Information Theory and Wordle
In 2022, mathematician Grant Sanderson (3Blue1Brown) published a viral analysis of Wordle using information theory to rank every possible starting word. The core concept was Shannon entropy: how much information (in bits) does each guess provide on average, across all possible Wordle answers?
A guess that eliminates exactly half the possible answers every time provides 1 bit of information. A guess that eliminates three-quarters provides 2 bits. The maximum possible information per guess is log₂(N) bits, where N is the number of remaining candidates.
The best starting word isn't the one that scores highest on any specific puzzle — it's the one that, averaged across all ~2,300 Wordle answers, leaves the fewest remaining candidates after the first guess.
Why Vowel Coverage Matters
The Wordle answer list is rich in common vowels. E appears in about 46% of answers. A appears in about 38%. O appears in about 30%. I appears in about 26%. U appears in about 17%.
A starting word that tests four of the five vowels is statistically likely to get at least one hit (green or yellow) on almost every puzzle. That initial hit anchors your strategy immediately.
But vowel-only words (like AEIOU — not a Wordle word anyway) leave you with no consonant information. The best openers balance vowel coverage with high-frequency consonants: R (appears in ~38% of answers), T (33%), L (28%), S (26%), N (25%), and C (23%).
The 15 Best Wordle Starting Words
1. CRANE — Information score: ~5.83 bits. Covers C, R, A, N, E — five of the most common Wordle letters. Elegant and well-known.
2. SLATE — Information score: ~5.85 bits. S, L, A, T, E. The S is particularly valuable because Wordle answers ending in S are common. Often ranked #1 in pure entropy calculations.
3. TRACE — Information score: ~5.81 bits. T, R, A, C, E. Nearly identical letter set to CRANE in different arrangement.
4. CRATE — Information score: ~5.83 bits. C, R, A, T, E. Another CRANE family member. The T in position 4 has high positional frequency.
5. STARE — Information score: ~5.86 bits. S, T, A, R, E. Often cited as the mathematically optimal opener. Covers the highest-frequency consonants alongside two common vowels.
6. RAISE — Information score: ~5.79 bits. R, A, I, S, E. Strong vowel coverage (A, I, E) plus two top-10 consonants.
7. AROSE — Information score: ~5.72 bits. A, R, O, S, E. Three vowels including the less-common O.
8. IRATE — Information score: ~5.77 bits. I, R, A, T, E. Four of the top five vowels partially covered plus R and T.
9. STALE — Information score: ~5.84 bits. S, T, A, L, E. Similar to SLATE, with the L in a different position.
10. SNARE — Information score: ~5.80 bits. S, N, A, R, E. The N is underrated — it appears in about 25% of Wordle answers.
11. STOIC — Information score: ~5.70 bits. Tests S, T, O, I, C — two vowels (O, I) plus three strong consonants. Less common but useful for players tired of the CRANE/SLATE family.
12. AUDIO — Information score: ~5.61 bits. A, U, D, I, O — four distinct vowels. Weaker on consonants but excellent for narrowing vowel positions in hard-mode puzzles.
13. ADIEU — Information score: ~5.59 bits. A, D, I, E, U — four vowels. The U is the rarest vowel, so testing it early is valuable. Classic choice.
14. OUIJA — Information score: ~5.41 bits. Tests O, U, I, J, A — four vowels and the rare J. Lower overall entropy but useful against low-frequency-consonant puzzles.
15. QUIRK — Information score: ~5.22 bits. Q, U, I, R, K — tests the Q-U pair and some useful consonants. A novelty pick but surprisingly effective.
The Debate: Pure Math vs Human Intuition
The mathematically optimal starter (STARE or SLATE by most analyses) is not always the most practical choice for human players.
Argument for pure math: If you're trying to minimize your average solve count over thousands of puzzles, use the highest-entropy starter every day without deviation. STARE or SLATE. Full stop.
Argument for intuition: Wordle is a daily 3-minute activity, not a competitive optimization problem. If ADIEU is your starter and you enjoy saying it every morning, the marginal information loss (roughly 0.2-0.3 bits) is meaningless in practice. Human enjoyment matters.
Hard Mode consideration: In Hard Mode, every revealed letter must be used in subsequent guesses. This constraint changes optimal strategy — openers that spread information across many positions (like STARE) remain strong, but some word families become less useful when yellows must be incorporated.
How to Choose Your Starter
If you care about optimization: use STARE or SLATE every day.
If you want variety without sacrificing much: rotate between CRANE, TRACE, SLATE, and SNARE. The entropy difference between these is negligible.
If you're a vowel-hunter who wants to identify vowel positions first: AUDIO or ADIEU then cover consonants on guess 2.
Use our Wordle Solver to track remaining candidates after each guess and see exactly how your starting word choice affects the solution space.