https://etherscan.io/chartsync/chaindefault
You can buy a 1 TB SSD for about a hundred bucks. There are various reasons people use Infura, but inability to run a full node locally is not one of them. (The devs are working on removing some of the reasons for using Infura, by improving light clients.)
Now if you want an archive node, that will use a lot more space. But a full node already has the entire transaction history and entire current state. An archive node adds the ability to do fast queries of historical states, like "how much of this token did that address own in block number X last year."
If you have a full node, then you can generate an archive node from local data, but if you don't need fast historical queries there's no need for it.
> You can buy a 1 TB SSD for about a hundred bucks.
And Ethereum is supposedly doing just over a million transactions a day [1] And it already requires a terabyte of data to support it.
That's 1% of Alipay [2]
That's 0.02% or thereabouts of Visa [3]
How... How many terabytes will it need for those kinds of volumes?
[1] https://etherscan.io/chart/tx
[2] https://www.adyen.com/payment-methods/alipay
[3] https://usa.visa.com/dam/VCOM/download/corporate/media/visan...
https://vitalik.ca/general/2021/01/05/rollup.html
There's also work on things like clearing data that hasn't been accessed for a long time, and allowing it to be restored by supplying a Merkle branch.
In the “world computer” known as Ethereum, operations are prioritized via “gas” payments in ETH. A bunch of people you don’t know on the other side of the world might get interested in trading drawings of cats with each other[1], and suddenly the “world computer” has no resources available, and the cost of inserting simple records into the “world database”, or moving some tokens to an exchange suddenly goes from $2 to $50, or $500. This can happen right in the middle of an operation so you have to be careful to consult “gas prices” and probably the I Ching wouldn’t hurt, and then kind of make a guess and hope your transaction doesn’t get stranded.
If you have some tokens you want to get out of your wallet and you don’t have any ETH in the wallet to pay for the transaction you may need to beg, borrow, or steal some, and you’ll almost certainly be left with some dollars worth in your wallet that you may never be able to empty completely.
Imagine what it would have been like to work on a big timeshare mainframe in the sixties where booking time to do simple things cost massive sums of money, except that the bids for resources can come in from all over the world and things are being repriced constantly.
People like to call this decentralized, because no one is in control, but it actually centralizes the allocation of resources in a way that makes things many multiple times more expensive than they should be.
[1] The “world computer” is not actually trading images, because that would be impossible given its limited compute resources, so all of the hard stuff like transmitting or decoding a gif is actually happening on the normal internet and discrete computers. The “world computer” is just spending tonnes of resources recording which user currently “owns” the cat via a database entry.
Transactions getting stuck because of gas prices spiking: This was one of the main drivers of the 1559 improvement proposal which was recently accepted for the summer upgrade. Hopefully it fixes the UX around transactions and estimating gas fees.
On demand for blockspace: as the linked interview goes into, scalability has been the main point of research for the past 5 years and has come on remarkably, we should hopefully see improvements over the next few months and the rest of the year leading up to pos.
On making things more expensive than they should be: the goal is always make it as cheap as possible whilst maintaining security. Now, you may not value that security of the ledger, but others do. It will be cheaper than it is use right now, but it will always cost more than mysql. Some people value that, others don't. I would argue the constant demand for blockspace (and so the high prices) show pretty clearly there is significant demand for that security.
The decentralisation is about resilience of the system, security of the information, and transparency (but not surveillance). We cant fix money and centralised forces of money and power overnight (through id argue theres a decent number of ethereum devs working on projects around UBI etc) but we can try create systems that allow people to work together with infrastructure they know wont be pulled out from under them when they need it. There are payment networks for sex workers who get relentlessly hassled and accounts blocked by banks, there are ubi scripts that you can adapt for your own community, there are quadratic funding mechanisms that can help us fund public goods. Beyond the stigma of the hype-moon-kids hustling money and crypto-scammers that seems to get most of the media attention, there is good work being done to actually help people. CryptoKitties is silly, but it proved something. the nft hype is a bit silly now, but it is creating interesting royalty and distribution ideas that allow me to e.g. sell my music without needing a stripe account and a bank account either of which could decide they dont want we anymore and screw me over. That is valuable to me.
Too often there is a lot of hand waving around this tech and a reluctance to admit and account for the serious trade-offs. It’s hard to have a realistic discussion because there are too many people talking up their book, either ETH holders themselves or even worse, VCs
I'm definitely in the camp of it being up to those of us who work in the space to prove our point and deliver on what we've been saying. I really believe we are getting to at least delivering the tech that might make the ideas possible to explore at scale.
There will be waves of mania, price bubbles, bad crypto ideas, and the general internet scammers, but hopefully something good comes from it all. And I expect the space to be belittled and rejected until it is able to show its value, probably rightly so, we have big promises to live up to.
Though contrary to what people think about this being a weakness for ethereum, there are still ~10000 nodes running, so pretty close to the amount bitcoin has.
Its an area under active research though.
for example NFT metadata/data is often stored in IPFS & only linked to from the contract.
but confusingly some people still call data stored on IPFS "on-chain data".
It does, which is why they are looking into scaling solutions (sharding, layer 2 solutions, etc).