This is also assuming that aliens will want to scout out the universe, and that probes are the best way for an advanced civilization to do so. We might be like someone from the past who comes to the present and determines that we don't have any long distance communication because they don't see any smoke signals.
But are we even dealing with a large sample size? Maybe there's "only" been several dozen large interstellar civilizations out there, and a few hundred small ones that never really got far beyond a handful of planets. Or maybe there were a couple of attempts, but they failed horribly. The problem with the "paradox" is that there seems to be a huge number of possible answers. It ends up being "how come this one scenario I'm thinking about hasn't happened."
P(probes) = 1 - (1 - P(a civ can launch a probe)) ^ N(civ)
If we don't see probes, and if we don't think P(can launch) is exactly zero, then N(civ) just has to be small.
EDIT: HN deleted my Fortran-style exponentiation.
In our Local Group of galaxies, however, we can confidently conclude that nobody has ever launched self-replicating interstellar probes. Which means nobody has ever evolved a technical civilization slightly more capable than our own.
Radio signals are ephemeral; probes are forever.
Perhaps the lack of probes means there is a fundamental flaw to the concept when applied to the actual, physical universe. Or that technologically advanced civilizations exist which aren't interested in sending self-replicating probes into interstellar space.
Not every civilization capable of trying may retain such technological capability forever, either. Even if attempting the trick is relatively rare in the life cycle of civilizations, those that can may not care to do so within the window of their capabilities. You could just as well argue that it's not possible for a civilization exactly as advanced as our own to prevent the launch of a deadly bioweapon which leads to their extinction. You could argue any number of logically sound things which may simply not shake out due to chance in the real universe. Maybe only a few such civilizations have actually arisen, and by pure chance, none of them have done it.
In another comment, I described the odds of these probes arising as
P(probes) = 1 - (1 - P(a civ can launch a probe)) ^ N(civ)
and unless P(can launch) is zero, not just really small, I think that we can conclude that N(civ) is not a very large number.
...and continuing to do so reliably for millions of years, somehow being able to efficiently fabricate arbitrary components of exceeding strength and complexity out of dust and ice, without any appreciable loss in quality of hardware or software, even in the radioactive environment of interstellar space. Just because a model is mathematically consistent doesn't mean it's practically feasible.
On the other hand, we are surrounded by earthly self replicating probes and we seem to be doing fine.