An Occupational Hazard

here's a couple of the frames:

and:

That first time, Murderbot changes itself so that it can pursue answers to its murky past. It wasn’t a life or death situation, and Murderbot felt like it could say no and its decision would be respected. In fact, it did say no to some of the suggestions. The two situations are so different on the face of it but when you get down to their hearts both reveal the same emotional/self-improvement hurdle that Murderbot struggles to get past.

When I first read Artificial Condition, it seemed pretty obvious that Murderbot doesn’t want to change itself because it doesn’t want to look human, and it doesn’t want to look human because it doesn’t have any desire to become more human in any way. But a few rereads later this line caught me by surprise:

So yes, a part of it is that it rejects humanity outright. but another part of the reason it doesn’t want to look like a human is because it thinks of humans as being people, and it still struggles to think of itself as person and not a thing. And looking more like a human makes it feel more like a person. There are a lot of things it allows itself to do/a lot of things it allows itself to be subject to expressly because it doesn’t think of itself as a person. Because to treat a person like that would be unacceptable.

And then we jump forward to Network Effect. Murderbot makes a copy of its kernel. Just as a reminder, earlier in this book, ART gets deleted, and then restored from a copy of its kernel. So we have textual evidence that a kernel is essentially the digital blueprint of machine intelligence. The original and the copied kernel are functionally the same being. Murderbot then starts to construct killware code so that the kernel can carry out the functions necessary for the mission. Since this is a pretty big task being done in a time crunch where lives are at stake, it asks ART to help construct this code.

So 2.0 is not a completely new being that they created together (which is the kind of thing I would consider to be their baby). They didn’t mix their kernels together to make a new person (another thing I would consider their baby). Murderbot never says that it thinks of 2.0 as its kid. Amena is the one who makes that claim.

This is how Murderbot describes their relationship right before killing it:

Two different iterations, with different capabilities. Same person, two versions. A few paragraphs previous to this, Murderbot uses that same word, iteration, to describe the different copies of TargetControlSystem:

It’s the same word it uses in System Collapse to describe the different copies of ART:

This was something that ART tried convincing Murderbot in that first conversation they have about constructing killware:

Murderbot tries telling ART ‘It wouldn’t even be me’, and ART is stunned speechless, and then says that Murderbot doesn’t understand how its own identity works. This is something ART has plenty of experience with, splitting its consciousness into different iterations. It knows that they would be the same person. It says ‘I didn’t mean you’, not ‘I didn’t mean our baby/kid/offspring’. ART considers killware with Murderbot’s kernel installed to be Murderbot.

This is how 2.0 identifies itself to Three:

A rogue secunit, working with an armed transport, currently present as killware. Not killware made by a secunit and armed transport. It tells us it’s unequivocally telling the truth here, not lying to make the situation easier to digest. So I think it’s fair to say that 2.0 does not think of itself as their offspring.

It seems to me that Amena is the only one to really consider 2.0 to be their ‘baby’ in the text of the book. Murderbot acquiesces to Amena that what they’re doing is kind of like making a baby. Because 2.0 is a newly conscious separate being from the original Murderbot, but it’s still Murderbot. It calls itself Murderbot. It distinguishes itself from the original copy by tacking on 2.0 at the end of its name. It refers to Murderbot’s memories as its own. And then at the end of Network Effect, Murderbot straight up says that the analogy that it was like their baby was wrong, but ART and Amena were right about it being a person:

I went through every moment that included or was regarding 2.0, to get a better understanding of how each of the machine intelligences involved classify its identity, and how it identifies itself, but that needs to be its own separate post or this will balloon to over 5000 words.

So Murderbot is being forced to change itself by the circumstance it finds itself in. At first glance it might seem as though this is an example of Murderbot exerting its autonomy, since this time it is the one to suggest altering itself. Except there’s no longer a choice that it can say no to, and they have no alternative ideas (can you call it a choice if you can’t say no? If you have no other options?). They don’t have the time to come up with any other solutions. Creating a person whose only function is to die should be unacceptable. But Murderbot allows itself to be changed in this way because when it’s put under pressure it still doesn’t think of itself as a person.

Murderbot doesn’t feel the same revulsion to altering its kernel as it did with its physical body, which should also feel like a point of progress, to be more open to changing itself. But unlike in Artificial Condition, these are not physical changes that end up causing it to appear/feel more like a person. It isn’t getting the same kind of reminder and reinforcement of its personhood as it did in Artificial Condition. Which leads to the conclusion of the book, when the dust settles and Murderbot has had time to think about what has happened and it admits, yeah, 2.0 was a person. Because IT is a person. Which is a lesson it already knew in theory, but is having trouble with in practice. Not only was 2.0 a person, but it was a part of itself. A part of itself died at its own hand (are these pronouns getting confusing for anyone else? No? 😂).

A big element of the following book is Murderbot having a negative reaction to what is essentially a waking nightmare about the trauma of losing a part of itself. System Collapse is a direct response to what happens in Network Effect, and I kind of wish that Network Effect and System Collapse were smushed together to make one larger book. I know that Martha Wells had a hard time writing TMBD after Network Effect, and I can see why. Network Effect ends with the highest stakes resolved, but none of the emotional fallout gets any resolution.

I mean, putting aside the whole suicide thing, it got left behind on a planet with no feed access! The only bot it ever called a friend was forced to act against it, and put the people it loves in danger! Its friend died! It’s worst nightmares came true. Even if it all turned out all right in the end, going through that kind of thing’s gotta leave some emotional scars. It’s come to the realization that there’s stuff that it doesn’t want to do (sacrifice itself to save others), but we have yet to see it actually lay down any boundaries with the people it intends to work with for the foreseeable future. Which is a shame, because I loved reading about grumpy Murderbot laying down boundaries with its PresAux people. That gave them room to show Murderbot that they were willing to make space for it. We didn’t get any of that with ART’s crew. No, actually I’m wrong. Murderbot does give them one boundary, and it’s a boundary that’s broken before the book begins: don’t send it back down to the planet.