Here are two cases:
1. A trolley is hurtling towards a person on a track. You can divert the trolley so that it plummets over a cliff. You know that there is a small (but not vanishingly small) chance that there is a person in the trolley.
2. Eating plant food involves, indirectly, being complicit in the suffering of animals that are killed during harvest. You could reduce the amount of plant food you eat by eating insects like crickets. You know that there is a small (but not vanishingly small) chance that crickets have whatever feature(s), F, make things into persons. (For instance, crickets might feel pain and pleasure).
Bob Fischer (in his 2016 paper 'Bugging the strict vegan', ) wants us to accept that these two situations are equivalent. In both cases, we might say, you 'might kill someone who matters'. Since you clearly ought to divert the trolley, you ought to eat the insect.
I'm not a huge fan of trolley cases. They're good at exposing our bare intuitions about, well, trolley cases. But they're often not so good at getting us to think about the underlying reasons or values behind those intuitions.
Nonetheless, I think Bob's challenge to vegans (and vegetarians) who don't want to eat insects is a reasonable one. OK, in eating an insect you might violate a right (at least, those are the terms in which I think of it). But if you don't eat an insect, additional rights (of field animals) definitely get violated.
Here is a third case:
3. A person is dying of organ failure. In the bed next to them is a human being who is being kept artificially alive while apparently in a permanent vegetative state (PVS). You could take the PVS patient's organs, and give them to the fully conscious patient. You know that there is a small (but not vanishingly small) chance that the PVS patient has whatever feature(s), F, make things into persons.
People will differ about this, I'm sure. But, as it stands, we don't typically think it's OK to use PVS patients as involuntary organ donors, just because we are unsure about whether they are still persons. And I think that many people (including myself) who would send the trolley over the cliff in case (1) will not be happy with cutting open the PVS patient in case (3).
It's true that all three cases could be described, as I did above, as cases where you 'might kill someone who matters'. But it's also true that there is a distinction within that description. In particular, while case (1) is a case where:
You might kill someone who definitely matters,
Cases (2) and (3) are cases where:
You will definitely kill someone who might matter.
The first case is thus an instance of what I call outcome risk: risk that derives from uncertainty about what will occur.
The second and third case are cases of what I call status risk: risk that derives form uncertainty about the moral status of those who will definitely be affected.
I'm not too sure what, exactly, differentiates the two. After all, someone might object that status risk still is a form of outcome risk: the outcome you're not sure about is whether you'll kill anything that matters. But I think this distinction may go some way towards explaining what's wrong with Fischer's analogy. We can't move directly from our thinking about outcome risk to claims about status risk.
One final thought. Perhaps part of the worry is about the epistemic limits involved in each case. In a case where I'm looking for people in trolleys, I know roughly what I'm looking for, what sorts of measures I can take to improve my accuracy, and so on.
In cases where I'm looking for (to speak metaphorically) people inside a body in front of me - whether that body be human or not - I'm more at risk of ignoring salient but alien factors due to a kind of chauvinism (this bug doesn't express pain the way I express pain, so it doesn't feel pain). And I may also be aware that this has historically led to ignoring the genuine claims of certain classes of individual. Indeed, vegans and vegetarians should be especially attuned to this worry, since (if my experience is anything to go by) many people still believe this about almost all other animals, including some where the evidence seems incontrovertible.
Is that awareness sufficient to warrant refusing to eat insects? As Fischer, and others, might press, we surely can't follow this logic all the way down, so that anything where there's even the remotest possibility that it has F gets treated as if it certainly does. But all I want to gesture at here is the idea that the kind of risk we confront might warrant a different approach than straightforward probabilistic accounting. Where we draw the line is a different, equally difficult question.