I love the idea, but I'm just not sure that this is the 'hard' part of the problem, so I end up feeling like these are solutions in search of problems. Maybe I just don't grok how these contracts are interestingly different from other types of contracts.
Your friend's site and this article suffer the same basic problem: blockchain technology is really only effective at solving these problems:
1. Timestamp validity is of utmost importance (proof-of-existence).
2. Transactions are being heavily censored and there is a need for truly peer-to-peer value transfer.
This article has a terrible introduction that explains blockchain as enabling an upside-down communication method allowing users to aggregate their requests among many apps. This isn't something that a blockchain is needed for. In fact, I'm finding it hard to see how a blockchain improves upon traditional solutions already implemented.
Also, as you said Afton, traditional contracts work just fine and are preferable to blockchain contracts in almost every real-world scenario.
In the case of agriledger specifically, I am wondering why ECDSA concepts aren't enough to solve the problem of authenticity of records.
Blockchain apps can really redefine many concepts and processes, just like Bitcoin's blockchain is redefining money right now. However, I'm hesitant to over-prescribe blockchain as a solution to all that ails the tech world.
Blockchains aren't expensive unless they're anonymous. If they're anonymous, you need something like "mining" as a proxy for reputation. If the players in a blockchain are a dozen known mutually mistrustful banks, that's not a problem. It takes a majority of the players, acting together, to make a change. Banks are interested in this sort of thing because they need clearing intermediaries, and those intermediaries are expensive organizations to operate.
The parent article, though, is a solution looking for a problem.
Anonymity and decentralization are two different things, you can have one and not the other. For example, Bitcoin and most other cryptocurrencies has decentralization (mining) but only weak anonymity (all transactions are public). The decentralization is most of the cost, but adding anonymity (like zcash) adds even more cost with larger transactions.
Specifically, it's going to be business discovering that transaction ledgers in Merkle trees are excellent and being sold that as "Blockchain(tm) Technology".
(If someone at work asks you about Blockchain(tm), tell them "oh yes, bits of Blockchain(tm) are excellent, we've been using it even before Bitcoin existed!" i.e., you use git.)
Incidentally, one of my friends is working on such a block-chain based app: http://www.agriledger.com/
I still don't understand it, even though it seems cool.