Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Medicine The Almighty Buck Businesses Earth

Walmart Tests Blockchain For Use In Food Recalls (bloomberg.com) 109

An anonymous reader quotes a Bloomberg article about Walmart: Like most merchants, the world's largest retailer struggles to identify and remove food that's been recalled. When a customer becomes ill, it can take days to identify the product, shipment and vendor. With the blockchain, Wal-Mart will be able to obtain crucial data from a single receipt, including suppliers, details on how and where food was grown and who inspected it... "If there's an issue with an outbreak of E. coli, this gives them an ability to immediately find where it came from. That's the difference between days and minutes," says Marshal Cohen, an analyst at researcher NPD Group Inc...."

In October, Wal-Mart started tracking two products using blockchain: a packaged produce item in the U.S., and pork in China. While only two items were included, the test involved thousands of packages shipped to multiple stores... If Wal-Mart adopts the blockchain to track food worldwide, it could become of the largest deployments of the technology to date.

America's Centers for Disease Control and Prevention estimates roughly their recalls affect roughly 48 million people annually, according to the article, "with 128,000 hospitalized and 3,000 dying."
This discussion has been archived. No new comments can be posted.

Walmart Tests Blockchain For Use In Food Recalls

Comments Filter:
  • by Anonymous Coward

    UPS, FexEx, etc. track packages with barcodes, no need for blockchain.

    • by fph il quozientatore ( 971015 ) on Monday November 21, 2016 @07:50AM (#53330949)
      Yes, but you need a blockchain to make your cloud-based internet of things 2.0 social mining semantic smart-city experiment.
    • Re: (Score:1, Informative)

      by Anonymous Coward

      Barcodes are ideal to identify classes of products but not individual products. For example 'One gallon of whole milk' only needs one barcode for all individual entities, but when you sell millions of bottles of milk every day you need a larger identity key, like a blockchain if you want to track individual bottles.

      • Re: (Score:3, Insightful)

        A barcode is not a Fedex tracking number. A barcode is just a way to encode numbers in a machine-readable way. A tracking number uniquely identifies a single package. Blame the person who modded the information-free anonymous comment +1.

      • by Anonymous Coward

        Dude, "blockchain" isn't some word that you can use to describe a random thing. Use "thing", "widget", "thingamabob" or "doohickey" for that. "A blockchain" is not a key and does not identify anything. A blockchain is used to make information tamper-resistant (and it doesn't even do that on its own.)

      • Re: (Score:3, Insightful)

        by Anonymous Coward

        if you want to track individual items, then slap a QR code onto each product with its serial number. Problem solved.

        there's absolutely nothing that a write-only ledger (block chains) solves that a regular ledger (in a database, or just an append only file, etc.) cannot solve...

        • by Anonymous Coward

          THIS.
          There is absolutely no reason for this silly use.

          It isn't hard to make a serial number where the left-most bits are significant, then any digits after X are the unique identifies for a certain region, identified by one of the left-most bits.
          This sort of serial number can expand to infinity on the right side without breaking any algorithms.

      • by Anonymous Coward

        WTF? How did this get modded to +5? There are plenty of products that already use barcodes to identify serial and revision number, particularly for electronics, which sell millions of items. Millions of packages, drivers licenses, VIN, MAC address, etc. Even with a block chain, you need some way to put a number on an product or receipt to identify it, and barcodes are capable of working with large enough numbers to do so, especially more modern versions.

        That is completely orthogonal to blockchains, as a

      • when you sell millions of bottles of milk every day you need a larger identity key, like a blockchain

        What you need is called a batch number, and they've been in use for decades.

    • Re: (Score:1, Interesting)

      by Anonymous Coward

      This.

      Blockchain is just a buzzword and adds nothing whatsoever to the process.

      Somebody at Wal-Mart obviously doesn't understand what they are being sold.

      • by Anonymous Coward

        The concept of storing all of the supply chain process from field to consumer is surely noteworthy at Walmart's scale, but it doesn't require a blockchain. Block chains are important for things like distributed transactions where there isn't one central trustworthy party with a vested interest in making sure the right transactions are properly entered and kept. This is just a complete logistics inventory database that contains all these records - presumably against a GUID per package that allows the machi

    • by davide marney ( 231845 ) on Monday November 21, 2016 @08:10AM (#53331025) Journal
      Sadly, the article is silent on some important details. If you dig into the IBM announcement, you find that they are putting the entire chain of custody records into a blockchain, from source to the consumer -- all the things that traditionally would have gone into production logs, shipping manifests, etc. right down to the final delivery to the home. So, much, much more than what can be contained in a tracking barcode.
      • by beaker_72 ( 1845996 ) on Monday November 21, 2016 @08:34AM (#53331095)
        OK so I've read the IBM announcement and you're right that is what they're using blockchain for. However I don't see how this adds any value over and above the use of a standard relational database. To track the source of an offending item of produce, that item is going to need some kind of unique identifier - regardless of where the chain of custody data is held. A unique identifier can be used to access the relevant information in a database. How does blockchain improve on that?
        • I don't see how this adds any value over and above the use of a standard relational database.

          Blockchain technology: 9 benefits & 7 challenges

        • A unique identifier can be used to access the relevant information in a database. How does blockchain improve on that?

          A relational database can be deleted, corrupted, or falsified after the fact. You are assuming everyone in the supply chain is honest. Incidents such as the melamine-in-milk [wikipedia.org] scandal show that is not always the case.

          • by Gorobei ( 127755 )

            And you don't need a blockchain to fix the relational db tampering issue. You just crypto-sign records and publish the (possibly encrypted) ledger. Sign-and-register is technology that works and has been around for long before computers even existed.

            The only thing blockchain solves is the double-spend problem. And that isn't an issue for WallMart's solution.

            • You just crypto-sign records and publish the (possibly encrypted) ledger.

              If the entire DB was encrypted and published every time it was modified (many times per second) how could that possibly be a better solution than a blockchain? If you only encrypted and published the diffs, then wouldn't the chain of diffs pertaining to a single item be functionally the same as a blockchain, other than being scattered and more difficult to verify?

              • by Gorobei ( 127755 )

                You just crypto-sign records and publish the (possibly encrypted) ledger.

                If the entire DB was encrypted and published every time it was modified (many times per second) how could that possibly be a better solution than a blockchain? If you only encrypted and published the diffs, then wouldn't the chain of diffs pertaining to a single item be functionally the same as a blockchain, other than being scattered and more difficult to verify?

                You're totally correct. A classic DBMS is just an efficient way to map a series of immutable transactions into a current state that can be queried efficiently: in it's most pure form it's just a complete transaction log, an initial empty state, a mapping from state+transaction->state, and a bunch of caching to aid in querying. So the transaction log is exactly equal to your chain of diffs.

                Is the chain of signed diffs the same as a blockchain? Not quite. They are both chains, but transactions in the diff

        • by bws111 ( 1216812 )

          Who owns this database? Who can add to it? Who can modify it? Who can see the contents of it?

          The way it is done today is with standard relational databases. That method is not ideal, as even the summary mentions (it can take quite a while to backtrack through all the individual suppliers databases to find a common thread). With blockchain, the entire history is right there, in one place.

          • by Lehk228 ( 705449 )
            right where? it still has to be stored somewhere. and blockchain is vulnerable to 51% tampering
          • by Lehk228 ( 705449 )
            or you can, you know, come up with a common schema and mandate that your suppliers update your database in real time. this will have all of the advantages, plus it will run faster, plus it will work with existing software systems.
      • Sadly, the article is silent on some important details. If you dig into the IBM announcement, you find that they are putting the entire chain of custody records into a blockchain

        There's nothing wrong with using tech to track the source, but at the source, if you dig into [sfgate.com] field practices, perhaps the problem could be addressed earlier.

      • OK, I'll be the one to say it: I still don't see the point.

        If you squint a little bit so the algorithms being used go out of focus, what you see is system into which information about events is entered and then extracted later to form an audit trail. This could be done on a stupendous scale with any kind of eventually consistent database.

        The usefulness of blockchain is that it implements authentication and non-repudiation without the need for (or indeed possibility of) of a central authority. For example

        • by bws111 ( 1216812 )

          Decentralization is precisely the point. A farmer may sell his crops to dozens of buyers. Whose database is he going to update to say which field this particular truckload of lettuce was grown in? One of those buyers may make pre-packaged salads, provided to dozens of wholesalers. Whose database are they going to update to say which truck supplied the lettuce, and what factory/date/shift processed it? Whose database does the trucking company that moved the product to a warehouse update? What database d

          • by hey! ( 33014 )

            I understand decentralization is what they're after; it's just that blockchain isn't the only way to do that. You have to justify specifically why you need to use that architecture otherwise it's overkill.

            • by bws111 ( 1216812 )

              Of course it isn't 'the only' way to do it. There is seldom only one way to do anything. They could do it all on paper if they wanted. They don't have to justify anything to you - they just reached a business decision that this is what they want to do. I would like to hear your reasons why you think it is 'overkill' though.

              • by hey! ( 33014 )

                The problem with blockchain is designed to make a tradeoff between a very specific requirement and high computational costs/implementation complexity. The requirement is this: in a system where literally no individual actor can be trusted, make it possible for everyone to be able to determine who owns a particular bitcoin. The solution is this: employ an audit trail of past transactions that is intentionally designed to be computationally expensive to construct.

                The strategy is that in order to repudiate a

                • by bws111 ( 1216812 )

                  Blockchain is not bitcoin. The only thing computationally expensive in blockchain is an attempt to forge an entry. The expensive thing in bitcoin is the mining operation - there is no mining operation in blockchain. Blockchain is a template of what data is associated with something, and controls on who can see and modify portions of the data. There are 'smart contracts' which specify what parties must agree to a transaction (consensus). The stuff is all controlled by encryption, and does not rely on a

                  • by hey! ( 33014 )

                    Never said it was bitcoin. I said it was designed to meet the needs of bitcoin, and does that very well.

    • by rednip ( 186217 )
      A blockchain is nothing more than a log of all transactions, including splits which would need to be 'attached' to a physical unit for the use described in this story. It's very much like a bar code, but individualized for each package and inclusive of the blockchains of ingredients rather than a pre-registered number dependent on individual company database for context. If properly done there are clear advantages in tracking ingredients from farm to table especially when it comes to recalled food, but it
    • by sjbe ( 173966 ) on Monday November 21, 2016 @08:46AM (#53331137)

      UPS, FexEx, etc. track packages with barcodes, no need for blockchain.

      A barcode doesn't track anything nor can it realistically be updated once it has been created. They use a database to track packages. The barcodes are merely a means of quickly "typing" a bit of data at a physical location - the database is what actually keeps track of things. Barcodes become cumbersome as a means of identifying specific packages when you get to very large volumes. UPS deals with about 15 million packages per day. A big number but nothing like what would be necessary for real time tracking of what Walmart is looking at doing. Walmart deals with tens of billions of individual product transactions so the complexity is substantially higher. There is a big difference in data and complexity between shipping a single box of 100 widgets versus knowing the entire supply chain history for each and every one of those 100 widgets.

      • Sounds like you think that they are going to print out the whole block chain on the actual package. Most likely they will use a *gasp* barcode on the package as the id to reference the package in the block chain.
        • by bws111 ( 1216812 )

          You are missing the entire point. The point of the blockchain (regardless of how it is identified) is that you have the ENTIRE HISTORY of that product in one place, and that history is cryptologically secure. Even if that one product had 5 steps in the supply chain, it is ALL there. Your magical UPS barcode tells you nothing except what UPS knows about the package. That is a HUGE difference.

          • by Anonymous Coward

            I think you missed the point.

            The barcode isn't going to tell YOU anything except that Wal-Mart uses barcodes and, if you have several of the same thing, that they use different ones for every single thing. This is the same thing with UPS, the bar codes don't tell you anything except that your package has a number.

            Wal-Mart is NOT going to be encoding something directly on each package that has its full history because that would mean altering this indicator every time the package was packed, shipped, unloade

          • by Lehk228 ( 705449 )
            this is different from a regular RDBMS how?
        • You only need the latest block in the blockchain. From the latest block, you can find every previous block associated with that item. That's the point of blockchains.

          For example, a package of broccoli might get a barcode or QR code of a new block as it gets put on the shelf. This is the only block directly associated with this particular package of broccoli.

          However, the block on the package of broccoli references the block that was placed on the box that the package of broccoli came out of. That block refer

          • That is assuming that IBM will build a distributed public blockchain for Walmart when the truth is probably that this will all be a hosted server by IBM for Walmart that no one else will get access to. So blockchain is just a magic word here, there is nothing that makes this easier to do with a blockchain than with a traditional database.
  • Word (Score:5, Funny)

    by Cyphase ( 907627 ) on Monday November 21, 2016 @07:47AM (#53330937) Homepage
    Blockchains in the big box chains gon' rock change. Strange.
  • Saving money (Score:4, Informative)

    by jbmartin6 ( 1232050 ) on Monday November 21, 2016 @07:47AM (#53330941)
    Not that there is anything wrong with it, but we see the crux of the matter with this quote: "It’s also the difference between pulling a few tainted packages and yanking all the spinach from hundreds of stores"
    • Sure, and if the pesky individual privacy rights can be eroded some more,

      a national DNA database could narrow it down to one field hand skipping a bathroom trip to the field potty to up his pick count.

      • by Anonymous Coward

        Yes, the well known "right to defecate on the production floor into the products being shipped to customers" right that so many fought to defend...

        • Re:Saving money (Score:4, Informative)

          by rmdingler ( 1955220 ) on Monday November 21, 2016 @08:59AM (#53331211) Journal
          The 2006 E Coli spinach outbreak was directly a result of defecating on the production field.
          • by Anonymous Coward

            The 2006 E Coli spinach outbreak was directly a result of defecating on the production field.

            Introducing the Intel 3cxx, The first colon embedded chip with wireless communication. For blockchain managers who actually care about shit.

      • There are no new privacy issues in this strategy.

        Currently, food-borne illness is identified when a consumer volunteers information about becoming ill from consuming a food. That illness is then cross-referenced with the matching food product the consumer purchased, based on consumer testimony or receipts. Enough of such information identifies a product, which we can then remove.

        The proposed method attaches the entire handling chain to the consumer's purchase. Once we identify the receipt and the foo

    • That's sort of the point, isn't it?

      Back in 1790, 90% of American laborers worked in agriculture--farm work. Women made clothing at home to produce additional income, using skills necessary for economic survival due to the feasibility (or lack thereof) of the average household to afford clothing in the way we generally view as normal. People grew substantial gardens and livestock, hunted, and otherwise supplemented their nutritional needs, due to a lack of affordability of food.

      In 1900, 28% of American

  • by Anonymous Coward on Monday November 21, 2016 @07:48AM (#53330945)

    Okay, reading through that article it uses 'blockchain' repeatedly without introducing the concept or defining anything it does/offers.

    It can be summed up as 'Walmart testing thing IBM wants to sell to everyone' gg, we're done here. All aboard the hype train!

    But, Anonymous Coward, it's a decentralized private cloud immutable ledger system! It's the future. Right, right, I give you it's a novel use of the underlying mechanism of bitcoin... take an open source project, wrap your head around it and sell it as IBM. Neat. It's still a cringe amount of buzzword bingo. It's still not clear from all the hype why taking business analytics + buzzword > business analytics.

    • take an open source project, wrap your head around it and sell it as IBM.

      I'm convinced! I'll have two IBMs, please.

  • by Anonymous Coward

    Gotta find a nail, gotta find a nail.. damn, that screw could be a nail!,

  • by Anonymous Coward

    Is there a reason walmart cannot just run a central database with a bunch of submitted information? The only thing good about blockchain tech is that its decentralized.

    • Re: (Score:3, Insightful)

      "The only thing good about blockchain tech is that its decentralized." Which is exactly the situation with our food chain, which involves hundreds of decentralized participants. So, blockchain makes a ton of sense.
      • So, instead of getting all this data together in a database with a FK SerialNumber, it stores all the information in something that changes each time it changes hands, thus allowing them to know all information up to when they obtained it, but not afterwards.

      • And in the real world every one in the supply chain have to report the logs to either Walmart or IBM, there will most certainly be no decentralized thing about it at all.
    • by Anonymous Coward

      Any one person with appropriate privileges (or someone who stole them) can change the entry for a barcode in a database. Or just delete it.

      It is harder to accomplish this using a blockchain.

    • Is there a reason walmart cannot just run a central database with a bunch of submitted information?

      Yes. First off there are few/no standards in place to do it nor much infrastructure for real time tracking of this information currently. All this will have to be created from scratch and so they are experimenting with various technologies for doing just that. This is one of those experiments. Second, our food supply chain is hugely decentralized and tracking the transaction records is currently very cumbersome. Since we need to know the origin of products it makes sense to adapt a technology which is

  • America's Centers for Disease Control and Prevention estimates roughly their recalls affect roughly 48 million people annually, according to the article, "with 128,000 hospitalized and 3,000 dying."

    I think you meant to say:

    America's Centers for Disease Control and Prevention estimates roughly their recalls affect roughly 48 million people roughly annually, according to the article, "with roughly 128,000 hospitalized and roughly 3,000 dying."

  • Stop importing pork from China. You wouldn't know what kind of filth would be in that.
  • Walmart was a huge proponent of AS2 to talk to its suppliers back in the day. And I remember the pharmas here in the US were playing with AS2 to do "pedigrees" that tracked product from endpoint to endpoint. So...why isn't Walmart looking at AS2 (or AS4) to do the same here?
  • and this is exactly why I eat all of my food direct from farmers. blockchain shmockchain, walmart kills 50 times more people than sharks. so many that they need this much technology just to track it! none of this actually stops the food from being tainted.

    Like I've said before, I don't want walmart to take back the bad food. I want them to be horrified that their food was tainted. Alas, it is not their food, and hence they don't give a shit. They care only about their dollars.

    So, I buy directly from

    • The problem is that the chicken seems to have a stronger educational background than you have.
  • by Baron_Yam ( 643147 ) on Monday November 21, 2016 @09:59AM (#53331573)

    So... a blockchain why? Because Wal-Mart doesn't really need anything more than a regular old database for this purpose.

    • by Vairon ( 17314 )

      Perhaps they want a system that can be used by multiple companies besides just Walmart. If I was a farmer whose food might end up at Walmart, Target, Whole Foods, etc why would I want to submit tracking data multiple times to multiple companies' databases when I could just submit it once to this 3rd party IBM based blockchain and then any end-company which uses it can have the info they need.

    • Blockchains don't do anything you can't do already. However, the blockchain makes it easy to get very high granualrity of data with no extra cost, whereas with traditional tracking methods there is a not-insignificant cost to obtaining all that detailed information.

      The idea is simple and really quite brilliant. To create a blockchain of physical products, all you need is a barcode scanner/printer set up to generate new blocks for the block chain and a basic database that fits the number of items you sell, a

  • Fact: tracking the source of a potentially tainted product from a single receipt, instantly, can be helpful in finding the tainted product, testing it, stopping its distribution if necessary, and notifying the purchasers.

    Fluff: rather than use a bunch of simple database joins, we used a distributed cryptographic system providing reliable attestation of a transaction without dependence on a particular server, together with anonymity. We also swat mosquitoes with atomic bombs.

    • by bws111 ( 1216812 )

      'Fluff' is claiming this could be done with 'a bunch of simple database joins'. For even a simple food product (say a bag a salad) there are many different companies involved. Farmers, trucking companies, processors, packagers, packaging suppliers, wholesalers, distribution centers, retailers. All with their own databases. All with their own database designs. All with their own database managers. All with their own database versions.

      Yeah, I guess it would be pretty simple to just get authorization to

      • I think you are not considering the scaling issues. In a perfect world, your box of pepperoni would have a blockchain by which you could trace back to that mad cow and the farm it came from, and the guys who undercooked it, and the plant they work at, once someone got CJD. In real life the processors co-mingle meat at every stage, and the accounting load doesn't scale.

        I own very many devices that are capable of having an embedded serial number. For some reason, most of them have "12345678" or something equa

  • Wal-Mart customer service employees never seem to have any sense of urgency when I have an issue. They appear comptent and are usually pleasant enough. They just don't move very quickly. How will Wal-Mart make sure that tainted food issues are handled immediately by people that are used to treating everything one way. Will there by a hotline, or are we expected to return with a receipt to customer service, who then may or may not act quickly. A plan doesn't work if the human training isn't complete.
    • Nominally, in a fully tracked system, a store could automatically let everyone know next time they check out with their shopper's card or use the same credit card etc.
  • You can always compute nested sets from adjacent lists like people are suggesting if the entire adjacent list is on one cloud or system, but you can lose a whole branch if you lose a link instead of only losing that one link if one supplier is non-conforming or offline. It's not fake or fluff, it's error minimizing.

If all else fails, lower your standards.

Working...