Thread by @DjangoWexler

Ooookay. NFTs. If you know this already mute this thread.

Context: I follow a lot of artists because I love seeing cool art in my feed, but it means I get a lot of art world discourse too, and this is a big deal there at the moment.

For the record: I am *not* a visual artist. But I *am* (by training) a software engineer.

So the chatter has been, basically, that some artists have been selling NFTs for lots of money but there's a lot of problems with them and should we do it or what?

(For those who don't want to read through the whole thread, the tl;dr is -- if you're an artist, go nuts, by all means extract money from stupid crypto people, just get cash up front and don't base your business model on this being around long.)

What is an NFT? I'll skip the detailed tech explanation of blockchain that should go here. Basically you can store small amounts of info in this distributed ledger.

For something like bitcoin that's transactions. But you can also create a token (on Ethereum for the most part) that contains a link to some piece of content and a hash of that content, which becomes an immutable unit that can be transferred between users.

THAT'S IT, and that's important. There's no magic. It's a URL and a hash to verify what's supposed to be at the URL, built into a digital "thing" that you can trade around.

The idea pushed by NFT promoters is that this "creates digital scarcity" because there can be provably few of these things and thus they can be valuable, like fine art or rare baseball cards. And crypto people have bought into this in a big way.

Someone paid $2.5m for an NFT encoding @jack's first tweet!

cnn.com/2021/03/07/tec…

Various others have jumped on the bandwagon.

So, you might be thinking, "this sounds kind of dumb."

You are 100% right! It is real, real dumb. It's actually kind of hard to express how bad an idea this is, because there are so many layers of stupidity.

Layer number one is that ownership of an NFT doesn't provide any benefits (the fancy word is "usufructs") *other* than the opportunity to hopefully sell it at a higher price. You can't do anything with your NFT that you couldn't already do, in other words.

VERY AND CRUCIALLY IMPORTANT it does not provide you with ANY KIND OF OWNERSHIP OR RIGHTS on the underlying asset.

Now, some artists have been selling rights *along with the NFT*. And that's fine! But the NFT isn't necessary or important to that process.

If I am an artist and I sell an NFT and also sign a contract saying "buyer X now hold the copyright to this work" or whatever rights I'm selling, the contract is the important part. The NFT has no legal standing except, maybe, as a kind of receipt.

A tweet that I now can't find had a good analogy: an NFT is a fancy ownership certificate, with neat printing and gilt edges. If I sell you something for real, I might provide one! If I sell you one for the Brooklyn Bridge, the gilt edging doesn't make it real.

(Caveat: the reason for the "maybe as a receipt" point is law is never cut and dried. If there was a case where X says "I never sold that!" and Y says "you totally did!" then an NFT *might* serve as evidence? We'll find out, maybe, someday! But so could a scribbled note!)

Anyway, that's layer one of stupidity. Layer two is that NFTs aren't actually scarce, because there are dozens of competing sources for them, and anyone can just make one from anything.

You can make them from tweets! Without the original tweeter's permissoin.

The people making the "rare baseball card" analogy kind of miss the point that cards aren't "rare" in an absolute sense, cards *made in a specific year by the Topps corporation* (or whoever) are rare. I can't just print a new Mickey Mantle and sell it.

In the absence of some kind of central authority for determining authenticity (and crypto haaaaates central authorities) then the scarcity is meaningless. Setting one of these up is not difficult technically -- I could do it in an afternoon.

I'll elide some of the further stupidities because they get more technical and get to the point.

This is not going to be the next big thing.

It has so many problems on top of problems that it seems unlikely it will lurch along long enough to attract the attention of regulators.

And IF IT DOES then there are so many MORE problems to deal with the whole thing will explode.

The only reason people are paying money for this is that crypto as a whole is in the euphoria stage of the bubble, a bit like internet people in late 1999.

If you're old enough to remember that, you'll remember investors throwing money at, just, the stupidest shit, with supposedly respectable industry publications intoning that this was the next great quantum shift in buying dogfood or whatever.

Because of this, I've waited until now to talk about environmental concerns. Which are ... not great.

Crypto as a whole is catastrophically bad. Every day, new crypto and fees are divided up among crypto miners, roughly on the basis of computing power.

(Yes, yes, I'm way oversimplifying.)

It's a fixed pie, so in order to get a bigger slice the crypto miners are engaged in a furious arms race of installing more and more machines to burn more and more power.

The result is that crypto now uses *more than half of all data center power usage in the world*. Literally more than Google, Amazon, Microsoft, every other tech company, and every major financial center combined.

BUT NFTs are a very small part of this space. Total NFT energy use to date has been estimated at 7.5 gigawatt-hours. Not great! But Bitcoin uses over 120 TERAwatt-hours per year.

(In other words, all of NFT, ever, uses one tenth of one percent of Bitcoin's *monthly* bill.)

Now, if NFT became the big deal its boosters claim it will, then that could rise dramatically and be a real problem. But it won't. I would be shocked if it's still a thing at the end of the year. (Though, I'm not in the business of making predictions about irrational markets.)

Sooner or later, though, reality will catch up and the NFT market will go up in smoke like Beanie Babies or Dutch tulips or a thousand other bubbles. So the environmental stuff, while bad, is not really something individuals need to fret about -- it'll resolve itself.

Of slightly more concern is that fact that a lot of these places are, basically, scams, with huge amounts of the money going to the promoter and not the artists. That's par for the course in the crypto space, unfortunately. On the plus side, the artists aren't losing rights!

Hence my earlier advice to artists:

If you can sell NFTs and people will pay, go for it. Do not sign anything giving up any rights, and you'll be fine on that front.

(Please be careful though. That's one bad scenario, a rights grab -- that there's a click-through agreement to do the NFT that includes some rights transfer, and the scammer later tries to enforce it. A potential nightmare.)

Yes, the environmental cost of crypto is high. But the crypto bros are doing it *anyway*. If we can transfer some money to people who actually create beautiful things, it's a net win.

Just make sure to get, you know, actual money and not ETH.

In the long run, it doesn't matter because there is no long run for this. (*Definitely* don't, like, mortgage the house to spin up an NFT company or something.)

Anyway. The advice is just my opinion, and everyone makes their own accommodation. The technical stuff is from watching from the sidelines, somewhat slack-jawed, as this idiocy unfolds.

(For those who might say, "Oh, but it would be so useful to have a ledger of who owns what, and maybe NFTs can get us there!" Maybe you're right, but if so we can do it without the goddamn blockchain and it'll be a million times more efficient.)

Anyway. Fin for now. If people want more technical explanation (e.g. WHY the environmental cost is so high) let me know.

ADDENDUM. This is another bad scenario for artists, obviously. These platforms are multiplying wildly and lots are scams -- it should go without saying, but obviously don't pay any fees up front either!

https://twitter.com/D_St0ckIey/status/1369723697821593604

ADDENDUM for those who care! Here's a basic sketch of "proof of work" and why it sucks.

https://twitter.com/jamari_oneal/status/1369778177560752132

The challenge the blockchain is intended to solve is "how to get many unconnected participants in distributed to agree on the 'true' version, if they may have diverging incentives."

To make this happen, the designer borrow a concept called "proof of work".

This was originally discussed in the context of anti-spam software -- the idea was to put in a slight roadblock that humans wouldn't notice but would stop robots from sending tons of spam. (We mostly went with AI filters instead.)

The idea is that you have a HASH FUNCTION -- a bit of math that you can plug in anything on one side and get 256 basically-random bits out the other side.

The input and the output are deterministic (its not actually random) but you can't predict the output from the input.

e.g. putting in the text of WAR AND PEACE gives you X, and changing one character in WAR AND PEACE changes it to Y, and X and Y are not at all similar. It's supposed to be impossible to go *backward* from the output to the input.

(caveat: I say "supposedly" because proving this is one of the great unsolved problems of computer science, related to P=NP. But that's a digression.)

So, for proof of work, I say to you: "Provide me some input that, when I run it through the hash function, produces a results that starts with 0000000."

The only way to find such an input is by trial-and-error. Since strings of leading zeros are rare, it takes lots of tries.

By adjusting how many zeros are required, you can adjust the difficulty. (More is harder.) So in the anti-spam case, I might demand this before accepting an email to cause the sender to spend, say, .1 seconds computing it -- enough to cripple a spammer.

Back to blockchain. The solution to the distributed trust problem looks like this. You have a "chain" consisting of blocks of data. For a block to be valid, it needs to start with an input that hashes to a result meeting a difficulty test.

People who add new blocks are rewarded with new coins and transaction fees. (What ETH calls "gas".) So lots of people do this trial-and-error process, and whoever lucks into a valid hash first wins and gets to add the block.

To figure out which chain is the "true" one, you just have to look at which one is the longest. Since we know that generating blocks takes lots of trial-and-error, you can't just *make up* a longer chain. Everyone is incentivized to extend the longest.

The difficulty parameter (how many leading zeroes the hash needs) is adjusted automatically so that the average time to find a new block is about 10 minutes.

This trial-and-error is the process referred to as "mining."

Now, in one sense, this is a clever hack. It does, in fact, establish distributed trust! Very nice, round of applause.

It also has a bunch of really serious problems, but we'll focus on one set: energy costs.

See, the number of new coins created per block is fixed. (With bitcoin it started at 50 and halves every so often, so it's now 6.25.) You also get some transaction fees, but those are also basically constant. So there's a fixed pot of money to be divided up every day.

It gets split up amongst the ~900 winners of this game, who happen to find the working hashes first.

This means, by installing faster computers, I can make it more likely to be *me* instead of you.

Because of the automatic difficulty adjustment, the overall rewards don't increase -- my gains are your losses and vice versa.

The result is a Red Queen's Race. I can install new hardware to get ahead only *relative* to what everyone else installs.

The limiting factor is the value of the coins the process produces. All that hardware costs money and burns electricity. All else equal, we should expect miners to build up until the cost to mine a coin is more or less equal to the sale price of the coin.

And *that* is where things get sticky. The price of a bitcoin (and most other crypto) has gone up enormously, and has *particularly* shot up since march of last year.

Thus it is more worthwhile for miners to install more and more hardware, and spend more and more power, in pursuit of the same number of coins. The higher the price goes, the worse it gets.

And it has gotten really bad. As noted, Bitcoin uses over 120 terawatts-hours of power yearly for this arms race, more than most countries. ETH uses much less because it's smaller. (~5 TWh)

As I said earlier, NFTs are only a small fraction of ETH, but the amount of energy they use per transaction is still startling.

(Note though that's just based on averages. The blockchain gets built whether the NFTs get made or not!)

Anyway, the bottom line is that blockchain uses this colossal amount of energy essentially by design -- in order to achieve distributed trust, it has to impose a big cost per block on the miners. The more valuable the coins, the higher the cost has to be.