The Naturalist (The Naturalist #1)(28)

MAAT is based on how I think, but much more advanced.

I built it using source code from a research project designed to find genes that contribute to longevity. It’s AI that builds better algorithms with each iteration. Each time becoming more and more complex.

I couldn’t tell you how the current version of MAAT works, just that she does. Sometimes.

When the researchers who developed the core AI behind MAAT asked it to figure out what gave a strain of fruit flies its longevity, it pointed to the genes that regulated resveratrol—the same chemical in red wine that has been tied to human longevity. When they tried to figure out why the software singled that out, the answer was a string of data that no human could understand.

What MAAT could tell you right now from the data points on my map is what is already obvious.

She’s really useful when you give her thousands or millions of points.

Points I don’t have. The killer is just two black dots in time and space. But . . . in the absence of firm data, the other trick is to give her assumptions.

If we were looking at mating cycles, and Juniper and her killer were two mountain lions, I could tell MAAT the frequency that a female is in estrus and an estimate of the male’s range. That information would give me an estimate about when they would encounter each other again. If a male mountain lion had multiple females it bred with and they had specific ranges, I might be able to predict where else he would show up.

And if there were general rules about the kinds of places they reproduced, I might be able to narrow down candidate spots based on available geographic information.

From all of this, MAAT could give me a dozen or so places where I could plant wildlife cameras and reasonably expect to catch the two large cats doing it, even over an area of dozens of miles—all of that based on three data points and general information not specific to an animal.

The problem is I don’t have any more data to put into MAAT.

I know nothing about the killer.

He was born at some point. He met Juniper. At some point after that, years or minutes, he killed her. His last appearance was getting her blood on Bart. Then he vanished from his graph.

I need more data than what’s on my map.

From where?

If I don’t have data, then I have to use the next best thing . . . which is also the worst best thing.

Assumptions.

I need to make guesses.

On a real graph these wouldn’t be black circles. They’d be half black, half white. They’re maybes.

Sometimes they lead you somewhere interesting. Other times they derail you for months . . . or years.

Our war on cancer has been filled with countless maybes. Billions of dollars and millions of human hours have been spent chasing after a pattern we can’t even begin to guess at.

Even still, we’ve made some progress. Many of those maybes have panned out. People live longer than before because not all that effort was wasted. And for every maybe that turns out to be a no, we still move forward.

I need some maybes and assumptions about the killer.

I can’t be worried if they’re wrong. I just need a starting point.

Let’s make some . . .

Juniper’s killer was clever because he got away with it. That’s a hard thing to do. He was either very lucky or experienced.

Okay . . . let’s go with experienced.

Oh, shit. Sometimes one assumption makes something else automatically true.

An experienced killer implies that he’s done this before . . .

I open up my laptop and do a search for bear attacks in the United States and Canada.

I’m not sure what I was expecting, but there’s only been a handful in the last ten years.

The Fish and Wildlife Service has detailed reports. Most of them are in deep woods. I look for any within a few miles of a highway.

There are two. In the first, three years ago, a self-proclaimed grizzly expert was killed. I’d personally rule that a suicide.

The other was six years ago. A woman was found bleeding to death on a road. She died on the way to the hospital.

Experts decided that she’d also been killed by a grizzly. The report shows diagrams of wounds and a photo of a tissue sample. There’s even a hair. But no DNA analysis was done.

The bear they caught was identified by the victim’s blood on its pelt.

That sounds familiar—just like Juniper.

The hair on the back of my neck raises. It’s my own animal sense telling me I’m looking at something dangerous.

I put a red and a black circle where the other victim was found and a black one where the accused bear was trapped.

It’s fifty miles away in a different county, making Detective Glenn and the others seem less suspicious to me.

This has happened before, somewhere else.

But two red dots don’t make a pattern. Not yet.

I need more data.





CHAPTER TWENTY-THREE


THE HUMAN CIRCUIT

A wider search of bear attacks is a dead end for me. They’re supported by finding human remains in the animal’s scat. This doesn’t mean the killer couldn’t have left the victim to be scavenged by bears. Apparently bears are not very picky eaters. It just means that these look exactly like bear attacks. There’s nothing suspicious to them, unlike Juniper Parsons or the other woman, Rhea Simmons.

I pull up an article on Rhea. She was twenty-two and apparently hitchhiking her way across the country. Born in Alabama, her family had no idea she was in Montana.