We've made plenty of drafts in the last month. Many of those haven't left the Drafts folder. When we're happy with the puzzle itself, they're moved to Polish, where they are tweaked, fixed, rearranged, combined or split, and in some cases changed up entirely. Those we're happy with and have a place for in the level sequence are moved to Final, given numbers, and put in the next build.
But what goes into polishing a puzzle? Well, a lot of things, actually. Visibility improvements, preventing alternate solutions which could result in players missing important lessons or techniques, drawing attention to solution elements or away from other areas... the list goes on. I thought I might take a look at a concrete example: Adv_LP_DualRoute, now at #19.
The first pic I have here isn't actually the original draft at all, but rather the version included in Playtest Build 1. No matter! I've recently polished the level up a bit, so let's take a look at what I changed, and why.
EDIT 28/11 - I've come back and put in a gif of the level's final version, to show how the solution works and make everything clearer.
First: the overhang has been shortened by one block. This improves visibillity significantly, as the previous version blocked the player's view of the second laser. Shortening the overhang, however, means the player can climb onto it with only two corpse-blocks in the 4th row. This can be fixed by moving the first laser back one space from the 1st column into the wall.
In playtesting after this change, some players were misjudging the gap at 2,2,2 and moving into that space, thinking they would fall onto the corpse-block at 3,1,2 and not the water. To solve this problem, I added another tile to the overhang at 2,2,2 – the space below it must remain empty so the corpse-block at 3,1,2 can't be climbed (it would be pushed into the water instead), but extending the overhang means the player can move around that area safely.
Broadening the overhang to this location makes the rear platform accessible from the overhang. This isn't the end of the world, as if the player is on the overhang they've already 'solved' the puzzle and no more deaths are required, but we'd prefer that they bounce on the corpse-block at 3,1,2 to reach the exit... for a couple of reasons. Firstly, we think it's just a more interesting solution for the player to execute than simply walking across the overhang. Second, it reinforces the idea that while the corpse-block at 3,1,2 can't be approached from the left side without pushing it into the water, it can still be useful to bounce on – this concept is important in a subsequent level, and we want it in the forefront of the player's mind.
To solve this one, we simply raise that section of the rear platform. That, of course, puts it right next to the window, which we prefer to avoid, so we considered flipping the puzzle laterally so the rear platform is on the right side instead of the left. However, because the camera is angled 5° to the right, the silhouette of the overhang is clearer with the current layout than it would be when flipped. To keep the best visibility possible for the player, then, I decided against the lateral flip – the platform next to the window is a very minor problem, and one I'm ok with leaving for the other benefits it buys us.
Finally, I changed the ground tile at 5,0,5 from rough to smooth (pushable) terrain. This one was originally intended to prevent an alternate solution to the earlier version of the level. We discovered in playtesting that the player could reach the long overhang by climbing the corpse-block on 4,1,5 from behind it. The player could then simply walk across a second corpse-block at 4,1,4 to climb the long overhang, bouncing across the third corpse-block at 3,1,2. Though inventive, this solution didn't demonstrate and reinforce the idea of using the corpse-block on the 3rd row as a shield to access the laser on the 4th row anywhere near as well as the conventional solution – the few players who discovered this solution had more trouble with later puzzles requiring that technique. Although the newer version of this puzzle (with the shorter overhang) invalidated that alternate solution, I left the tile as smooth terrain. This was partly for aesthetic reasons, but mostly because it drew further attention to the line of the fifth column, which is central to the first parts of the puzzle's solution.
Well, that ended up being pretty long, but hopefully demonstrated how even small changes can affect a whole bunch of other things! This is a big part of the reason why polishing a draft can take such a long time – there are so many interconnected elements, and so many influences on presentation and player behaviour that need to be considered. And to think: one of the reasons I chose this level to talk about was that the polish changes were relatively straight-forward!