Instead, for Platform Decay, what I want to talk about is voice. Murderbot - who doesn't use that name at all in this book, by the way - is a bioengineered sapient construct officially called a SecUnit, basically a productized corporate-security mook, one of what may be thousands or millions of similar units across this medium-future galaxy. This unit, though, hacked its governor module, for good and sufficient reasons we learned in previous books, and is now a fully autonomous and self-directed individual...that calls itself Murderbot in its own head, also for good and sufficient reasons.
Murderbot narrates all of the books in the series, in a direct, colloquial, stream-of-consciousness style that is immediately compelling and must be tough for Wells to pull off so seamlessly and gracefully. Everything we know about this universe comes to us through the lens of what you could semi-reasonably call a slacker killer robot - a person who is not human, who doesn't like humans all that much, no longer has to obey them, and wants to be left alone to watch media most of the time.
(I don't know if I've made the direct connection before, but Murderbot's voice reminds me a bit of John Sladek's Tik-Tok in its tone and repressed anger at humanity - and obviously comes as the latest in a very long line of killer robots in SF.)
Like a lot of voice-driven books, Platform Decay is tightly constructed in ways that I don't know if most readers notice. Murderbot books tend to begin in medias res, run through one specific complex situation, and end immediately when the danger is past - and that's what happens here, as well. There have been glimpses of Murderbot in less-fraught moments, during the course of the series, but that's the standard: what is it like to be in Murderbot's head during what I might as well call a mission.
For example, about halfway through this particular escapade, Murderbot encountered, well, pirates, who were looking to rob or do worse things to the unwary humans they captured, but were not at all prepared to face an annoyed Murderbot. On pp.130-131:
I really wanted to kill them but they were so whiney and screamy, it was hard to get started. I don't like helpless humans in my media, but in reality, it's... Provoking them hadn't worked so I tried to be encouraging. "Maybe if you all run at me at once you could kill me."
Target 4 looked tempted, but the others all shook their heads rapidly, 3 sobbed louder, and 5 whimpered.
Great.
Being good at violence - and, more subtly, in control of dangerous situations where violence is likely - is one of Murderbot's core competencies. Or, I should say, the competencies Murderbot wants to emphasize and think about and focus on, the parts of its skill-set that it's most comfortable with. (There's a reason it named itself Murderbot.)
More complexly, since we experience all the events in these books via Murderbot's POV, the telling of those events features a lot of cascading and intersecting thoughts - and, often, notes from semi-separate processes, since Murderbot is not a normal organic human, but a created and hacked and self-modified mind - which Wells often uses parentheses to set off and encapsulate. This is funny and insightful and piercing - and, I'm sure, fiendishly difficult to get to work this quickly and precisely (p.157):
I needed a plan, too. Option (1) I could tell Farai that Leonide's family were all dead.
It was safer, it was sensible. My job was keeping my clients safe.
(Emotion check: Then why does it feel so awful.)
Option (2) I could tell Farai they were all dead and alter my memory structure so I believed it was true.
Risk assessment just hit the roof.
I'm not sure I like having risk assessment be more accurate. It keeps scaring the shit out of me. I knew doing something basically terrible and then lying to myself about it by altering my memory archive would change me, not in a good way... Yeah, when you put it like that, it seems really obvious.
Reading Platform Decay, I was struck by just how much work Wells was doing in few words, in this short book. Every line is precise, even descriptions are filtered through what Murderbot sees (dangers, human fripperies it thinks are pointless) and what they could imply (where hostiles could be hiding, where it can send its own drones to scout things out) and how it feels about everything.
Because Murderbot is close to an open nerve, even eight books into the series. It's not comfortable being a mind that has emotions - in this book, it's built in that "emotion check" I quoted above, to flag those thoughts. Murderbot doesn't like the emotion check, that's clear: but it needs it; it needs to be able to understand its own emotions, to live with them, to regulate them.
Like any of us.
Murderbot has only had emotions and self-will for a few years at this point - it thinks sometimes unkind thoughts about "human juveniles" in this book, but it's basically a juvenile itself, still working towards emotional regulation and integration. I don't know if this series will last long enough for Murderbot to get there - and, in a way, I hope not - but that's the arc. That's what we're seeing, over the course of the series: a new mind exploring the world and its own feelings, trying to do right when what it really wants to do is sit in a corner and binge Sanctuary Moon for twenty hours.
And what actually happens in this book? It opens with Murderbot, along with its compatriot Three, another self-hacked SecUnit, sneaking onto a torus - a gigantic human-built station that completely encircles a world and contains many separate, individually-controlled sections within it. Several of what Murderbot calls "its humans" are held hostage there, by Supervisor Leonide of the unpleasant Barish-Estranza corporation, and Murderbot is going to find them, extract them from their situation, get them across the torus to a port, and leave there with Three's help.
That's what happens, more or less, though there's a lot of complications along the way. But that's the joy of reading it.















