Your ownership is now on a distributed ledger, with a cryptographic hash of the content, paired with long term storage of said digital artwork. The Internet Archive's costs are ~$2/GB to store content in perpetuity, which seems insanely cheap to carve off as part of a transaction (Eth gas fees aside).

Nope, you're not missing anything. NFTs are the world's most convoluted and expensive way to store a bookmark.

You are missing something - a huge position in crypto. Like the article points out, your existing investment would benefit from all the hype that a slew of crypto-oriented services and products could give. Irrespective of whether those same services could be implemented "better" using standard centralized tech. And - amusingly - irrespective of whether those services offer products that you would ever in a million years have paid for without the novelty of crypto sprinkled on top - e.g. paying big bucks for receipts for jpgs.

(There's all kinds of problems with it, sure, but they aren't paying for the link.)

With an NFT, you don't get that. It's equivalent to your county clerk's deed registry, including the $100 filing fee, and excluding the legal machinery which gives the deed registry its value.

Sure, I’m not saying an NFT is substantively like a deed, I’m saying the link in an NFT serves a broadly similar purpose to the address in a deed.

An NFT is perhaps more akin to a certificate from one of those star name registry outfits that were popular for a while, but with less specificity as to what you supposedly bought with respect to thing it describes.

https://blockzeit.com/opensea-nft-marketplace-stops-hacker-f...

There is another example in the article - his nft was deleted from the marketplace, and nobody buying monkeys cares what is on the blockchain.

If you’re worried about the government forcing you out of your home at gunpoint, what makes you think they can’t seize a private key or force a few keystrokes?

Again, you’re seriously arguing that it’s harder for the government to take your house rather than give up your password?

Then the people with guns now have to expend resources to maintain and enforce those amendments. If they are not somehow just discarding the entire blockchain subsequent to their amendment, they're maintaining an every increasingly complex set of merges. Furthermore their amendment (very probably) isn't a cryptographic blockchain, so it's subject to all the problems that the actual blockchain list are not (forgery for example).

What makes blockchains unique is that they are the first example of these various records (ledgers, titles, etc) that physically cannot be manipulated in certain ways.

Their amendments are theirs. This is like saying that keeping your own accounting is worse for you than putting it on a blockchain, since someone might forge your own accounting books - it just makes no sense.

"They" can do just about anything they want. They can make their amendment. They can declare the blockchain null and void. They can hold a gun to your head and tell you to sell your NFT. They can even pull the trigger, in an attempt to make an example out of you for the next fool that tries to defy their authority. But the one thing they cannot do is seize your NFT without your volition. Not without breaking some of the fundamental mathematical ideas behind encryption.

Is there value in that in present day society? Maybe not. But there is undeniably something special about it.

That’s not true.

I mean, even if the access to the NFT relies solely on material in your head, there are pharmacological approaches, among others, that while not necessary reliable, can cause you to give up information without meaningfully willing it.

Even a pharmacological approach is a side channel attack which no one seems to care to distinguish between attacks on or flaws with the underlying idea. When discussing the merits of blockchain technology we are allowed to take for granted its very obvious underlying assumptions. Namely that there exists private information held by a user of the system.

No a link isn't a description of its content, just like the article demonstrated the content can change to anything, anytime, in many ways. Even if the URL contains the hash of the content like with IPFS URLs it's not a description of the content but one step better because you can check if it's pointing to the content it supposed to be.

Yes, one of the “all kinds of problems” I mentioned upthread (this one isn't an inherent problem with NFTs, but seems to be a practical one with many current NFTs) is that while NFTs certify ownership of something with regard to the linked content, exactly what that is (beyond the certificate that is the NFT itself) is often not clear, even, AFAICT, to the purchasers.

More precisely you have a certificate that says you own something (often ambiguous, though this could be precise; ambiguity is a choice in the minting of an NFT rather than a fundamental issue with the technology) relating to the content described by means of a link (the NFT may or may not include additional description of the content via metadata.)

So I suppose it is more accurate to say they own that particular citation of the url embedded in the blockchain, for certain values of own.

People that earn money on NFT don't have feeling that they miss something.

Correct, because it's clear storing the content in web2 Internet Archive is superior ("you’d have to store the NFT data on chain, which is prohibitively expensive"). They will persist regardless of web3 shenanigans, and hash addressing ensures content integrity. You could even use a torrent to store and serve the content (again, which uses hashes to identify and preserve integrity of content).

Why would one trust a distributed ledger over a centralized archive run by folks whose primary focus is on preservation of the bits they're storing? The economic benefit of running storage nodes of encrypted content is unlikely to ever be sufficient to provide the same economic incentives a corporation or non profit realizes by offering the durability a centralized service provides (due to scale).

EDIT: @Ragnarork It seems like web3 is making some promises it can't keep?

Isn't that the polar opposite of the promise of web3...?

Then when I want to interact with a centralized NFT marketplace, I can upload the preimage to their server. They'd verify the hash and store the image. I'd continue storing it myself though, so if that marketplace goes away, I can follow the same process with another one.

Obviously you'd want to keep a copy yourself, but at least you could then prove to others the file you have really is the one the creator sold, no?

No expert at these crypto things, in either sense, am I missing something?

Its common for image files to be modified, many times even automatically by the hosting service. They might compress it, remove unnecessary metadata, or add metadata for themselves. Any of that would break the hash, so you'd need to make sure any host you use to store the original absolutely never changes the file.

Then what? Well the image exists and you can verify it wasn't changed off-chain since the transaction finalized, so that's good. There's now an image publicly available online BUT a specific block chain says you own it, so that's also cool.

But wait, that hash isn't guaranteed to be unique so really anyone could make another NFT pointing to the same URL and file hash, now they also own it? And anyone could just download the file, so they own it to? And there are no legal protections for NFTs, so what was the benefit of paying to have one block chain transaction say you own it in the first place?

Which is trivial, just download the file. The place where you bought the NFT would ideally have some facility where they guarantee you can download the correct file, otherwise why buy from them?

> But wait, that hash isn't guaranteed to be unique so really anyone could make another NFT pointing to the same URL and file hash, now they also own it? And anyone could just download the file, so they own it to?

Preimage attacks are quite hard to accomplish from what I understand against modern, secure hashes. If the hash used is later broken and a preimage attack is possible then yeah you're screwed. That's a risk you take.

As for exclusive ownership, I forgot in my initial reply to add another aspect I thought about which was the license. That is, some well-defined licenses should be specified, similar to the Creative Commons stuff, and the NFS should specify one of them. Then you know if you get copyright or not etc.

Enforcement of the license would of course be similar to other digital assets, ie hard to do unless you're big, that's just the nature of digital things.

Now, just to be clear, please don't take this to mean I'm advocating NFTs. I just think the way they're currently used seems to make them completely worthless, while in theory it might be possible to make them not quite worthless.

And then what am I spelling later? A transaction immortalized in a block chain with nothing more than a broken URL and, at best, a hash of the original file?

Edit: I realize I sound a bit dickish in how I'm replying. Don't take it that way, I'm really confused at how NFTs solve anything but really appreciate the conversations here and am glad to hear differing opinions!

Not much as far as I can tell. I mean it would kinda be like a signature on a painting, in that it's a visual indication of who made it. But the proof would be in the digital, cryptographic signature.

> I'm really confused at how NFTs solve anything

I'm in the same boat. I'm just trying to figure out how they might be useful if they implemented them differently.

Anything in them can be copied trivially, so on their own they are per definition not unique hence fairly worthless.

If they're only useful when two parties agree they are worth something during an exchange, how are they different from plain cryptocoins?

I mean this is a bit similar to the GPL, it would be useless if courts declared it can't be enforced.

And what does it mean for the transaction on the block chain if both the URL and the hash no longer match? Is it worthless now and unsellable? Or do you sell it with a note that says ignore the URL, ignore the hash, or both?

I did point out other issues and that may have been unnecessary, but a hash doesn't solve the rug pull problem if the art isn't part of the encrypted and (mostly) immutable transaction block.

Not really. The hash would prevent someone to pull the rug unnoticed, but it wouldn't prevent rug pulling in the first place.

With a hash, you would be able to prove that what's currently at that url isn't what you bought, but (since hashes are by definition non-reversible) you wouldn't be able to show or see what it was you bought (unless you stored it somewhere else yourself).

Which is usually trivial.

https://www.arweave.org/

Though, I'm not sure how this will "scale".

It could be cheaper if someone can finally make a P2P network which becomes and stays popular[1] but it’ll always require more than a one-time payment. That could be donor funded (Internet Archive) but I’d be leery of assuming anything long-term unless you’re paying for it.

1. Abuse is the hard problem here: if I host a node, when the police download something illicit my IP is the one they see and I have to prove that it was done without my knowledge. This is why nobody does this except for known sources.

That's what they are trying to solve with their tokenomics model.

The value of token will appreciate over time whereas the price of storage will keep getting cheaper.

It's simpler than s3 in many aspects so I'm not sure you would need a system administrator. Everyone can run a node and things are replicated many times over. The failover model is to look for the next node. There are no API, security, access, etc consideration to be maintained at the node level.

Data itself is public by default.

> Abuse is the hard problem here: if I host a node, when the police download something illicit my IP is the one they see and I have to prove that it was done without my knowledge. This is why nobody does this except for known sources

Yeah, that's important.

That's not a given, however, and it's not just raw storage but also network bandwidth and operator time which all require regular ongoing payments. Expecting newcomers to pay for the early adopters' storage in perpetuity is tricky because you need high demand for an otherwise useless token but there's a limit on the price for most users in the form of all of the competing options, which are currently faster and more reliable.

> It's simpler than s3 in many aspects so I'm not sure you would need a system administrator. Everyone can run a node and things are replicated many times over. The failover model is to look for the next node. There are no API, security, access, etc consideration to be maintained at the node level.

It's not that simple: anyone running much storage will need to spend time replacing failed drives, managing their bandwidth relative to demand, etc. That time needs to be paid for. Massive replication is necessary to deal with the reduced node reliability but that means the network needs to pay for considerably more storage in total than, say, Amazon does and adds significant scaling issues managing all of those extra nodes with more frequent status changes.

This has been tried a number of times before and it always founders due to being slower and less reliable, with considerably more complicated software required to deal with all of those issues which the competitors don't have. It's possible that this will be more successful but I think it's really important to look at how the market pressures have consistently gone in the other direction. Amazon didn't end up with exabytes of storage in S3 because it started there — people migrated their data there because it was faster, cheaper, and easier to have it there — and that is a competitive challenge for a replacement trying to build on nodes which aren't maintained with comparable levels of service.

It fundamentally can't - you need X amounts of storage * replication factor to store X amounts of data * replication factor.

Sure you can back up an image, but the backup is worth the same as a copy of the NFT: zip. You now own a pointer on the blockchain to nothing and a jpeg on your disk. I’ve got a lot of that going on already with zero expenditure.

Similarly, you can prove to others that the version on your disk the version pointed to by the blockchain by having people check the hash.

It seems like this would be absolutely trivial to implement, right? Just... add a separator token (say `#`) and a content hash (say with `sha1:` prefix, urn-style) to the end of the URL that's already in NFTs.

I don't really understand why NFT's don't already do this. I don't understand why they didn't do it from the start. It seems an obvious choice to me in designing such a thing. Like, it's so easy, and such a step up in making NFT's do something closer to what people think they do... it leaves me thinking that the design of NFT's just wasn't done seriously, and nobody using it really cares.

What am I missing?

It’s like that crypto thought-leader on Twitter who didn’t know his NFT’d pfp was being served to various web clients over http.

It’s also why web3 startups are throwing huge cash at engineers from “web2” companies because, while they may not be crypto experts, they know how to build scalable systems, how web tech works etc. That knowledge is sorely lacking in the crypto space.

https://twitter.com/cat5749/status/1476813266462539779

A permenantly verifiable has still doesn't really solve it though. Someone can still change or remove the file later, even if you downloaded the original before you now have a transaction with a bad URL but a good hash. You can't update the transaction to change the URL, so what would that mean for anyone wanting to buy the NFT from you?

There's also the much bigger issue - say we solve the above problem as well. There are no legal protections for NFT ownership and there is nothing stopping people from just copying the artwork you own. What's the point of paying so much money for the right to kind of own a piece of art that anyone can legally copy and use?

I don't fully understand the "collector" mindset. But let's assume there are people, similar to whales in free-to-play games, that are willing to pay ridiculous large sums for what the majority would not be willing to pay anything for.

Now, think of those collectors as being willing to pay for ownership over original artwork.

The Mona Lisa itself has many replicas, you can buy prints of it, and you could probably easily find paintings of it for much cheaper. Those are all copies as well, but their monetary value is much lower, because people know they are not the original.

Now, think of photography, there are people collecting prints, sometimes of digital photography. Similarly, the 1st print is worth a lot more. Think of Vinyl records, or CD/cassette tapes for music, the worth of the 1st pressed record is a lot more, and collectors are willing to pay a lot for them.

Now think of complete digital art, that which is not even printed. Which is the "original"? Unless you were to own the HDD or the RAM stick where it was first recorded, all instances are perfect copies of the same bits. So instead, the "original" is the first person the artist publicly acknowledged as the owner of the "original". It is like the artist signing the print. This is recorded in a public ledger, that people trust and believe to be very hard to manipulate or fake. That is what an NFT is.

You might find it absurd, but is it anymore absurd than paying lots of money for the 1st print of a photo? Or the first pressed vinyl? Or the first book as signed by the artist?

The value is in people's head and emotional attachment. Someone was given by the artist themselves recognition of the piece signed in a public ledger. That's now the "original" and people assign it value.

You can think of it a bit how a lot of collectors offer public showing of their collection, the fact others can "see" the artwork for themselves isn't what make it valuable, it's the emotional knowledge around it, that of having it handed directly by the artist itself.

This is what I've understood of it at least.

Edit: Now the article still makes good point, that as it stands, some NFTs are ambiguous as to what artwork they even relate too or if they were truly created by the "artist".

I think the idea of NFTs is that you know that the original artist (Beeple or whoever) issued the NFT, they clicked the buttons and saw the same hash you see on your screen.

Like if Leonardo da Vinci sent you a cryptographically signed email with something in it indicating that you specifically owned it, you'd probably find that valuable even though it's "just pixels" and the email can be copied - the ownership is embedded in the signed email (your name or public key, let's say) and can't be copied.

I think that's the point, anyway, I still don't think I really get it...

There's probably a whole industry around recognizing a true or a fake painting. I'd say if you possessed the Mona Lisa, you might still doubt its authenticity, or find yourself in a big debate with others who claim to also possess the "true" Mona Lisa. In a way, NFTs don't (or could be made not to) have this problem. I think this is actually something that people in the market of art collecting and trading actually value. I think especially in private collections, you can claim to have sold me the original bible of Pope Pius XII for 10 million and hand me a bible that is a fake, I believe to now have the real one. And then I can go and resell it to someone else for 11 million, while you also go and sell the real one you still have for 20 million to another person, and now three people believe to all have the real one. The NFTs being in a global ledger, it would be clear who owns it truly, even if three people have a copy of the same PDF.

> But even then the vinyl is physically old and unique. I can take it out and know that it was pressed in 1972.

That's because you value the artifact. But I'd say in this case the NFT IS the artifact. The NFT is what will live on, because in 2125 (assuming the chain still exists), someone will have this token tied to their own wallet. They can know that it was minted in 2021 with the same certainty (and possibly even more certain) that it was truly minted in 2021 by the artist himself (or at least the person whose key society believes was the true artist).

Finally, if the NFT contains say an IPFS URL, or some other content describing attribute, its even more clear. You know you own the first "copy" if you want.

Let me put it some other way. I create some JPEG drawing. I then hash it and have a hash of its content. I then register my art (the JPEG) on some chain by creating an NFT for it which contains said hash (maybe in the form of an IPFS URL). At this point, the world through the public blockchain ledger knows about my JPEG art, and as the first in the chain, I prove to be the creator, or it is known that I am the creator through some other means, like posting it to my blog.

I own the NFT for my own JPEG art at this point. I can host it myself on IPFS, or maybe I just post it on my blog, or even keep it secret on my computer. Now you want to buy it from me. At that point you pay me money and I transfer the NFT to you, the ledger now says that the token started from me and was transferred to you. You now own the token that says that the IPFS hash URL or the hash of my JPEG art belongs to you and was given to you by me, the artist. I also give you a copy of the JPEG itself through whatever means, maybe you download it from my IPFS hosting, or I send it to you by email, or you download it from my blog, etc.

In the digital world, it is all copies, but only you have the token.

Ya, if the token doesn't include the content description like a hash, it's a bit fuzzy and a lot crappier, because while it would show you got some token from me the artist, its not clear which of my artwork would be the one you have, assuming in 100 years the URLs were to no longer exist for example, or to point to something else. But I think this will become the norm eventually to have the hash or use IPFS.

I agree with you, I still would prefer a physical artifact, something that you can see the wear and tear, something from an old era, maybe it doesn't even look the same, maybe bits of it are gone and forgotten. But that's just me and what I'm willing to value. If people are willing to value a digital good the same, knowing the token traces back to the original artist, and they see the value in that, then it can be worth just as much.

Any hash can match virtually unlimited number of different turd images.

And even if you trust the hash function to never be broken or brute forced with future technology, it can only verify the image, not prevent it from being deleted or altered, rendering the NFT broken and useless. Verifiably broken and useless, but still...

And why are NFT links so common, because they just seem short sighted to me and borderline dumb considering how volatile everything in the crypto space is?

Nothing about NFT's seems long term viable as they are now.

Within the smart contracts themselves is a read function for that content uri that provides all the data needed (from what I've seen, a hashed string) to generate an .svg file. But it obviously taxes the system and costs a lot more in gas fees (not to read it, that doesn't cost gas fees, but to deploy the contracts and mint), especially the more complex those are, which is why you mostly see it with 8-bit or very low-res artwork.

Cryptopunks being the most well-known (and also the most valuable) NFT project is all on-chain, and Anonymice being the most open and forked project that does this. EtherOrcs does it a little differently but is also on-chain and has completely open contracts you can refer to as well.

There's quite a few more besides this, but I don't know what percent it is, probably pretty small. Some people won't buy anything that's not entirely on-chain. But you're right that most people don't really care, they just care about the price or the image.

I've been digging through the Anonymice and EtherOrcs contracts to get a better understanding of the different approaches they took (and I still wouldn't say I completely understand it yet). It's pretty interesting, though.

[1]: https://www.larvalabs.com/blog/2021-8-18-18-0/on-chain-crypt...

[2]: https://anonymice.org/

[3]: https://etherorcs.com/

EDIT: Sorry, you only said cryptographic hash. Cryptopunks started by providing that, but then moved to entirely on-chain (so above and beyond that), where you could query and get a full SVG file or stream of pixels for any given image directly from the contract.

But it could also mean that the Internet Archive creates a special page, say, "Owned URLs", where they list a username owner for each URL that someone has payed for. If you wanted to trade your URL, the IA would get a small cut to modify the contents of that page with the new owner.

This is 1:1 equivalent to the proposed scheme, but cuts out the inefficient "mint NFT on Ethereum blockchain" step, replacing it with a simple database on the IA side.

Or those NFT contents (and the URL domain!) are guaranteed not to disappear unless many web 1.0 and 2.0 services people was paying for and went out of business?

An article about this: https://www.theverge.com/2021/3/25/22349242/nft-metadata-exp...

In fact decentralized storage being integrated into web3 is I think an element that the OP missed in his analysis.

You can attach a IPFS hash to your NFT (or w/e) while still using clients that use a more reliable gateway.

It’s almost like you want some centralised entity to preserve copies of the images these NFTs link to.

I wonder how many IPFS-backed NFTs are only being seeded on nodes run by the big players like OpenSea?

It seems a substantially less silly idea than pointing a token at a url that you don't control. I guess I'm surprised that NFTs aren't all hosted on IPFS or something like it, if only as a backup. Like, have these people not heard of linkrot?

But I guess as long as the buyers don't realize yet that their immutable ledger entry can become a dangling pointer in a puff of smoke, it doesn't matter.

I was surprised too, but only for a moment. In the end it's basically just a record that you "own" a small amount of data (url, ipfs hash, 'coin'). Unless my ownership gets me some utility (like exclusive access to the jpeg, maybe? Ability to transfer the ownership to El Salvadorian govt to pay my taxes?), I don't see how it has value

Many NFTs are hosted by NFT platforms, and also by services such as https://nft.storage/ (backed by IPFS & Filecoin). It's quite trivial though to take the IPFS CID and pin it somewhere else (local computer, a pinning service like Pinata, etc.), and anyone can do it at any time. If all you want to do is be able to prove ownership at some point in the future, you don't really need to host the content indefinitely on IPFS...just host it when you need to.

Arweave is also a one off fee to have the file mirrored forever, the hosters are paid from the yield earned on that fee.

You are correct about IPFS, it's just like torrents. There are services like Arweave, Sia and Filecoin where you can pay...

Every user of the network has to provide some storage for the network itself. If there's not enough storage to safely store your new content on the network as highly available, the network would just say sorry, can't do right now, please wait on the line while we get new storage (users).

Sure, it would need some massive network effect to work at scale, but we have now, what, billions of devices connected to Internet? That ought to be enough.

I never really understood this current "decentralized" tech. Decentralized hashes with centralized gate keepers, and mixed with "old school SPOF tech", e.x. the VPS's that store the actual content. wat.

edit: 10GB per device/user and 1 billion devices. That's 10 exabytes. https://www.wolframalpha.com/input/?i=10GB+*+1+billion

And nobody wants to participate. These projects are doomed to be extremely niche. As TFA points out, even nerds do not want to run their own servers at this point.

It could have worked in the days of casual piracy (kazaa, napster, certain private torrent sites etc had a shitton of users) if you managed to sell it as a way to do exactly that..

But getting people to install apps today to donate their bandwidth and disk space for.. what cause? Let alone when they figure out that gasp your storage may then be used for illegal material. Nah, it just doesn't work.

Then forget about the anonymization features of Freenet, and build something that ties to your Google Auth, Facebook ID, Government ID, whatever.

And let LEA access all of the content and seize/prosecute illegal content. Really not that different than storing your content on any of the cloud storage providers. With the exception that your data would be always guaranteed to be highly available, and not on just one or two centralized cloud storages.

>But getting people to install apps today to donate their bandwidth and disk space

That's just a marketing headache. ;)

It’s not that simple: if you host anyone’s content, you’re taking on personal risk (do you want to have to convince law enforcement that the pirated Disney movie or child pornography served from your home IP was served entirely without your knowledge?), giving up your resources (“Netflix is slow, turn off the mirror and see if it gets better!”), and getting slower performance/reliability (e.g. why OpenSea uses GCP instead of IPFS) immediately in the hopes that it will at some point in the future become worthwhile.

Note also that cloud storage is centralized administratively but distributed for reliability. I would give very long odds that you’re more likely to lose data through random IPFS nodes disappearing / dropping your data than on S3, and if you have to run your own geographically replicated nodes it’ll cost more in your time until you have a very large amount of data.

Statistically nobody does that, and because P2P networks need to significantly over-provision to compensate for unreliable nodes it’s hard to get anywhere close to competitive. The Linux world has the freedom ethos, no concerns about copyright/malware/etc., and still few people torrent ISOs because it’s usually slower.

This is why we can't have nice things. :D

It's true that most NFT buyers have zero idea how this works. In 2 years multiple shitty NFTs are going to turn into 404. This is fine - people will learn to only buy images that use ipfs.

I always thought IPFS just as a BitTorrent but with blockchainy tech stack.

But if it can indeed guarantee that my content would always be available, then IPFS is the answer.

Ultimately, the owner has to host it, or pay someone to host it, or hope someone else hosts it. Although nfts are small enough that any semi-popular ones may stay alive potentially forever as long as someone, somewhere, hosts it on an ipfs node. Potentially long forgotten by literally everyone alive.

This doesn’t fix any issue. If the URL changes, your NFT is worth nothing and you have no way to get the object back.