What are some tips for making difficult levels?

[Mobiethian]

I am going to be pushing the difficulty meter, so to speak, in my next rank. I'd like to make it (as well as some of my existing levels) have more of a sense of difficulty.

Level design is no big deal for me at all, but trying to come up with a good, tough strategy or two for a level is where I am struggling. Any suggestions and ideas for level difficulty would be greatly appreciated!

[Proxima]

Something that has often worked for me is to start with a relatively complex terrain -- sometimes from an existing level -- and find a solution, either for the original save requirement or for 100%, trying to keep skill usage balanced (at most N of each for a suitable value of N) and in particular to minimise high-value skills like builders and miners. Then make a copy, trim the skillset to what you used in your solution, or N of everything, or a patterned skillset like 3-2-3-2... and you have a level!

Of course, if you are doing this based on an existing level, keep in mind that you can modify the terrain as well. For instance, in "Rhapsody in Blue", my 2-of-each copy of Insane Steve's "Rhapsody", I cut out the pipes along the top platform, because it would have required extra destructive skills to get past them, ruining the 2-of-each pattern.

[Simon]

One approach is not to worry:

• Make several levels of medium difficulty.
  
• Have others playtest your levels.
  
• What levels did they find much harder than you anticipated? Those are hard levels.
  

My take on Proxima's idea:

• Place some terrain and give plenty of skills.
  
• Solve your level several times. Optimize the solution for skills.
  
• Trim the skillset to match your optimal solution.
  

Or:

• Play lots of levels by others.
  
• Build a level that combines two ideas from existing levels.
  

geoo's favorite:

• Play lots of levels by others.
  
• Submit your replays to the author.
  
• The author will judge some of your solutions backroutes. Sometimes, he'll also fix his levels to prevent your backroutes.
  
• Which backroute did you like the most? Were you surprised that he judged a particularly cool solution a backroute?
  
• Build a new level that has your cool backroute as the intended solution.
  
• I recommend to make the new level look sufficiently different from the original level. After all, the idea is now different, and it's nice to give different ideas different looks.
  

Draw levels on paper, mainly to brainstorm, but you can also plan details.

It's not necessary to start with an idea. You can make your level look like art first, then draw inspiration from that.

Not only the level can look like art, even the solution can look like art. E.g., when you solve a level by Pieuw, ask yourself: Where can we place the basher/miner so that it continues the longest? Now, if your level has only one miner, has no other destructive skills, and has lots of obstactles to mine away, it'll be easy to spot the miner placement. Can you disguise* the beauty? Will the player have an a-ha effect when he finally finds the idea?

*) By disguise, I don't mean invisible things, e.g., don't hide exits/traps fully inside terrain. Everything should be clearly visible and its behavior should be obvious. Still, you can make the player wonder which parts are important and which are merely decoration. As Icho describes: Hiding things in plain sight.

You can make levels harder by adding extra problems to a medium-difficulty level, but you'll have to test for backroutes from the extra skills.

-- Simon

[WillLem]

To add to what others have already said, I would suggest also aiming to find interest/difficulty/play value in something other than a limited skillset.

Maybe provide lots of skills, but make the map itself difficult to navigate. Or, make use of some of the many objects that are available in the modern engines to create a puzzle which involves interacting with them; buttons & locked exits are a favourite of mine, teleporters are good but be careful not to over-use them!, updrafts, splat pads, pickup skills, splitters, and of course the classic one-way walls... SuperLemmix has also recently added collectible items, to encourage designers to consider adding side-quests to their levels.

SuperLemmix also has a Classic Mode which removes all of the player assists in order to re-introduce the original game’s execution difficulty - another option for you if you want to add difficulty that isn't just "find the intended solution".

At the very least, consider switching things up a bit - have a level with a limited skillset and single intended solution, sure, but then follow it up with something else. Get creative, it's your level pack!

[IchoTolot]

Difficulty comes mostly from 2 factors in my opinion.

1.) The unexpected usage of tricks.
2.) Entropy. Or a bit more precise: Having many meaningful possible things to do with the limited resources you have.

To further explain point 2:
Let's say you have 1 builder and 1 water pit in the only path to the exit with no other workarounds. The builder is 100% locked in that position and is therefore no mystery. 
Having multiple possible routes where that 1 builder can really help you out on the other hand makes the choice unclear. Bonus points if the meaningful usage is even hard to spot in the first place (hidden in plain sight).
If this holds true for not only one but multiple skills (+ adding some unortodox trick usages) the level can quickly become a very hard nut to crack (if there are no glaring backroutes of course).

[Armani]

Looking into other people's pack with your desired difficulty would help you a lot.
Playing some packs is the best way to do so but from my experience, just watching the solutions to them could be helpful too.

[Mobiethian]

These are all some wonderful suggestions and amazing advice! I truly thank you all, I really want my next rank to have a some more difficulty in the levels compared to the other levels in the pack.

The many, many ways there are to achieve this, it's intriguing and again, thank you all, greatly! I'm glad that I have a white board in my home office.

[Pieuw]

I usually try to find new or interesting ways to use skills, or to make cool "choreographies" happen. Difficulty tends to come naturally when you try to get off the beaten path. That being said, don't make hard levels for the sake of difficulty only. Your main goal should be to make your levels engaging and/or surprising and/or gorgeous to look at

[Mobiethian]

Well said! I'm watching more of your replays today and you do such an amazing job with Lemmings. I am wondering how you knew that level 1 solution, I didn't even know it. hah

[JRT]

To further explain point 2:
Let's say you have 1 builder and 1 water pit in the only path to the exit with no other workarounds. The builder is 100% locked in that position and is therefore no mystery. 
Having multiple possible routes where that 1 builder can really help you out on the other hand makes the choice unclear. Bonus points if the meaningful usage is even hard to spot in the first place (hidden in plain sight).
If this holds true for not only one but multiple skills (+ adding some unortodox trick usages) the level can quickly become a very hard nut to crack (if there are no glaring backroutes of course).

This is how I see difficulty too. When we look at a level and try to solve it we in our heads are performing some kind of Depth First Tree Search as we imagine the use of skills and combos in various locations in different orders. When we think down a branch but then we can see we have one builder left but require the height of two then we know this branch has probability to be a solution drops to zero so we move onto the next branch, maybe by tweaking elements of this branch, for example, can we walk up a bit of terrain for the extra height, if so, think down that branch, if not, move onto the next wholly different solution.

Essentially, a difficult level has on actual solution but a large enough possible solution tree that requires that you chase through the tree a fair way until you realise you have to move on to the next.