Some letters repeat more often than the others (Question)

[willwade]

See:

googlecreativelab#9

Set of letters {A, E, H, I, M, N, O, S, U} repeat at near about 5-6 times the frequency of other letters such as {B, C, F, G, P, Q, V, W, Y, Z,}. Infact the moment letter Z was first shown, and I entered the code, the learning sequence ended for having reached the count of 26 letters and switched over to the numbers.

I'm not sure if this is true (I havent personally got this far) - but if it is we need to fix this

[Ajedi32]

I've noticed this as well. The game doesn't end immediately after reaching Z, but it's still sooner than I'd expect.

I suspect that whatever system it uses to judge whether you've learned each letter isn't granular enough. It seems to assume you've "learned" a letter after you get it right a few times, and doesn't do a good job of systemically revisiting recently learned letters to make sure you haven't forgotten them, or of avoiding wasting time on letters you've already gotten right dozens of times with no mistakes.

A suspect making the game track progress statistics a bit more granularly (#38) might help with this.

[willwade]

I think this is still kind of valid

The word list heavily favors common letters (e/t/a/...) and underrepresents z/q/v/y/x.
In current words.js rough corpus counts are:

e: 196 occurrences
z: 12 occurrences
q: 12 occurrences
v: 9 occurrences
So some letters are >10x more common.
The “finished when Z first appears” is not necessarily a bug.
Completion is currently “all letters score >= threshold” (LEARNED_THRESHOLD = 2), see:

• config.js
• game-space.js (checkAllLettersLearned)
  So if Z was already at 1 and that hit made it 2, ending at that moment is expected behavior.

We could:

• alter minimum attempts per letter, and/or
• minimum recent accuracy per letter,
  so low-frequency letters can’t slip through too quickly.

Not sure