Is it okay for producers to overwrite user-contributed data with bad data daily?

So here I’m specifically thinking of Fleury Michon which apparently has a process to bulk import their nutritional information daily. See for instance the changelog for their “Rougail Saucisses” product :

That they import data would be fine. But it does not match the nutritional table printed on the packaging. So either they put incorrect data in the OpenFoodFacts database, or they put incorrect information on the packaging which would be false advertising and maybe worse. At least it’s not big discrepancies, and it does not even impact the Nutriscore.

Also, it looks like they create a ‘commit’ even if they have nothing to change. This pollutes the product changelog and presumably wastes space in the database. Just look like at that product’s changelog: it has 561 entries, 465 of which are by the fleury-michon user!

It’s also not a single product but rather a relatively large subset of their lineup:

Note that the insane number of commits issue does not happen across their whole lineup. For instance there is no issue with “Le Saumon à l’Oseille” (3302740003837, not a single fleury-michon commit) and “Les Ravioles du Dauphiné” (same, no fleury-michon commit).

So does anyone know which data source is correct: flauery-michon’s imports to OpenFoodFacts or what’s written on the packaging?
If the data imported by fleury-michon is correct and the nutritional table on the product is wrong, what should go into the OpenFoodFacts database? Is it okay for OpenFoodFacts to not match what the packaging says?
Is it acceptable for bots to create hundreds of empty commits?
More generally, what’s to prevent a producer from using a bot to overwrite any information they don’t like?

I also find it pretty ironic that I have no trouble editing the Fleury Michon products when for other products my edits are silently dropped by OpenFoodFacts. It feels like I am only allowed to edit products if my contributions will be overwritten within 24 hours.

2 Likes

This is very important and thank you so much to point this out.
In my opinion, we need to block this kind of bot if we dectect this kind of errors.

No, it’s not ok. Data should always match what the packaging says.

Thanks for the report. We’ll investigate this and let you know.

cc @manoncorneille @stephane @alex

Could it be that the same barcode is used in two different countries but with different nutritional information? For instance maybe fleury-michon is sending the Belgian fat content, 6.6 g, while in France I see 6.5 g for the same 3302748427024 barcode.
Is that something that OpenFoodFacts can handle? It clearly has provisions for multiple languages but I’m not sure if that applies to the nutritional table.

Just trying to envision all possibilities.
(but my bet would be more on Fleury Michon’s database being wrong, or that they ordered a huge batch of packaging which is now outdated)

1 Like

We track different values for languages just for a few fields, and nutritional value is not one of them. On the other hand, I don’t think the measurement varies between countries.

Maybe they could have a new formula, but the packaging photos could still be old. As they only upload the text fields (AFAIK) we have no idea of what’s the right answer.

Hi @fgouget , thanks a lot for your message.

It’s not exactly a bot : Fleury Michon sends us data, and I have written a script to import it. It’s possible that the script is buggy, it has been years since I wrote it.

It looks like Fleury Michon has not sent us new data since 2021, I will deactivate the script while we investigate.