Recipes feel like magic and they scare me, so I figured out how to program myself for cooking.
In Cooking for Geeks by Jeff Potter, he describes a process of recipe-making where you use the previous efforts of others to come up with the common ratios and their variances, as well as common modifications. By combining them together, you can come up with your own version of a recipe that fits your needs without having to do all the initial work that got us here. Why reinvent the bread? For most things I like to work from a "first principles" approach, but cooking is already challenging enough that I like Potter's suggestion to lean on the experience of others.
I do have a couple of other "needs" for a recipe:
Smallest feasible serving count. The minimum is usually based on indivisible things like "one egg", but when it isn't, I'd like to know what it takes to make only six dumplings so I can scale up, rather than dividing sixty down. I think this is because lots of recipes will base their standard amount on package sizes like "one pound of ground beef" or "16oz of condensed milk", but it feels like I'm being less wasteful from the start if I know the smaller numbers.
Smallest mental impact. I've got a lot on my mind, so the easier a recipe is to keep in my head, the better it is. Sometimes this means fudging the numbers a little, but I'm okay with interpreting "1/3 cup" means "more or less", instead of having to comprehend "5 tablespoons and 1 teaspoon".
This would actually be solved by switching to metric units, so I'm trying to do that!
Here's how I distill out the Platonic ideal version of a recipe:
Figure out what I'm trying to make.
This can be harder than it sounds, because "coffee cake" can mean an American-style with streusel topping, a version with a glazed topping, something more traditionally German that involves nuts, or European-style coffee cakes that are an entirely different end result. This is when I get the chance to realize that when I say "spaghetti" what I really mean is "spaghetti with red sauce".
Bring up at least half a dozen recipes.
I look around in the published recipes of cooks who have flavors that I like, but once I've added their version, I try to find ones that feel like they might be bringing something different to the table so I have a variety of opinions. This usually doesn't end up going anywhere interesting, but sometimes I find that there are regional variations of what "cornbread" means, and I get to learn that there are new ways to enjoy something.
Enter data in a spreadsheet.
I list out the major ingredients, seasonings, and cooking instructions into a column for each recipe, trying to normalize the quantities. Flour gets measured by weight, pretty much everything else gets measured by volume or unit, and I convert everything into single units. This is another place that would greatly benefit from converting to metric measurements, because I can consistently say "ml" instead of mixing between "tsp", "tbsp", "c", and "fl oz", which is easy to make mistakes with.
Look for patterns.
This is usually when I notice there's a "chewy" versus "soft" cookie recipe, or a version with butter versus vegetable oil, and several of the recipes will follow one pattern or the other with minor variances. This is my favorite part of the process, because it's where I notice that a miso rice recipe I thought was going to be straightforward has a schism between white miso users versus red miso users.
Devise a Platonic ideal recipe.
Once all the data is in place, I calculate the median value and the standard deviation to get a rough idea of what the "right answer" is, and come up with the first theoretically ideal recipe. Then I test it. If it works, then I have a new recipe!