What Makes a Puzzle Game Feel Clever Instead of Frustrating?

Those come through playtesting.

Lots of people try it. Some find it easy, some find it hard. It should be a series of progressive challenges most people find difficult but ultimately complete it.

Generally the longer spent the easier it becomes. More powerful items, more potions, more keys, etc. The top players can get through it quickly, but slow, inexperienced players can eventually grind through. The only way to verify it is to get a variety of people to play and give feedback.

As the designer you know all the secrets, so you are a terrible judge of difficulty.

As a puzzle game player, I find player performance and fun highly dependent on taste. You cannot please everybody.

For example, solving a Sudoku follows an unavoidable structural pattern: initially looking at every row, column and square for low hanging fruit and obvious partial constraints, then iteratively deducing less obvious constraints from coarser ones, filling cells and adding constraints from those cells, with difficulty peaking in the middle at some particularly challenging step and crashing near the end.

Each of these three phases could be unappealing or abnormally challenging for some players: some are unable to look at every part of the puzzle systematically, losing concentration and missing constraints they need; some might be so used to processing certain patterns that they find them boring, or conversely they could be unable to make a certain deduction, or require an exceptional effort, for lack of technique; some might be bored by the final collapse and treat filling the last cells like a chore.

Computer puzzles are often larger and more complex than a Sudoku, with more opportunities for having good and bad parts. For example, in unlocking type puzzles like Sokoban where pieces need to be moved to reach a certain final state, I enjoy "untying the knots", but I don't bother with the traditional optimization of the number of moves: after beating a level I want a new challenge.

Difficulty spikes are going to happen unpredictably when the player lacks some skill that they should have learned by that point of the game. In a computer puzzle you can usually detect difficulty spikes reliably (many failures and aborted attempts, long thinking times, many moves and/or errors, "bad" errors that should never happen) and you can offer tutorials and suggestions that are relevant for the level, or if possible relevant for the player's specific failure modes; it should be less irritating than excesses of preemptive teaching.

realtechnolab wrote:

How do you decide when a puzzle is "challenging enough"?

What techniques do you use to avoid difficulty spikes?

Do you rely more on playtesting, analytics, or your own intuition?

I haven't shipped an abstract puzzle game, but I have play balanced tactical wargames and 4X games. They have a puzzly quality to them. I've also played lots of old interactive fiction titles, and the problem of a "stumper" puzzle was quite a big design challenge.

I'm fine with giving different difficulty levels for a game, because I don't expect all players to have the same level of ability. Especially, even a very bright 8 year old kid, is not going to have all the analytic capacities I have as a middle aged man. At least people within the normal bell curve of the human population. I was that 8 year old kid; I've had plenty of time to hone all those skills.

Offering different difficulty levels is more work though. Consider if it is cost effective for you to do that extra play balancing, to reach more people than your core demographic. Don't forget that your core demographic is smart people who can handle this stuff. You shouldn't pander too much. You should endeavor to bring people up to the level of those who are actually good at these things. It's a kind of instruction and training, if they're lacking some of the skills and haven't bothered or happened to develop them themselves. You never know what someone's capabilities may really be, but... you don't have to bend over backwards for people who just can't work things out.

If you're doing an abstract puzzle game of limited format, I wouldn't expect the surface area of the game to be all that large, as far as where difficulty spikes come from. Then again some kinds of abstract games like chess or Go have serious sticking points in them. I don't think you're making one of those, based on your description, but if so I reserve the right to backpedal lol.

In contrast, adventure games were awful because the surface area of what anyone has to reason about is so broad, that it would often become a game of "guess the author's mind". Like WTH were they on about? Back in the day this led to things like "the player's bill of rights", things that you as a designer should or shouldn't be doing in such games. The game industry mostly didn't absorb this learning curve, and would make these same authoring mistakes over and over again. It got boring and that segment of the game industry imploded in the late 90s. There weren't enough people to pay for "stumpers" anymore, not with the visual production values of games in general going up and up.

Unity Source Code wrote:

I think the biggest factor is making sure players understand why they failed. If a puzzle feels unfair or relies on guessing, it quickly becomes frustrating. I like introducing one new mechanic at a time, then combining it with previously learned mechanics so players build confidence. Playtesting is probably the most valuable tool for spotting difficulty spikes, because players often get stuck in places the designer doesn't expect. Games like Portal and Baba Is You do a great job of teaching mec...

Agreed. Valve does an excellent job at playtesting.

When a playtester (in your demographic) plays a beginner level, they should only be stuck for very short amounts of time before continuing to progress. This develops flow. As the game goes on, it is acceptable to get stuck at longer and longer intervals.

If a playtester gets stuck for (subjectively) too long or doesn't understand the mechanics and thus uses a very awkward solution, you know something needs to be fixed. As Valve puts it, they keep developing/playtesting until "no longer excruciatingly painful to watch the playtests."

There's a great article on Valve's playtesting here: https://gmtk.substack.com/p/valves-secret-weapon.

Interesting. good puzzle design seems to come down to making the player discover the solution through clear clues and gradual difficulty, rather than simply making the puzzles harder.