Nightmare - Impossible?

First of all, I'm not sure if this belongs here or in the Demo's forum.

I'd like to say that I'm a huge fan of demos. And the ones that I most enjoy are the NM Speed and NM100 categories. Now, I have a doubt about that. We all know that there are levels that are obviously impossible in Nightmare, but what about the classic megawads? I'd like to stick to the best/more praised ones(10 Years, Cacowards) to keep this simple. Do you think there's any impossible level there? For example, HR MAP18. Is it impossible, even a TAS demo, carefully done? MM MAP28?

What do you think?

EDIT

Enjoy reading this very discussion.

I recently rewatched Icarus MAP16 done by Rizera and was impressed again.

N.B.: if you use prBoom+'s feature which enables you to continue demos from savegames, you could scum-save and resume until you get a map right. In the end, the completed -seamless- demo will be proof enough that a map is technically possible -but this can also be "proven" by using TAS tools. The hard part is actually replicating the necessary actions in real time and without mistakes ;-)

Despite being considered "cheating", at least if you try completing a map with TAS tools or scum saving and studying the result, you can get an idea of what would work and what would not. The only case where you'd be in trouble, is if both a scum-saved version and a TAS one only succeed by a very small margin and both perform essentially the same actions and routes. Then that might mean that there's only one "correct" approach, and even that will have plenty of risky 'single failure points', and doing it "for real" might be at the limits of human possibilities, if not impossible/too dependent on luck.

Thanks for the link vdgg, I was aware of that thread but couldn't find it, the demos there are absolutely impressive. But my question was, as Maes stated, if they are possible. Maybe too hard for anyone to even bother with it, but possible. It seems that, at least for some levels, the only practical way is a TAS built demo.

Thanks for your replies.

EDIT: Do you think that levels like HR MAP18 are impossible, even TAS?

HR MAP18 is impossible because of the starting room, but no-one prevents you to try with -fast -respawn instead, which should make a similar impression, only the first cyberdemon would actually teleport out and not kill you.

I think that means that there are levels for which we'll never watch NM demos, I'm afraid. Regarding this, do you think it's a good idea to test levels on Nightmare, even if it's just to make sure they are possible? I know that almost no one, including me, plays on skill 5, so I think it may not be that important, after all.

There are so few people who are truly able to test maps for NM playability that the point is largely moot in any case. Consider how ball-breakingly hard some ordinary iwad maps become on NM. If you arrange a map to be playable on NM for average players, then it's unlikely that UV will be very interesting for them.

One idea is to make it so skill 2 is the true "UV", and skill 4 is easier, leaving NM to be relatively more manageable. However, many players never bother to read text-files and just play on skill 4 by default, no matter what, so this idea is pretty much a non-starter.

If you want to make a map so that it might be physically possible on NM without having a tame Doomgod to test it for you, then the main things to avoid are perhaps:
1) Hitscan hell: areas where you can't avoid coming under very heavy hitscan attack.
2) Large numbers of monsters that must be dead at the same time to trigger an event. HR map07 for instance. KS map07 is near-borderline, but has been done.
3) Having a mandatory route that forces the player repeatedly to backtrack through heavily populated areas. The chaos from having respawned monsters in them is very hard to manage, as planning doesn't tend to help much.
4) One-off things like HR map18 are hard to anticipate unless you are very familiar with the workings of the engine and skill 5.

But yeah, NM should not be viewed as skill for which every map should be expected to be playable. It's essentially a demo category, rather than a skill level for standard play, and I doubt even those who record a lot of NM demos go into it with an automatic expectation of playability. Like for Pacifist and Tyson.

NM doesn't have to be even remotely fair ;-)

Of course, not all people need to have lives and heed that advice, hence the NM category was born.

@Grazza: That's interesting. It seems to me that there's really not a good "solution" to this, so in practical terms, UV is the way to go. But as you said, almost nobody plays NM it seriously, so it's largely a non-issue.

@Maes: That's right. Maybe one of the "cool" aspects of NM is the fact that it's practically a joke. It reminds of some mods for Midtown Madness, where you had like 500 cops behind your ass, and absolutely no way to avoid them, so you were constantly "stuck" between 10 or 15 at any given time, and getting away from them successfully was more a matter of luck than a matter of skill.

Nightmare is another one of those things that was stuck on (minimal effort) with little evidence of what was intended (or deep thought on the matter). I would guess .. it was just to give players who finished the maps something more to do.
I think this is another area where the source ports could come up with a better implementation.

This could be discussed in source ports, but then I would not get feedback from players on how desirable such alterations would be.

Ideas:
1. Adding another difficulty between Nightmare and UV would work, but wads that alter the difficulty menu graphics would be compromised.

2. Limit the total number of Nightmare respawns using an options selection, 1..100. This has the advantage that you could actually win by outlasting the respawning.

3. Limit the number of Nightmare respawns per monster using an options selection, 1..50.

4. Multiply the time before Nightmare respawn by monster strength.

5. Only allow Nightmare respawn when the player is farther than 1024 map units from the respawn point.

6. Only allow Nightmare respawn when the player does not have line-of-sight to the respawn point.

I was trying to identify what makes Nightmare so difficult, and I think your post has made it clear: Infinite Respawning. When monsters come back again, and again, and again... things start to become chaotic, to say the least. What you propose is a good idea, I don't know why ID didn't thought about limiting the number of respawns.

That's why I just simply play on -fast mode when I want a challenge, but sadly not all custom wads support it.