Hi Zeke! I'll tell you a little secret upfront: As small as the community of this forum is, there are actually several sub-communities.
And each of them prefer different things.
1) NeoLemmix (made by namida and Nepster), which is currently the engine that seems to have the largest userbase in this forum, has a reputation of sticking to the principle "puzzle first, execution second". The idea is that the rewinding and framestepping features that allow for pixel-precise assignment of skills make any kind of execution difficulty in a level uninteresting as a challenge - if anything, execution difficulty is merely a nuisance, since it forces the player to rewind again and again, even though they already know exactly what they have to do on a conceptual level.
Sometimes, execution difficulty may be a necessary evil, if a few frames / pixels difference in the timing of assignment of a skill actually make a conceptual difference, too. For example, whether you place a Builder in such a way that the lemming turns around (by building into terrain) vs. so that he doesn't turn around. But in general, people try to keep execution difficulty as low as possible.
Rewinding and framestepping are therefore not seen as "excuses" to include lots of execution difficulty; rather, ideally, a player only has to rewind if they've made a conceptual mistake, i.e., a wrong decision in planning out their solution, rather than merely having failed to click at the right spot.
Some players take this to the extreme and plan out the entire solution before the hatches even open. This is the reason why labels were added to hatches, to show e.g. pre-assigned skills, or lemming status (regular, neutral, zombie). So that you can already tell what kind of lemmings come out of which hatch, before the level even begins.
This makes NeoLemmix more similar to other puzzle games, like Star Wars Pit Droids or Lego's Alpha Team, in which you have to plan the entire path ahead of time (in these games, this happens by placing arrows on the ground that influence the direction of movement of the characters). And then, you only hit play and watch the solution play out, but can no longer interfere. In NeoLemmix, of course you can interfere at any given moment, but you shouldn't be forced to reconsider your solution halfway through playing a level.
This is also why the NeoLemmix community frowns upon "troll features" like hidden traps or exits, even though these things did appear in the original Lemmings game. Even though there is a clear-physics mode in NeoLemmix that shows you all the trigger areas, thereby revealing hidden objects, much like with the rewinding and framestepping features, the idea is that the player should never be forced to use them in order to solve a level.
2) Meanwhile, there is currently a fork of NeoLemmix in development called SuperLemmix, made by WillLem. This engine provides the player a wider range of options, including even more skills than NeoLemmix (such as the Spearer and Grenader, with the only "skill swap" so far being that SuperLemmix has the Freezer skill as a replacement for the Stoner skill in NeoLemmix; Freezer and Stoner are very similar in their main application, and only differ in the details). WillLem himself has said that ideally, there would be no single design philosophy for SuperLemmix. So execution difficulty is seen as an alternative form of challenge here, not necessarily less valid than puzzle difficulty.
This is because SuperLemmix also has a Classic Mode, which makes the game more akin to original DMA Lemmings again (except of course there's still the extra skills), by removing a lot of the convenient features NeoLemmix offers, such as rewinding, framestepping, and skill shadows. This however is optional; you can still use all of the same features as in NeoLemmix by simply not activating Classic Mode. WillLem however has made the suggestion (not a requirement) that level designers test whether their levels are solvable in Classic Mode, for those players who choose to use it. This goes to show that excessive execution difficulty, in form of several pixel-precise skill assignments, are not appreciated in the SuperLemmix community either. As said above, sometimes it's inevitable, and/or a necessary evil, in order to enforce a given puzzle solution. But it should be avoided where possible.
I for one am currently not using Classic Mode, and are therefore also not testing for it in my level design.
Nevertheless, I try to keep pixel precision on a minimum, in both my SuperLemmix and NeoLemmix levels.3) I can't say too much about Lix, which is an engine wholly separate from NeoLemmix and SuperLemmix. It features some skills that NeoLemmix doesn't have, like the Runner, the Cuber, the Knockback Bomber, and the Batter. Some of them might make it to SuperLemmix eventually, though (the Runner is currently being discussed as a "Hoverboarder", which would be a little different; the Cuber essentially became SuperLemmix's Freezer; and the Batter is still up for debate).
My impressions so far are that a) Lix is often used for Multiplayer games and b) the communities seem to be a little more separate. There are lots of players shifting freely between NeoLemmix and SuperLemmix, but there's more of a dedicated Lix userbase. Of course, there are still also a lot of long-time forum members who frequently use both NeoLemmix and Lix.
4) There are also Lemmix, Lemmini, and SuperLemmini. All of these engines, as far as I am aware, are restricted to the classic eight skills. They differ in the details (there are certain skill behaviours that only work in some of these engines; for example, a Miner in Lemmini can hop over small gaps, which doesn't work in NeoLemmix). Some packs that exist for both Lemmini and NeoLemmix therefore required some changes to certain levels, and some levels didn't make it into the NeoLemmix version of those packs at all.
5) Finally, there are other engines, like Lemmings 2: The Tribes and Lemmings 3D, for which level creation is also possible. Lemmings 3D for custom levels is called "Loap", which includes a level editor - this was developed by namida just a few years ago. However, both Lemmings 2: The Tribes and Lemmings 3D require replacing the original level files with the custom ones, so backups of the original levels (or any other custom packs for those engines) should be made whenever swapping them out.
With all of that out of the way, I'll try to briefly address each of your questions. However, given the above disclaimers, some of these answers may only apply to some of the subcommunities, while others might have a different stance on them.
I'm mainly using NeoLemmix currently (with a bit of SuperLemmix), though I have never been that strict about "puzzles only". For example, I'm fine with hidden traps or exits, as long as they are in-flavour (I like placing traps behind terrain shapes like animals, in order to make them eat the lemmings), or are hidden for aesthetic purposes - just as long as they aren't hidden exclusively to troll the player. I have shifted more towards enjoying puzzles than execution difficulty over the years, though. So I may criticise some things now that I used to defend in my earlier posts on this forum.
Should levels have one, or multiple ways to solve them?
If a level can be solved in multiple ways, it is called "open-ended" or "semi-open-ended". Having a little bit of leeway in a level, with regards to which skill belongs where, can actually make it more challenging to figure out, as this increases the "entropy" of the level.
For example, if you have bottomless pits that the lemmings need to get across, that immediately tells you where at least some of the Builders have to go - thereby making at least parts of the solution more obvious. The Builders then serve as the corner pieces of the puzzle. Thus, it may be more challenging if the designer does
not include bottomless pits (at least not in the path between hatch and exit), but pits with solid terrain - which the lemmings could get out of using Builders, Miners, Bashers...
The difference between an alternative solution and a backroute (and unintended solution that's too easy) is usually assessed by the number of skills the player has left on the panel. One or two skills leftover may not be that dramatic. But if a solution skips the main trick, takes a shortcut through the level landscape, and there are a bunch of skills leftover, that's most likely a backroute that the level designer should then fix in a later version of the level.
Should creators be lenient or stingy with how many roles you can give Lemmings?
If by "roles" you mean "skills", this depends on the type of challenge you're going for.Open-ended levels often offer X of every skill on the panel, leaving it to the player to decide how to get through which section in the most efficient way. This can still be turned into a challenge (famous examples include the final level of NepsterLems, "Final Frustration"), when the level is so big that even having 20 of every skill is not nearly enough to make it through, unless you think very carefully about which skill to use there.
Thus, in general, I'd say puzzles feel more satisfying if the number of skills (and instances of each skill ) are accounted for (i.e., when level designers are, as you said, "stingy" with the skills they provide). Whether that means having just a few skill types (even if you have several instances of each), or just a few instances of a broad selection of skills (such as on 1-of-everything levels, which are generally very popular). Either way: When you use the last skill you have available and the solution is complete, it creates that satisfying feeling of everything having just fallen into place.
X-of-everything levels, meanwhile, are more about resource management. Yes, you can make such levels deceptively difficult (like "Final Frustration"), where, even though the skills aren't technically accounted for, you will still run out of quite a few of them (usually, the Builders are the fastest to run out).
However, Final Frustration is not your typical X-of-everything level. Many level designers place these on the lower ranks of their packs, and I have been a vocal critic of that. I'd much more prefer simpler puzzles on the low ranks, with just a few skills, including 1-of-everything levels, than huge open-ended landscapes with X-of-everything. The latter always feels to me like the designer didn't want to think of a solution for a given terrain layout, and simply leaves that task to the player.
If however by "roles" you mean "how many different purposes can the same skill or lemming serve?", then that's indeed a way to make a level more challenging.Again, by increasing the entropy of the level (=the degrees of freedom): The more places in a level a skill could potentially go, the more the player has to try out (either in practice, or in their head) before they figure out which skill needs to go where to make the solution work.
How hard should a level be?
As easy as possible on the execution side, as hard as possible on the puzzle side.
Generally speaking, at least. Of course, both will depend on player skill. But player skill in the department of level execution doesn't receive a lot of appreciation in the puzzle-heavy communities.
With puzzle difficulty, meanwhile, there isn't really a "too hard"; if you can create a lot of hard puzzles, that will simply result in a different group of players playing your pack (usually the more experienced ones). Many of us agree that it's easier to create a hard level than to solve a hard level. Nevertheless, making levels hard by enforcing one specific solution isn't that easy either - simply because the player will always go for the path of least resistance. Be prepared for players to find lots of backroutes in your levels. While level designers aren't forced to patch up their levels to block these backroutes, a lot of level designers do: They go back, change the level to the extent necessary, and upload new versions of the packs.
This is also why a lot of packs go through a period of pre-release testing. Where a level designer will handpick a few volunteers and send them the pack via private message.So they can already find some backroutes, which the level designer can then fix, before releasing the pack to the wider audience. Of course, you can do the latter, too - but then more players will discover backroutes, and not all of them will return to play the "proper" version of your levels, i.e., after you've fixed them.
What themes do you like or dislike?
Good question for you to ask this early!
Indeed, I'm fine with a lot of themes, as long as there is one I can recognise. I feel themes are important to keep a player engaged - much like themes in a story. I need a reason to care about a given pack, considering how many different ones are already available. If the level titles feel like a generic mix of everything on the designer's mind, it's harder to see why I should engage with it than if the level design adheres to an overarching concept, or tells a story etc.
A few packs by other level designers that do include themes I picked up on are:- SubLems (by Colorful Arty)
- ArtLems (by Colorful Arty)
- SEB Lems (by Flopsy)
- Lemmings Migration (by Nessy)
The go-to theme for my own packs is usually music, which is why most of the levels in my flagship packs are named after songs.
But eventually, that theme will get a little old, as I'm already starting to run out of well-known song titles.
I actually once made a thread about potential pack themes, since I had other ideas for options, but wasn't planning to use all of them:
https://www.lemmingsforums.net/index.php?topic=3929.msg71909#msg71909Fun fact: In this thread, Nessy actually said his first packs
didn't have any themes - which would include Lemmings Migration. But I would count it in the "overarching story" category, as in, telling the story of a group of lemmings on a journey. Sure, it's not a particularly detailed theme, but it's still more than just "(insert pack creator name here) Lemmings".
In the end, it's a learning-by-doing thing.
Don't regard these as "rules" you have to consider all in advance. Just try your luck, perhaps release some levels in the "random level sharing topic" first, and you'll learn from the community feedback. Some people are prone to accidentally include lots of pixel precision; others are prone to leaving backroutes in their levels; yet others might create easier levels, but ones that look nice, etc.
