I think turning Lesswrong posts into NFTs might be a good idea
headline not generated through gradient descent to create most techbro sentence would you believe
A certificate of impact is a cool idea where when someone does something good they mint a separate certificate which grants the bearer credit for doing the thing. If someone wants a good thing to happen, they can either do the thing themselves, or buy the certificate of impact for the thing.
To first order this works like a grant or a bounty - an interested grantmaker could buy the impact certificate for up to the amount they would be willing to pay to make the good thing happen. In a world where certificates of impact are commonplace, Open Philanthropyâs grant reports would be a list of purchased impact certificates along with their prices.
But because that would give you a certificate of impact that you can later sell, there are some neat second-order things, like how if I donât care much about biosecurity but you do, and my friend is doing a biosecurity project that you donât know about, I could buy a certificate of impact off my friend, funding that biosecurity project, and then sell the certificate to you.
(Aside: I could make EA Funds But Weirder⢠by getting a lot of people to pay in, buying a bunch of impact certificates, and then allocating those randomly or proportionally to donors. The certificates of impact version of an index fund?)
Anyway I was thinking about these because recently people have been talking about a thing that also has no real value other than by social convention that corresponds to an actual real thing and that you can buy and sell and thatâs used to fund the real thing. Currently theyâre mostly just being used for digital art and also tweets, but their uses are much greater. This thing is the NFT.
One exciting hint of this something greater is Erik Brynjolffson selling an NFT corresponding to his paper The Productivity Paradox of Information Technology. A scientific paper is a new piece of information that can be used in the service of optimising the world and we know that venerable Mintable user zwadia values a corresponding NFT at 1 ETH. Neat! When I saw this it was very exciting because1 it feels imported straight from dath ilan. Funding and credit assignment (*NFT ownership), identified with each other!
Specifically, NFTs give a natural way to implement certificates of impact. The creator of a piece of research mints an NFT, which is the impact certificate. The current holder of the NFT is the person who holds the certificate of impact. Since ownership of the NFT is verifiable by looking at the blockchain, we always know who to credit for funding the research. People are paying millions for NFTs already so an NFT corresponding to credit to funding isnât implausible to coordinate around as a convention.
However, getting everyone in traditional academia to fund their research with NFTs sounds difficult because NFTs are weird and complicated and scientists get most of their money from grants from universities instead of internet cryptocurrency people so theyâd have to switch their style to not optimise for universities which is strange and risky if NFTs donât pan out. 1 ETH, isnât enough for a scientific paper anyway, so itâs just a guess and a hope whether big spenders would want to get involved allowing it to work at all for anyone ever.
But for people who are doing things that I think are valuable that could plausibly get money from internet cryptocurrency people, Lesswrong seems to be up there. Also the EA Forum. And the Alignment Forum, which is what I would have put in the title if it wasnât generated using gradient descent to find the most techbro sentence. Lesswrong has nice neatly deliminated artefacts that correspond to both good things, that is the Lesswrong post, and hence one could make NFTs out of these.
Alignment Forum posters seem to already be kind of optimising for being funded by internet cryptocurrency people: the first box on MIRIâs donation page is for âdigital currencyâ with debit card payments languishing in the distant second box, and Vitalik Buterin (inventor of ETH) is the third-highest MIRI donor at $1,033,057, so if they adapted their writing style to impress internet cryptocurrency people thereâd probably be less bridges burned with their normal sources of funding. This is the central use because each Alignment Forum post corresponds to a new piece of work in AI safety so somebody interested in reducing x-risk from unfriendly AI would hopefully be in a good place if they searched for impact certificates to buy there.
Lesswrong posts are mostly done pro bono so any funding raised from impact certificate sales is just purely extra money to the author of the post in order to incentivise making good posts. So at least for Lesswrong it would be even more costless for someone to try to make their posts the kind that people would want to buy impact certificates for, that is to say basically free.
There are some ways that you might want to do this. Firstly, the NFTs would have to be a proof of stake cryptocurrency, because proof of stake is so obviously better that it would massively discredit the entire enterprise if one did something as foolish as implement your LW/AF/EAF impact certificates NFTs on a proof of work blockchain. Seriously, look at the gas prices! Why ever subject yourself to that? Tezos is the first proof of stake cryptocurrency I heard of so because of network effects and the difficulty of coordinating a switch to a different cryptocurrency, even if that cryptocurrency might be better, letâs just talk about Tezos.
The easiest way would probably be to have the impact certificates posted on an existing platform, a list of which is nicely compiled here. Hic et Nunc is one of these (chosen because itâs at the top of the list I linked), and it supports NFTs that correspond to HTML documents, so if a Lesswrong poster wanted to they could, right now, save their Lesswrong post as a HTML file and upload it there and sell it. Iâve half a mind to do this myself but my Coinbase account still isnât verified so it would be mildly inconvenient to get hold of any Tezos and also if uploading .html files to Hic et Nunc turns out to not be a good way to handle impact certificates I wouldnât want to either lock in a bad implementation of my idea or have the impact certificate that I sell to be discounted by the chance that this method of handling impact certificates becomes a widely used method.
To expand on this there could be some kind of Lesswrong authentication thing going on. Having a frontend or browser addon that lets you log into Lesswrong and mint the impact certificates on Hic et Nunc from your Lesswrong account would help one verify the identity of the minter and ensure that one impact certificate created with the frontend corresponds to one Lesswrong post and vice versa, which would help cement the This Token = The Impact Certificate idea that might otherwise make an implementation useless as impact certificates and maybe even fade away to unuse before anyone uses it.
To expand on this there could be a totally different LWIMPACTCERT token with its own NFT smart contract and everything on the Tezos blockchain. Then I can simply define a social convention such that LWIMPACTCERT token, regardless of the social conventions around other types of NFT, is definitely for sure 100% the impact certificate. I donât know much about what smart contracts can do, but the metadata attached to a LWIMPACTCERT could be specialised to correspond exactly to a Lesswrong post (e.g. they could have an ID thatâs equal to the ID in Lesswrongâs database), and maybe even have checks for identity of the minter and uniqueness in the contract itself? I tried looking into smart contracts a few weeks ago but accidentally my laptopâs boot partition (everything was fine) trying to install Linux and gave up on it. Maybe something to pick up later.
Anyway, I just think that making NFTs that correspond to Lesswrong, Alignment Forum and EA Forum posts, that are understood to give the holder credit for funding the creation of the posts, would be a good idea.
except for how dath ilan doesnât really have âscientific publishingâ (having suggestors and replicators) or âcheap GPUsâ (machine-learning based AI is unalignable) or âproof of work cryptocurrenciesâ (they just all used proof of stake instead)