I often get told that blockchains provide a single source of truth, or create trust in the system. But what do we mean when we say trust or truth? I wrote a blog article (Original Link) discussing truth, trust and consensus for the physical supply chain. The body of that article is reposted below. Let me know your thoughts!
The Meaning of Trust and Blockchain in Food Systems
"Blockchains don’t guarantee truth; they just preserve truth and lies from later alteration, allowing one to later securely analyze them, and thus be more confident in uncovering the lies. Typical computers are computational etch-a-sketch, while blockchains are computational amber."
-Nick Szabo, in his blog
In our paper on why blockchain, I briefly discuss truth and trust in data recorded in a blockchain but such a topic deserves more discussion. Blockchains often get referred to as a 'single source of truth' or 'trustworthy'. In a protocol such as Bitcoin, transactions are mathematically verifiable- transactions are checked to have sufficient funds to execute prior to being included in a block, and rejected otherwise. The Bitcoin blockchain was designed to not increase trust, but to enable transactions while minimizing the amount of trust needed in order to transact.
It is important to delineate the kind of trust we mean when we say that a blockchain is a trusted record. Blockchains are cryptographically secured so as to be computationally difficult to change after a block has been accepted, creating immutability. That is to say, one can generally trust that records are not alterable after a certain time period has elapsed (such that the chain length has grown enough where creating a new chain of blocks longer than the existing chain is functionally computationally impossible). However, this does not mean that the contents themselves should be trusted. 
Image from Pixabay
In data posted in blockchains, information posted in the system is not inherently trustworthy. Supply chains and the physical nature of their goods present a complicating factor for the validity and 'truthiness' of data in a blockchain system. Prior to being immortalized in a blockchain, the data needs to be first submitted to the system, presenting a vulnerability to tamper with the information. In many cases, and often where the data has the most value beyond a pure traceability solution, the data is generated and submitted by a single actor. As a result, such information is difficult to validate, especially on the cost basis and scale of the food system where billions of pounds of food move.
For example, a truck's temperature log may be generated by a sensor in a pallet in that truck. Such information is difficult to verify and prove validity on large scale, as the sensor may have been tampered with to whitewash readings, only providing values within specification. Another case should also be considered in which a sensor in the truck may provide a reading in disagreement with what the trucker's own measurement device. Further, what happens if a sensor is tampered with maliciously to shift blame to the trucker? Coming to agreement, or consensus, as to what is 'true' in such a system with conflicting unilateral actors has been yet unsolved.
The adage "garbage in, garbage out" applies here as in any data system. Asserting that the majority of the system will generally behave well is insufficient for a solution. Food fraud is already estimated to affect about 10% of food in the UK, and poses a 10-15 billion dollar problem globally. Many transparency solutions promise premiums for increased food information, especially in areas of sustainability such as organic, sustainability, fair trade, local or animal welfare. Promising higher prices or increased revenue for information without effective deterrent mechanisms or methods to validate information only further incentivizes malicious actors within the system. Transparent supply chains offer some amount of deterrent in allowing scrutiny, but transactions and movement of goods can always happen off-chain- essentially becoming legal smuggling. Transparency will be discussed in another essay.
The elephant in the room when examining blockchain solutions for the food system is the issue of how to validate information. Consensus in most protocols means agreement upon which blocks have been accepted into the chain, and as a result which transactions are considered to be valid. However, most current supply chain protocols skim over the consensus mechanism, as a robust validation method has yet to be developed for physical objects. The low margin and flexibility in the food industry further restricts the range of possible solutions, as physical validation quickly becomes prohibitively expensive on scale. As a result, we should be careful what we mean when saying information in a blockchain form trustworthy records.
No legal precedent has been set on the implications of data records stored in a blockchain yet, but at a minimum business relationships would likely be negatively affected if an actor is found out to by lying which may carry just as much deterrent weight as a lawsuit. In the case of buyer driven implementation, the threat of loss of contracts poses a high deterrent value to upstream actors. Nevertheless, honesty is not inherently solved by having immutable records in a blockchain, although having records makes it possible to find out where the transgressions occurred.
The value of coordinating the various actors along a system as complex as the food supply chain cannot be understated. Although physical items present a point of vulnerability for the system, the physical nature of supply chains and inherent trust relationships between some parties provide a set of simplifying assumptions which can be beneficial. We remain optimistic for the value of blockchain systems in food and supply chains. Stay tuned for a discussion of transparency in the system, and reach out to [email protected].