Inaccuracies and lack of context lead to misleading info

[vertigo220]

I love the idea of this project, as people rarely know much about nutrition and about what's in their food, but unfortunately I find it to not only not be useful, but to actually be potentially harmful due to a multitude of issues with how the data is interpreted. Some examples:

1. It doesn't seem to differentiate between added and non-added sugar in the score (and many products' nutrition facts don't list them separately), counting any sugar as bad. This is seen e.g. when looking at That's It fruit bars that contain apples and various other fruits and nothing else, yet the score is dragged down due to sugar. Yet if you look at the grams sugar vs grams per serving, it's actually a better ratio than it is looking at the whole fruit. It should not only take this difference into consideration and not penalize a food simply because it has high sugar (according to the app, an apple would be moderately unhealthy at a "C"), but it should potentially penalize more for added sugars in certain situations (like added sugar to sweet potato fries), though that of course would be tricky.
2. Also on the topic of sugar, it doesn't differentiate or really make much of a deal about the use of HFCS vs regular sugar. Not only should this count as more negative points per gram, but it should be a prominent red flag and should provide more info about it. It should also be smart enough to look for any of the many names used for HFCS and group them under that. As it is, something like Ritz crackers with 6.25g added sugar per 100g only has 1 negative point for sugar whereas sweet potato fries with 3.57g added sugar (total of ~14.25g) has 3 negative points, despite the fact the Ritz crackers not only have nearly twice the added sugar, but contain HFCS.
3. It should also consider other factors, particularly fiber content, when ranking the sugar content (perhaps it already does but it didn't seem to).
4. Sweet potato fries have a low score due to low nutrition content including the fact they contain no fruits/vegetables (per the app). Arby's fries, which are certainly not a "healthy" food, are relatively decent as far as fries go, also have this issue, with 0/5 positive points (2.9%) for fruits/vegs/legumes, despite being made from potatoes. And I suspect this is a common issue with various other food types and products.
5. Calories seem to be simply regarded as bad, causing reduced scores simply due to their presence, seemingly ignoring the fact they are ultimately what's needed for fuel to survive. Yes, empty calories are bad, but it doesn't seem to make that distinction. A calorie-heavy food that's also nutrient-rich is perfectly fine, so long as the expected calories per "meal" are reasonable (which of course can get tricky, and may require having the user enter details like height/weight/sex/age to help determine that). Note that's not per serving, for what should be obvious reasons. But it's not as simple as x vs y calories, where a food containing x is automatically worse than a food containing y simply because x > y, and that's what seems to be going on here.
6. It's unclear where the estimated ratios of ingredients is coming from, and a) doesn't seem likely that it's very accurate given it doesn't seem to be based on anything other than random guesses, and b) it should be more clear/transparent about that or about what's driving those estimates.
7. It should factor in different oil types (mono- vs polyunsaturated, omega ratios, etc, i.e. oils vary in how healthy they are) and, on that note, should point out and, IMO, deduct points for the all too frequent cases where ingredients state "one or more of" or "and/or," essentially meaning you don't actually know which oil(s) are being used and in what ratios, which is a personal pet peeve of mine.
8. It seems to regard fat and saturated fat as strictly bad, similar to sugar (note that sugar is certainly "bad" in general but, as mentioned, there's a difference between added and not) and calories, despite the fact sugar is often used in place of fat, and between the two, fat can often be considered the lesser of two evils, especially unsaturated fats (and especially the healthier options of those). I'm not saying it shouldn't be rated poorly, but again, it requires context and shouldn't be so overly simplified.
9. Finally, also regarding fats, it appears to completely ignore trans fats. While trans fats are supposedly illegal now, that doesn't seem to have stopped companies from making food that contains them, I suspect due to the loophole that allows them to claim no trans fat so long as it's < 0.5g per serving. This is the reason I have refused for years to buy any Nabisco product, as they spent years prominently displaying the lack of trans fat on their packaging (Ritz crackers), when really they contained, and still contain, trans fats, just under this absurd limit. And while you should already know this, the giveaway is the presence of a "partially-hydrogenated" oil in the ingredients. Any food that contains this (or frankly any "hydrogenated" oils that don't explicitly state "fully-hydrogenated," as I simply assume their lack of using the latter term is due to trying to be sneaky, as food companies have no integrity and are constantly coming up with ways to deceive consumers, and so I have zero trust in them and therefore have to make such an assumption and will not buy any food with any non fully-hydrogenated oil) should be severely criticized and downgraded. This should be it's own scoring criteria, probably with an automatic loss of a full letter grade (and personally I feel it should be more than that if there's deception).
10. And I guess that's one more point: deception. There are a lot of tactics employed by food companies, including the trans fat reporting, the and/or use with oils, the multitude of names for HFCS, the use of 2-3+ different forms of sugar to be able to list them further down the ingredients list, things like calling a mayonnaise "Olive oil mayonnaise" or "Avocado oil mayonnaise" and then having soybean oil as the main ingredient/oil and simply having olive/avocado oil in some unknown, lesser quantity in it as well (or even having olive/avocado oil first and still containing other oils, which is still diluting the preferred, better oil people are paying more thinking they're getting). There should be a mechanism for calling out these products/companies so people can choose to go with alternatives that aren't trying to trick consumers at every turn.

On an unrelated note to all of this, the search takes forever and seems like it shouldn't be so slow. Also, the serving size should be stated on the nutrition facts page.

[VaiTon]

Hiu @vertigo220 , sorry for the extremely late reply. I can see this is more a discussion that an issue in using the platform, so I'm moving it to the discussion section.

[Freso]

Hi vertigo.

A lot of your points seem to be regarding Nutri-Score, which is not a scoring OFF controls, but one created by the French public health agency, and, in fact, many of your critiques of it have also been echoed by others. If you have suggestions for other scoring algorithms we could implement (esp. ones backed either by legal frameworks or peer reviewed studies), please let us know!

Regarding point 4, I just want to note that sweet potatoes (Ipomoea batatas) is a wholly different plant to potatoes (Solanum tuberosum). Sweet potatoes have recently, like potatoes, been legally declassified as a vegetable due to their amount of starch content. Whether we, personally, individually agree or disagree with this, it is still something we will reflect in calculations based on legal frameworks.

Regarding 6, there’s a wiki page on recipe estimation, but there’s also currently work on-going to improve the recipe estimator, using more signals (e.g., nutrition values; if there’s 7g/100g sugars, it likely doesn’t make sense for a product to be 50% sugar, for instance). I think @john-gom and @aleene are the main ones doing this work, but I can’t find an issue where you can follow this progress, or get involved.

Regarding points 9 and 10, I think these could be a great subjects for knowledge panels. “This product contains non-fully hydrogenated fats, which are extremely likely to contain trace amounts of trans fat acids.”, “This product is marketed as an avocado oil-based product but also contains other oils/fats.”, etc.