nmenon10

I liked playing the game! A few things that worked well for me - a gradual ladder meant I knew word length was going to increase by 1 and was prepared mentally, the timer was not in my face distracting me as I tried thinking, the clue comes around the 25s mark to help me get unstuck. What didn't work for me was valid words not being accepted, leaving you a little demoralized. I see a lot of suggestions already around this and I don't have anything more specific to add to that.

But overall, this was fun!

This is awesome! Looked at the palette definitions in draw.ts - nice that it flows through all 18 recipes from a single array. Ran the 7 palettes through deuteranopia and protanopia simulations and 4 fail (clay-sea, ember-slate, rose-forest, coral-teal), mostly where a and b have similar luminance and color information is doing all the work. The good news is that your architecture makes it a clean fix: one new entry in the PALETTES array and all 18 recipes inherit it! Are you open to PRs?

Do I live in London? No. Do I find this fun and therapeutic? Absolutely yes!

One small thing: clicking between trains causes a jarring jump atleast for me. Felt like if I was going to select a few trains in a row, it might start making me feel disoriented. A smoother eased transition might make it feel a lot less whiplash-y.

Got to 13 before blanking on what turned out to be a pretty obvious word. Personally, I think the timer mechanic that would work best here might be a running bank rather than a per-word reset. Start with X seconds, and when you solve a word add back N seconds scaled to word length. Fast solves on easy words give you buffer for the hard ones. Keeps the pressure without the "lose a potential perfect score on one brain-freeze" frustration.