I was going to put out a newsletter to cover this subject, but it quickly became apparent, as I was doing the research, that there will be quite a bit of controversy, regardless of how “fair and balanced” (haha) of an approach I take.
I would therefore like to, instead, encourage discussion among the major stakeholders (project founders, collectors, and content creators / artists, alike).
The question is, how much data related to the NFT needs to be on-chain for it to be a “valid” NFT? Does a simple token suffice? Do we need metadata that can be processed and then mapped to the NFT? Does the metadata have to contain the instructions that create the art? Or do we have to have the actual image (or code that generates the exact image) on chain?
How do we classify these types of NFTs? Grade A, B, C? Type 1, 2, 3? Please discuss! This is an important issue that needs resolution as the NFT economy continues to grow rapidly, and serious money becomes involved (even moreso).
would you ever buy an Editional NFT now that this has happened? i wouldn't. the sad reality is, nearly every NFT project to date has designed their metadata+art storage in the same way Editional did, which is going to mean they suffer a similar fate on a long enough timeline.
“We [SuperRare-Pixura] see IPFS as the kind of least bad solution right now. It is open and everyone can host a node if they want, making it better than say a cloud DB. But for now we host our own node to ensure uptime since the incentivization layer (Filecoin) is still being worked on” https://twitter.com/jperkinsx/status/1197204888184381440
i do like that IPFS can potentially work with immutable solutions in the future and you recognize it is the "least bad" rather than "good enough". you guys are at least in a position to make your token art truly immutable in the future, once that incentivization layer exists or other solutions provide a way to migrate from IPFS. it's a struggle though, as a potential collector, to make a large art-nft purchase with this uncertainty.
I’m with Jim here. We can give older projects a pass as they were pioneering the technology, but we need an emphasis on future proofing NFTs, especially art NFTs. The chain needs to be the absolute source of truth. Having to depend on a creator to host the images will inevitably lead to future failures.
As for naming schema, I think Type makes the most sense. We’re not necessarily talking about a grading system here, but rather illustrating that certain types/classes will have a higher probability of longevity.
A: All the determining attributes are recorded on-chain (for example the CryptoKitties genes). Metadata can be hosted off-chain, but attributes should be an exact match with the on-chain data.
B: The metadata is off-chain, but immutable:
1. Metadata is verifiable by cryptography (for example when using IPFS)
2. Metadata URI cannot be updated via contract functions
C: The metatada is off-chain and links to a modifiable resource.
This doesn't cover questions on how to handle the actual images. Whether an image is plain for a token or can be reconstructed from attributes (compisition of different images), there is still the concern of how the images are hosted: mutable or immutable.
on-chain / immutable images need to fit into this - otherwise, pretty good classifications. on-chain/immutable art nft projects are coming. autoglyphs inspired a few different teams.
This is emerging as a a key and hot topic. There are different problems related with today's NFT, which other comments points. Bur for me the most disturbing thing is the “ERC-721 paradox.” By ERC-721 paradox I mind the fact that image (inseparable with collectible) is absolutely out of the standard (!?). If you see the standard: http://erc721.org image is absent. Images aren't in the blockchain (Autoglyph are the exception that confirms the rule)
my understanding is the team that created 721 and cryptokitties wanted to put the art on-chain, but could not come up with a feasible way to do so, and in turn probably thought it impractical to include in the standard. the standard line of thinking has been to store as little as possible on-chain, and the vast majority of eth devs i have spoken to share the same group think around images and data on-chain - "you don't do that". it's a hard hurdle to get over when the people capable of getting over it won't consider the efficient ways to do it, and instead say "no".
i am going to take the extreme end of the argument here. art+metadata=nft. and what i mean by that is the art and metadata are the nft, and without either, the nft is incomplete. by storing either of these things off-chain/in a mutable way, it makes the NFT vulnerable to being disconnected from the metadata and art, which defeats the entire purpose of art-based nifties.
I like Chavrouse classification though I really don't see a significant advantage (or reason for pricing premium) for Grade A NFT over Grade B1 if there's a 99.99999% guarantee that offchain image & metadata isn't lost over the next 100 years. Unsure what the guarantee for Filecoin will be.
To put it another way, all physical art could also disappear in some freak catastrophic event. The odds of that happening are non-null. With proper incentives we can have the odds of the same thing happening to all IPFS-hosted NFT content smaller, maybe
The closing of "Editional" makes this discussion even more urgent. They're hosting the art's meta-data, literally "for now." But how long will that last? https://medium.com/editional/sunsetting-editional-f0f3a49ffb6e
would you ever buy an Editional NFT now that this has happened? i wouldn't. the sad reality is, nearly every NFT project to date has designed their metadata+art storage in the same way Editional did, which is going to mean they suffer a similar fate on a long enough timeline.
And that is an IPFS hash stored in the metadata? Or are they running private servers for storage?
i don't know for certain but i believe they are just storing the images on a private server. most projects are just using their own storage.
“We [SuperRare-Pixura] see IPFS as the kind of least bad solution right now. It is open and everyone can host a node if they want, making it better than say a cloud DB. But for now we host our own node to ensure uptime since the incentivization layer (Filecoin) is still being worked on” https://twitter.com/jperkinsx/status/1197204888184381440
i do like that IPFS can potentially work with immutable solutions in the future and you recognize it is the "least bad" rather than "good enough". you guys are at least in a position to make your token art truly immutable in the future, once that incentivization layer exists or other solutions provide a way to migrate from IPFS. it's a struggle though, as a potential collector, to make a large art-nft purchase with this uncertainty.
It's challenging to come to a resolution right now because current tech constraints prevent blockchain from supporting various forms of visual arts.
Ironically, a considerable amount of trust is still required for users to collect & appreciate NFT art on par with their physical counterparts.
I’m with Jim here. We can give older projects a pass as they were pioneering the technology, but we need an emphasis on future proofing NFTs, especially art NFTs. The chain needs to be the absolute source of truth. Having to depend on a creator to host the images will inevitably lead to future failures.
As for naming schema, I think Type makes the most sense. We’re not necessarily talking about a grading system here, but rather illustrating that certain types/classes will have a higher probability of longevity.
So that would be "immutability grades":
A: All the determining attributes are recorded on-chain (for example the CryptoKitties genes). Metadata can be hosted off-chain, but attributes should be an exact match with the on-chain data.
B: The metadata is off-chain, but immutable:
1. Metadata is verifiable by cryptography (for example when using IPFS)
2. Metadata URI cannot be updated via contract functions
C: The metatada is off-chain and links to a modifiable resource.
This doesn't cover questions on how to handle the actual images. Whether an image is plain for a token or can be reconstructed from attributes (compisition of different images), there is still the concern of how the images are hosted: mutable or immutable.
on-chain / immutable images need to fit into this - otherwise, pretty good classifications. on-chain/immutable art nft projects are coming. autoglyphs inspired a few different teams.
This is emerging as a a key and hot topic. There are different problems related with today's NFT, which other comments points. Bur for me the most disturbing thing is the “ERC-721 paradox.” By ERC-721 paradox I mind the fact that image (inseparable with collectible) is absolutely out of the standard (!?). If you see the standard: http://erc721.org image is absent. Images aren't in the blockchain (Autoglyph are the exception that confirms the rule)
my understanding is the team that created 721 and cryptokitties wanted to put the art on-chain, but could not come up with a feasible way to do so, and in turn probably thought it impractical to include in the standard. the standard line of thinking has been to store as little as possible on-chain, and the vast majority of eth devs i have spoken to share the same group think around images and data on-chain - "you don't do that". it's a hard hurdle to get over when the people capable of getting over it won't consider the efficient ways to do it, and instead say "no".
i am going to take the extreme end of the argument here. art+metadata=nft. and what i mean by that is the art and metadata are the nft, and without either, the nft is incomplete. by storing either of these things off-chain/in a mutable way, it makes the NFT vulnerable to being disconnected from the metadata and art, which defeats the entire purpose of art-based nifties.
ICYMI this thread is related with this debate:
https://twitter.com/pet3rpan_/status/1215076733201702913
I like Chavrouse classification though I really don't see a significant advantage (or reason for pricing premium) for Grade A NFT over Grade B1 if there's a 99.99999% guarantee that offchain image & metadata isn't lost over the next 100 years. Unsure what the guarantee for Filecoin will be.
To put it another way, all physical art could also disappear in some freak catastrophic event. The odds of that happening are non-null. With proper incentives we can have the odds of the same thing happening to all IPFS-hosted NFT content smaller, maybe
Costs associated with on-chain:
https://link.medium.com/UEcw0Y5Go2