Comparing data from both iterations, I can draw some conclusions about what constitutes appropriate progression design.
Hand-Holding Is Dangerous
One of the obvious features of Iter-1's experience graph is a drop in enjoyment and difficulty between Levels 1 and 2. These levels were reserved to explain - very slowly - two basic interactions in Circle Puzzle to new players; movement, and locked pieces. It was all a precaution against players not understanding how to play the game, and was done under the idea that only one new concept should be introduced at a time. But spending two whole levels to introduce two simple ideas was clearly boring. Meanwhile, where the two levels were combined into the first level of Iter-2, no significant drop in enjoyment was seen, and difficulty did not rise at all.
At the same time, Iter-2's Level 2 seems to have increased in difficulty too quickly. Many players struggled on this level more so than on the following Level 3. I think this is because Iter-2's Level 2 was added as a more complex single-gem puzzle, as I was unable to design any simpler ones. But, clearly, the opening levels of the game demand more attention.
Making Tutorial Stages Fun May Be Impossible
I was disappointing to see lower enjoyment ratings for tutorial levels (Lev 1-2 in Iter-1, Lev 1 in Iter-2), but I wasn't surprised. Though a consistently enjoyed game would have been preferable, it's hard to imagine how you can maintain the player's interest while teaching them the mundane functions of the game. The fun in a puzzle game is solving a difficult problem. In tutorial levels, it was important to avoid distracting the learning player, so any difficult problems were avoided. Instead, Levels 1-2 in Iter-1 and Level 1 in Iter-2 were mostly brief tests that demanded the player to learn basic mechanics.
It would be interesting, however, to see if more puzzle-like challenges could be worked into these segments of the game, without disturbing the player's learning process.
More Challenge Sometimes Means Less Completion
Completion rates for both iterations can be seen below:
Where most players were able to complete all of Iter-1, only half could finish Iter-2. If the game was commercially released, and assuming that players who didn't finish the game on their first try would never go back to it, half of the game's players wouldn't enjoy all of its content. Does this indicate a shortcoming, or simply a more challenging game?
One of the main things I've learned from the development of this game is that levels were rarely received as I expected them to be. In Iter-1, I assumed that Level 5 would definitely be perceived as the most difficult. Instead, it was rated to be roughly as difficult as Levels 3 and 4. Level 2 was considered to be easier than Level 1, despite my assumption that the added mechanic and very minor puzzle-solving element would demand more effort from players. This suggests that I need to take more care with how I order my game's progression, and utilize playtests to check for these mistakes.
After practicing the difficult process of structuring a game to achieve flow, I'm more aware of what I need to do to make an engaging game, but I wouldn't say that I've accomplished flow. The game was not phenomenally well-rated; average ratings were between 3 and 4 out of 5 (Where 1 was "Not At All" and 5 was "A Lot"). It's not bad, but the game can always be improved.
Though I did see some enjoyment in the playtesters I was watching directly - and though these moments of enjoyment did occur in the circumstances I'd designed for, i.e. on the triumph of beating a puzzle - there's always room for improvement. In the case of Circle Puzzle, I suspect that work could be done in areas beside progression/tutorials. For example, many players progressed through the game without actually understanding how to approach each puzzle. I would ask a victorious player if they understood the significance of move-ordering; they'd indicate that they did not, and that their moves had mostly been guesswork. Additionally, the game's art, UI, and kinaesthetics were quite poor (not being the focus of my work), and I felt that the lack of interactive feedback/game-feel harmed the game's enjoyment.
I look forward to employing flow, pacing, and skill-gates in my future games. Furthermore, the language I've learned to refer to these concepts will serve well in communicating with other game designers.