Show HN: Create a perfect meal with linear programminga | Hacker News Reader

Show HN: Create a perfect meal with linear programminga

mathfood.com

27 pointsjwally26 comments

Nice job! However I don't clearly understand the Results section. I mean, I can work it out, but it should be obvious.

With your supplied example, here was my train of thought when I hit solve: I get "Egg whites: 239.05". Am I supposed to eat 239 eggs? Probably not -- that's a lot for one meal. So there must be a unit; grams? well I don't know how much an egg weighs, plus I don't know the proportion of the white within a full egg. I certainly won't open Wikipedia right now to solve it!

That said, I love the design, especially the top illustration, so my point was just: if you can make the results more human-readable, that would help.

I don't get the final result. It's in grams? I then have to figure out how many eggs or chickens make up 300 grams? Weighing everything gets annoying. I'd rather eyeball.

For me it's clearer to work by pictures of the food, like http://office-k-chiro.com/wp-content/uploads/2014/02/office-... that's what I'm dealing with IRL and it's easier to memorize 5 jumbo eggs vs 300 grams.

And by that logic, I'd rather drag and drop pictures to build up the meal, get the nutritional information after than the other way around.

It is in grams.

Watch the video demo from around right here: https://www.youtube.com/watch?v=irTdDJ6fu-c#t=97

When I'm doing the same problem in the actual tool, there's a "details" button which pops up a screen telling you the meal's details and how much of each food to eat.

You can click on the foods in this area and it'll tell you how many eggs 300g egg whites requires (I actually click on cheese in the video, but you get the idea).

Yea that's pretty well hidden in the UI. My attention span didn't last to the end of the video.

This is generally quite nice. To add to some of the other feedback, I felt it could do with more verbose help text and more importantly for my brief play with it, more focus bringing basic staples to the top of results for food search.

I tried to add just bacon and it took rewatching the demo video and some guesswork to figure out searching for 'bacon raw' would get to basic bacon. Trying just 'bacon' or 'pork' only gets lots of babyfood and soup related answers in the generated dropdown.

The search definitely needs work. The food database is from the USDA so the descriptions are wordy. For what its worth, when you're searching you can use an exclamation mark to negate a term. For instance "!baby !cere banana" gets me all banana related foods that don't have "baby" or "cere" in them (which for some reason is a lot).

I understand the minimum amount for protein while building muscle and the maximum amount of carbs for keto, but shouldn't the general use-case be to keep the amount of calories in check per meal?

Making a meal consist of 1400 grams of carrots because there was an automatic ceiling on chicken due to protein seems silly.

Seems like its a good app, but it needs sane defaults.

Thanks for the feedback! I tried showing it in the video (maybe needs more explanation) but you can set a max and a min on a diet item. Say you want to eat Egg whites, cheese, potatoes, and ketchup. Because ketchup and potatoes are both carbs the tool will probably choose one or the other; not both. What you could do is set a max on the amount of ketchup @ say 28 grams. That way all of your carbs aren't coming from ketchup.

This looks really cool but... Carbs, protein, fat and... cheese?

:D Was trying to demonstrate that its possible to set boundaries on the food items you pick. If I set the cheese requirement @ 0 (technically min=0, max=0), I can double the amount of bacon I get. It allows you to decide which is more important: Cheese or Bacon.

Cheese is Cheese. It's a different nutrient group because it's cheese.

It looks nice, but I am not sure as to what new things it brings over, for example, eatthismuch.com (who generates menus from an already populated food database).

Thanks. I hadn't seen eatthismuch.com before. Very clever.

Eatthismuch (from what I've just now looked at) appears to approach the problem like this: "What are your diet goals? What kinds of foods do you like to eat? Ok. Here's a menu telling you what to eat."

There's a thing (that's starting to die down now) with nutrition called "if it fits your macros" meaning that you can eat anything that you want to eat (pizza, hot-dogs, hamburgers, tacos, etc) as long as your ratios are kept sturdy. Mathfood helps you do that.

The only thing I didn't like was, the text over the graph paper. It was a little hard for me to read. But its a cool idea, I like it.

This thing rocks! I always thought about building something similar - didn't know it's called linear programming though.

Looks interesting... How is this different than MyFitnessPal + a Macros calc? is open source?

Open source? The LP engine is, but the site probably has a bit of code stink in it @ the moment. Once I get some time to clean it up, I might open source it. Benefits to open sourcing?

Not super familiar with MyFitnessPal, but MathFood allows you to say "I want waffles, eggs, and bacon for breakfast but I want to keep to my super strict diet ratios. How do I build this meal to hit those ratios?".

What's the LP engine please? Thanks!

Can't promise its pretty but: https://github.com/JWally/jsLPSolver

Yeah. Or how is it different from DigitalOcean + A custom web app that I build?

Would love some feed back on my first web-app. Thanks!

Technically, I should be really interested in this app as I've been spending the last few months trying to hack my diet. However, I tuned out after a few minutes. Specifically: * There is a huge wall of text. Most users (or at least me) want to learn what the product is by playing with it, not by reading. * Lots of form fields before you get value. I looked at the start point and immediately walked away. A large form scares away users. * It's weird that your height input only takes total inches instead of the normal feet + inches (users don't want to do math) * Not sure what the value is here - if I'm a consumer and I actually know the % calories across carbs, protein, and fat then why would I need this app? people who have a metric for that (like I do) probably fall into a really small niche of health obsessed people who already know how to make their diet hit their targets. I use myFitnesspal and I know exactly how to hit my targets.

Basically my summarized feedback is focus on the UX and a broader demographic. This is an awesome problem to tackle and I like your approach but I don't think many people could benefit with it the way its currently setup.

FYI this is awesome first web app - the amount of functionality you've provided is awesome.

Awesome points. Thanks for the feedback. You're right, target audience is pretty niche. Bodybuilders, figure athletes, gym rats, etc.

I probably didn't communicate it well but the value of the tool is this: How do you build a meal that meets your diet goals (fat, protein, carbs) given that foods have a mix of each type.

Interesting way to go from problem to solution ... not sure if I can apply it (I don't think of my diet in terms of "I need x grams of protein today") but I bet there are people who do.

I think it would communicate better if it explained why "linear programming" is interesting here, and briefly what it is, rather than just linking to Wikipedia. Also, there are a lot of basic typos, you should run a spell-checker or something.

Awesome idea, it would be nice if the application proposes some meals that are a good match for the daily needs. It should be possible since you have already a big database of foods.

Very nice, but it's missing units in Results.

Reply on news.ycombinator.com