“The version of the Darwin Algorithm that was stolen—version 3.4, which is almost a year old now, had a number of issues that make it particularly dangerous. When I used it to simulate evolution across thousands of generations, it tended to generate anomalies that skewed further and further away from the intended target. For that reason, it was quickly discarded. Subsequent versions were far more controlled and effective, and even so, we continue to tweak them to this day.

“Those in possession of this algorithm would also be unable to make the necessary adjustments to correct its deficiencies. The algorithm itself is merely the end product of many underlying calculations that cannot be easily reverse-engineered. It’s made even more difficult because the code contains hundreds of thousands of lines of manually entered entries—and without knowing how those entries relate to each other and why, you’d be at a complete loss to make sense of them. In fact, you’d have to be skilled in genetics, have done the same types of research—which I’ve yet to publicly document—and of course be a cryptanalyst.

“Which is all to say, whoever stole version 3.4 is stuck with it. Which… is unfortunate.

“This version of the algorithm resulted in a high mortality rate for Rattus norvegicus, the common brown rat. The problem arose in what I like to call line noise in the genome. The algorithm was effective in that after two hundred thousand generations, some of the genes would express exactly what I wanted—but by then there was so much chaos, so much nonsense in the rest of the genetic material, that any attempts to insert the simulated genome into a live specimen resulted in the rat’s death.

“It has taken thousands of clinical experiments—work that is still ongoing—to isolate what we’re looking for. A genetic expression that will kick cancer’s ass.”

A beep sounded behind Juan, and he stopped the dictation. He turned to the rightmost workstation and smiled as the message “Pattern Matched” flashed on the screen.

Workstation 4 had found a matching sequence. That settled it. Whoever was responsible for the avian sample—the feather that resembled one from a Gouldian finch—had been using version 3.4.

Juan began scrolling through the results summary.

He was shocked at what he found.

The evolution on the avian sample had been run for two hundred thousand generations. That was what Juan had done for the rat evolution, but that was because their generations were so short. Two hundred thousand rat generations equaled only a four-thousand-year span of evolution—and even that had taken immense work to get right.

But sexual maturity on the Gouldian finch was between six and nine months, which meant someone had simulated over one hundred thousand years of evolution.