I'll reply to the general arguments at greater length when I have time. For now, a quick look at the specified levels.
Detailed discussion follows in spoiler tags, as it reveals several intended solutions. A brief summary:
* Most levels will be okay with fixed SI.
* The main exception is levels with a time limit, where fixing the SI would prevent enough lix spawning.
* Compression Method 2, The Hotel in Hell and The Continuum Hypothesis have other problems, which may be fixable.
* As a separate issue, Buridan's Lix should have initial SI raised to 60 due to new digger behaviour.
Compression Method 2: The entire puzzle here is to use the SI to merge lix into a single location. Unsalvageable without variable SI.
Thomas the Climber: Solution requires sending up two climbers very close together. Easiest way to achieve this is setting SI to 4. However, without changing SI from its initial 50, you can wait until at least 19 lix are out, when there will be two close enough together. (Could set initial SI to 4 but this might make the trick too obvious?)
Changing of the Guards: Originally intended solution is to set SI to 4 initially so that last lix from left hatch can start digging a.s.a.p. allowing floater from right hatch + others from left hatch to be saved. However, digging with second-last lix from left hatch allows this solution to work without changing SI.
Buridan's Lix: Actually has no problems, except that initial SI must be raised to 60 due to new digger behaviour.
The Hotel in Hell: The problem here is that the initial SI must be kept, to allow the gap to be bridged without losing lix (there is no way to hold the crowd), but the SI must then be changed to 4 to allow all lix to spawn in time. (This actually should affect several other levels, such as Toccata.) Geoo has a crazy elegant solution involving tight SI control to merge most of the crowd.
The Continuum Hypothesis: Tight SI control is necessary to merge most lix (SI 49) then send up climbers for the I am A.T. trick appropriately spaced to gain the maximum possible height (SI 36). The only way to deal with this with fixed SI would be to give more climbers and diggers, or else to lower the height of the wall. Not the best level ever due to highly demanding execution, so I'm okay with it being toned down.
Thanks for the response. Gave me lots of opportunity for reflection. After sleeping over it, I feel I can dissect it fully. >_>;
why this couldn't simply be left to the level designer's discretion as it is today.
It is an unfitting feature that has not been pulling its weight.
To go from the game with everywhere-fixed SI to one with rarely-used variable SI, we incur these costs:
1. Allow the player to change the world in a new way, unlike anything else in the game. We may be used to it because L1 has it, but the difference of variable SI to skill assignments is glaring. If wanted, I can elaborate on the value of simplicity and conciseness in game design.
2. Make two buttons in the game UI, and display the current value. The buttons alone are highly nontrivial, taking left and right clicks. Setting a number freely in a game is UI bloat -- this is what city simulation games offer, or "Myliege, how much grain do thou wish to distribute to thy peasants?"
3. Introduce these buttons to new players. There is a ton of problems hidden here that have been bugging me for years.
- A time limit is self-explanatory, provided you notice it's there. These buttons are not.
- They aren't needed in 98 % of the levels, and don't scream "use me" once they are.
- They stop affecting the world after all lixes are out. Even when they do something, it's not obvious what they do -- people will change the SI by 1 and see no difference. They violate discoverability.
- As a comparison, the time limit doesn't advertize itself at level start either -- this bad visibility has been reported as a bug.
4. Force even experienced players to check another level environment setting -- whether or not the SI is variable -- each time they play a level.
5. Make a second setting in the editor that depends on another editor setting, and must be changed along with it. (UI improvement would be a checkmark in the editor for fixed/variable SI, fixed by default. But this doesn't leverage any other numbered reason or bullet point, and it a rather high amount of work compared to the benefits.)
What benefits do we get?
Levels don't need variable spawn interval to be solvable. Even if they were thought to need it, they don't need it, or can be adapted surprisingly easily.
An exception are levels where SI fiddling is absolutely the main idea. And here, I value game design over pulling along every bit of accumulated tradition. If only 5 levels out of 500 would have to be culled, it's already a success.
Nepster has claimed how it helps execution. Yes, if the level designer can speculate on the player using SI, the level designer is less inclined to ease execution otherwise. I would like to see levels that can't be adapted accordingly.
So, the variable SI brings way less than the time limit -- unless unfixable levels are dug up, all we lose are the very few levels designed explicitly around SI changing. Its cost is much higher, and doesn't afford this.
We handle these issues by recommending best practices for level designs.
Time limit: This isn't a method to affect the world as a player, which is the main reason I want to get rid of variable SI. As discussed, I have claimed this has more benefits in return, at lower cost. Also, I'd compare "all levels have unlimited time" closer to "all levels have SI 32" than "all levels have a fixed SI".
Builder stretching: Allowing builder stretching in the physics is simpler than disallowing it. Disallowing variable SI, or disallowing time limits, removes complexity from code, and from the player's mental model. But disallowing builder stretching would introduce code that checks for how long a walker has been walking since being a shrugger last, and then disallow builder assignments here.
Alternatively, the shrugger would have to walk to the tip of the brigde before shrugging. I've considered this before, I have shunned it due to looks and tradition.
I'll concede that even the historic evidence suggests the game designers at DMA do not value variable SI as a level solving tool, starting with the fact that they removed it as soon as they got to Lemmings 2
Hmm, L2 and L3 don't have it. They don't make up for crowd control too nicely, but that's an issue with level design and other game features.
Lemmings 3D and Lemmings Revolution weren't designed by the same team anymore.
I assert that the two teams behind L3D and LR fell into the same trap as most developers of Lemmings-like games: Copying features from L1 without reflecting on their merits.
eliminate otherwise fine levels in the original game like "Just a Minute (Part II)" and "Flow Control".
Flow control can be fixed with runners and longer walkways. :-) Just a minute 2, I'd have to take a look.
I'm not trying to cull variable SI from Lemmings 1, but from Lix, where there are better features for the desired effects.
-- Simon
PS. Conundrum can be adapted by setting the SI to 4 and giving one runner. (Or two runners, as red herring.) The level is short enough that the runner can't build over the lava without stalling the crowd.
Nepster has claimed how it helps execution. Yes, if the level designer can speculate on the player using SI, the level designer is less inclined to ease execution otherwise. I would like to see levels that can't be adapted accordingly.
In any sufficiently complex system (such as Lix), you can remove any non-essential part and still adapt everything. E.g. you could remove all time limits or all traps or all other objects like springs, steams, ... or the ability to rotate terrain pieces or level wrapping or... Nothing essential would have to change!
The main point is not, that SI changes cannot be removed, but that they help level creators and players.
To get you some data, I manually went through (most of) the collected replays for the community level pack (version 2015-06-22, viewed with an old version of Lix) to look for replays where SI changes are an essential part of the player's solution, but were not needed in other (respectively intended) solutions. Here is the list:
everywhichway_Akseli
fearofheights_sekti
halfwaydownthestairs_clam
halfwaydownthestairs_geoo3
impetus_geoo2
jacobsladder_sekti
labyrinthofdespair_Akseli
labyrinthofdespair_proxima
lixcannon_sekti
lixferanda_Nepster
lowprofile_Ramon (same lowprofile_geoo)
overmyhead_mobius
slippingagain_Akseli
slippingagain_Ramon
snowjump_geoo
survivalofthefittest_geoo
thebottomless_Ramon
thehotelinhell_Akseli
thisisastickup_Ramon
trapeze_clam
waitwhyisthereatree_Ramon
And here some levels, where all solutions use different SI changes. These levels will likely make problems when adapting to fixed SI:
thatpeskygap
ticktoxic
toofartowalk
To go from the game with everywhere-fixed SI to one with rarely-used variable SI, we incur these costs: ...
If anything, points 2 and 3 are bad UI and not an intrinsic problem of SI changes. And it's not that bad in my opinion - you display ingame help texts!
Point 4 doesn't hold because players will have to look at the SI even when it is fixed.
Point 5 I don't understand: Assume no SI change is required in a level. Why is fixing the SI then better than allowing changes?
Flow control can be fixed with runners and longer walkways.
There you have the same problem as with We're in this one together: I would rather change the SI than find the correct placement for the runners. We changed the bombers to untimed ones to remove placement problems, so why reintroduce them?
One more point: If you are creating a game without level editor, then you can expect that the level creators are experts and impose such restrictions like fixed SI, ... Experts like us here will surely find a way around it.
But having a level editor, you want to give everyone enough different tools to make their desired level. And SI-changes is one tool here that simplifies the level creation process. And I prefer people creating levels with SI-changes (even when the level can be modified to remove them) over people getting frustrated that they cannot produce a level featuring their level idea. Or they would create levels needing very precise skill assignments. I must confess that two years ago, I would have kept the old version of Heed the Traffic Light even with the current mechanics and fixed SI - just try it and see what I mean :lix-evil:.
Suggestion: Why not keeping SI-changes possible, but try to modify all current levels so that one doesn't need that feature? Then players may choose whether they prefer SI-changes or solutions via better skill usage.
Nepster, thanks for making the list to try.
I'll be meeting with Icho tomorrow, he's gonna show me a couple more examples of his own.
I've begun going through the suggested replays. Here's my results for the first 5 levels, and Pesky Gap. My interpretation: The players use the variable SI because it's available, but it doesn't make their solution easier to execute. Every time, once the solution idea is formed, were the players subjected to fixed SI, they'd have executed the idea similarly fast and precisely. Concrete reasons for this interpretation:
Every which way: The key is to have exactly 2 turn in the miner hole. Once you know that you want to do this, it seems to be of similar difficult with/without spawn interval. Of the 5 levels examined, this the only one where I'm unsure about variable SI; all others are clear cases how it's unneeded.
Fear of heights: Spawnint change is purely a fast-forward here -- give fewer lems, or a faster SI.
Halfway down the stairs: People use this to cluster the lems faster, which would cluster anyway. In particular, if you do nothing but wait, you can still bat about 7 lems to the top of the first step.
Impetus: No matter what you try, the crowd is safely contained. This level should have SI 4, it can't lose anything, but allows easier timing.
Jacob's Ladder: Many spawnints seem to work, and they all lead to equally beautiful solutions, with the same idea. Let's fix one of these working spawnints at random -- that cannot not kill the puzzle, because most guessed spawnints would have worked, too.
Pesky Gap: The straightforward adaption works: Give fewer lems, give runners. I mentioned this in IRC on the day I made the thread. I find the entire level a bit lush, but that opinion is off-topic.
I found this general argument of yours to pull most weight:
There you have the same problem as with We're in this one together: I would rather change the SI than find the correct placement for the runners. We changed the bombers to untimed ones to remove placement problems, so why reintroduce them?
This is an excellent point. Is it applicable to levels other than mostly-SI-centric levels, which I don't care about much?
But having a level editor, you want to give everyone enough different tools to make their desired level.
When a level requires too much precision, it sucks. People will complain to the level designer, and the level designer will improve. When several levels require finding out the correct SI first, they suck. People will complain at the level designer.
People should not accept badly-set SIs, even when they can change it during play. They should report the bad design.
If anything, points 2 and 3 are bad UI and not an intrinsic problem of SI changes. And it's not that bad in my opinion - you display ingame help texts!
No, it is worse than fixed SI. If you want variable SI, you have to implement it somehow. No matter what you do, it's more cluttered than no button at all.
Point 4: I don't mean that players need to look at the exact number. I mean that players need to look at whether or not the spawnint can be changed. If it's fixed, the exact number isn't relevant too much; you have to experiment in the level anyway.
Point 5: Fixing the spawnint makes the player concentrate on the core mechanic of the game, assigning skills to lixes. So far (5 levels of your list examined), I haven't seen a good example for how variable SI gives power or comfort over the well-designed fixed-SI version. If this holds true for the remainder of the list, then fixed spawnint is comparable to giving unlimited time, instead of an overly generous time limit: Less to worry about, without losing puzzle scope.
Suggestion:
We shouldn't give options that don't do anything useful, or are mere placebos. Again, this depends on further examination of your list; so far, I'd lump 5/5 use cases in the placebo category.
-- Simon
We're in this one together is a good nontrivial example.
On first sight, it's a prime example for runners: Compress lems to walk through a trap. Nepster has warned explicitly against this idea, and I see first problems with runners, too. Several runner groups would have to cooperate.
The first idea is SI about 30-40, give 1 extra floater and 6 runners. Send two clusters of 3 lems through the leftmost trap. Let the trap kill 2/3 of each group. The level would lose a bit of its spirit, because one idea of the intended route is to have these groups coordinate, losing in total only 3 lems to the trap, not 4.
There is still an extremely interesting puzzle left: Bashing and bombing the solid part is a level-worthy idea on its own. You don't guess that from your inital look at the level. This seems to be the core idea. Everything else is built around it, and this core is left fully intact. But I don't want to make this call for Steve, the designer.
Change of opinion from example levels: Lower probability than 3 days ago. When I made the thread, I expected more levels to pose design problems to remain solvable, keeping their spirit. That assumption turned out wrong. Then I expected more levels to rely on it for comfort. My small personal sample suggests there are few such levels.
Nonetheless, I want to look into Nepster's list further, and meet with Icho tonight.
-- Simon