The resources required to attack every life bearing planet only really becomes super expensive compared to waiting for civilization to arise if life is very common but civilization is not, which is admittedly a possible scenario, but by no means guaranteed. But consider: any civilization capable of launching an attack on another, particularly one that can be considered highly likely to completely destroy the target in a single strike (which you absolutely need, because if a target survives your attack, it now knows that you exist and even if it did not also follow your "attack everyone" doctrine, will see you as an existential and hostile threat) must necessarily have interstellar travel technology. The amount of time needed to develop this, and the amount of energy and resources this capability implies, make it highly likely that they have very good automated manufacturing as well. With those two technologies, you dont need to listen for radio signals or similar. You can send a tiny and difficult to spot or trace probe to every star out there (potentially at almost no cost if you can make a probe that can extract raw materials and build copies of itself, but even if you cant do that, the probes can be much smaller and lighter than a planet killing projectile and so if you can build one of those, you can at least launch a probe to every world with atmospheric compositions indicative of possible life, to observe from close range and tell you if civilization arises there. Thus, any civilization that wants to follow this policy is impossible to hide from, it doesnt matter if you send radio signals or not, or if you build structures that are visible or not, because your position was compromised before you even considered that there might be something to hide from. If youre a civilization that worries that aliens might be hostile, then, trying to hide makes no sense, because it wont help. What would make sense instead then is to try to grow as far and as fast as possible, in the hopes of acquiring enough redundancy that your neighbors dont have the capacity to destroy you, or at least enough that they arent sure that they do. This kind of growth doesnt seem to be the policy of anyone we see either though, because it should be visible even to us (we can see stars, and if you want to grow as much as you can and have automated manufacturing, you could start to build dyson swarms and similar structures that would visibly change the amount of light that reaches us from a given star).

Now, there are a few responses to this line of thinking that I've seen: The first is that a civilization this paranoid might not want to expand to other stars, because a colony in another system is so distant as to be effectively a new civilization, which might turn hostile to you, and is right at the next star over, and so civilizations might just stay in one star system and launch attacks from there. This doesnt really help them hide, as for the reasons Ive just mentioned, they should be easy to spot by anyone that actually has the ability to threaten them, but it might make it less likely for us to see, which is all that matters for the fermi paradox. But these aliens would still be able to send out probes to spy on our planet, so if theyre within around 5000 light years or so, they should have been able to see us develop civilization, and so if thats what they want to destroy, we again, shouldnt exist to contemplate this right now (and if theyre much further away then this, and theyre still worried that we might be a threat, then they really do need to destroy us before civilization ever arises, for reasons Ill get into in a moment). These hypothetical aliens must in order to make sense have a different policy than just "destroy anything smart enough to develop civilization". The next most obvious trigger then is "destroy anyone that makes it into space". Suppose then that you're these aliens. Your probes report some aliens on a planet 500 light years away (given the galaxy is in the ballpark of 100000 light years wide, this is in your cosmic backyard, relatively speaking), or if youve not done the probes, you hear some radio signal indicating this. You decide to launch an attack. But, you have a problem. That signal from your probes (or the radio signals if you hear those instead) was sent out 500 years ago, and even if your attack moves at lightspeed (it wont unless its something like a laser, but you probably want to launch something at a large fraction of that speed, so lightspeed is still a decent estimate) its still going to be 1000 years between when that species started going into space, and your attack arriving. Thats almost certainly enough time to colonize a lot of their solar system, so just attacking their homeworld probably isnt enough. Do you attack every large celestial body in that star system just in case? They could also have significant habitats and industry and such in orbit of various objects, or in orbit of their sun, so even that might not be enough. Worse, that thousand years of space could be enough time to start to get into interstellar travel themselves, so you might need to target every system within a certain radius of their home star, and even worse than that, if theyre just as paranoid as you, its enough time that they could conceivably begin to launch their own interstellar attacks, and if they happen to see your home star and think "that looks like it might have life, lets attack it", then your policy was insufficient. Youre not launching an attack against a newly space fairing civilization, youre launching one against whatever exists in that area of space when your attack finally arrives. Thus, you really need to attack well before civilizations start to go into space. If we're anything to go by, the time between early space exploration and industrial revolution is only a matter of a couple of centuries, so something like industry or radio is also too late, unless your targets are in a very narrow window of distance. If you're within a few thousand light years (still relatively close compared to the size of the galaxy) then you really should be attacking by no later than the first sign of early civilization, and if youre farther than that, as I mentioned earlier, you really should attack before civilized life ever even arises, because there would be time for a planet to go from having literal cavemen to an emerging interstellar empire before your attack even got to them, and once they have interstellar travel, you dont know exactly where they're all going to be, and they have the capacity to at least try to attack you. So, you either attack before civilization in which case we shouldnt be here, or you colonize the galaxy yourself to have outposts nearby to any emerging aliens, which is fundamentally not stealthy, and which again means we shouldnt be here because our planet should have been colonized by said aliens before we could ever evolve.

The other response Ive seen before is that maybe, nobody is actually willing to engage in a policy of genocide at first sight like this, but everyone is afraid that someone might be doing so, and so everyone hides despite there being nothing to hide from, and so we see no aliens. But this assumes that everyone considers this possibility, deliberately holds back their own development by trying to hide, despite probably also realizing that hiding is futile anyway, and that nobody across the galaxy ever, or has ever, not done this and so had the galaxy to claim for itself, which seems absurdly unlikely, especially given those hiding still have the option to send the probes, and potentially discover that everyone else is doing the same and so knows of them anyway.

The TLDR of this, because I know this was a rather lot of text for just a response to this, is: The dark forest assumes that one isnt a target unless one takes action that reveals oneself, and that one can be sure of destroying any civilization one knows about in a first strike, and that these are the only options available with no way to try to make oneself a less vulnerable target. None of these assumptions seem reasonable upon further pondering, and if they do not hold, the scenario does not make sense.

Even the game theory analysis fails, as it doesn't consider a sufficient number of outcomes nor their branching over time.