In late 2020, Bitcoin passes $15K. A first since December 2017, a period that coincides with the launch of most play-to-earn projects. 2021 is a huge bullrun in which anything sells, and it’s the famous golden age of NFTs. It’s also the year the famous Bored Ape Yacht Club comes out, its apes changing hands for several million dollars within a few months. And all the jokes follow:

• My kid could do that;
• 4 million for a monkey;
• I just right-click and save it, blah blah blah.

Because behind this nonsense, which looks like a Marcel Duchamp social experiment, lies a real use case: the blockchain really can certify the link between an object, a QR code, a certificate, etc., and a person, via their wallet. Enough to imagine outlets in concert ticketing or in tracking the authenticity of luxury clothing, for example. Or to dream of one day replacing the land registry (and not notaries, as some claim) with the blockchain.

In reality, there could be uses in video games… but the big publishers have no real interest in pushing them. It could start with the game licence itself: you can resell your PlayStation game at GameStop or on eBay, but not the digital versions. Likewise, money spent in one game (a character, an item, an accessory, points, etc.) is lost in the next one, and can rarely be transferred or used in another. Something the blockchain would allow, since it can bypass the proprietary systems of publishers and distributors.

And here we already hit a small problem (or detail) at Sorare.

Officially, Sorare uses the Ethereum blockchain and its NFTs use the ERC-721 standard. All fine on paper. Well, on the blockchain.

But writing something onto Ethereum is expensive. And slow. The connoisseurs will talk about Zooko’s triangle, where each corner stands for security, decentralisation and speed. You can have two, not three, or at any rate you always drift away from one. Ethereum favours security and decentralisation, so inevitably it crawls and the fees sting. It isn’t built to handle thousands of real-time micro-transactions, like those of a game where you swap cards or make moves on the market.

That’s where Layer 2 comes in. The idea is simple: you carry out the transactions elsewhere (faster, cheaper), and just record them periodically on Ethereum, a bit like a monthly bank statement. That’s what Sorare does with StarkEx, a solution (French, no less) from StarkWare.

Right, this is the rough version, so experts, no nitpicking.

StarkEx is a sort of cryptographic co-processor that runs on a technology called validium. To keep it simple:

• StarkEx executes transactions off the main chain (Ethereum), which allows thousands per second;
• It then proves cryptographically to Ethereum that these transactions are valid, using so-called STARK proofs¹;
• But unlike a rollup-type Layer 2 (such as zkRollup), the data itself is not on Ethereum: it is stored off-chain. Hence the term validium, because that chain does the validating.

So why am I frying your brain with all this?

Because it affects decentralisation and transparency. With validium, you have to trust the operator to store the data properly (here, Sorare/StarkWare). If tomorrow they shut up shop or lose the data, you could technically be unable to prove you own your card, unless you have explicitly minted it and then withdrawn it to the main Ethereum blockchain.

To limit that risk, StarkWare set up a DAC². This committee, made up of independent entities, keeps a copy of the data and can, in theory, force its publication on Ethereum if StarkWare were to fail. It’s a sort of safety net that adds a layer of trust to the system, without offering the same guarantee as a 100% on-chain solution.

So in practice:

• By default, all Sorare cards exist only in the StarkEx ecosystem, with just cryptographic proofs anchored on Ethereum;
• No card is automatically minted on Ethereum - it is up to the user to decide to “mint then withdraw” their card if they want to use it outside Sorare (on OpenSea, for example);
• This transfer to Ethereum requires paying fairly expensive gas fees, and can take a few hours;
• All in-game actions (auctions, trades, team line-ups) stay on StarkEx, which lets Sorare offer a fee-free, near-instant experience day to day.

If you look at what happens on OpenSea, you see that in reality this only happens for the rarest cards, the ones that can have a market value. To the point that right now barely a dozen cards a month change hands this way.

That said, you can still see one of the limits of the blockchain’s promise, supposedly public, transparent, traceable and blah blah blah. Validium is the opposite of all that: it’s a centralised architecture. Which leads to the problem we saw earlier, contrary to the original spirit of the blockchain (trustless): nobody knows what goes on in the kitchen, and in theory Sorare or StarkWare could block certain transactions or certain users. Whereas on Ethereum, nobody can stop you selling your NFT card, even if the platform that issued it no longer exists or has decided to ban you.

So, should we panic and scream scam? Not really. For the average user who plays Sorare for fun and maybe to win a few e-euros, these risks are largely theoretical. STARK proofs guarantee that nobody can forge your transactions, and the system works very well day to day.

On the other hand, you can legitimately ask, once again, the blockchainwashing question: couldn’t all this actually have been done with a plain old database?

The answer is probably yes. But would Sorare have been able to raise as much without blockchain?

To that question, a VC who took part in the most high-profile of the rounds answered “no”. A crypto entrepreneur answered “surely not”. And just so we’re sure, I’ll say “in my opinion, no”.