Hacker News Reader

Thanks for your feedback.

I realize this won't ever work perfectly for every type of dish. But the dishes I have tried using the given ratios have all come out much better then my previous unaided versions of the same dish.

The ingredient database I am working with definitely needs an overhaul. The original focus of this db was matching ingredients broadly, and that is now becoming a major problem.

I think the ingredient database you are looking for is http://www.ingredientpairings.com/. This brings forward more of the broad matching ingredient problems though. When I redo the ingredient database this should be more useful.

A pretty basic point to make here is that many recipes --- perhaps most --- don't scale linearly with ingredients. There's usually a core ratio you can distill out that can scale arbitrarily, but then tweaks to other parts of the recipe.

The bigger problem I have with every service like this is that the recipes are untrustworthy. For instance, I'll generally avoid recipes from Epicurious. The curation problem here seems hard.

Can you please share why you avoid Epicurious?

The recipes they have from Gourmet seem to me to be the best large collection of online recipes: diverse, well-written, tested, proofread, and forthright in their descriptions of each dish. They often have helpful user reviews as well. Some recipes tend to be more time consuming or require rare ingredients or equipment. But generally I'm relieved if (assuming I've already exhausted my proper cookbooks) Google gives me an Epicurious/Gourmet recipe as opposed to another site. The Bon Appetit, Shape, and other recipes on Epicurious aren't as uniformly good.

What are other sites that you do prefer? I'd be happy to discover something better.

Cooks Illustrated is pretty much the only website I regularly pull recipes from. I've taken some things from eG, since so many pros used to be on there. On the rare occasions that I ask Google to find a recipe, I'll take a FoodTV recipe if it's from a show by someone who actually owns a restaurant.

The rest of the Internet is rife with people who think mayonnaise goes in Caesar Salad.

Genuine question on this. Why can't you just scale a recipe directly? Are there ingredients that play a binary role (so to speak) so they just need to be present regardless of the quantity? What am I missing here?

I cook a lot and cannot think of any totally non-scaling ingredient, although some things go out of proportion a bit; e.g., the relative volume of water used to cook rice goes down as the recipe grows.

I feel like this is pretty well known; not that recipes can't be scaled, but that you can't reliably do it by multiplying the ingredients.

I just flipped through Ferran Adria's "The Family Meal", which is a record of every house meal served at El Bulli, each of which is scaled to 2, 6, 20, and 40 people. Sure enough, almost none of them scale every ingredient linearly (Caesar salad and hamburgers did, because you just make more dressing or more patties for more people).

In particular: oil, butter, lemon juice, onions, and intensely flavored seasoning ingredients (herbs in particular) often scaled by 2x when the servings scaled by 3x. In some places, an ingredient scaled by more than 3x to make 3x as many servings. Salt went in both directions.

It's interesting. Wholesale scraping of thousands of other peoples' recipes is ballsy, clever, but makes me a bit queasy.

You probably already know this, but the ingredients lists are not copyrightable according to [http://www.copyright.gov/fls/fl122.html]. It seems from reading that site that the list of steps that gets loaded when you import a recipe may be subject to copyright, depending on how closely those steps resemble the original recipe's instructions, and whether the original instructions were copyrightable.

I'll agree that common flavor pairings, frequently associated spices, and best techniques for a given ingredient are the more interesting data. It'd be great to ask the database what spices are frequently used in Thai food, for example. Combine it with a list of local suppliers of hard-to-find or imported ingredients and you have a winner.

It's a long, long tail before they call it "stealing". -Josh Strike

So, what you really want is _The Flavor Bible_ (you can get it on your Kindle), which is an intensely well curated, expert-driven dictionary of flavor pairings. For instance, Thai: Thai basil, bell peppers, CHILE PEPPERS, cilantro, coconut, coriander, cumin, curries, fish, fish sauce, garlic, ginger, herbs (fresh), lemongrass, lime, mint, noodles, peanuts, rice, shirmp paste, sugar, turmeric, vegetables.

The basic database you're looking for already exists (and, if you cook regularly, at a very reasonable price). You should just buy the book. Yes, it does beg to be a web app, but you need the expertise behind the book more than you need the convenient interface. I'm wary of convenient- but- bad- answers to important questions.

I typed in duck confit, and was quickly able to remove duck, and add duck legs and duck fat, to the recipe.

That is a feat that makes this app almost as impressive as Notepad.exe. "Almost", because Notepad didn't start out by lying about the ingredients.

(Sorry to be so snarky about it).

With only three clicks, using the suggested ingredients, I removed one ingredient, added two others, and was told the proportions automatically.

Notepad neither suggested the alternatives nor updated the amounts required.

You're saying that, knowing what the ingredients for duck confit were beforehand, you were able in 3 clicks to make this app say what the ingredients for duck confit were.

A better test (I make a lot of duck confit): do you know how to make duck confit now? Did you learn from this app? What are roughly the steps you learned?

Well, what I'm saying is that, knowing roughly how to make duck confit but being unsure of the proportions, I was able to find out what I needed to know easily. You, on the other hand, weren't able to find out something you already knew; that doesn't sound like a problem to me. If you know exactly how to make something (or you don't know at all how to make something at all) then your right, don't bother using this app.

A good test for an app that is designed to teach you how to make a dish is: do you know how to make it now? But that's a bad test for this app, which helps you build recipes that you already know something about.

I had the other impression about the app; that it was intended to tell you what you needed to make for a dish you didn't know how to make (because if you did, you wouldn't need the app to tell you what you needed).

to get technique, you have to click through to one of a zillion recipes

Sounds like the Cooks.com approach that was one of the first sites I blocked when Google instituted their site blocking functionality.