Pages

Friday 16 November 2018

Robopocalypse by Daniel H. Wilson

If I really liked Charlie Jane Anders' take on the emergence of artificial intelligence, alas, the same cannot be said for Daniel H. Wilson's Robopocalypse. I wanted to like this, really I did. I was hoping to include it in a proposed theme for my SF/F book club of science fiction and fantasy written by Indigenous authors. And while there are a lot of interesting Native characters here, unfortunately, the book itself just wasn't that good, and had a few major problems that drove me bonkers.

I mean, this book and I got off on the wrong foot right away. (I almost wrote wrong foot right off the bat, but that was a little too mixed-metaphor.) I am not fond of AHHHH KILLER AI!!! books. I don't ask that all AIs be benevolent, but I do like a less extreme view than THE ROBOTS ARE COMING TO KILL US ALL.  It feels a little lazy. And a lot extreme.

But really, my main problem is that Wilson sets up a central conceit for the format of the novel, and then breaks it, repeatedly, in very distracting ways. It doesn't work, and if the format means you can't tell the story the way you thought you were going to without throwing in stuff that could not be known at the time by the people who knew it, then throw that format out the damn window and find something else. Experiments in form are great. When they hamstring and undercut your story? Not so much.

So what is the format? Well, we begin at the end - we know, from the first chapter, that humans have just defeated the master AI, known as Big Rob. This is also just a weirdly disconcerting name, but I'll let that one pass. It was a long and gruesome war, but we won. (Which does interesting things to the narrative, but it does undercut tension, which, given that this is more a thriller than anything else, is kind of a strange choice.)

Last out of the hole in Alaska that the humans just bombed the shit out of is a black box, with footage of most of the robot war. So the rest of the book is the narrator narrating what he sees on the black box, sometimes supplemented with his own recollections of these events. It's a neat conceit, but...it doesn't work.

It doesn't work because these characters often seem to have knowledge they cannot have had, and for no good reason. It doesn't work because while sometimes Caleb (I think that's his name, I'm not that sure), the narrator, just tells what he sees, more often as the book goes on, we're inside the heads of humans and robots, hearing things in the first person, in a way that is not possible given the central conceit of the book. If this had just been a collection of narrators, that would have been fine. Trying to shoehorn them into this structure caused more confusion and annoyance than it added sophistication.

Take, for instance, Paul. Paul is Lakota and a soldier of the United States in Afghanistan. He operates a mobile unit that interacts with the Afghani population, and so sees the moment when the robot is infected with the AI's virus and turns violently on the people among which it's supposed to live. We see Paul's congressional hearing into the incident, and that all works fine. Then he returns to Afghanistan, and is there when communication between humans is cut off, and survives by allying with Afghani hill fighters, taking out the robots from a distance and triangulating their sources. While he might hear the broadcast that comes from England that tells the world about Big Rob, it doesn't make sense for him to then have the level of knowledge about the AI, its location from what we see him able to do, and it doesn't make sense for us to be in his head.

Because he and his closest ally manage to triangulate where the strongest signal coming to the local robots is coming from, which is Alaska, and it would be fine if he just broadcast that - strong signal, going to all robots, don't know what it is. But for some reason, the author has decided to have him broadcast far more information that he has no way of knowing and is actually irrelevant. Let the people in Alaska put two or three pieces together themselves. There's no reason this guy who hasn't had contact with anyone but a few fighters for months to be the strategic commander when there is no back and forth.

And then later, we get sloppier. A robot who is not under control of Big Rob (god, I hate that as a name for the AI. The one it gives itself, Archos, is at least not irritating), joins a human contingent. He says, he says, when he joins that he found out where they were from the human girl named Maria who can interface with the robots. But then, one chapter later, someone introduces him to Maria and he and Maria both make nice, and there's no evidence he ever knew about her. I started to flip back and forth to make sure I wasn't losing my mind, but nope! It's a bad editing mistake.

There are a few good things in here - Maria's story is interesting, although the recounting of her mother's final hours is marred by the same issues, coming from a first-person perspective that a person watching satellite or video footage could not have known. The humans are remarkably willing to accept other humans who have been altered by the robots to include robot parts, and to accept the free robots who join them later. I like the sense that not everything would become xenophobic, but given the rest of the world that Wilson has posited, I'm not sure I buy it. I bought it in Charlie Jane Anders, because that was the world she'd created, and I loved it. I don't buy it here, because it feels incongruous, or at least in need of further explanation than it gets.

I'm disappointed. I wanted to really like this book, but I was far more frustrated by the inconsistencies and overall feel than anything else. I mean, this book drove me to italics. What else do you need to know?

No comments:

Post a Comment