Palabrilla Extreme never tells you which individual tile succeeded. Each five-letter guess produces three row-level totals: correct letters in the correct positions, correct letters in different positions, and incorrect letters. Eight accepted attempts may sound generous, but two poorly designed guesses can consume a quarter of the game without revealing what caused their different scores.
The reliable approach is not to imagine hidden colors behind the grey tiles. It is to create pairs of guesses whose differences are small enough to interpret. Instead of asking only, “What might the answer be?”, ask, “What controlled change would resolve my most important uncertainty?”
That leads to three separate jobs: discovering which letter occurrences belong in the answer, arranging confirmed letters, and testing whether any letter is repeated. The method below tells you which job to prioritize and how to design a guess around it.
Treat the comparison—not the row—as your basic clue
A single result is a constraint, but it rarely identifies the letters responsible for it. A score of 2 / 1 / 2 says that three occurrences from the guess are included in the answer: two correctly placed and one misplaced. It does not name those three occurrences.
Track one derived number for every row:
- Included occurrences: correctly placed plus misplaced.
- Excluded occurrences: the reported incorrect total.
For 2 / 1 / 2, there are three included occurrences and two excluded ones. During early letter discovery, the sum of the first two categories is often more useful than their division. Once you start arranging known letters, the division becomes crucial.
A second row becomes especially informative when most of the first guess is preserved. If one letter is replaced and the included total rises from three to four, the swap provides focused evidence. If the same five letter occurrences are rearranged, any change in the first two totals concerns position rather than membership.
Use this central rule:
Replace letters to investigate membership; move letters to investigate position.
Late guesses may need to do both, but combining the operations too early makes the feedback harder to explain.
The PLACE method for choosing every guess
Before submitting a word, follow five steps: Parse, Link, Assess, Change, Evaluate. Together they form the PLACE method.
1. Parse the totals without assigning them to tiles
Write down the word and all three numbers in their displayed order. Do not convert a row-level result into an unsupported statement such as “A must be green.” Extreme has not identified any tile.
A compact note can look like this:
Guess 1 — 1 placed / 2 misplaced / 2 incorrect — 3 included
The final figure is your own calculation. It is not a fourth clue; it simply makes membership comparisons faster.
2. Link every new result to the complete history
A new row adds a constraint rather than replacing the previous ones. Any viable answer must reproduce the exact three-part feedback for every submitted guess.
Keep provisional groups for:
- letter occurrences likely to be present;
- letters likely to be absent;
- unresolved relationships, including possible duplicates.
Do not force a binary verdict before the evidence supports one. A comparison may show that the letter entering a guess performs better than the letter leaving it. If two letters were changed together, however, the result initially describes the two groups, not their individual members.
3. Assess the kind of uncertainty you currently have
Classify the problem before searching for another word:
- Membership uncertainty: you do not know which letters belong.
- Position uncertainty: you know much of the letter set but not its order.
- Multiplicity uncertainty: a letter is present, but it may occur more than once.
- Final compatibility: only a few candidates remain and they must be checked against every row.
This classification prevents a common strategic mismatch: rearranging an incomplete set, testing a duplicate before confirming the original letter, or covering new letters after all five occurrences are already known.
4. Change one controlled variable
Find an accepted word that makes the smallest change capable of answering your question.
For membership, preserve four letters and replace one whenever the vocabulary permits it. If you must change two letters, record the change as a block. The next score compares the outgoing pair with the incoming pair; it does not automatically evaluate all four letters separately.
For position, preserve the same multiset of letter occurrences and rearrange as little as possible. Swapping two positions is easier to interpret than scrambling all five. If a pure rearrangement is unavailable, retain as much as possible and explicitly mark any new control letter.
For multiplicity, replace a probably absent letter with a second copy of a confirmed letter while keeping everything else stable.
5. Evaluate the possible outcomes before submitting
Complete conditional statements in advance: “If the included total rises by one, then…” or “If membership stays fixed but the placed total rises by two, then…”.
If you cannot explain what the main possible results would mean, the experiment is probably changing too much. Also verify that the proposed word remains compatible with established constraints. Near the end, an interesting but incompatible probe may leave too little room to use what it discovers.
Worked example 1: isolating a letter substitution
Imagine two accepted guesses with these abstract patterns:
- guess A:
C A N T O - guess B:
M A N T O
Only the first letter changes. Suppose A scores 2 / 2 / 1, giving four included occurrences, while B scores 3 / 2 / 0, giving five.
Because A N T O stayed unchanged, the increase in membership must come from the C-to-M substitution. M supplies an included occurrence and C does not. That is the robust conclusion.
Do not immediately add “M belongs in position one.” The correctly placed total also rose, but the feedback remains aggregated. You must interpret the complete comparison without pretending that a specific point has been attached to M. Other rows or a controlled positional test must support the location.
Now suppose both guesses produce four included occurrences. The swap has not cleanly separated C from M. Depending on the established constraints and multiplicities, both might be excluded, the substitution might exchange equally useful occurrences, or one effect might be offset by another. Record the unresolved relationship and design a third comparison rather than choosing the most convenient story.
A score difference can produce a direct deduction. Equal totals are still evidence, but they frequently establish an equivalence or unresolved pair rather than proving that a particular letter is present.
Worked example 2: using a small rearrangement
Now consider two guesses containing exactly the same five occurrences:
- guess A:
A B C D E - guess B:
B A C D E
Only positions one and two are swapped. A produces 2 / 3 / 0; B produces 4 / 1 / 0.
The included total remains five, as expected, because no letter occurrence entered or left. Correct placements increased by two. Since positions three, four, and five were held fixed, both swapped letters must now be correctly placed in B. The experiment is local enough to support a positional conclusion.
If the placed total had risen only from two to three, B would have gained one net placement. You could not automatically say which of the two moved letters is now correct. One may have entered its correct position while the other remained misplaced. Another controlled swap—or a restriction inherited from an older row—would be needed.
This shows why “same letters, different order” is not automatically an excellent probe. Rearranging all five positions can improve the score while leaving many possible explanations. A two-letter swap or a small rotation produces feedback that is easier to localize.
Test repeated letters only with a reason
Repeated letters require you to count occurrences rather than merely distinct characters. Evidence that one A is included does not prove that the answer contains two As. Because multiplicity is limited, a second A in your guess contributes to the included total only if the answer can match that second occurrence.
A controlled duplicate test needs three ingredients:
- a confirmed or strongly supported letter;
- a probably absent letter that can be removed;
- four other positions stable enough to preserve the comparison.
Use this pattern:
- baseline row: one A plus a control letter X thought to be absent;
- duplicate test: a second A replaces X, with everything else retained.
If the included total rises by one, you have evidence for a second A occurrence. If it does not rise and X was genuinely absent, the extra A also fails to add an included occurrence. If X was never securely excluded, however, you have merely compared X with a second A and may still be unable to evaluate either one independently.
Do not introduce duplicates simply because repeated letters are possible or because an appealing candidate contains them. A duplicate uses a slot that could have tested a new distinct letter. Establish the likely membership set first; investigate extra copies when multiplicity becomes the actual unresolved question.
Know when to switch from letter discovery to placement
After each row, calculate the included total. If controlled substitutions can still improve that number, you are primarily in the membership phase.
When a row reaches five included occurrences, that hypothesis no longer needs new letters. Preserve its multiset and start making small positional changes. Replacing one of those occurrences with an unrelated letter would reopen a question you had already answered.
You may switch earlier if the available vocabulary and accumulated constraints leave only a narrow family of candidates. Even then, mark any provisional letter honestly. A plausible set is not the same as a proven set.
Manage the eight attempts by strategic phase rather than a rigid turn schedule:
- early guesses should provide broad distinct-letter coverage and interpretable substitutions;
- middle guesses should close membership gaps and begin controlled rearrangements;
- late guesses should be candidates satisfying the full history, unless one precise experiment is essential to distinguish them.
The principle is exploration first, localization second, and compatibility checking throughout.
Common comparison failures
Replacing all five letters to “get more information”
A completely unrelated word may cover many new letters, but the difference between its score and the previous score has too many possible causes. Use a broad reset only when the initial result was poor enough that coverage is more valuable than isolation.
Painting imaginary colors onto grey tiles
A total of two correctly placed letters does not authorize you to choose two positions. Mark a specific position only when a controlled comparison has held the alternatives stable enough to localize the effect.
Forgetting an old row after a useful breakthrough
A candidate can fit the latest comparison and still contradict the opening guess. Every row remains active until the solution is found.
Testing duplicates before membership
Repeating an uncertain letter sacrifices distinct-letter coverage. Duplicate a letter to answer a multiplicity question, not as a general gamble.
Using a probe that changes every dimension
A probe need not be your most likely answer, but it must serve the current phase. If it changes membership, order, and multiplicity simultaneously, its feedback may not resolve any of them.
Advanced decisions when a perfect pair is unavailable
Natural vocabulary will not always supply two accepted words differing in exactly one position. When that happens, compare blocks.
Suppose C and R leave a guess while M and L enter. If the included total rises by one, the incoming pair contributes one more net occurrence than the outgoing pair. That does not identify M or L individually. Your next word should cross the groups: keep M, restore R, and alter as little else as possible. The combined constraints can then separate effects that the first block comparison merged.
You can also use a known absent letter as a control. If the five answer letters are largely established but no pure rearrangement exists, inserting a confirmed nonmember creates a known membership loss. Any positional improvement can then be interpreted alongside that deliberate loss instead of being confused with an unknown new letter.
When two final candidates remain, do not select one merely because it looks more familiar. Mentally score each candidate against every previous guess. Eliminate any word that fails even one three-part total. If both survive, seek a guess that would produce different feedback for them. A surviving candidate that also distinguishes the pair is normally more useful than an external probe because it retains the chance to finish immediately.
A ten-minute practice routine
You can train controlled comparisons with completed games rather than playing many new ones:
- Take a finished game and cover the solution.
- Copy the first two rows and calculate each included total.
- Circle exactly which letters or positions changed.
- Write one conclusion that is logically secure and one tempting claim that remains unsupported.
- Design the ideal next experiment, even if you cannot immediately find an accepted word matching its pattern.
- Reveal the actual next row and check whether your reasoning mixed membership, position, or multiplicity.
- Repeat with two rows involving a duplicate letter.
The purpose is not to reconstruct a supposedly perfect solution path. It is to practice forming a precise question before choosing a word.
Before every submission, return to PLACE: parse the totals, link all constraints, assess the uncertainty, change one variable, and evaluate the possible outcomes. With that process, Palabrilla Extreme’s three anonymous numbers stop behaving like vague hints. They become a growing system of tests in which each carefully chosen comparison narrows the answer for a reason you can explain.