The Last Days of the Barcode(theatlantic.com)
theatlantic.com
The Last Days of the Barcode
https://www.theatlantic.com/podcasts/archive/2024/01/the-last-days-of-the-barcode/677185/
6 コメント
We do already know exactly what will be in this QR code. My software has been supporting GS1-QR for several years. The content will be exactly the same as in GS1-128 (formerly known as EAN-128: a special interpretation of Code 128) and in GS1-DataMatrix (a special interpretation of DataMatrix).
The content is defined in GS1's General Specifications: https://www.gs1.org/standards/barcodes-epcrfid-id-keys/gs1-g... . Basically, it's a set of key/value pairs, where the keys (known as Application Identifiers) are 2-4 digits. The values for some of the early well-known Application IDs are a fixed length. Others are variable-length, separated by a marker character. In GS1-128 this is the FNC1 symbol, which isn't directly an encoding of an ASCII character, but is usually reported by the scanner as ASCII 29 (Group Separator). Any newly-added Application Identifiers now have to be terminated with the separator so that older software can skip over Application Identifiers that it doesn't recognise.
For retail, the core Application Identifier is AI 01, the Global Trade Item Number. This is just the 12-digit UPC-A, 13-digit EAN-13 or 8-digit EAN-8, zero-filled to 14 digits. There are 14-digit GTIN-14s, I'm not 100% sure on how they're allocated, I've never seen them in general retail.
The additional power of GS1 codes is that they can contain things like the weight of an item that's sold by weight, and/or the best before or expiry dates, and/or batch numbers. You just concatenate the appropriate Application Identifier and the value (plus separator if required).
For GS1-128, there's a practical limit of around 50 characters before the barcode gets too long to be readable. (Though note that there's a sub-mode of Code 128, Code Set C, that encodes two digits in one 'module', the space that one letter takes up in Code Set A or B.) That's probably why it hasn't really taken off. For greater amounts of data, you want to use DataMatrix or QR. Or GS1's previous suggested retail symbology, GS1 DataBar, previously known as Reduced Space Symbology (RSS).
The problem as always is chicken-and-egg. Retailers won't support a new symbology if there isn't a critical mass of products that use it; manufacturers won't support a new symbology if not all retailers can process it. GS1 DataBar has had limited take-up for products that need a very small barcode, e.g. individually labelling fresh fruit or veggies that are sold by the unit.
QR Code does have the advantage that it's easier to read by imaging software than UPC-A or EAN-13, due to the targeting patterns in the corners. That could increase the speed of scanning items.
The content is defined in GS1's General Specifications: https://www.gs1.org/standards/barcodes-epcrfid-id-keys/gs1-g... . Basically, it's a set of key/value pairs, where the keys (known as Application Identifiers) are 2-4 digits. The values for some of the early well-known Application IDs are a fixed length. Others are variable-length, separated by a marker character. In GS1-128 this is the FNC1 symbol, which isn't directly an encoding of an ASCII character, but is usually reported by the scanner as ASCII 29 (Group Separator). Any newly-added Application Identifiers now have to be terminated with the separator so that older software can skip over Application Identifiers that it doesn't recognise.
For retail, the core Application Identifier is AI 01, the Global Trade Item Number. This is just the 12-digit UPC-A, 13-digit EAN-13 or 8-digit EAN-8, zero-filled to 14 digits. There are 14-digit GTIN-14s, I'm not 100% sure on how they're allocated, I've never seen them in general retail.
The additional power of GS1 codes is that they can contain things like the weight of an item that's sold by weight, and/or the best before or expiry dates, and/or batch numbers. You just concatenate the appropriate Application Identifier and the value (plus separator if required).
For GS1-128, there's a practical limit of around 50 characters before the barcode gets too long to be readable. (Though note that there's a sub-mode of Code 128, Code Set C, that encodes two digits in one 'module', the space that one letter takes up in Code Set A or B.) That's probably why it hasn't really taken off. For greater amounts of data, you want to use DataMatrix or QR. Or GS1's previous suggested retail symbology, GS1 DataBar, previously known as Reduced Space Symbology (RSS).
The problem as always is chicken-and-egg. Retailers won't support a new symbology if there isn't a critical mass of products that use it; manufacturers won't support a new symbology if not all retailers can process it. GS1 DataBar has had limited take-up for products that need a very small barcode, e.g. individually labelling fresh fruit or veggies that are sold by the unit.
QR Code does have the advantage that it's easier to read by imaging software than UPC-A or EAN-13, due to the targeting patterns in the corners. That could increase the speed of scanning items.
> QR Code does have the advantage that it's easier to read by imaging software than UPC-A or EAN-13, due to the targeting patterns in the corners. That could increase the speed of scanning items.
The UPC and EAN symbols have huge targeting patterns, the entire symbol is effectively a targeting pattern. That single dimension massive redundancy makes finding and aligning them very easy. When you have a 2D code you need robust dedicated targeting patterns, but locating thick parallel lines in an image is easier. The massive price paid is information density, not read speed. In the 90s there were CMOS imager barcode scanners built on now antiquated embedded PowerPC that could scan most linear barcodes in milliseconds.
But this "interview", yeah it's shite. QR codes are "barcodes" in the meaning that is important here: extremely cheap, conspicuous, accurate machine readable labels. The barcode isn't being replaced at all in this regard.
The UPC and EAN symbols have huge targeting patterns, the entire symbol is effectively a targeting pattern. That single dimension massive redundancy makes finding and aligning them very easy. When you have a 2D code you need robust dedicated targeting patterns, but locating thick parallel lines in an image is easier. The massive price paid is information density, not read speed. In the 90s there were CMOS imager barcode scanners built on now antiquated embedded PowerPC that could scan most linear barcodes in milliseconds.
But this "interview", yeah it's shite. QR codes are "barcodes" in the meaning that is important here: extremely cheap, conspicuous, accurate machine readable labels. The barcode isn't being replaced at all in this regard.
> Desai: And so basically what these people decided is that if they let any individual company profit off of a barcode, then that would really be a huge impediment to this actually becoming universal. And obviously the whole idea of a universal product code is that it’s universal. So what they did is they chose seven finalists—seven companies—that would create a code in a way for it to be scanned, and none of these companies would get any sort of profits off of it.
> They would all agree to put the code in the public domain, and they would all just, you know, basically make money by selling scanners. That was the idea.
> Rosin: Wait. I just want to pause here. It’s, like, unimaginable. It seems completely logical when you explain it, that they should come up with a reason why this thing would be widely accepted, and that reason is sort of Marxist.
What's that, American ignorant author's defining a standard? (Sorry, cannot help but think that anybody else would call that sort of idea "Marxist").
> They would all agree to put the code in the public domain, and they would all just, you know, basically make money by selling scanners. That was the idea.
> Rosin: Wait. I just want to pause here. It’s, like, unimaginable. It seems completely logical when you explain it, that they should come up with a reason why this thing would be widely accepted, and that reason is sort of Marxist.
What's that, American ignorant author's defining a standard? (Sorry, cannot help but think that anybody else would call that sort of idea "Marxist").
> They would all agree to put the code in the public domain, and they would all just, you know, basically make money by selling scanners. That was the idea.
It's no different than giving away the razors to create demand for, and selling, the blades. Which is necessary in this case with intellectual property in order to overcome the chicken-and-egg problem of competing, non-interoperable standards that don't gain significant market share, no one uses, and thus no one makes money on. A code that can't be read by the scanner you have and a scanner that can't read a code you have is useless.
Calling this Marxist makes little sense and reads like the author has an agenda.
The following bit that referred to an "Elon Musk of barcodes" and inventors getting notoriety is pretty uniformed too.
It's no different than giving away the razors to create demand for, and selling, the blades. Which is necessary in this case with intellectual property in order to overcome the chicken-and-egg problem of competing, non-interoperable standards that don't gain significant market share, no one uses, and thus no one makes money on. A code that can't be read by the scanner you have and a scanner that can't read a code you have is useless.
Calling this Marxist makes little sense and reads like the author has an agenda.
The following bit that referred to an "Elon Musk of barcodes" and inventors getting notoriety is pretty uniformed too.
Not gonna lie, I’m not convinced we are in the “last days” of the barcode. Large institutions and associated standards change slowly.
As far as I can tell, the article is two employees of The Atlantic chatting about a topic they have no direct industry experience with but one of them has done some light research on. The only real information in the article is that the GS1, the organization in charge of UPC, has set a 2027 goal for switching over to some type of custom QR code, whose specific functionality either hasn't been fully defined yet or the employees of The Atlantic don't fully understand but have some guesses about.
Yeah the majority of their focus was on the consumer facing side of barcodes and then how QR codes replaced them (they haven't).
I can't see any of the warehouses in the middle of nowhere that I deal with changing their setup to scan everything with "AI" like Amazon is supposedly doing. It's hard enough to get them to use barcodes right now.
I can't see any of the warehouses in the middle of nowhere that I deal with changing their setup to scan everything with "AI" like Amazon is supposedly doing. It's hard enough to get them to use barcodes right now.
6R1M0R4CL3(1)
They say that a modern replica of bar code could never be made today. But we are full tilt into the open source movement with Linux, Chrome, Docker, PHP, Apache, Nginx, Makefiles and more all underpinning modern code.
Even QR codes are basically open source.
------------
Open standards have won the market today in practice. If a modern barcode re-competition were to exist, I'd bet on an Open Source one over any proprietary solution.
There seems to be a lot of understanding of the barcodes history here. But also a lot of cynicism that seems completely unwarranted.