Wait, is crypto already back?
Happy Friday! It seems like we were just asking if “crypto” had perished. Now that “dangerous piece of madness that everyone hates” is stirring again. All we know is that things are about to get a lot more Glitchy. You’ve been warned!
Winter may be thawing. Look, we don’t really care about crypto-coin prices. But we can’t ignore how much power the crypto market has over the attention market. So, let’s do it: we seem to be on the verge of the next crypto hype cycle. Is winter over? Can’t say yet. But some of the crypto media headlines out there suggest that at the very least the weather is warming.
Take this one from Decrypt: Rookie Sensation Victor Wembanyama Set to Make NBA Top Shot Debut. Remember NBA Top Shot? Dapper Labs’s NBA-licensed NFT trading cards were among the stars of what crypto people call the NFT Summer of 2021. But since everything crashed, well …
Two things must be said about the 19-year-old, 7’4’’ Wembanyama (other than that he’s really tall): he is a phenomenal basketball player—and the hype around him has been stratospheric. Up until recently, that hype was all about his skills at sportsball and had nothing to do with crypto. Then, five days before the Top Shot headline, someone paid $110,120 for a Wembanyama NFT trading card from a company called Sorare—so Decrypt may have been tapping into some FOMO. Even so, a Top Shot debut making headlines at this point in the timeline is kind of wild.
Maybe Sam Bankman-Fried getting convicted of fraud was like the crypto world coughing up an extremely pretentious hairball—and now everything’s good again? Or maybe it’s been long enough since the last wave of catastrophes that “crypto” doesn’t seem so bad anymore to the mainstream consciousness. Whatever is going on, it’s mostly in the mind.
But let’s come back to Sorare for a second—the company raised $680 million in 2021 for its NFT-based fantasy sports platform. Unlike a lot of the rest of the world of NFTs, though, it didn’t crash and burn shortly thereafter. Instead, it’s been quietly gaining in popularity. Companies like Sorare are a reminder that crypto isn’t really back … because it never really left. While non-believers tuned out when the promise of quick profits dried up, true believers kept chiseling away at the underlying technologies—zero-knowledge proof systems, autonomous worlds, next-generation NFTs, new kinds of decentralized organizations, and other things we haven’t heard about yet—that will determine where it all goes next. —Mike Orcutt
Speaking of autonomous worlds: Project Glitch’s own Lucy Harley-McKeown will be on hand next week at the inaugural Autonomous Worlds Assembly in Istanbul! If you find her and say hi, maybe she’ll give you a limited edition Glitch sticker:
Why Eve Online’s creator is building an on-chain game. Eve Online is the massive multiplayer online game. Its players devote much of their real lives to it, building genuine relationships with others they meet in the vast, space-based world. Many of the folks trying to build autonomous worlds count Eve as inspiration. So it was particularly intriguing when CCP, Eve Online’s creator, announced in March that it had raised $40 million from Andreessen Horowitz to build a blockchain game.
Most of the so-called Project Awakening is still under wraps. But in a recent article at gaming news site Rock Paper Shotgun, CCP’s CEO Hilmar Veigar Pétursson shed some light on the rationale. “We live in a world where a large part of the experience is co-created by the people playing the game, but the business model is not a very inclusive setup,” he said. Whereas platforms like Roblox and Fortnight “take all the revenue and then share it back through some agreement,” Pétursson apparently envisions smart contracts that will automatically pay third-party contributors when someone uses something they created. He talked of “an ecosystem more than a nation that has its own currency.”
Sure, the blockchain may not be necessary. “I can cobble together a Frankenstein’s monster of various technologies” to devise a novel model for compensating third-party creators, Pétursson said. “I want to do something cooler than that, and we’re going to use this dangerous piece of madness that everyone hates, and I’m going to show the world you can do amazing things with it.” —Mike Orcutt
Tornado Cash takes another L in court. Back or not, crypto is still taking a beating from US regulators and federal judges. A district judge has dismissed a lawsuit that crypto and blockchain policy advocacy group Coin Center had filed against the Treasury’s Office of Foreign Assets Control (OFAC) over OFAC’s sanctioning of Ethereum-based privacy platform Tornado Cash.
OFAC has argued that its actions are legal under national security law, given that North Korea has allegedly used Tornado Cash to launder hundreds of millions in stolen crypto. Among other things, Coin Center had argued that OFAC overstepped its legal authority by sanctioning 29 of Tornado Cash’s smart contracts that are not under the control of any human controllers. That meant no one could have a “property interest” in the code, a condition Coin Center said was required to be an “entity” that OFAC could sanction. The judge didn’t buy that argument, calling it “built on a faulty premise” because the “operative language” in the relevant law is “any interest.” Tornado’s founders, DAO, and others who held TORN tokens had a clear interest in the smart contracts, making them collectively an “entity” that can be legally sanctioned, the judge said. Coin Center plans to appeal.
Meanwhile, the questions we raised before still stand: What if Tornado Cash’s founders hadn’t launched a DAO, created a token, and done all the extra stuff they did in addition to deploying smart contracts? Would judges still be able to make this argument? Then again, rhetorical questions are cheap. When econ blogger JP Koning raised similar questions on Twitter, Ameen Soleimani, an Ethereum developer who trying to build a more “compliant” version of Tornado Cash called Privacy Pools, had a sobering response: “(Y)ou realize the cost of attempting this and getting sanctioned as a result is probably prison, right?” —Mike Orcutt
ODDS/ENDS
The US Treasury’s Financial Crimes Enforcement Network (FinCEN) aims to use the Patriot Act to label crypto mixers a “primary money laundering concern.” While the specific implications aren’t clear, many have pointed out that FinCEN’s definition of “mixer” would apply to many more services than what crypto folks typically call mixers.
In the 34th annual Halloween special of The Simpsons, which aired November 5, Homer accidentally turned Bart into an NFT, leaving him trapped inside the “blocktrain.” Chaos ensued—as did a slew of obscure NFT art references that only crypto folks would recognize. Honestly, we’re still not sure what to make of it.
World App, the wallet you need to hold WorldCoin, has more than 1 million monthly active users, according to its developer Tools for Humanity. According to The Block, that would place it “among the most popular crypto wallets in the industry.”
Someone used crypto to buy a billion-year-old black diamond for $4.3 million at a Sotheby’s auction.
The US Securities and Exchange Commission said it faces challenges recruiting crypto specialists because candidates are “often unwilling to divest their crypto assets.”
Joe Biden issued an executive order on “the safe, secure, and trustworthy development and use of artificial intelligence.” The document contains much for people to talk and argue about. But this line stood out to us: “Agencies shall use available policy and technical tools, including privacy-enhancing technologies (PETs) where appropriate, to protect privacy and to combat the broader legal and societal risks — including the chilling of First Amendment rights — that result from the improper collection and use of people’s data.”
Here’s how the order defined PETs: “any software or hardware solution, technical process, technique, or other technological means of mitigating privacy risks arising from data processing, including by enhancing predictability, manageability, disassociability, storage, security, and confidentiality. These technological means may include secure multiparty computation, homomorphic encryption, zero-knowledge proofs, federated learning, secure enclaves, differential privacy, and synthetic-data-generation tools.”
In comments submitted to the US Copyright Office, Andreessen Horowitz argued that attempts to impose copyright liability on the creators of AI models will “either kill or significantly hamper their development.” That’s according to a report from Business Insider. The USCO recently posted a request for public comments on the issue to “help assess whether legislative or regulatory steps in the area are warranted.”
Meanwhile, a new tool called Nightshade gives artists the power to add a digital poison pill to their work so that if it is used in an AI training data set it will cause the model to break. “Dogs become cats, cars become cows, and so forth,” reports MIT Technology Review.
An “AI agent” developed by Nvidia Research uses large language models to create algorithms that can teach robots how to perform complex tasks. According to Nvidia, the algorithms performed better than human-written ones on 80% of the tasks, which were based on “open source dexterity benchmarks that require robotic hands to demonstrate a wide range of complex manipulation skills.”
WARNING: TECHNICAL
Name: EKILA (a reference to a customary tribal sharing framework used by the Bakaya tribe of Western Central Africa, though the paper doesn’t draw this connection directly)
Brain(s) behind it: Researchers at the University of Surrey and Adobe
What it is: In its creators’ words, EKILA is a framework that combines a “visual attribution technique” with an emerging “content provenance standard” and a “tokenized representation of rights.” All of that stuff combined is supposed to give human artists recognition and credit for their contributions to “art” created using generative artificial intelligence.
Generative AI processes massive amounts of data—images, text, and audio information—to “learn” how to create new things in response to prompts. For example, tell popular generative AI platform Midjourney, which was trained on large datasets of text and images, to create an image of a “chef cooking with a mouse on his head,” and here’s what it produces:
Now imagine if I were a kitchen equipment seller and I used this image as an advertisement. Here’s a problem: the image was only possible because (likely) the company scraped millions of images off the internet, and (probably) paid for none of them, or as few as possible. If you are an artist or a photographer, and your work finds its way into my ad, you’re shit out of luck. I’m making money from your creation, and you can’t even prove it.
Many projects and companies are trying to solve this problem, taking a range of approaches. EKILA’s approach combines an emerging cryptographic standard for identifying the provenance of images with a payment system that uses NFTs.
First, it uses a “matching method” that can attribute synthetic images (or even parts of them) to the images a model is trained on and apportions those images with relative credit. Basically, the tool scans the synthetic image, finds the region that requires attribution using something called a “patchified fingerprint model”, and scans the training library to “match” where it came from. The researchers measured EKILA’s performance across millions of images, and also in comparison to other systems, including CLIP, an approach developed by OpenAI. Here’s an image from the paper explaining how it works:
Once it has identified the source images, EKILA uses an emerging provenance standard called “C2PA” to figure out who created it, who owns its rights, and whether or not it's been edited over time. C2PA is the brainchild of companies like Adobe, Intel, Microsoft, and the BBC, who own a lot of media and want to make sure that they’ll be paid when those models show up in generative AI. They seem to imagine a future in which generative AI models would be trained using image repositories modified to comply with the CP2A standard, which cryptographically encodes all of these images with provenance data.
EKILA takes it one step further, envisioning that each CP2A-ized image should also be turned into an NFT, which can be connected to crypto wallets that would be automatically compensated each time the image is used to create a new synthetic image. The NFT’s smart contracts would be able to handle multiple layers of rights and use standards built in, the authors reason, creating a system similar to the one used in the music industry where artists can receive financial rewards depending on how they contributed to a project. —Sam Venis