How a nine course menu actually gets written

People assume a tasting menu is a fixed composition that a kitchen rehearses until it is perfect. Ours is not. It is written every morning, cooked once that night, and thrown away. This is an account of one of them: Thursday, November 13, at Tribeca. I kept notes specifically so I could write this, because when I describe the process out loud it sounds either romantic or chaotic, and it is neither. It is mostly a scheduling problem with taste attached.

6:40 AM, the phone call

Every morning starts with Danny at the fish supplier, who calls me rather than the other way around because he knows I will buy whatever is genuinely good and not argue about the list I sent yesterday. On this Thursday he had eleven pounds of monkfish tail landed at Montauk the previous afternoon, four pounds of very good sea scallops, and no turbot at all, which mattered because turbot had been course five all week.

So course five was gone before I got out of bed. This is the normal state of affairs. Roughly two of nine courses change for supply reasons on any given day, and one more changes because someone in the kitchen has a better idea by mid afternoon.

7:20 AM, the produce list

The Rhinebeck delivery arrives with a printed sheet and about forty percent of what I asked for, which is again normal in November. What we got: black trumpet mushrooms, unusually good, twelve pounds. Celeriac. Two crates of small beets with the tops intact. Cabbage. Winter squash I did not order and did not want. No chicories, because the frost had come early and they were not ready.

The black trumpets are the decision point of the day. Twelve pounds of a mushroom that only appears for about three weeks is not something you use as a garnish, so somewhere in the menu there has to be a course that is mostly about them.

8:45 AM, the constraints get written down

Before anything creative happens, Marcus and I write the constraints on the board. This is the part nobody photographs and it is ninety percent of the work.

Ninety two covers booked, in four seatings. Fourteen of those covers have dietary notes, including three full vegan menus, two shellfish allergies, one severe tree nut allergy and one guest who has told us honestly that they do not like anything raw. Twelve of the covers are the same regular party who ate here nine days ago, so at least four courses need to be different from that night's menu, which is framed on the wall and easy to check.

Then the physical constraints. The oven can hold four roasting trays. The soufflé needs twelve minutes and there are only six copper moulds, which caps the dessert at thirty covers per seating unless something else carries part of the load. The chef's table takes two cooks out of the main brigade for the whole service. And the second seating is fully booked at 7:45 PM, which is the worst possible time because it collides with the first seating's meat course.

9:30 AM to 11:30 AM, the draft

Writing the menu means arranging nine plates so that the room's energy rises and falls twice, the kitchen's workload stays flat, and no two consecutive courses use the same technique or the same temperature. Those three requirements fight each other constantly.

What went on the board, in order: oyster with frozen mignonette, because it is cold, fast and needs no oven. Charcoal gougère, because the pastry section has capacity at the start of service and nothing to do. Scallop crudo, using Danny's scallops raw. Then the black trumpets, as a warm tart with a barley koji custard, which is the dish the whole menu is really for. Monkfish over binchotan for course five, replacing the absent turbot. Celeriac and pickled walnut as a palate reset, and immediately a problem because of the tree nut allergy, so a second version without walnut gets written alongside. The aged duck for course seven, twenty one days that morning. Cheese. Then the soufflé, capped at thirty covers, with the chocolate tart as the second dessert to spread the load.

That is nine, and I drafted eleven, because two of them are always wrong.

1:00 PM, family meal and the argument

The kitchen eats at one, and this is where menus get better. Junsu said the soufflé and the tart back to back read as two chocolate courses even though only one contained chocolate, because the burnt honey ice cream drank so dark. He was right. The tart became the yuzu sorbet, which also solved a problem I had not noticed: there was no acid anywhere after course six.

Marcus argued the celeriac should come before the monkfish rather than after, because the fish is the better dish and should sit closer to the duck. He was also right, and it cost nothing to swap.

4:15 PM, the tasting, and two dishes die

Every plate is cooked and tasted before service by everyone who will be serving it, standing up, at the pass. This takes about twenty minutes and it is the only quality control that actually works.

The squash dish, which I had put in as a tenth option, was bad. Not broken, just pointless: sweet on sweet, with nothing to argue against. It went in the bin and the squash became staff meal for the next two days.

The scallop crudo was worse in a more interesting way. The scallops were excellent and the dish was fine, but it sat two courses away from the oyster and read as more of the same, cold and saline and clean. So the scallops got seared instead and moved to course four with the black trumpets, and the tart became a garnish rather than the centre of the plate. That is a real change made ninety minutes before the first guest arrived, and it is the reason we do not print menus until 4:35 PM.

7:45 PM, none of this is visible

By the time the second seating sits down, the menu is a single sheet of paper with nine lines on it, most of them three or four words long. The guest sees monkfish, squid ink, wild garlic. They do not see that it exists because a turbot boat did not come in, or that the course before it moved twice, or that the person plating it argued for that order at lunch and won.

I do not think guests need to see any of that. But I have been asked what we actually do all day often enough to want it written down once, and the honest answer is that we spend the morning finding out what we have, the afternoon deciding what to do about it, and twenty minutes at 4:15 PM throwing away the things that did not work.