Live data from Hacker News

Walmart Requires Lettuce, Spinach Suppliers to Join Blockchain

blogs.wsj.com

191–196 of 196 posts

Re: Walmart Requires Lettuce, Spinach Suppliers to Join Blockchain

#191
post #28

Earlier quoted context omitted.

I don't get that either. Absent proof-of-work (which is however terribly energetically wasteful), blockchain is just a public GIT repository where every commit is signed, no?

I agree that it's a buzzword, but at least intuitively, a signed blockchain (in the true sense that each block contains a hash of the previous block) does seem to confer some additional security over a plain database. A plain database would give you a log of events and who logged them, but security and trust resides in the database and how it's authenticated. Anyone with sufficient permissions could modify the databa…

> security and trust lie in the database and how it’s authenticated

And it’s any different on a private blockchain?

Re: Walmart Requires Lettuce, Spinach Suppliers to Join Blockchain

#193
post #141

Earlier quoted context omitted.

As I understand it, a blockchain is a decentralized database which does not require the individual instances to have perfect connectivity and perfect trust all the time. I can see why that would be something worth doing in a food chain. To trace, for example, a particular load of soy beans to a customer eating a steak, there are a couple of completely autonomous companies/sole proprietors involved. From the original…

I’ve been thinking a model for such a distributed ledger is Usenet/nntp where multiple partial copies of the data is stored on various servers. If one were to deploy a non-blockchain version of such a thing, what would it actually look like?

You'd quickly start messing around with signing, encrypting and checking packets of information... Which would be a lot like a blockchain.

Re: Walmart Requires Lettuce, Spinach Suppliers to Join Blockchain

#194
post #79

Notice how this article never clearly states specifically how a block chain solves the supply chain problem, nor does it even clearly state the problem in the first place. A very common response to these block chain/supply chain stories is that "a block chain is just a database." But even that statement is overly optimistic. A block chain is a log file. It keeps a record of a sequence of state changes. All of the tam…

> Notice how this article never clearly states specifically how a block chain solves the supply chain problem

Oh, I noticed. In fact it's the first thing I go looking for in these articles about "blockchain solves problem X".

And then I read the words "IBM Hyperledger". I've tried to find out what Hyperledger is, but penetrating IBM's blizzard of marketing bullshit about it has so far proved beyond me. All I know it has some relationship to a think called a "permissioned blockchain".

From what I can gather (as it is also surrounded by a cloud of marketing hype) a permissioned blockchain doesn't use proof of work, proof of stake or anything related. Nor is there a race to append the block. (Thus is it prompted as solving bitcoin's energy "problem", and because it doesn't need time to accumulate the proof of work it's not limited in the number of transactions it can process). Instead a permissioned blockchain appoints (ie, gives permission to) some miners to process the blocks. I have no idea what happens next, but I'm guessing in practice there is only one miner who checks the incoming transactions and assembles then into a block, signs the block using a public key, and publishes it by appending it to the database where it remains unchanged forever more.

To me the word blockchain doesn't just mean "append only published database". If that was true every log file would qualify as a blockchain. It also involves a clever set of rules for appending to that database that have one clear goal: they ensure the log is append only by making alterations computationally expensive.

In doing that a blockchain makes the "trust" quantifiable, eg to rewrite an entry 24 hours old in the log you are going to have to expend roughly 100 G Watt hours. Then by using more magic (the consistency checks aren't just rules, they are computer programs that can include assertions on future events, and thus control what happens in the future), the assertions can extended into "I promise to do X in the future, and if I don't I loose Y", and the only way to undo that promise is to expend that 100 G Watt hours.

A permissioned blockchain shares none of those properties, so to me it isn't a blockchain name notwithstanding. What it does look like is the old fashioned titles office - if I want to transfer my house title I have to give a notice to the Titles Office, and when they give it their official stamp of approval and append it to their records, and it's done. In that case the currency is property, the transactions are record of change of ownership of that property (which always originate with the state owning the land), the blockchain database is the records of all changes of ownership (which happens to be public in this case), the single miner is the Titles Office.

The amount of trust you have in the system boils down to much you trust the Titles Office to do its job (are they easily fooled, can someone be bribed, could a fire destroy the chain, and so on). This is not easily quantifiable. As an example I live in Australia whose civil systems are about as strong as you can get, yet every few years we have a new headline of someones house being sold from under them (usually while they were on an extended overseas holiday) due to title fraud. There is no recourse when this happens - the transaction is as irreversible as a bitcoin transaction and the state have legislated to ensure they are not responsible for any losses. However unlike bitcoin, it's impossible to extend the trust you put in the Titles Office to other applications. Or to put it another way: they are not going to look after your Ethereum ICO's for you, whereas Ethereum could do a very good job of looking after titles.

So it isn't a blockchain. It looks to me like IBM is selling "Titles Office" software for other applications, given it the fancy name "Hyperledger" and added blockchain to marketing blurb for extra pizzazz. 10 points for marketing prowess I guess.

It does however sound like a reasonable way to track food supplies. The currency is the food with some certifications. The transaction are changes of ownership, the miner is someone appointed to collects the sad transactions, stamp them as official, and append them to the log, the database is public so everyone (including the state and consumers) can verify everyone has made the right promises on that strawberry they are eating so they can sue there arses off if it contains needles. And the entity doing the certification (IBM?) gets to make a bit of money for their efforts.

Re: Walmart Requires Lettuce, Spinach Suppliers to Join Blockchain

#195
post #102

Earlier quoted context omitted.

Is it really write-once though? I'm pretty sure there's no PoW involved (didn't bother to read) so all bets are off.

You can replace PoW by a simple signatures, or a quorum of signatures. This is quite centralized, but still write-once. In any case, with a 'hash-linked-chain-of-blocks' attempted forks are really easy to prove.

I get that, but there are no guarantees it is write-once, because it might happen that the "quorum of signatures" changes their mind.

Re: Walmart Requires Lettuce, Spinach Suppliers to Join Blockchain

#196
post #186

Earlier quoted context omitted.

Sorry, I misquoted: it's 75 million cases of food borne illness in general, and 5,000 deaths, using older statistics from smaller population sizes. Using your stats, this outbreak was responsible for 0.1% of e.coli illnesses and 1/6th the deaths. But it's also 1/300,000 of the overall food poisoning illnesses, and 0.1% of the deaths. So it hit the lettuce industry significantly harder than normal, even though the act…

Latest data is Estimated annual number of episodes of illnesses caused by 31 pathogens transmitted commonly by food Domestically acquired foodborne [1] 9,388,075 (6,641,440–12,745,709) Deaths [2] 1,351 (712–2,268) [1] https://www.cdc.gov/foodborneburden/pdfs/scallan-estimated-i... [2] https://www.cdc.gov/foodborneburden/pdfs/scallan-estimated-h...

It’s pretty hard to compare the value lost to the lettuce industry vs the illnesses and deaths.

The EPA values a life at $7.4M for policy purposes [0]. So 1351x7.4M is about $10B. In 2015, the value of all US farm output was $136B. [1] So you could argue that you’d want to save more value from lives than from market impact, if you wanted to be purely utilitarian (that I hope no one in thread does).

But there’s a big impact from illness. I think many people underestimate the impact of food borne disease because they are young and healthy. I’m glad you pulled up the impact because it’s a big deal.

Hopefully as people better understand the danger, we’ll be able to avoid reductive math where we only worry about things when they are big, and way harder to fix.

[0] https://www.epa.gov/environmental-economics/mortality-risk-v...

[1] https://www.ers.usda.gov/data-products/ag-and-food-statistic...

Post reply on HN