As burrowing monsters, it's probably fair to say that the ground is providing them the fictional context to justify a Dexterity (Stealth) check to hide.
It is exceedingly easy to justify how and why a monster might be hiding and waiting in ambush, especially a "lurker in the earth" like an ankheg with its tremorsense and a party bumbling headlong into its territory at normal pace with feet pounding the ground all the way. Other monsters could similarly be given fictional justification to be in the position to surprise adventurers. Fiction is easy to come up with.
However, my position is that it is not a fair challenge (and is fake difficulty) when the DM fails to fictionally telegraph the threat in a way that provides an opportunity for the players to discern that there is some threat in the area and that caution is warranted.
As I mentioned above, I would have described collapsed earthen tunnels, a husk of molted chitin, a partially eaten orc in a puddle of acrid bile, or the like. This would encourage the players to engage with the exploration pillar of the game - check out the tunnels, the husk, the corpse - the investigation of which may allow them to make deductions about threats in the area based on what they learn (with or without ability checks depending on the uncertainty or certainty of their stated actions). Based on their findings, they might opt to be more cautious and slow their pace. Or they might ignore that telegraphing or make the wrong deductions and carry on. Or they might decide to haul ass away from the area.
The key thing here is that the DM created an opportunity to interact with the environment and for the players to make meaningful decisions about what to do. The ankhegs might still get the drop on the PCs even in this scenario depending on what the players say they want to do, but at least they had a chance to avoid or prepare for the danger. It turns a potential "gotcha" into a scene that builds the tension and allows for meaningful decisions.
Now, the OP seems to not do this (or at least not all the time as I would), relying upon meta communication (if you will) to warn players that they should be more cautious. The player skill then becomes reading your DM and taking action accordingly. That's fair enough if that's how they normally do things at that table. It's just not what I'd do.
4) Yours. I give out clues and if the players understand my clues, then they get chances to make decisions before initiative is rolled and surprise is determined based on those decisions and the normal rules.
I like your way and have used it in the past, I just think that with all that is going on during a game, I (as an old DM who forgets a lot of stuff) will often forget to do that and hence, I rely on the variability of the dice to fix my oops. I also have a Cleric PC in the group who has the Alert feat, so my way works good for her because there will be more encounters with at least some form of surprise. I can also understand that a given player from your table might wonder why I am not giving the PCs enough information (in his mind) to interact with the environment (i.e. why am I not foreshadowing an encounter?) if your player was to play at my game.
Thanks for the feedback. To me, it's easy to remember because I have a relentless focus on the basic conversation of the game (Basic Rules, page 3). The first step is that the DM describes the environment. Without that context, the players cannot describe what they want to do meaningfully. It ends up being just procedural stuff - fighter in the front, secondary tanky guy in the back, squishies in the middle, slow pace if we're not in a hurry, ten foot pole engaged, check every door or chest for traps, listen at doors, and so on. To me, that's not meaningful - that's rote memorization for the purpose of failure mitigation.
As well, I don't use passive checks in most of my games. I think that simply comparing two numbers to each other and then hitting the PCs with a surprise attack is "fake difficulty" in that the outcome is not reasonably determined by the player's actions. To be fair, the passive check does represent the average result of performing a task repeatedly - in this case, "keeping an eye out for hidden threats." But given that there are only a few things that will cause you to be unable to do this (see "Other Activities," Basic Rules, page 65), most of which won't see play, it kicks most of the decision-making back to character creation alone, encouraging players to pump Perception rather than interact with the game environment in a meaningful and interesting way. Surprise does still happen in this approach; however, it's not presented or received as a "gotcha" situation. The players had a real opportunity to take action before the threat is revealed, even if they botch the job (or fail to take action at all).
Well, I do describe the environments. I just might not necessarily do it with the level of foreshadowing that you might. Also, I definitely understand the concept of:
DM: "You have been slogging through the swamp for three days. You see a dead animal carcass, just like the early dozen or so carcasses that you have seen before. What do you do?"
as some DMs thinking that this is too much of a "wink, wink, nudge, nudge" that something special is going on at this location and giving the player a little bit of metaknowledge that they should not have. In other words, if the PCs had already encountered a dozen (or even hundreds of) carcasses in the swamp, and precautions that they took the first few times are not resulting in encounters, wouldn't they eventually come to think that the sign of a carcass rarely means an encounter?
I do not disagree. And I rarely use passive perception or passive insight myself. The reason for the "passive stealth" houserule of mine is that I do not want to roll 5 PC active perceptions against 4 NPC active stealths, and 4 NPC active perceptions against 5 PC active stealths. That's a lot of rolling.
Nor do I want encounters to just be about:
Player 1: "The DM opened his mouth and described an environment. Everyone be alert."
Although my system is not perfect, it does avoid that issue a bit. And if I combine it with some foreshadowing (when I remember to do so), the players do not get into this mode of "hey guys, something is about to happen" because sometimes, encounters just spring up on them and sometimes, I'm putting out a lot of detail for the environment which may or may not result in an encounter.
But, I don't think that the DM should go beyond a general description and into a lot of detail for overland travel encounters. If monsters are hiding, PCs really shouldn't get a chance to do something out of the ordinary. Describing an environment, then asking what the PCs are doing results in "metagame decision making" instead of "fake difficulty".
But I definitely get your point.
DM: "You have been slogging through the swamp for three days. You see a dead animal carcass, just like the early dozen or so carcasses that you have seen before. What do you do?"
as some DMs thinking that this is too much of a "wink, wink, nudge, nudge" that something special is going on at this location and giving the player a little bit of metaknowledge that they should not have. In other words, if the PCs had already encountered a dozen (or even hundreds of) carcasses in the swamp, and precautions that they took the first few times are not resulting in encounters, wouldn't they eventually come to think that the sign of a carcass rarely means an encounter?