swestrup | Food Thoughts and Business. (Reply)

Since I've been on Atkins I've been making heavy use of the really lame database system at the USDA website. Its just not too useful for planning a meal. On the plus side though, the raw data is freely downloadable, and a number of Atkin's and other-diet related sites have used it to jump start their own database efforts. The problem is, most of those sites are truely lame too.

As I've been using the USDA database and googling the web, I've been imagining what *I* would like to have in a food website. Here's what I've come up with:

Give the thing a hierarchical, user-extensible database of food nutrition facts, along the lines of IMDB or FreeCDDB. Let folks add data that they read off of nutrition packages and have a dispute mechanism to handle typos and whatnot. (Wiki-style might be good enough). Let folks add new brands as they wish. Thus, when looking up Cheddar, there could be average values as well as particular values for particular varieties of, for instance, Kraft Low-Fat Cracker Barrel.
The database system would try and wing-it whenever it had inadequate numbers. Thus, if you wanted the amount of fiber in Red Onions, it might well give you the numbers for Yellow Onions, if it didn't have them for Red. Such guesstimates would be CLEARLY LABELED as such.
Folks could also enter their favorite recipies via a form system. All foods used in recipes would get looked up in the database, and unknown foods or unknown varients would get flagged for explanations. People could also enter varients of a given recipe and/or substitutions that they had tried, as well as comments and ratings.
The database would try and generalize substitutions it has seen to other cases, and be willing to suggest off-the-cuff recipes by taking known recipes and varying them in ways it has already seen.
With this flexibility, it should now be possible for a user to enter one or more general food preferences (I prefer organic food, I'm celeriac, I'm diabetic, I have Crohn's, I'm alergic to legumes) and then customize them with specific preferences (I looove garlic. I hate lobster. Mushrooms are okay, but I woudn't want a dish based on them, polysorbate 80 isn't food), and have searches take this into account.
It would be able to do recipe searches based on particular nutritional requirements (I need a meal for 4 with less than 20 carbs per serving, I want a snack that's high in fiber, I'm looking for a good recipe for getting my potassium levels up), and would rank them based on estimates of how much you'll enjoy them.
Recipe searches would include being able to say 'Here's what I have in my fridge, what can I make?' and 'I'm makeing quiche, and I'll have lots of egg whites left over, what can I make with them?'
And, of course, it would be willing to present recipes to you formatted for printing on various sizes of index card, as well as willing to sell you a recipe book customized to your preferences.

This could make money in a bunch of different ways. Folks could have to pay a small yearly fee to be registered (although I don't think that's a good idea), and the website could make money off of selling customized mugs and recipe books. The main thing I think this site could make money off of is advertising. If this works right, it should get a LOT of page hits, and that might well equate to reasonable income from food advertisements and possibly cookware endorsements.

It would be a lot of work, but a Version 1.0 would not have to have any of the harder stuff implemented, and it might still fly.