Is Murderbot an Unreliable Narrator?
Long answer: Not really, in my opinion, and I want to try and explain why.
Definitions
The term unreliable narrator has become somewhat watered down here on tumblr and in pop cultural discussions in general but I am working with what I learned at uni here. (Disclaimer: In academia too there are different definitions and approaches to unreliability but I'm going with the basics here)
For Wayne Booth, who first coined the term, "an unreliable narrator is one who tells lies, omits or misrepresents important information, or presents inaccurate interpretations of events as objective truth." Peter J. Rabinowitz says "An unreliable narrator however, is not simply a narrator who 'does not tell the truth' – what fictional narrator ever tells the literal truth? Rather an unreliable narrator is one who tells lies, conceals information, misjudges with respect to the narrative audience." (emphasis mine)
In other words, an unreliable narrator deceives the narrative audience (that is the implied in-universe audience), knowingly or unknowingly, in a way that significantly changes their interpretation of the narrative.
According to Greta Olson, there is "a differentiation between fallible and untrustworthy narrators" and "all fictional texts that employ the device of unreliability can best be considered along a spectrum of fallibility that begins with trustworthiness and ends with unreliability."
That means: It is not a binary of reliable vs. unreliable narrator. It is a sliding scale, a spectrum, and I would argue that on that spectrum, Murderbot falls closer to the trustworthy side than the unreliable.
General Thoughts
By and large, Murderbot presents the facts of the story to us truthfully. It does not withhold information from us that impacts our understanding of the events (see below for exceptions).
Are there moments where Murderbot's narration is fallible? Sure. It makes mistakes, particularly with judging people's intentions and their opinion of it. It consistently assumes people (especially "its" humans) think worse of it and trust it less than they actually do. It considers itself not important, and therefore assumes it is also not important to others. It also often rates it competence as much lower than it actually is.
But, and this is crucial: It does also tell us when it makes mistakes and often tells us the objective events in addition to its subjective assumptions or interpretations. By that I mean: It tells us its assumptions about the humans not thinking much of it and considering it expendable. But it also tells us the humans' words and actions that prove the opposite is true. It tells us about how incompetent and useless it is (especially in System Collapse) and then also tells us the sequence of events showing it actually is still capable of doing its job. It learns alongside us that its negative thoughts about itself aren't always true.
That's not unreliable narration.
Now, Murderbot's refusal to admit/discuss its emotions is the most common argument I've seen for MB being unreliable, and that is the one that probably bothers me the most.
Because
a) A narrator isn't obligated to discuss everything equally with the audience. MB refusing to discuss its feelings is a characterisation choice, but it's not keeping any vital information from us that we need to understand the story. There's gonna be narrators that focus more on outside descriptions and some that focus more on internal thoughts and perceptions and neither is inherently more unreliable than the other.
b) that being said imo MB's narration actually focuses A Lot on internal perceptions and feeling, in spite of it being so reluctant to talk about them. I think MB gives us a lot more emotion than we give it credit for. I would argue it doesn't actually hide all that much. It may often not be able to state its feelings outright, but that doesn't mean it doesn't tell us about them
As an example, let's look at the moment when Gurathin reveals its name to the PresAux group because that's always one that sticks in my mind with regard to MB and how it communicates its feelings
I opened my eyes and looked at him; I couldn’t stop myself. From their expressions I knew everything I felt was showing on my face, and I hate that. I grated out, “That was private.”
Murderbot doesn't actually tell us anything about what it's feeling here. It only tells us it's feeling something intense, and that it shows in its face, that the others can perceive it and react to it, and that it hates it. We can extrapolate what it's feeling of course: it's feeling betrayed and exposed and embarrassed and vulnerable (and angry but that's a given for MB). But I would argue it's not hiding that from us. It's given us the context we need to extrapolate what it's feeling because a) it doesn't want to talk about it directly so it does it indirectly and/or b) it wouldn't even know how to name the emotions it's feeling in that moment.
So I would say that MB does the best it can given its limitations (in experience and willingness to talk about feelings) to give us the most accurate view of its emotional state without having to talk about it. After all, we almost always know how it feels, or can at the very least guess at it. If it really was trying to hide its feelings from us (not from the other characters, it probably is actually pretty opaque to them most of the time) the narration would be reading very differently.
Looking at it from a Doylist perspective for a second: This is also stylistic choice on the part of Martha Wells, it's basic "show, don't tell": She doesn't simply tell us what Murderbot is feeling, she is showing us through its thoughts, actions and reactions and the reactions of other characters. That doesn't make Murderbot untrustworthy.
Now, I want to look at a few other passages and examples that are often brought up in this debate.