(1) There exists an omnipotent, omniscient, and morally good Creator (a ‘true God’).
(2) Beings who suffer have been caused to exist.
(3) A true God meets any requirement entailed by benevolence.
(4) Benevolence requires that harm be avoided unless its avoidance implies greater harm or deprivation of benefits that outweigh the avoided harm.
(5) The never-existent cannot be deprived.
(6) The falsity of (2) avoids harm.
(7) The falsity of (2) entails no harm discouraged by benevolence, nor any deprivation of benefits (from (5)).
(8) A true God could have caused (2) to be false (from (1)).
(9) Benevolence requires that (2) be false (from (4), (6), and (7)).
What's called benevolence here is what's usually called nonmaleficence; this would be a purely stipulative matter except for the fact that nonmaleficence is very often considered only a presumptive or prima facie obligation capable of being overridden by more robust moral principles, showing that the argument is assuming that there are no other relevant moral principles on the table. Karlsen assumes that the only or even most obvious way to reject (3)+(4) is to take divine goodness to be 'inscrutable', but this is very obviously false; there are a lot of different ways one could go here; (4) is not as innocent of presuppositions as it is taken to be. Giving someone a free immense and ongoing benefit in exchange for having risked (and, as it happens, received) a painful papercut today would violate 'benevolence' understood in this way: the only reason (4) allows for harm not being avoided (note that it is not 'harm being caused' but the much weaker 'harm not being avoided') is avoiding worse harm or an overweighing deprivation of actual benefits. (If one were to interpret it as allowing not giving the immense and ongoing benefit as deprivation of benefits, it wrecks validity of the argument, because (6)+(7) would no longer cover all the options that need to be considered to draw a conclusion. Thus it has to be deprivation of benefits already in hand, and not deprivation of future benefits.)
Karlsen states flat-out, "The primary objective of morality is to avoid harm." While it is uncontroversial that this is an objective, Karlsen actually needs it either to be the only objective or the primary objective in such a way that it can never under any circumstances or conditions be subordinated to another objective. That is, simply holding that it is the usually-most-important of many objectives is too weak, because then (3) would immediately be in doubt -- perhaps there are exceptions to the requirements of benevolence consistent with being morally good.
What (4) is effectively doing in the argument is making it so that one is not allowed to consider what benefits might follow from existence or non-avoidance of harm. Take a very simple example like letting one's children play a sport. Letting one's children play a sport is not avoiding harm; the risks of harm involved in playing a sport are many, extensive, and ineliminable from the game. In sports people will occasionally get hurt and, again, (4) does not require that one be causing the harm oneself, but only that one not be avoiding the harm. (This interpretation is not only the most obvious interpretation of the words, it is required by (8)'s having any function in the argument. (4) in this argument requires that one avoid harm not just by not causing it but also by eliminating risks for it.) It is almost never going to be the case that not letting your children play sports implies greater harm or deprivation of benefits than a serious sports injury; there are plenty of children who do just fine not getting anywhere near sports, and while one makes the worst sports injuries as rare as possible, the only way to avoid them entirely is not to let any children play sports ever. Now we see here a case where at the very least where people in fact regularly consider the benefits of non-avoidance of harm in their reasoning, and, indeed, clearly think it is defective not to do so. But even if we set that aside, while we could obviously discuss the prudence of the options on the table, is it really a violation of the requirements of benevolence to allow one's children to participate in activities with a risk of harm on any other grounds at all except to avoid greater harm or deprivation of benefits in hand? Are we really all committed by the requirements of morality to strict tutiorism? There are excellent reasons to reject such an idea, and, in fact, the number of serious views in ethics consistent with it are very, very few.
So the argument seems to run into problems from the get-go. It's worth noting, incidentally, that the argument overall is in fact simply an anti-natalist argument taken to its limit. Anti-natalism is a philosophical position that is, as they say, trending upward. The reason that Karlsen's argument runs into this problem is simply that anti-natalist arguments usually do. It's relatively easy to argue that people should be cautious about having children; it is remarkably difficult to argue that everyone has presumptive or prima facie obligation not to have children, and the arguments for such obligations usually depend on highly controversial ethical assumptions. Karlsen needs something even stronger than this sort of prima facie anti-natalism; and, in fact, his argument is a rigorously anti-natalist argument, requiring that no one procreate (which is a non-avoidance of harm involving causing beings who suffer to exist) if they can at all help it. The reason (4) ends up being odd the way it is, is that it's the way Karlsen is getting something like the asymmetry central to rigorously anti-natalist arguments like David Benatar's; it is a way to avoid the question of whether a being who suffers in any way can also have a genuinely good life.
This, of course, is all not even getting into the vexed question of whether (5) is true. However, the primary questions about (5) are really questions about (4), as well. (5) is clearly not true as a matter of how 'deprived' is used in colloquial speech, so the question is whether it still turns out to be true in whatever very specific sense is required to make (4) true, assuming it can be made true.
