Although I always tell my students that they should read every recipe three times before starting, I’m definitely guilty of not following my own advice. Case in point, on page 274 of the fifth volume of Nathan Myhrvold’s Modernist Cuisine is a recipe entitled “Crispy Goat’s Milk Ricotta Dumpling.” The recipe is listed as having been adapted from one by Wylie Dufrense of the restaurant wd~50, in New York City. When I tried the recipe, it kind of worked but not necessarily as planned.

When I saw the recipe, I thought that it was approachable—many in the book are not to be started without thousands of dollars of equipment. I had all the ingredients on hand except for the Ultra-Tex 4. Maybe I could use a different starch? After all, one of the aphorisms I tell my students, “You cook with the ingredients you have, not with the ingredients you wish you had.” (These words of wisdom are not original. I learned them from another cooking teacher, Adam Weiner.)

So I preceded to mix the ricotta, cream, salt, and pepper together as instructed. I measured out the instructed amount of starch, substituting tapioca starch for the Ultra-Tex 4, and folded that into the previous mixture. I should have realized that all starches are not modified equally. The mixture did not come together as a dough as the recipe implied that it would—one of the recipe illustrations showed the completed mixture being rolled out in ropes as I do for gnocchi. This mixture I had created was definitely not roll-able.

But that didn’t mean it was lost. I remembered the process of making gnocchi à la parisienne where pâte à choux is piped into boiling water to produce the gnocchi. So I loaded the dough into a disposable pastry bag fitted with a 12-mm (1⁄2-in) plain tip. I held the bag at the edge of a pot filled with hot oil so the tip extended over the edge of the pot. Using a small knife, I cut off 2-cm (3⁄4-in) long sections of dough so they dropped into the oil. These were cooked until they browned and then transferred to a plate with absorbent paper to drain. They were actually pretty good, but there was one problem.

When I started the recipe, I had in my mind that it would be nice to stuff a piece of a different cheese into each gnocchi before cooking. By using the piping method I wouldn’t be able to stuff the individual pieces. So I went online and ordered some Ultra-Tex 4.

Ultra-Tex 4 is a modified food starch produced by National Starch and Chemical Company. The company’s data sheet describes it as “a premium cold water swelling modified food starch derived from waxy maize.” It goes on to state that mixtures made with the starch “are extremely resistant to harsh processing conditions such as intense heating, high shear and low pH.” So that’s why the recipe uses this starch. Plain starches requires heat to begin to swell.

A few days later when my bag of Ultra-Tex 4 showed up in the mail, I prepared a new batch of the dough. (This time I increased the salt slightly and left out the pepper. Other than that I use the proportions as written.) Sure enough, this time the mixture could easily be rolled into ropes like in the picture accompanying the recipe. But upon further reading—remember I have a habit of not reading recipes—I found that, according to instruction 3, the rope should have been “3 cm ⁄ 2-1⁄4 in. in diameter.” I had rolled mine closer to 12 mm (1⁄2 in). Whoops. No wait! 3 centimeters is equal to 1-3⁄16 inches and 2-1⁄4 inches is equal to 5-3⁄4 centimeters. And the finished pieces shown in illustration 5 are nowhere near either of those two sizes unless the fingers in the picture belong to a giant. The same dimensions are given for the length in instruction 4, which also doesn’t match the picture. I cut mine about 2 cm (8⁄10 in) long.

Since the recipe doesn’t state a temperature for the deep-fry oil, I assumed 180 °C (355 °F), which is pretty standard. After chilling the gnocchi as specified, I fired up my oil, and threw a couple of pieces in for a test drive. Crash! The oil sputtered and the gnocchi split and started to rapidly exude little white ricotta globules. In no time, the oil was black and smoking. A mat of little brown globs coated the surface. In less than of few seconds, my beautiful ricotta gnocchi transformed into a gloppy mess.

I transferred the remaining uncooked gnocchi to a small baking sheet and placed it in my freezer. It was time for some serious thinking. I still had at least 20 gnocchi left. How could I save them?

The answer came that evening in the form of fried fish. I often panfry fish fillets using one of many coatings. One of my favorite is to just dust the fillets with flour or starch. Flour produces a nice crispy surface. Cornstarch also is mostly crispy. Tapioca starch is both chewy and crispy. It also does the best job of sealing all the moisture in the fillets. What if I coated the gnocchi with tapioca starch?

The next morning I thawed a few of the gnocchi. Now, instead of the surface being dry like a dough, it was moist and a little sticky. The Ultra-Tex 4 may have been exhibiting some syneresis. No problem, moist and sticky is good. I dumped the gnocchi into a bowl filled with tapioca starch and rolled them around a bit. Then I let them sit in the starch while I heated the oil.

After the previous day’s fiasco, I decided to start with just one of the gnocchi. I transferred it to the oil without disturbing the starch. This time, no explosion. The gnocchi sizzled in the hot oil, and when it had browned, I transfer it to a plate lined with absorbent paper. Towards the end of its cooking, the gnocchi had split a bit, but no cheese spilt out. I popped the hot gnocchi in my mouth. Wow! It was hollow. It seemed like there was nothing inside except for an intense cheese flavor. Hollow, but flavorful. I couldn’t even sense the stick of cheese I had inserted into the gnocchi.

I decided to try something that I normally criticize others for: layer flavors. I had, in my refrigerator, part of a wheel of cheap brie I had been given at school. I sliced some very thin, maybe 1-1⁄2 mm (1⁄16 in) thick, strips from the edge of the wheel. I laid these out flat on a baking sheet and while the next batch of gnocchi was frying, I blasted them very briefly with my butane torch. The edges charred a little, and the cheesy part in the center softened and warmed. When the gnocchi came out of the oil, and after a brief draining, the brie strip was wrapped about the gnocchi. The taste was, as my long-deceased German uncle would say with a rolled “r,” marvelous. I decided that these were so good that every serving should have two portions.

The proportions that I wound up using for the basic dough were similar to the original recipe: