Whoami? I’m unique, dude (or why blockchain is useful for provenance)
A review of the critiques made about using blockchain for proving provenance and authenticity
This post is an answer to a blockchain critique from Sebastien Meunier. I think it’s good when people try to take some distance from hype. Well, except when they’re saying something obviously wrong for the sake of making an argument.
Blockchain bashing has become a sport, but let’s be fair: it needs more convincing use cases, beyond the bitcoin world. Too many ICOs have led to nothing or are purely and simply scams. Many projects have simply failed (for instance, RChain).
So more scrutiny is required and blockchain should only be taken as a mean, not as a goal. But some sectors, such as supply chain, are generally considered as potential targets. The intuition to use a blockchain lies in the immutability of the underlying ledger, which can therefore be used as an audit trace during the life cycle of goods (from the manufacturer, to a processor, a transporter, a retailer, one or several users up to destruction). As such, people have started to design systems to fight a pressing problem: counterfeiting, representing an estimated loss of 500.000 jobs in the EU alone. It also represents a risk for safely and health, especially for medicines sold online for instance.
Sebastien Meunier declares blockchain is not a good fit, obviously (notice the obvious part). He has generally made his mission of showing the inability of blockchain to provide value in pretty much any domain. Yet, most of the arguments are just wrong, as we’ll see.
Summarizing the critique
The original article says:
The author doesn’t see any added value into the current solutions to track physical goods, such as VeChain Foundation, Provenance, Chronicled Staff, BlockVerify or Arianee. The only exception he cites is DUST Identity, which has an interesting technology based on nanodiamonds spread over the surface of the object to make it unique.
But apart from that, a good deal of cheat & lie (which I will call reason f).
What is true about the critique
Let’s start with the wine (reason f), as it gives a real life example everyone can understand easily. Again the problem is huge. Being French, you can imagine how pressing this is an issue ;-)
Having traveled quite extensively in Asia, I can testity that it’s sometimes difficult to find an authentic bottle of Champaign, even in good restaurants. So finding a way to trace back to the wine fabric would seem like a good idea.
So the idea is that the end user could authenticate a wine, by scanning a QRCode or a NFC tag afixed to the bottle.
Alas, that’s not easy.
Issue 1: the oracle problem
Linking blockchains to the real world is arguably a difficult task (even theorically), known as the oracle problem (see for instance, chainlink or witnet or this blog post for a general overview of why that’s difficult).
Issue 2: physical protections (such as QRcode or NFC) protections can be broken or bypassed
That’s not even complex to do in most cases, many hacks exist. There are actually many competing technologies that try to enforce authenticity.
That’s also why DUST Identity can claim a much better protection.
I won’t detail much more in this current post, to keep the argument relatively focused (although already fairly long).
However, some solutions are actually disrupting, so let me know if interested in a further article about that in the comments. As in any security system, attention should be drawn to the weakiest point.
Issue 3: wine is about the liquid
The case of wine is even more complex: there’s a bottle. But what I’m really interested in is actually the liquid inside the bottle. When you buy the forged wine in the restaurant, it may actually come from an original casing from a Grand Cru (so a legitimate bottle), but the wine put inside the bottle has been replaced. Therefore, even with VeChain as described (and supposing no one actually breaks the physical protection, see issue 2), the counterfeiter may buy a single real bottle and sell a forged content many times over.
So to know for sure, we would need to find a way to know the bottle has not been opened (i.e. detect that the stopper has not been removed or replaced), or maybe even build an expert synthetic tongue.
Well, DUST Identity says they use “non-toxic, inert diamond material. Carbon itself is FDA-approved as a safe additive for oral dosages.” So maybe you could put the identifier on the cork or even in the bottle.
Issue 4: wine is about the user experience
Ever tried to make your grandma use a blockchain app? Good luck. The problem is that she invited you and insists on paying the bill, so she needs to be sure whether that wine is for real.
Yet, in most cases, UX related to blockchain really sucks if you’re not a geek. A notable exception is the ZenGo wallet, but still my grandma wouldn’t make it. You want me to pay the bill, is that what you want?
Now is the funnier part. Since we want to have peace of mind, suppose we did replace the “VeChain hardware tag” by some nano-diamond dust. The problem now is that my phone can’t be the scanner (as for NFC), so I now need to bring my “universal scanner”:
Don’t worry, the good folks at DUST Identity seem to be targeting mostly the markets of electronical parts, and are funded mostly by the military (you know, to fight the mighty chinese chip producers against attacks like the one reported by Bloomberg). And anyway, it would require a significant change in supply chains to implement at a large scale.
So I bet you won’t be able to drink nano-diamonds before a while.
Issue 5: who pays?
Well, this brings us directly to a last issue. Last but not least. Suppose we do find a perfect technical setup.
Someone still has to pay for the change. To have the full life cycle for the product, it would need to be the brand (to get the trace from the origin, the sooner the better). The rationale is that a brand would do it if what it looses due to counterfeits (both indirectly, due to products which weren’t sold and from the cost to make lawsuits) is larger that the cost of the (blockchained?) system, related to the design and integration of the system in the supply chain, as well as the diffusion to the end-customer.
So would the brand really do that? We’ve seen some early experiments (e.g. the Aurora project from LVMH). However, in practice, legitimate buyers can simply go to a LVMH physical store and they’re pretty sure they will get a true product (albeit the manufacturer mandated by the brand could actually produce more products than the command made by LVMH, therefore produce “true” fakes that cannot be detected; for the end user, this extreme case would not matter, as the quality is exactly the same). LVMH, at least for now, doesn’t really seem to care about what comes after their direct buyers, because that’s a different business. Still the strong growth of the luxury resale segment (e.g. Vestiaire Collective, Collector Square, Rebag, etc.) may at some point ignite some change.
Those other players, especially online market places or resellers (e.g. selling pharmaceutical products online), might be very interested because the risk is life-threatening for them. They spend a lot of money trying to make sure products are authentic, but that’s a real difficult problem. Even with artificial intelligence, you’re not sure 100%. There’s always a “what-if”. What if the algorithm or the expert is wrong? The good news is: it becomes a traditional insurance problem. And everything that comes after the certificate by the online reseller is then logged in the chain.
So in the present case, the best is the enemy of the good. The best would be to trace from the original manufacturer, but this would take time and money to implement, for actors which don’t always have a clear interest to make the effort. Yet, it can be a game changer if you’re an online business, just like when people started to pay online merchants using their credit card. You need to prove that you can be trusted before it can happen.
intermediary score (reason f): sbmeunier: 1 / blockchain: 0
Reviewing reasons a-e
a) There is no unit of value. Let’s state the obvious: no accounting ledger, no blockchain!
That’s false: the unit of value is the produced good (physical as a bottle or digital as a certificate), which has been recorded put in traditional accounting ledgers for centuries (even without blockchain, in credit/debit format). So a decentralized ledger is also a natural fit.
intermediary score: sbmeunier: 1 / blockchain: 1
b) Nobody is trying to double-spend anything. It’s not like you want random people to buy and sell certificates of authenticity.
Is that so? The wine forgery example in the Chinese restaurant is clearly a double spend attack (even a x spend attack, as you can replay as much as you want).
More generally speaking, there is a market for the resale of goods. Think Vestiaire Collective for instance, in the luxury sector. Even LVMH is working on its own blockchain. As a consumer, you may wish to own an authentic handbag for a short period of time (for instance, Rebag has a program called Infinity) or even just rent it for the night. And guess what, this is already a huge market just by itself (guess what, in China too) and very prone to counterfeiting.
As an example, Chanel voiced concerns regarding the authenticity of its resold products several months ago with a lawsuit against The RealReal. In the lawsuit, the French luxury brand raised those questions with the retailers authentication experts.
So there’s a real need for a market of certificates of origin, and this applies to a large variety of goods (luxury, but also pharmaceutical products and so on, see the following article).
intermediary score: sbmeunier: 1 / blockchain: 2
c) There is no need for a consensus. It’s not like universities are fighting each other to certify the master that John Doe did at Harvard
This argument doesn’t make sense. The consensus is never applied at the origin of the information. Of course, Harvard can certify the master of John Doe by itself. But what limits any other John Doe to display on LinkedIn (or on a similar platform) a Harvard degree?
What we want from the blockchain certificate is to provide a proof that John Doe did graduate from Harvard, without having to actually ask Harvard every time if that’s true. That’s typically a use case for self-sovereign identity (see the indy example for a short primer, and DIF for more advanced specs).
intermediary score: sbmeunier: 1 / blockchain: 3
d) Users are identified (registered firms contributing to the supply chain, students)
intermediary score: sbmeunier: 1 / blockchain: 4
e) There are natural authorities (the product owner owns the IP and is liable, the US Department of Agriculture, organic/non-GMO accredited certifying agents, Universities, etc.)
Yes indeed. And that’s a very good thing, otherwise there would be no way to know which information to trust initially. In self-sovereign identity systems, those natural authorities serve as claim issuers, which can later be checked by other parties (called verifiers).
But I guess the underlying argument is to say that we can always ask the natural authority, without relying on a blockchain. That’s true, but burdersome and often impractical:
- In the case of supply chains, it’s important to notice that current online platforms go well beyond the reach of traditional supply chains, disrupting business models such as drug distribution without physical retail stores, or through peer to peer exchanges that include secondary re-sales between end-users without the original brand being involved. The life cycle of products cannot be controlled entirely by any single entity or its close partners any more. Users are in the loop, whether you like it or not.
- In the case of Harvard, would they respond to any request from a recruiter or a linkedIn viewer? A ledger provides the infrastructure to respond automatically to those verification requests.
total score: sbmeunier: 1 / blockchain: 5
So the total score shows that the counter-argument is pretty weak. Still, a careful reader would notice that the point made by @sbmeunier about the link between blockchain and the physical world is still an open issue. So realistic people avoid the hype and actually rely on simple albeit limited methods, for instance by asking those natural authorities as discussed in reason e (such as the original brand or FHS for watches), waiting for better days to come. But it’s better than nothing.
We can do better, in the spirit of DUST Identity but with a simpler practical setup and a good dose of SSI, but that is out of scope from the current post.
Still not convinced? To kill the debate, so to speak, Sebastien Meunier cites 2 counter-examples:
I don’t see how this would make an argument. You can ALWAYS build a system without a blockchain, and it’s usually much easier not to use a blockchain if you can. The question is whether an implementation is more efficient, technically and functionally, than an other one.
To blockchain or not to blockchain, that is the question. An example is the tracking of pharmaceuticals. Both the EMVS (in the EU) and Mediledger (in the US) are trying to fight against fake medecines.
- EMVS is a centralized database and has been developed by a single entity, mandated by the EU. The Falsified Medicines Directive (FMD), came in to force in February 2019, and requires a unique serial. An official tender has been made, leading to the development of an IT system, available at a cost per user and per month (just like any SaaS offering, except that you don’t have the choice).
“Solidsoft Reply won the contract to build and operate the EMVS. Starting in February 2019, every pack of prescription medicine entering the European pharmaceutical supply chain must, by law, bear a unique identifier (a two-dimensional Data Matrix barcode). The EMVS verifies every pack identifier at the point where the pack dispenses. Every drug manufacturer, parallel distributor, wholesaler, hospital, and pharmacy operating across 32 nation states must connect to the EMVS.”
- Mediledger is implemented on top of a blockchain protocol, which seems provided by Chronicled Network (most probably at the origin of the private consortium to extend their supply chain offer). In this specific instance, the exact specifications aren’t very detailed (see https://github.com/mediledger). The system is subject to the benefits and issues of Ethereum smart contracts. The main difference compared to EMVS is that the network participants share the verification, instead of relying on a centralized database system owned by the regulatory autority (like the FDA) and operated by a third party. The fees are paid to the network nodes that operate the consensus verification, instead of a fee per user.
The comparison shows that it’s probably quicker to implement a traditional system, but it doesn’t help establishing the eco-system around it. Last, it’s important to note that despite these systems, it still remains difficult to limit the propagation of fakes, as noted by Europol during the COVID crisis. The physical protection of goods and their various forms of packaging remains a difficult problem and the answer is also largely driven by regulation.
Note: this review is independant of any interest group, I have no affiliation to any of the mentioned projects.