Gina took her time, prodding the names of half a dozen dishes to view the finished products, and screens of data on the design of the ingredients. She said, “You can work it all out, if you pay attention. If you know what genes they moved from where, and why, you can make a fair stab at predicting the taste and texture.”

“Go ahead, dazzle me with science.”

She hit the CONFIRM ORDER button. “The green leafy stuff will taste like spinach-flavored pasta – but the iron in it will be absorbed by your body as easily as the heme iron in animal flesh, leaving spinach for dead. The yellow things which look like corn will taste like a cross between tomato and green capsicum spiced with oregano – but nutrients and flavor will be less sensitive to poor storage conditions and over-cooking. And the blue puree will taste almost like Parmesan cheese.”

“Why blue?”

“There’s a blue pigment, a photoactivated enzyme, in the new self-fermenting lactoberries. They could remove it during processing, but it turns out we metabolize it directly into vitamin D – which is safer than making it the usual way, with UV on the skin.”

“Food for people who never see the sun. How can I resist?” I ordered the same.

The service was swift – and Gina’s predictions were more or less correct. The whole combination was actually quite pleasant.

I said, “You’re wasted on wind turbines. You could be designing the spring collection for United Agronomics.”

“Gee, thanks. But I already get all the intellectual stimulation I can handle.”

“How is Big Harold coming along, anyway?”

“Still very much Little Harold, and likely to stay that way for a while.” Little Harold was the one-thousandth-scale prototype of a projected two-hundred megawatt turbine. “There are chaotic resonance modes turning up which we missed in the simulations. It’s starting to look like we’re going to have to reevaluate half the assumptions of the software model.”

“I can never quite understand that. You know all the basic physics, the basic equations of air-flow dynamics, you have access to endless supercomputer time…”

“So how can we possibly screw up? Because we can’t compute the behavior of thousands of tonnes of air moving through a complex structure on a molecule-by-molecule basis. All the bulk flow equations are approximations, and we’re deliberately operating in a region where the best-understood approximations break down. There’s no magical new physics coming into play – but we’re in a gray zone between one set of convenient simplifying assumptions and another. And so far, the best new set of compromise assumptions are neither convenient nor simple. And they’re not even correct, as it turns out.”

“I’m sorry.”

She shrugged. “It’s frustrating – but enough of it’s frustrating in an interesting way to keep me from going insane.”