“Yeah, That Feature Should be Easy to Do”

August 6th, 2013

Today, Basil received a nice 1.6 update. This update brings a really obvious, simple feature that many other recipe applications include: ingredient scaling. It’s really convenient while in the kitchen to be able to adjust a recipe’s yield according to your needs (the amount of jambalaya you need for two on a quiet weeknight is a little different than the amount you need to make while having guests), and it’s conceptually very simple.

Except it’s not so simple.

It sounds like it, of course; when you think of doing ingredient scaling, it’s just taking an amount at the beginning of a line (“1 item,” “2 tablespoons”) and scaling it by some factor—cutting it in thirds, doubling it, whatever. That’s the approach most applications take for ingredient scaling: they assume all ingredients are phrased such that there is an amount at the beginning of the line and then a food item. Based on that assumption, they identify a number at the beginning of the ingredient line and scale it by whatever scale the user selects. In the basic cases, like “2 cups white sugar”, this works OK. Not great, but OK.

But the problem is that not all ingredients items are the basic case, because there are many ways to express things. What about ingredients like “1 (12 ounce) box of pasta” versus “2 medium cloves garlic (about 1 teaspoon)”; in the first case, the measurement amount shouldn’t be adjusted, but in the second, it should be. In the second case, assuming that you only have to adjust the amount at the beginning of the line can be very confusing while cooking: After I double the recipe, do I use 4 cloves of garlic, or 1 teaspoon? This assumption—made because parsing imprecise language like this is difficult—actually makes cooking harder for the user, because they have to remember how their application fails at correctly scaling ingredients.

A new feature for cooking isn’t helpful if using it actually makes it harder to cook. There are many other cases; what about scaling amounts expressed as words, like “half”? Or what about when you cut “1/2 cup” in fourths and it tells you to use “1/8 cup” rather than “2 tablespoons?”

Scaling ingredients isn’t useful if it makes you think about these things while cooking or doesn’t work for some ingredients. It’s a distraction.

I avoided building a scaling feature for Basil for that reason: I didn’t want to add more noise if it had all of these caveats. But I’m happy to announce that Basil now has ingredient scaling, and it does its best to do the right thing in all of these cases. It handles the cases discussed above, and many others, too. It will do its best to present ingredient amounts in a way that makes sense (no “1/9 cup sugar” or similarly bizarre things).

It isn’t perfect. There are probably cases I’ve missed, or cases it’s not handling properly, but I am committed to addressing those.

Working on ingredient scaling, though, reminded me that problems which seem exceedingly simple and easy to solve on the surface are usually icebergs waiting to sink you. They may look simple, but they tend to conceal much complexity underneath the surface that is only obvious once you’ve thought through the problem. These “yeah, that should be easy to do” features have, more often than not, been the hardest problems to solve in my experience.

This experience is also another lesson that while users like receiving new features and they’re exciting to announce, over the longer term users only want a new feature if it improves their net-experience in some way. If it ends up not being useful in many cases, or adds complexity to the process of using your application, then that feature could make your application less useful than it was before.