolutionary upfront price for data storage, it's a huge bet on this trend.
The price of ETH goes to the moon. ETH price is determined not only by speculation but also by actual demand to pay to put data on chain. If it becomes commonplace to store data on chain, it will greatly increase demand for ETH.
I've made https://clonk.me to have conversations and share files on chain as NFTs and https://newgeocities.com to upload web pages on chain.
It doesn't matter if Eth price goes to the moon, the economic model is fundamentally flawed.
> Users start dumping data on-chain that must remain forever
> Eth price goes to the moon
Where in this equation are nodes being compensated for storing data? The fees that users pay to *add* data do not count -- that's a completely different cost than storing data. The nodes that join any time in the future are burdened **without payment** for all data that came before - the payment went to the nodes that added it and, apparently, Eth holders, if your solutions is 'Eth price goes to moon.'
If your argument is that people won't be able to add such data because it's too expensive -- just mete it out. Add it during low demand. It really doesn't matter; nodes get paid to *add* data, nobody gets paid to *keep* data; at least they aren't paid anything like a market rate. The solution now is just let fees increase to compensate the node cost, meaning new users pay for old uers - a truly awful economic model.
Anticipating the next argument: "the chain will just stay small enough so that storage costs decrease more quickly than archive costs." You are now a slightly better version of small-block Bitcoin (same economic model: don't allow arbitrary scaling) i.e. you have economic parameters set by developers and not the free market.
Automatic Transaction Rebroadcasting solves it. It gives you market prices for storage and pays that money to nodes actually storing it.
That’s a very fair take. I’ve always thought one of the best qualities of crypto is its potential use as like a digital proof of history.
But I never really considered the continuous cost of keeping data around forever.
Would either one of the next upgrades for the network help solve this issue by allowing some data to not be kept forever? I forget if it’s blob space or danksharing but one of these allows for data to be removed after some time right?
Then what are the chances that our need for data storage does not grow as fast as our data storage capabilities? That seems to be the trend over the years with most technology. So if our capacity is increasing and ethereum is becoming more efficient with the data it stores is there a case it is sustainable to keep this data on the network for a very very long time? That sounds realistic to me.
Many protocols take the easy route and just say storage will get cheaper as more data is placed on-chain.
The problem is that this isn't a market pricing mechanism - it's developers forecasting prices and hard-coding variables based on those estimations. Not only does this put huge economic power and continuous reliance on developers, but it will never be as accurate as the free market.
When you start guessing prices instead of letting the market decide, you get problems on either side of the error. Over-price and users are paying nodes extra for no extra utility, under-price and nodes will not provide utility at all (unless centrally subsidized, like Infura is now).
The only way to get consistently low fees is the let the market price storage. Now when storage prices go down the cost for using the blockchain will immediately follow. Actually, it's much better than that...
Because nodes are rewarded for storing data more efficiently, they actually earn profits they can use to research and develop even more efficient forms of storage - so they will be enhancing that trend instead of merely relying on it.
You don't get any of this when you don't pay nodes the market rate for storage and just hope that adoption will happen more slowly than storage technology happens to become cheaper.
It's incredibly inefficient as a data archive. It's also incredibly expensive. Why would you spend $5000 to store some data on blockchain that would cost $0.01 using other methods? And there is no guarantee of the permanence of blockchain. At some point, the size of blockchain will become so large, it will be a liability hosting such decentralized databases.
Yes but if the promise of permanence doesn't offset that cost then there is a fundamental problem somewhere: blockchain is supposed to be permanent, no?
If it's expensive to store data, then this is a problem even for a blockchain's normal everyday use. You are correct to generalize, I simply took the starting point OP suggested.
But it is a more general issue around pricing all storage. Automatic Transaction Rebroadcasting is a general solution - so when you zoom out and abstract away the single examples, it remains the only sustainable direction to keep fees low and let the blockchain survive forever without intervention.
There's no guarantee blockchain will be permanent. In fact, because it's so expensive and inefficient to maintain the blockchain, it's highly likely it will not be permanent. There are 30,000 blockchains that are already dead because they were not economically viable to maintain.
And right now, both BTC and ETH are not economically viable to maintain. The only way they remain online is because a tremendous amount of resources are being dedicated to marketing their adoption -- resources that aren't getting paid back, resources that only see a return if you can continue to recruit more people to buy in higher. But after 15 years, blockchain still hasn't proven to be uniquely good at anything in particular. So it's not looking like it will be around much longer.
Look at all the crypto exchanges that have collapsed. These are not isolated incidents, despite everybody wanting to claim they are. The sad reality is there is no way to operate an ethical, legal exchange because the market doesn't really create any substantive value. The only value being injected into the market is from dark markets and fraud. While that seems like it might be a limitless resource, it's not really, and various authorities are clamping down on the fraud.
I have no idea what you mean by "automatic transaction rebroadcasting." This just seems like another meaningless buzzword.
We have existing systems that are part of solid business models that are self sustaining that can ensure their servers and databases are online and operational in perpetuity. Crypto/Blockchain doesn't do that. It's not self-sustaining right now and there's no sign it ever will.
I get that you guys are sitting on top of tokens you hope will go up in value -- so you want to believe I'm wrong. Your "faith" in crypto is not backed up by real world evidence.
You are only proving my point.
The reason I bring up ATR is because it actually fixes a massive portion of the economic model of blockchain that makes it inefficient and unsustainable.
You can cast the pejorative "buzzword" on anything - what do you expect, new solutions won't get new names? At some point you have to dig a little deeper than the fact that something has a name you haven't heard of before.
Anyways, and to your point: blockchain can't be expected to survive if the people who get paid for running it add zero value back into it. We say we are paying security, and we are, but that simply isn't good enough. If you can't pay security out of the same funds as infrastructure, then the infrastructure fails without central intervention and your objective security cost is pointless because you can find equivalent vectors on the monopolized infrastructure.
ATR takes money that used to go towards hoarding wealth or crunching numbers and forces nodes to hold data - all data, that should be on chain. It solves the problem you are keen to point out - that infrastructure will collapse even if the chain promotes itself as permanent. ATR gives you economic guarantees of what is only ideological in current chains.
If you're not too sensitive to new terms, look into Saito. It solves the other problems of sustainability and infrastructure centralization (as an attack and monopolization vector -- not to mention scale bottleneck).