When it launched almost two decades ago, ‘Second Life’ promised an environment where residents could carve out an existence without many of the restrictions of physical reality. Even gravity was optional in a world where you would simply make your avatar fly to wherever it needed to be.

Though ‘Second Life’ provided the ability to create a new virtual identity and all the looks to go with it from scratch, designing and making your own stuff is time-consuming even if it is easier to do in a virtual space where all the components are just bits of data. Many users opted to stick with 18^th-century economist Adam Smith’s invisible hand and instead of DIYing their new life bought what they needed from specialist creators. With that came an economy.

In principle, ‘Second Life’ is a near-zero marginal-cost environment. As long as the servers can support it, you can copy its digital objects at will and pass them onto others freely. What is the price of an individual object in that space? That also winds up being near zero, it seems, but not necessarily because people follow Karl Marx’s theory that we pay for human labour.

To encourage users to act as creators, Linden Labs gave them the option to disable copying. In the autumn of 2006, when the program CopyBot made it possible to duplicate just about anything, some retailers stopped selling their goods because they feared what they sold would be rendered worthless. For them, artificial scarcity was the key to staying afloat in ‘Second Life’’s economy.

Launched a few years after ‘Second Life’, the developers of the game ‘World of Warcraft’ found an economy that quickly developed around the built-in scarcity needed to prevent characters becoming too powerful too quickly, and dispatching all the monsters. Participants rich in real-world cash and poor in terms of the willingness to put in the hours needed to give their characters better weapons would simply buy them from the armies of farmers who would take on the burden of either ‘grinding’ for Warcraft gold or just stealing it from hacked accounts. Blizzard Games finally decided to sanction in-game payments through WoW Tokens in 2015. Once again, the invisible hand had been made official.

Artificial scarcity has continued to help deliver big profits even for games that are notionally free to play. Take ‘Fortnite Battle Royale’, developed by Epic Games, as an example. Despite being a combat game, the cartoon-like look and feel encourages players to treat it to show off to others. Players pay Epic for V-Bucks so they can buy skins and ‘emotes’, scripted moves and dances, for their avatars, many of which only go on sale for short periods of time. And yet, just under 70 per cent of players have coughed up for V-Bucks to acquire at least something in-game, according to a 2020 survey of Americans commissioned by US financial website Lendedu, with more than half going on avatar looks. Most understood the skins and emotes confer no advantage to the player.

The rarity value of virtual objects soon crossed over into the physical world. Designer Benoit Pagotto met his co-founders at branding company Rtfkt when they came together to create a custom pair of trainers to be given away as a prize for a ‘Fortnite’ e-sports tournament in late 2018. They styled the shoes after a skin for the game’s Raven skin, taking the black-and-purple colour scheme onto a version of the Yeezy product line, developed in a joint venture between Adidas and rapper Kanye West.

The experience convinced Pagotto and colleagues they could take designs into the virtual world, where in-game avatars could dress in their designs and pick them up in the real world, not necessarily to wear but to collect and flip to other collectors. After the ‘Fortnite’ project, the Rtfkt team took shoe designs into the virtual world and connected them to digital tokens to demonstrate to customers they had strict limits. They encouraged customers to pull images of the shoes into Snapchat to see how they might look in real life or on their own virtual avatars. That was enough to raise more than $3.1m for the virtual shoes in a matter of seven minutes.

Though the trainers were designed for the virtual world, buyers also received the right six weeks after the auction to collect a physical pair to wear in real life, or more likely display on a shelf as shoe collectors and flippers mostly do. Damien Hirst’s studio is using the same technique to determine who receives a copy of his limited-edition print The Empresses, auctioned earlier this year. Over the next three years, an owner of the digital token can opt to sell it to another buyer, who may also choose to flip it, or take delivery of the physical print.

The approach to generating electronic tokens for these sales dates to a demonstration performed by Anil Dash, CEO of developer community Glitch, and artist Kevin McCoy at the Seven-on-Seven conference in New York in May 2014. They used a derivative of the Bitcoin blockchain to record the creation of a short digital document they called a monegraph describing an animated GIF of cars created by McCoy’s wife Jennifer.

What Dash and McCoy did not have at the time was an easy way of automating the creation, sale, and transfer of what have become known as non-fungible tokens (NFTs). Blockchains like Ethereum added the ability to run a full programming language, albeit slowly and far less efficiently than if executed on a conventional computer because the programs need to be run on the same mining computers that compete to add each transaction to the blockchain’s permanent record. The overhead and need to pay miners for their processing time means so-called smart contracts tend to be as simple and small as possible. That also goes for the data an NFT is meant to represent.

Though immutability remains one of the main reasons proponents argue blockchains are the future of trade, the way NFTs are handled means that this is by no means guaranteed (see ‘Object permanence’). All that exists on the blockchain itself is an ID, a link to the smart contract code that generated it and a link to some URL off-chain that, hopefully, describes or contains the item itself.

Because you can link to just about anything, NFTs rapidly became prone to forgeries, fakes, and frauds. Last summer, by hacking into Banksy’s website and creating a page for an NFT auction that purported to be for the work ‘Great Redistribution of the Climate Change Disaster’, one scammer managed to collect more than £200,000, though they later refunded the money minus about £5,000 in transactions fees when the buyer claimed to have tracked down their Twitter account.

In a warning of the potential for scams, a hacker going by the name of Monsieur Personne demonstrated ‘sleep minting’, another possible avenue for making fake NFTs look more authentic. An NFT creator can easily add a new token to anyone’s account by specifying its ID in the original transaction before transferring it back out again. In a demonstration of the attack, Personne made it look as though Beeple (real name Mike Winkelmann), creator of the collage ‘Everydays: The First 5000 Days’, which auctioneer Christie’s sold as an NFT for $69m in February last year, had minted a new version of it and put it up for sale.

Start-up HitPiece simply went for an industrial approach to minting links as NFTs at the beginning of the year without any attempt to make them look as though they came from creators. The company listed links to thousands of tracks on the music site Spotify as NFTs, claiming that buying the tokens might result in fans being able to meet the artists. The almost immediate backlash, in which the Recording Industry Association of America called it a “scam operation”, led to the company abandoning its plan in a matter of days.

When describing the possibilities for monegraphs in 2014, McCoy asked rhetorically whether the tokens have any intrinsic value. “That is a job the market can decide,” he argued, adding: “If you buy a house, you buy a title search and insurance against title fraud. The problems that those titles have, this could have as well.”

For the moment, the community of buyers has decided links are enough, at least for the moment. And there is no NFT insurance. However, one advantage that blockchains transactions do have is one of relative transparency. Enterprising developers have built tools that detect common frauds or potential problems with NFT, such as whether a creator can alter the documents behind the scenes. Start-up Forta claims to have more than a hundred developments using its tools to create warning apps for NFT and blockchain exploits and has launched a search engine that lets potential buyers look for evidence of scams like inflating a token’s apparent value through sleep minting or wash trading, where an item returns to the original seller through a circle of connected accounts

Unless regulations do arrive in the world of NFTs, customers will have to rely on their wits and tools like this, as well as a willingness to believe that NFTs link to items of value at all. Even that might wind up being automated. Pastel, another start-up in the blockchain space, is working on techniques that use machine learning to create fingerprints of the target images in collectibles and art NFT that can be compared against each other to identify possible fakes.

In the long term, NFTs of objects sold for use in the metaverse alone may prove easier to assign a tangible value simply because the algorithms that control the in-world mechanics can enforce contracts associated with the NFTs that allocate land, clothes, or digital objet d’art to a particular participant. It remains a small market by comparison to other NFTs so far. Even so, sales for items intended for use in metaverse worlds hit the equivalent of $1bn in cumulative sales – most of them as cryptocurrency transactions – at the beginning of March.

Despite the large amounts of money making their way into NFTs, work by Chainanalysis at the end of last year found that more people own items in ‘Second Life’ – amounting to half a million – than those who own any NFTs at all, which stood at just over 350,000 at the time. And the distribution is more like that of financial wealth than of income. Most of the NFTs circulating today are owned by ‘whales’ – consumers with an appetite for collecting and with pockets to match. Just over 30,000 wallets accounted for 80 per cent of the value of the whole NFT market.

Whether NFTs make more of an impact at all on the metaverse depends on the attitudes of the operators of the different worlds and how their own much larger customer base feels, which is far from positive. On an earnings call in November, Electronic Arts CEO Andrew Wilson indicated that NFT-based trading might find its way into games such as its successful ‘The Sims’ and ‘FIFA’ properties. Players, he says, “want more digital experiences outside the game: e-sports, NFTs, broader sports consumption. And they want us to move really, really quickly.”

By February, Wilson was more cautious. “I believe that collectability will continue to be an important part of our industry and the games and experiences that we offer our players. Whether that’s part of the NFT [ecosystem] and the blockchain? Well, that remains to be seen.”

For the far smaller games company Team17 the reversal was far faster. At the end of January, the company said it was developing a spin-off NFT from its ‘Worms’ games. An immediate public backlash from its own developers and fans saw the company cancel the plans within 24 hours. In October, Valve banned games that use NFTs from its Steam platform, declaring that in-game objects should not have a real-world value. A few weeks earlier Epic CEO Tim Sweeney wrote on Twitter that NFTs were too tangled up “with an intractable mix of scams, interesting, decentralised tech foundations and scams” to be of interest. The company has since said it will allow third-party games onto its online store that do involve NFTs but has no plans to use them in its own properties.

Though there are blockchain-based virtual worlds and games such as ‘Decentraland’, which on its own is responsible for much of the metaverse trading in NFTs, many other developers may simply choose to continue operating their own storefronts and, like ‘World of Warcraft’, funnel all real-world cash transactions through them. They may continue to try to close unofficial auctioneers, as Valve did in 2018 with one site that was trading avatar skins for its ‘Counter-Strike’ game. Some may embrace the concept if they see opportunities to grab some extra revenue by taking a cut of NFT sales or by letting avatars cross over from one game or world to another. Epic has already gone partly down that road by licensing characters from Marvel comic books and bringing them into ‘Fortnite’ events.

Market analyst Newzoo said in its most recent report on gaming and metaverse trends that companies may simply add NFT-like technologies but choose to brand it differently to avoid controversy. Though scarcity seems set to remain a part of the metaverse, the NFT may ultimately prove to be a sideshow despite the apparent surge in transaction value over the past couple of years. If nothing else, the rise and possible fall of NFTs will provide economists with some more insights into why people pay for anything, whichever world they are in.

When Anil Dash and Kevin McCoy demonstrated their monegraph prototype in 2014, the creators acknowledged it had flaws that would need to be ironed out. One important problem was the need to use a link to hold the actual image or video itself. What happened if that link simply disappeared because its host website expired?

This continues to be an issue for many of the non-fungible tokens (NFTs) sitting on blockchains. Storing data on blockchains is an expensive option. It costs 20,000 gas units on Ethereum to store just 256 bits of data: 1Kbyte of data needs 32 of those blocks. The gas price varies wildly based on network usage, but you could expect to pay $1,500 just to store an average JPEG of 100KB or so even at quiet times. Not surprisingly, anything in big files gets pushed off-chain. Often the link will be to a regular webserver that as soon as its owner forgets to renew the domain can just disappear. Or a new owner may substitute the data at the end of the link for something else of their choosing, making for an expensive rickroll for some unlucky buyers.

One answer used by some issuers of art and collectible NFTs is to use the Interplanetary File System (IPFS) and similar services, which are peer-to-peer analogues of the Hypertext Transfer Protocol (HTTP) used by most web pages. Like BitTorrent, IPFS addresses are innately linked to the content and designed to have the data spread around multiple nodes for better resilience. You also cannot change the data represented by the link without changing the link address itself, making the data as immutable as the blockchain entry itself. IPFS is now one of the popular choices for NFTs, used, among others, for the Bored Ape Yacht Club, though not all its variants, as well as Damien Hirst’s The Empresses NFT deeds.

The problem for users is that they must determine how unsafe the storage chosen by the NFT issuer is, though the Looksmutable search engine set up by developer Javier Ramírez is one relatively easy way to show whether NFT data can be changed after minting.

In principle, the software routines that render the images off-chain today could be incorporated into blockchain smart contracts but would cost significant amounts of cryptocurrency whenever it was used. It is far cheaper to do it off-chain but leads to the issue that despite being promoted as a distributed application it is not entirely true.

The solution to the off-chain processing problem may arrive with layer-two blockchains, which will trade the trusted but computationally and network-intensive consensus mechanisms of core or layer-1 blockchains for more streamlined processing. Another option may be to use homomorphic encryption to protect code stored using something like IPFS but executable on any machine on a network without anyone being able to see what it is doing directly. Homomorphic encryption comes with a high hardware cost, but it may prove to have less overhead than using a layer-2 blockchain protocol.

If the fraud and exploitation of artificial scarcity were not enough to toxify the NFT space, we also have the issue of the blockchain’s enormous demand for electrical energy to contend with. Here, at least, there are some answers on the immediate horizon.

The big problem with the original blockchain protocols was that though they demonstrated you can have trust in a system that can involve numerous untrustworthy participants, that came at a high cost. The Bitcoin protocol and the many similar variants that now populate the blockchain space rely on the idea of proof of work, which demand miners compete for work by performing essentially useless but difficult calculations that favour those with the most powerful hardware. As more miners arrived wanting a slice of the action, first using expensive GPU cards and then even custom silicon tuned to the hashing algorithms in these protocols, the need to push the computational cost of mining spiralled upwards. Today we have cryptocurrency-mining warehouses with electricity bills that make indoor illicit cannabis farms look positively environmentally friendly.

The network of choice for many NFTs, Ethereum is switching to an alternative known as proof of stake, which shifts the incentives towards one based more on establishing trust for individual participants. In this system, miners become validators who pledge coins they have built up to buy tickets for a lottery with the prize of executing a transaction and receiving a fee for doing so. Validators are encouraged to not perform fraudulent transactions as, if caught, they will lose the coins they staked to buy their winning ticket. The theory is that this will be enough to convince ‘whales’, users with large holdings, to not simply buy their way into a market they can rig. The payoff in terms of energy is huge: Ethereum claims it will cut energy use for processing transactions on its network by 99.95 per cent. But at the same time, there is the risk that high-stakes participants will simply subvert the network. Like Bitcoin, there is a point at which one or a group of participants can buy effective control and, like Russia on the UN Security Council, become impossible to remove without breaking up the network itself.

There are other problems with basic proof-of-stake schemes: privacy and scalability. As Ethereum collects more transactions, trying to do everything on a single blockchain becomes ever more onerous in terms of memory overhead, which will reduce the pool of miners willing to commit resources and, in turn, lead to networks becoming dominated by a few powerful players.

This is where fundamental research into cryptography algorithms may help. Several teams are working on technology originally developed in the mid-1980s by ACM Turing Award winners Shafi Goldwasser, Silvio Micali and Charles Rackoff. Their concept of the zero-knowledge proofs makes it possible for one computer to demonstrate to another that it can solve a problem without providing the answer to the problem itself. The original concept relied on an interactive exchange of information in which the user looking for proof gradually establishes confidence in the answers to the point where they accept the other party’s claim. In the past decade, one branch of zero-knowledge proof research has developed non-interactive protocols. These are the forms that people such as Ethereum co-founder Vitalik Buterin expect to make their way into blockchain protocols.

A key advantage that zero-knowledge proofs can bring to blockchain is the ability to run programs off the main chain and incorporate the results at the end without users losing trust in the whole system. Buterin regards these layer-two protocols as fundamental to reducing transaction costs on blockchains and that greater use of crypto will be stymied if those high charges cannot be reduced. “We need to stop looking at chains as chains and start looking at them as ecosystems,” he said in recent online interview. “I think layer-2 scaling is actually better for experimentation. You don’t have to put all your eggs in one basket.”

When it launched almost two decades ago, ‘Second Life’ promised an environment where residents could carve out an existence without many of the restrictions of physical reality. Even gravity was optional in a world where you would simply make your avatar fly to wherever it needed to be.

Though ‘Second Life’ provided the ability to create a new virtual identity and all the looks to go with it from scratch, designing and making your own stuff is time-consuming even if it is easier to do in a virtual space where all the components are just bits of data. Many users opted to stick with 18^th-century economist Adam Smith’s invisible hand and instead of DIYing their new life bought what they needed from specialist creators. With that came an economy.

In principle, ‘Second Life’ is a near-zero marginal-cost environment. As long as the servers can support it, you can copy its digital objects at will and pass them onto others freely. What is the price of an individual object in that space? That also winds up being near zero, it seems, but not necessarily because people follow Karl Marx’s theory that we pay for human labour.

To encourage users to act as creators, Linden Labs gave them the option to disable copying. In the autumn of 2006, when the program CopyBot made it possible to duplicate just about anything, some retailers stopped selling their goods because they feared what they sold would be rendered worthless. For them, artificial scarcity was the key to staying afloat in ‘Second Life’’s economy.

Launched a few years after ‘Second Life’, the developers of the game ‘World of Warcraft’ found an economy that quickly developed around the built-in scarcity needed to prevent characters becoming too powerful too quickly, and dispatching all the monsters. Participants rich in real-world cash and poor in terms of the willingness to put in the hours needed to give their characters better weapons would simply buy them from the armies of farmers who would take on the burden of either ‘grinding’ for Warcraft gold or just stealing it from hacked accounts. Blizzard Games finally decided to sanction in-game payments through WoW Tokens in 2015. Once again, the invisible hand had been made official.

Artificial scarcity has continued to help deliver big profits even for games that are notionally free to play. Take ‘Fortnite Battle Royale’, developed by Epic Games, as an example. Despite being a combat game, the cartoon-like look and feel encourages players to treat it to show off to others. Players pay Epic for V-Bucks so they can buy skins and ‘emotes’, scripted moves and dances, for their avatars, many of which only go on sale for short periods of time. And yet, just under 70 per cent of players have coughed up for V-Bucks to acquire at least something in-game, according to a 2020 survey of Americans commissioned by US financial website Lendedu, with more than half going on avatar looks. Most understood the skins and emotes confer no advantage to the player.

The rarity value of virtual objects soon crossed over into the physical world. Designer Benoit Pagotto met his co-founders at branding company Rtfkt when they came together to create a custom pair of trainers to be given away as a prize for a ‘Fortnite’ e-sports tournament in late 2018. They styled the shoes after a skin for the game’s Raven skin, taking the black-and-purple colour scheme onto a version of the Yeezy product line, developed in a joint venture between Adidas and rapper Kanye West.

The experience convinced Pagotto and colleagues they could take designs into the virtual world, where in-game avatars could dress in their designs and pick them up in the real world, not necessarily to wear but to collect and flip to other collectors. After the ‘Fortnite’ project, the Rtfkt team took shoe designs into the virtual world and connected them to digital tokens to demonstrate to customers they had strict limits. They encouraged customers to pull images of the shoes into Snapchat to see how they might look in real life or on their own virtual avatars. That was enough to raise more than $3.1m for the virtual shoes in a matter of seven minutes.

Though the trainers were designed for the virtual world, buyers also received the right six weeks after the auction to collect a physical pair to wear in real life, or more likely display on a shelf as shoe collectors and flippers mostly do. Damien Hirst’s studio is using the same technique to determine who receives a copy of his limited-edition print The Empresses, auctioned earlier this year. Over the next three years, an owner of the digital token can opt to sell it to another buyer, who may also choose to flip it, or take delivery of the physical print.

The approach to generating electronic tokens for these sales dates to a demonstration performed by Anil Dash, CEO of developer community Glitch, and artist Kevin McCoy at the Seven-on-Seven conference in New York in May 2014. They used a derivative of the Bitcoin blockchain to record the creation of a short digital document they called a monegraph describing an animated GIF of cars created by McCoy’s wife Jennifer.

What Dash and McCoy did not have at the time was an easy way of automating the creation, sale, and transfer of what have become known as non-fungible tokens (NFTs). Blockchains like Ethereum added the ability to run a full programming language, albeit slowly and far less efficiently than if executed on a conventional computer because the programs need to be run on the same mining computers that compete to add each transaction to the blockchain’s permanent record. The overhead and need to pay miners for their processing time means so-called smart contracts tend to be as simple and small as possible. That also goes for the data an NFT is meant to represent.

Though immutability remains one of the main reasons proponents argue blockchains are the future of trade, the way NFTs are handled means that this is by no means guaranteed (see ‘Object permanence’). All that exists on the blockchain itself is an ID, a link to the smart contract code that generated it and a link to some URL off-chain that, hopefully, describes or contains the item itself.

Because you can link to just about anything, NFTs rapidly became prone to forgeries, fakes, and frauds. Last summer, by hacking into Banksy’s website and creating a page for an NFT auction that purported to be for the work ‘Great Redistribution of the Climate Change Disaster’, one scammer managed to collect more than £200,000, though they later refunded the money minus about £5,000 in transactions fees when the buyer claimed to have tracked down their Twitter account.

In a warning of the potential for scams, a hacker going by the name of Monsieur Personne demonstrated ‘sleep minting’, another possible avenue for making fake NFTs look more authentic. An NFT creator can easily add a new token to anyone’s account by specifying its ID in the original transaction before transferring it back out again. In a demonstration of the attack, Personne made it look as though Beeple (real name Mike Winkelmann), creator of the collage ‘Everydays: The First 5000 Days’, which auctioneer Christie’s sold as an NFT for $69m in February last year, had minted a new version of it and put it up for sale.

Start-up HitPiece simply went for an industrial approach to minting links as NFTs at the beginning of the year without any attempt to make them look as though they came from creators. The company listed links to thousands of tracks on the music site Spotify as NFTs, claiming that buying the tokens might result in fans being able to meet the artists. The almost immediate backlash, in which the Recording Industry Association of America called it a “scam operation”, led to the company abandoning its plan in a matter of days.

When describing the possibilities for monegraphs in 2014, McCoy asked rhetorically whether the tokens have any intrinsic value. “That is a job the market can decide,” he argued, adding: “If you buy a house, you buy a title search and insurance against title fraud. The problems that those titles have, this could have as well.”

For the moment, the community of buyers has decided links are enough, at least for the moment. And there is no NFT insurance. However, one advantage that blockchains transactions do have is one of relative transparency. Enterprising developers have built tools that detect common frauds or potential problems with NFT, such as whether a creator can alter the documents behind the scenes. Start-up Forta claims to have more than a hundred developments using its tools to create warning apps for NFT and blockchain exploits and has launched a search engine that lets potential buyers look for evidence of scams like inflating a token’s apparent value through sleep minting or wash trading, where an item returns to the original seller through a circle of connected accounts

Unless regulations do arrive in the world of NFTs, customers will have to rely on their wits and tools like this, as well as a willingness to believe that NFTs link to items of value at all. Even that might wind up being automated. Pastel, another start-up in the blockchain space, is working on techniques that use machine learning to create fingerprints of the target images in collectibles and art NFT that can be compared against each other to identify possible fakes.

In the long term, NFTs of objects sold for use in the metaverse alone may prove easier to assign a tangible value simply because the algorithms that control the in-world mechanics can enforce contracts associated with the NFTs that allocate land, clothes, or digital objet d’art to a particular participant. It remains a small market by comparison to other NFTs so far. Even so, sales for items intended for use in metaverse worlds hit the equivalent of $1bn in cumulative sales – most of them as cryptocurrency transactions – at the beginning of March.

Despite the large amounts of money making their way into NFTs, work by Chainanalysis at the end of last year found that more people own items in ‘Second Life’ – amounting to half a million – than those who own any NFTs at all, which stood at just over 350,000 at the time. And the distribution is more like that of financial wealth than of income. Most of the NFTs circulating today are owned by ‘whales’ – consumers with an appetite for collecting and with pockets to match. Just over 30,000 wallets accounted for 80 per cent of the value of the whole NFT market.

Whether NFTs make more of an impact at all on the metaverse depends on the attitudes of the operators of the different worlds and how their own much larger customer base feels, which is far from positive. On an earnings call in November, Electronic Arts CEO Andrew Wilson indicated that NFT-based trading might find its way into games such as its successful ‘The Sims’ and ‘FIFA’ properties. Players, he says, “want more digital experiences outside the game: e-sports, NFTs, broader sports consumption. And they want us to move really, really quickly.”

By February, Wilson was more cautious. “I believe that collectability will continue to be an important part of our industry and the games and experiences that we offer our players. Whether that’s part of the NFT [ecosystem] and the blockchain? Well, that remains to be seen.”

For the far smaller games company Team17 the reversal was far faster. At the end of January, the company said it was developing a spin-off NFT from its ‘Worms’ games. An immediate public backlash from its own developers and fans saw the company cancel the plans within 24 hours. In October, Valve banned games that use NFTs from its Steam platform, declaring that in-game objects should not have a real-world value. A few weeks earlier Epic CEO Tim Sweeney wrote on Twitter that NFTs were too tangled up “with an intractable mix of scams, interesting, decentralised tech foundations and scams” to be of interest. The company has since said it will allow third-party games onto its online store that do involve NFTs but has no plans to use them in its own properties.

Though there are blockchain-based virtual worlds and games such as ‘Decentraland’, which on its own is responsible for much of the metaverse trading in NFTs, many other developers may simply choose to continue operating their own storefronts and, like ‘World of Warcraft’, funnel all real-world cash transactions through them. They may continue to try to close unofficial auctioneers, as Valve did in 2018 with one site that was trading avatar skins for its ‘Counter-Strike’ game. Some may embrace the concept if they see opportunities to grab some extra revenue by taking a cut of NFT sales or by letting avatars cross over from one game or world to another. Epic has already gone partly down that road by licensing characters from Marvel comic books and bringing them into ‘Fortnite’ events.

Market analyst Newzoo said in its most recent report on gaming and metaverse trends that companies may simply add NFT-like technologies but choose to brand it differently to avoid controversy. Though scarcity seems set to remain a part of the metaverse, the NFT may ultimately prove to be a sideshow despite the apparent surge in transaction value over the past couple of years. If nothing else, the rise and possible fall of NFTs will provide economists with some more insights into why people pay for anything, whichever world they are in.

When Anil Dash and Kevin McCoy demonstrated their monegraph prototype in 2014, the creators acknowledged it had flaws that would need to be ironed out. One important problem was the need to use a link to hold the actual image or video itself. What happened if that link simply disappeared because its host website expired?

This continues to be an issue for many of the non-fungible tokens (NFTs) sitting on blockchains. Storing data on blockchains is an expensive option. It costs 20,000 gas units on Ethereum to store just 256 bits of data: 1Kbyte of data needs 32 of those blocks. The gas price varies wildly based on network usage, but you could expect to pay $1,500 just to store an average JPEG of 100KB or so even at quiet times. Not surprisingly, anything in big files gets pushed off-chain. Often the link will be to a regular webserver that as soon as its owner forgets to renew the domain can just disappear. Or a new owner may substitute the data at the end of the link for something else of their choosing, making for an expensive rickroll for some unlucky buyers.

One answer used by some issuers of art and collectible NFTs is to use the Interplanetary File System (IPFS) and similar services, which are peer-to-peer analogues of the Hypertext Transfer Protocol (HTTP) used by most web pages. Like BitTorrent, IPFS addresses are innately linked to the content and designed to have the data spread around multiple nodes for better resilience. You also cannot change the data represented by the link without changing the link address itself, making the data as immutable as the blockchain entry itself. IPFS is now one of the popular choices for NFTs, used, among others, for the Bored Ape Yacht Club, though not all its variants, as well as Damien Hirst’s The Empresses NFT deeds.

The problem for users is that they must determine how unsafe the storage chosen by the NFT issuer is, though the Looksmutable search engine set up by developer Javier Ramírez is one relatively easy way to show whether NFT data can be changed after minting.

In principle, the software routines that render the images off-chain today could be incorporated into blockchain smart contracts but would cost significant amounts of cryptocurrency whenever it was used. It is far cheaper to do it off-chain but leads to the issue that despite being promoted as a distributed application it is not entirely true.

The solution to the off-chain processing problem may arrive with layer-two blockchains, which will trade the trusted but computationally and network-intensive consensus mechanisms of core or layer-1 blockchains for more streamlined processing. Another option may be to use homomorphic encryption to protect code stored using something like IPFS but executable on any machine on a network without anyone being able to see what it is doing directly. Homomorphic encryption comes with a high hardware cost, but it may prove to have less overhead than using a layer-2 blockchain protocol.

If the fraud and exploitation of artificial scarcity were not enough to toxify the NFT space, we also have the issue of the blockchain’s enormous demand for electrical energy to contend with. Here, at least, there are some answers on the immediate horizon.

The big problem with the original blockchain protocols was that though they demonstrated you can have trust in a system that can involve numerous untrustworthy participants, that came at a high cost. The Bitcoin protocol and the many similar variants that now populate the blockchain space rely on the idea of proof of work, which demand miners compete for work by performing essentially useless but difficult calculations that favour those with the most powerful hardware. As more miners arrived wanting a slice of the action, first using expensive GPU cards and then even custom silicon tuned to the hashing algorithms in these protocols, the need to push the computational cost of mining spiralled upwards. Today we have cryptocurrency-mining warehouses with electricity bills that make indoor illicit cannabis farms look positively environmentally friendly.

The network of choice for many NFTs, Ethereum is switching to an alternative known as proof of stake, which shifts the incentives towards one based more on establishing trust for individual participants. In this system, miners become validators who pledge coins they have built up to buy tickets for a lottery with the prize of executing a transaction and receiving a fee for doing so. Validators are encouraged to not perform fraudulent transactions as, if caught, they will lose the coins they staked to buy their winning ticket. The theory is that this will be enough to convince ‘whales’, users with large holdings, to not simply buy their way into a market they can rig. The payoff in terms of energy is huge: Ethereum claims it will cut energy use for processing transactions on its network by 99.95 per cent. But at the same time, there is the risk that high-stakes participants will simply subvert the network. Like Bitcoin, there is a point at which one or a group of participants can buy effective control and, like Russia on the UN Security Council, become impossible to remove without breaking up the network itself.

There are other problems with basic proof-of-stake schemes: privacy and scalability. As Ethereum collects more transactions, trying to do everything on a single blockchain becomes ever more onerous in terms of memory overhead, which will reduce the pool of miners willing to commit resources and, in turn, lead to networks becoming dominated by a few powerful players.

This is where fundamental research into cryptography algorithms may help. Several teams are working on technology originally developed in the mid-1980s by ACM Turing Award winners Shafi Goldwasser, Silvio Micali and Charles Rackoff. Their concept of the zero-knowledge proofs makes it possible for one computer to demonstrate to another that it can solve a problem without providing the answer to the problem itself. The original concept relied on an interactive exchange of information in which the user looking for proof gradually establishes confidence in the answers to the point where they accept the other party’s claim. In the past decade, one branch of zero-knowledge proof research has developed non-interactive protocols. These are the forms that people such as Ethereum co-founder Vitalik Buterin expect to make their way into blockchain protocols.

A key advantage that zero-knowledge proofs can bring to blockchain is the ability to run programs off the main chain and incorporate the results at the end without users losing trust in the whole system. Buterin regards these layer-two protocols as fundamental to reducing transaction costs on blockchains and that greater use of crypto will be stymied if those high charges cannot be reduced. “We need to stop looking at chains as chains and start looking at them as ecosystems,” he said in recent online interview. “I think layer-2 scaling is actually better for experimentation. You don’t have to put all your eggs in one basket.”