• Projects

Ever thought to yourself - I wish Quordle was harder?

Status: Closed early access. Prototype with 40 levels.

Rules

I am regularly playing Quordle, and found my favourite part was often late in the game when the played words had resulted in interesting constraints on the final puzzle.

I also liked the challenge of trying to choose which of the 4 puzzles to solve to optimize your chances.

So I wondered - could we construct a Quordle like puzzle such that you could always solve it in 4 steps? Could I make a series of overlapping puzzles, such that at each step, there is a single unique answer for one of the puzzles?

And the answer is - yes, it is possible.

Crucially, there seems to be a large number of possible games. My initial fear was that the constraints may be too much and I end up with a ludicrously small number of playable scenarios.

Development

Like the architecture of the original Wordle, I am keeping it all client-side, with a static site deployment.

The initial challenge was to find and create valid levels.

After that, I have been focusing on play-testing and making the game more fun.

I still have to do some work with on-boarding. The rules are hard to explain, so current play-testers need a lot of guidance. I’m looking at improving this through tutorials and introductory levels.

Alphabetical -> Frequency -> Prevalence

Initially my word list was alphabetical. With simple algorithms, this resulted in my puzzles all being “A” words. ACORN, ABACK, ARENA etc featured in multiple levels.

I adapted here by adding randomness to the algorithm.

But I came across the issue of puzzles being too hard, with too many obscure words. So I need to be able to tune the difficulty of puzzle generation.

I updated my word list to be sorted by frequency, based on how often the word is seen in written text. But I noticed some common words at the bottom of the list - ERODE, PIXEL, FUNGI. And unusual words relatively high up - PRIMO, VOILA, HARRY, SALLY.

I then learned about “word prevalence”, which is how likely a word is to be known by a speaker.

With these changes, I’ve been able to create a “easy-meter” algorithm, which measures how easy a particular puzzle will be. By using this I can find easier puzzles for onboarding users.

Eventually, this could be used for increasing the difficulty of daily puzzles during the week. Crosswords are known for “Friday is the hardest” for example.

What’s in a name?

The established convention for Wordle like games is that they have to end with “dle”.

Game about flags? Flagdle. Game about things? Thingdle. Making a list of all Wordle inspired games? Listdle.

During early testing, I found 4 puzzles seemed to be the best number. To me this evoked a “Grand Slam” motif, like tennis or golf, giving the user 4 “major” puzzles to solve. And with this, a nice clear goal of achieving a grand slam.

Just another Quordle / Wordle?

I have made some good quality of life improvements over the existing games out there.

• Click on a puzzle to set the keyboard for that puzzle
• Customisable settings like letter size, position of enter button
• Hint system, allowing a user to be given a letter.

Learnings

• That colour and usability and accessibility are subtle.
• Test early
• Difficulty is very variable, which can feel cruel.

• Projects