Originally Posted by 10tonsLtd
We did go through all the level type variations during development. We ended up trying to go for plenty of variety instead of emphasizing a single method of creating levels of different level of challenge. It's true that several of the trickier levels in the game are those where the grid is limited and/or broken, but it's not all there is to it. As you cut the grid down, you're also removing possibilities for wrong solutions, which in turn makes the level easier. In other words, if the player chooses to solve levels simply by going mechanically through all the likely solutions, the levels with the least possible solutions are the easiest.
We also just randomize the starting positions of peablins at the start of each level. If the level is very simple, sometimes the peablins end up being close to the solution right from the start. This could of course have been eliminated by making sure the peablins always start as far away from the solution as possible (or, indeed, from manually fixed locations as in the tutorial levels), but we chose to dedicate that amount of work elsewhere.
Finally, large levels facilitate the player making things difficult for him/herself better. Sometimes the player gets a whiff of a solution, and starts to work towards that... And then keeps working on it to make it work... And then works a bit more just to find out that it's never going to work. This is sort of optimal gameplay experience for a lot of players. It's of course entirely possible the player goes for the correct solution the first time around and breezes through the level.
As for optimal solutions, or regarding single group vs. several groups of peablins at the end, it's not considered at all. We wanted to keep the basic rule as simple as possible: everyone must hold hands, all hands. That's it (except with Comfort Zones). We did consider developing the rule set in various ways, including some grouping preference, but we didn't feel it fit well with the main goals of the game; more Joining Hands, just prettier, more forgiving and with new characters.
I also can't stress enough how huge an impact experience makes to this kind of very limited gameplay mechanics. We've had test players take 25 minutes to complete the first chapter. And they're not stupid people, they just haven't played this kind of game ever before.
This actually exposes a very strong characteristic of the basic design; once you crack it, it gets really kind of easy. Then again, if you haven't cracked it, even the easy levels are really quite hard. The learning experience is a great rush for a lot of people! The basic mechanic is quite stretchy, so it does allow quite difficult level designs, but the spectrum is really wide for a single game. Or then we'd just have to double or triple the level count, and in that scenario we just hit development cost wall pretty much head on. Naturally, if the game performs well, we probably will make more content, and the new content will be mostly or only of the most challenging kind.