Hi folks,

[The Lix keyboard layout] is for people like you, not like me.
-- GigaLem

But once you make it second nature, SC2 feels smoother, snappier than ever before. Welcome to the world of enhanced control.
-- JaK (Starcraft 2 hotkey layout designer) about an advanced SC2 keyboard layout

We've had a lamenting discussion in IRC 2 days ago, and then in Lix in-game chat last night. Giga doesn't like the Lix hotkey system. Some definitions:

Skill keys: This is what Lix and Clones have. For each skill, you can map a key of your choice to it. That key will always select that skill.

Positional keys: This is what L2 and Neolemmix have, and L++ (= old Lix) had it until 2010. For each skill button on the screen, you have a key. These keys are arranged similarly to the on-screen buttons, in a row or grid. The key will therefore select different skills on each level. You can't remap them in any of those games, but remapping doesn't belong in this definition.

Matching keys: This is what L1 and old Lemmix have. It satisfies both definitions above: F3 is always the first key, and always climber.

I consider skill keys by far the superior solution. A modern game must use skill keys. Here is a ton of reasons:

• Muscle memory. You don't have to look at the skillbar. Each key is the same skill, always. That makes you faster.
  
• Comfort. You don't have to move the hand, you map skills to left-hand typing letters. If you mouse with the left hand, you map them all over to the right hand. That makes you faster.
  
• Ergonomic rolling motions. You want common composed actions as a smooth finger roll, like walker-jumper. The Lix default layout is not optimal for this, but the Lix layout can be amended (<-- link to topic). That makes you faster and removes human errors.
  
• Directional select. This must be in reach at all times!
  
• Extra surrounding keys. You can have pause on space, reload or fast-forward on the number row, ..., all quick to reach.
  
• The strong players all like speed/ergonomy-maximizing layouts, in both Lix and Clones. And Starcraft 2, see initial quote of this topic. And Quake, and, and, and...
  

What are the drawbacks of skill keys?

• It takes longer to learn.
  

About this drawback, I agree with the linked Starcraft article:

[Our custom layout] does thousands of actions, and every common action, faster than standard or grid [= positional/matching keys], giving you more actions per minute. It minimizes strain, increasing endurance. It is not what you are used to. It gives you the edge. It’s time for better mechanics.

The time has come to choose between what is right, and what is easy.

[...] It greatly depends on your natural ability to adapt to change, but players often report between the range of 30-100 games [each takes 20 minutes on average] for a basic understanding and feel of the layout and 100-200 games for a complete understanding and feel of the layout.

And that's for a game which has about 5 to 10 times more keys to learn than Lix. Learning the Lix layout is faster than learning SC2 layouts.

Your choice of hotkeys is printed on the on-screen buttons, to help you learn it.

Yes, learning a new layout makes you play worse at first. Nobody cares. :-) I've learned a different keyboard layout and ate 1 to 2 months of sucking at typing. I've remapped and relearned in Lix and Clones several times, and took the temporal suckage.

geoo likes skill keys so much that he hates Lemmix's default, unremappable keybinds. He has remapped them with an external program to match his Lix layout.

To compare, what happens with positional keys? Giga suggested the F1 through F12 for the 12 skill buttons. Maybe it's faster to understand at first.

Then, you will end up distracting yourself from the game by looking at the skillbar all the time. You must look at your hand while you move it in position, but looking at the hand shouldn't happen at all. You will never have more than 3 or 4 keys in easy reach. While you can do rolls, by setting the hand up for them before they happen, they will never feel nice!

<SimonN> anyway, he knows what he wants, but he doesn't understand what it would bring him
<SimonN> I know that he might be slightly faster than now, but can never have the good advantages of the clustered layout, and I'll have to implement it and support it as a user option
<NaOH> shh, pat pat
* NaOH ruffles your hair with her bunny paws

Making Lix keys matched?

That requires to arrange the on-screen buttons in a cluster similar to the keyboard keys. Probably not feasible, even with a higher resolution.

Making more skill keys (we have 12 at any one time, but there are more skills in the game than 12) doesn't solve the problem too well, when they're all aligned in a row. There is no good set of keys all in a long row.

A clustered on-screen display of the buttons is possible with a higher resolution, but it would look a little unwieldy. Also, a row of on-screen buttons is best for the mouse: Each button is at the edge of the screen, where the mouse cursor can't overshoot, making the buttons easy to hit.

More data from new users

When Proxima tried Lix for the first time, he wanted keys to move the skill selection left/right. This is also a positional system, and therefore slower than skill keys. Proxima, what is your current opionion on all this after using the game for a longer time?

NaOH's brother has tried the game once. He wanted a matching keyboard layout immediately.

Who has additional experience as/from a new user?

-- Simon

The crash of Lix upon loading geoo's replays is a bug.

Instead, the game should dismiss the replay as faulty: The replays aren't replays. They are level files -- identical to the levels in the contest. ;-) Akseli got this archive from möbius, so möbius or geoo should find the proper collection of replays.

-- Simon

It has also been suggested that some semblance of my likeness be used for the "ZTIMER RAGE!!!!1" image warning players of the occasional timed puzzles in Lix builds.

I consider not warning the player about a timed level, a bug (!) in Lix, not only a missing feature. Bugtracker entry. That, and the next-level button, are interesting development aims -- for when I feel like going back to C++ again. ;-)

I'm fine with this. *shades descend* 

Yes, people shall deal with it.

-- Simon

Yes! And make the list longer, 20 or 30.

-- Simon

challenge that timing can bring

For singleplayer, many consider that a chore, not a welcome challenge.

Singleplayer needs tools and aids to easen execution. This design guideline has been very well received in the past years. Untimed exploders fit right in.

Multiplayer is not touched at all -- except maybe for proposition (3), i.e., consistently untimed L1 bombers, which are not used anyway in multiplayer.

The choice between (2) and (3) is whether consistency or ease-of-use is more important.

-- Simon

Hm, why would you prefer Lemmix or Lemmini above DOSBox?

Better controls, better user interface. Action replays combat the artificial difficulty of executing solutions for long levels. With Lemmix, you get 100 % DOS physics nonetheless.

If they're compared against L1/ONML and not Custlemm, then they're also faster to access the desired level. Level passwords have become obsolete technology.

This doesn't make Dosbox a bad choice, use that if you're used to it already.

-- Simon

Thanks for the replies so far. I'll annotate the main reasons I could make out.

New player screws up his solution by fling-exploder: Even if the player hasn't played the tutorials, not expecting the flinging at all will happen a few times at most. The fling-bomber's approximate effect is easy to anticipate afterwards. It's a normal learning process. Right now, the player has to learn both exploders anyway. Culling bomber 1 will make it easier, not harder, to learn the game.

Solvable levels become harder/easier: This can happen both with untimed bombers and/or replacing them with fling bombers. I'd eat this one, reorder the levels if necessary, and benefit from the better game design.

Solvable levels become fiddly: I don't expect this to happen much, unless the level is already fiddly. Counterexamples are welcome.

Timed bomber levels become obsolete: Not at all. The game helps with execution more than any other Lemmings variant already. Losing the timing fits in line here, and therefore supports the whole-game experience. Here are my opinions on the named levels:

• Follow yellow brick road: Seems good enough with untimed exploders. It might move towards the easier side a little bit, but it will remain a fresh level in the pack. Could work with L2 exploders after slight terrain changes, and is immune to backroutes.
  
• Dr Strangelix: Still entirely nontrivial with L1 bombers, and a great puzzle. This would need heavy terrain changes for L2 exploders.
  
• Once you pop, you can't stop: Would love untimed bombers here. Again, I deem the main challenge to be the planning here. There is a slight chance of backroutes near the exit with L2 flinging.
  

Solved levels get backroutes: This seems the most weightful argument. I can't tell how many levels are liable to backroutes here. When you design something with the L1 bomber in mind, you can be sure right now there won't be any flinging. This guarantee will be gone if the game auto-replaces L1 with L2 exploders. Even unlikely backroutes like in "Once you pop, you can't stop" are risky, because the lixlfpack has been strongly tested.

Making L1 bomber and L2 exploder more different from each other: This is in order if we can't get rid of either one. Important design goals are simplicity and ease of learning. Scrapping the L1 bomber contributes most to this goal, and making it more different from the L2 bomber contributes somewhat.

-- Simon