Open Products Facts Insights: Difference between revisions

From Open Food Facts wiki
No edit summary
No edit summary
 
(3 intermediate revisions by the same user not shown)
Line 48: Line 48:
# Nobody has had the incentive to abstract circular APIs at planetary level and open them
# Nobody has had the incentive to abstract circular APIs at planetary level and open them
# We can’t scale scoring to 36K categories alone
# We can’t scale scoring to 36K categories alone
 
==== Complicated data acquisition ====
==== 1. Product loose their barcodes after purchase (or never had one) ====
===== Product loose their barcodes after purchase (or never had one) =====
# We will allow guided search to find your product (or approximate it / create it)
# We will allow guided search to find your product (or approximate it / create it)
## TV >> Sony >> Flat
## TV >> Sony >> Flat
# We could imagine integration with popular object recognition frameworks
# We could imagine integration with popular object recognition frameworks
==== 1bis.Getting REALLY all objects online ====
===== Getting REALLY all objects online =====
* Getting REALLY all objects online (even without a barcode)
* Getting REALLY all objects online (even without a barcode)
** Creating a reusable open object graph
** Creating a reusable open object graph
Line 61: Line 61:
*** Allow association of this QRCode with a barcode, giving extra-benefits for the users
*** Allow association of this QRCode with a barcode, giving extra-benefits for the users
* Grow micro-communities around objects
* Grow micro-communities around objects
==== 2.Collaboratively mapping carbon emissions, starting with top categories ====
===== Very slow lifecycle =====
* Food is a "fast moving consumer goods" in the sense that it gets bought, used and discarded quickly, thus increasing the speed of data collection on Open Food Facts
* You don't own 20 TVs in your lifetime, and so core contributors of Open Food Facts will see their impact limited.
* Here's [https://docs.google.com/spreadsheets/d/1p_wGUQlyolvMbXLEQIpMkSiXRQmVM4loJNwVa4wWtoY/edit#gid=0 a spreadsheet analysing that]
==== Collaboratively mapping carbon emissions of 46K categories, starting with top categories ====
* Carbon emissions assessments exist for many categories already
* Carbon emissions assessments exist for many categories already
* They are not mapped with public knowledge graphs and standard vocabularies, limiting their use
* They are not mapped with public knowledge graphs and standard vocabularies, limiting their use
Line 67: Line 71:
* Document unpublished LCA results made by companies & researchers on Wikidata
* Document unpublished LCA results made by companies & researchers on Wikidata


==== 2.Scaling to 36K goods categories thanks to UN/GS1 Vocabs ====
==== Scaling to 36K goods categories thanks to UN/GS1 Vocabs ====
* Scaling to 36K goods categories thanks to UN/GS1 Vocabs
* Scaling to 36K goods categories thanks to UN/GS1 Vocabs
** Those describe anything from a rubber duck to surgery instruments
** Those describe anything from a rubber duck to surgery instruments
Line 73: Line 77:
** They provide a foundation for collaborative augmentation around circularity
** They provide a foundation for collaborative augmentation around circularity


==== 4.We can’t scale scoring to 36K categories alone ====  
==== Scoring: We can’t scale scoring to 36K categories alone ====  
* Turn product analysis and deciphering into partner platforms
* Turn product analysis and deciphering into partner platforms
** Creating a bespoke analysis and scoring system for each of 36K categories is a lot of work
** Creating a bespoke analysis and scoring system for each of 36K categories is a lot of work
Line 80: Line 84:


==== A new hope: Large Language Models ====
==== A new hope: Large Language Models ====
* Ability to generate data models
* [https://github.com/openfoodfacts/openfoodfacts-ai/issues/296 Use LLMs to create data models for Open Products Facts categories]


=== 3.Creating the missing open APIs for the Circular Economy ===  
=== 3.Creating the missing open APIs for the Circular Economy ===  
Line 119: Line 125:
Key objective: allowing to collect relevant data for up to 36000 different kinds of products, and display it without having to evolve mobile apps constantly
Key objective: allowing to collect relevant data for up to 36000 different kinds of products, and display it without having to evolve mobile apps constantly


 
==== Customizing product addition and product information for many heterogeneous product categories ====
* To achieve that we need to create a list of criteria for each of the 43000 product categories that are listed by the UNSPSC (A joint effort by the United Nations and GS1).
* That way we can store relevant information for each product category. For instance an adaptor can be USB-C Micro-USB or Apple Lightning for instance. Being able to collect and to show the right criteria is essential for providing value-added.
* Beyond those specifications, we will make sure users can easily add new criteria as product categories evolve overtime.
* [[Folksonomy Engine/How to use it on Open Products Facts]]


=== Some stuff never finds an owner ===
=== Some stuff never finds an owner ===
Line 133: Line 143:
** relevant nearby addresses to either repair or recycle the product.
** relevant nearby addresses to either repair or recycle the product.


== Aggregating Supply & Demand for circular options ==
=== Aggregating supply and demand for used products ===
=== Aggregating supply and demand for used products ===


Line 169: Line 180:


=== Achat ou seconde main: le cas d’un jouet en bois ===
=== Achat ou seconde main: le cas d’un jouet en bois ===
* Des objets fonctionnels sont jetés car ils n’ont plus d’usage, et qu’il est compliqué d’écouler une réplique de cuisine en bois
* Functional objects are thrown away because they no longer have a use, and it is complicated to sell a wooden kitchen replica


=== Lending stickers ===
=== Lending stickers ===

Latest revision as of 14:23, 10 June 2024


Has anything changed since “The Story of Stuff” (2007) ?

Source: https://www.storyofstuff.org/movies/story-of-stuff/

LINEAR SYSTEM ON A FINITE PLANET

“But the truth is it’s a system in crisis. And the reason it is in crisis is that it is a linear system and we live on a finite planet and you can not run a linear system on a finite planet indefinitely.”

OBJECTS DON’T HAVE A LONG LIFESPAN

“Guess what percentage of total material flow through this system is still in product or use 6 months after their sale in North America. Fifty percent? Twenty? NO. One percent. One! In other words, 99 percent of the stuff we harvest, mine, process, transport—99 percent of the stuff we run through this system is trashed within 6 months. Now how can we run a planet with that rate of materials throughput?”

OBSOLESCENCE (PLANNED & PERCEIVED)

“She looks like she is driving in spaceship central and I look like I have a washing machine on my desk.”

“Stuff” has a large impact on the planet

The manufacturing industry is one of the largest contributors to greenhouse gas emissions worldwide. In fact, in 2017, according to the Environmental Protection Agency, industry accounted for just over 22% of emissions in the United States.

Insights

  • Facilitating the listing of used products on e-commerce or donation services
    • Ability to sell/give products you don't want anymore is one of the tenets of a circular economy
    • taking photos, filling out details about it… is a time-consuming effort.
  • What if all the effort could be done by Open Products Facts?
    • The only thing left ? Indicate the state of your product.
      • eg: for a computer, for instance, it would enquire about the state of the battery the state of the screen etc
  • Repairing a product is something complicated for consumers and producers
    • You don't have the manual, the parts to do it
    • You don't know nearby professionals able to repair it
    • Professionals can’t tell the general public which products they are able to repair
  • Recycling a product is complicated for consumers and organisms
    • If you want to dispose of a product, you are often unaware of specialised systems to recycle it or reuse it

Extend our environmental impact to non-food products

  • Transform our Open Products Facts platform into an essential platform for the circular economy and non-food carbon rating
  • Create the conditions for development that is radically more open to the community and to third parties

Governments are moving forward on Environmental labelling

  • The European union is moving forward (Digital Product Passport)
  • France has moved forward (Compulsory labelling)

The Open Products Facts Vision

  • Build the data platform to power the circular economy by reducing friction on product data
  • Comprehensive product data about all products gives an additional chance to be able to reuse them

Challenges

Hard challenges

  1. Product loose their barcodes after purchase (or never had one)
  2. Environmental impact LCAs have never been linked to barcodes or to an open graph
  3. Nobody has had the incentive to abstract circular APIs at planetary level and open them
  4. We can’t scale scoring to 36K categories alone

Complicated data acquisition

Product loose their barcodes after purchase (or never had one)
  1. We will allow guided search to find your product (or approximate it / create it)
    1. TV >> Sony >> Flat
  2. We could imagine integration with popular object recognition frameworks
Getting REALLY all objects online
  • Getting REALLY all objects online (even without a barcode)
    • Creating a reusable open object graph
    • Allow users/apps to list objects that don’t have a barcode (packaging thrown away), or even that might be decades old (and predate the barcode)
      • Generate a private QR-code unique to the product you own
      • Allow association of this QR-Code with a category (minimum XP, helped by AI categorization)
      • Allow association of this QRCode with a barcode, giving extra-benefits for the users
  • Grow micro-communities around objects
Very slow lifecycle
  • Food is a "fast moving consumer goods" in the sense that it gets bought, used and discarded quickly, thus increasing the speed of data collection on Open Food Facts
  • You don't own 20 TVs in your lifetime, and so core contributors of Open Food Facts will see their impact limited.
  • Here's a spreadsheet analysing that

Collaboratively mapping carbon emissions of 46K categories, starting with top categories

  • Carbon emissions assessments exist for many categories already
  • They are not mapped with public knowledge graphs and standard vocabularies, limiting their use
  • Link the various LCA results with Wikidata and Open Products Facts to unlock broader usages and bridge gaps
  • Document unpublished LCA results made by companies & researchers on Wikidata

Scaling to 36K goods categories thanks to UN/GS1 Vocabs

  • Scaling to 36K goods categories thanks to UN/GS1 Vocabs
    • Those describe anything from a rubber duck to surgery instruments
    • They are already adopted by logistic players / e-commerce
    • They provide a foundation for collaborative augmentation around circularity

Scoring: We can’t scale scoring to 36K categories alone

  • Turn product analysis and deciphering into partner platforms
    • Creating a bespoke analysis and scoring system for each of 36K categories is a lot of work
    • However, many organisations are at work on some of those important categories
    • Turning our core expertise (implementing algorithms to score products at scale) into a platform is key to solving this.

A new hope: Large Language Models

3.Creating the missing open APIs for the Circular Economy

  • Creating generic APIs for 36K product categories
    • Basic info, Characteristics, Images, Average Carbon emissions, Alternative categories (eg Hair Dryers vs Towels), Manufacturer support by country, Manuals, Donation option, Resell options, Second hand purchase options, Local Borrow/Lending options, Local community repair options, Repair services options, Repair Videos/Documents, Spare parts options, Recyclability status, Local responsable recycling options
  • Making generic existing APIs discoverable
  • Stretch: Abstracting / Making category-existing APIs discoverable (over time)

A non limitative list of circular APIs

  • READ/WRITE themes: product data, circular services, scoring systems (with standards)
  • Generic APIs for 36K product categories:
  • Basic info
  • Characteristics
  • Images
  • Average Carbon emissions
  • Alternative categories (eg Hair Dryers vs Towels)
  • Manufacturer support by country
  • Manuals
  • Donation option
  • Resell options
  • Second hand purchase options
  • Local Borrow/Lending options
  • Local community repair options
  • Repair services options
  • Repair Videos/Documents
  • Spare parts options
  • Recyclability status
  • Local responsable recycling options
  • Eventually category specific APIs (eg: car spare parts availability)


Flexible APIs for flexible mobile apps

  1. Leveraging our new display API called Knowledge Panels (trialled on the great diversity of food categories)
  2. Creating a new “Category-driven Editing/Contribution API”

Key objective: allowing to collect relevant data for up to 36000 different kinds of products, and display it without having to evolve mobile apps constantly

Customizing product addition and product information for many heterogeneous product categories

  • To achieve that we need to create a list of criteria for each of the 43000 product categories that are listed by the UNSPSC (A joint effort by the United Nations and GS1).
  • That way we can store relevant information for each product category. For instance an adaptor can be USB-C Micro-USB or Apple Lightning for instance. Being able to collect and to show the right criteria is essential for providing value-added.
  • Beyond those specifications, we will make sure users can easily add new criteria as product categories evolve overtime.
  • Folksonomy Engine/How to use it on Open Products Facts

Some stuff never finds an owner

One Amazon warehouse reportedly throws out 130,000 products a week, including some that are brand new. An expert blames its giant third-party retail business. If a product isn't selling on Amazon, the cheapest option for third-party retailers is to pay Amazon to get rid of it.

Facilitating the work of recycling/reuse companies and charities

  • Each product category has some specific recycling requirements
    • For instance anything with a battery has to be disposed in a certain way
    • We have hundreds of “eco-organismes” recycling companies and charities in France
  • Solutions APIs, based on the product category
    • Integrate relevant expert knowledge from the right “eco-organisme”
    • relevant nearby addresses to either repair or recycle the product.

Aggregating Supply & Demand for circular options

Aggregating supply and demand for used products

Aggregating supply & demand or cannibalized spare parts for products

  • Companies have a legal obligation to produce spare parts for a given number of years. 
    • It is often hard to find where to buy those parts (and the right one) when they are still produced.
    • It is even harder to find alternative producers of spare parts (and still the right one) when the legal time frame is past. 
    • It is often very complicated to get spare parts for your product. 
  • We could list official spare parts purchase websites and shops, list alternative shops, store 3D schematics of spare parts (which would thus offer a repairability warranty), aggregate offer and demand for cannibalized spare parts.

Making better, more informed choices when purchasing

Let people make choices that have will have consequences in the future by comparing products not only on classical specs… but also on:

  • repairability
  • production
  • carbon impact
  • recycling carbon impact.

Open Products Facts will unlock other use cases

  • Normalising product recalls - Currently product recalls are not associated with barcodes. Very few users actually hear that their product was recalled, especially for low value products. Linking and storing recalls in Open Products Facts would enable governments to more easily reach consumers
  • Automating insurance claims - Potential users also include improving the way insurance claims are done. One could imagine scanning all the products about and being able to easily hand that list over to the insurer in the event of a fire or a theft.
  • Better public action - We also hope that it will be useful for public actors, to be able to have a more informed discussion with manufacturers, to improve Customs Management, to levy taxes on non repairable products…
  • Fostering new business models - This could potentially help a new kind of startups emerge, based on the resale of old products. What for instance BackMarket has achieved or product electronics could be more easily achieved on over product categories, at larger scale.

Externalized costs: the toy radio example

Excerpt from "The Story of Stuff"

  • I was thinking about this the other day. I was walking to work and I wanted to listen to the news so I popped into this Radio Shack to buy a radio.
  • I found this cute little green radio for 4 dollars and 99 cents. I was standing there in line to buy this radio and I wondering how $4.99 could possibly capture the costs of making this radio and getting it to my hands.
  • The metal was probably mined in South Africa, the petroleum was probably drilled in Iraq, the plastics were probably produced in China, and maybe the whole thing was assembled by some 15 year old in a maquiladora in Mexico.
  • $4.99 wouldn’t even pay the rent for the shelf space it occupied until I came along, let alone part of the staff guy’s salary that helped me pick it out, or the multiple ocean cruises and truck rides pieces of this radio went on. That’s how I realized, I didn’t pay for the radio.
  • So, who did pay?

Repair options: the example of textile

Achat ou seconde main: le cas d’un jouet en bois

  • Functional objects are thrown away because they no longer have a use, and it is complicated to sell a wooden kitchen replica

Lending stickers

  • Support the P2P lending initiative digitally
  • Allow proximity sharing
  • In Cities, allow people to stick QR-Codes to their mailboxes