Principle 1: Answer the crowd, not the question
Every prompt is secretly two questions. The visible one — "Name a European capital" — is a decoy. The real one is: "Which European capital will the fewest players name?" Your first instinct answers the decoy. Trained players spend their first five seconds on the real question: picture the crowd. Paris, London, Rome — the Plankton layer names itself. Only then do they start answering. This inversion — from what do I know? to what does the room know? — is the whole of backwards thinking, and it is a skill, which means it compounds.
Principle 2: The Too Clever trap is the game's centre of gravity
Between the obvious answer and the rare one sits the most dangerous tier in Noillirk: Too Clever, 15 points. It is the answer that feels like a deep cut — the "underrated" classic everyone calls underrated, the quiz-night capital, the foodie's first fermented food. It feels like strategy and scores like a rounding error. The tell is emotional: if an answer arrives fast and makes you feel smart, it is probably Too Clever. Genuine depth arrives slower and feels slightly uncomfortable — you're less sure it's even valid. That discomfort is the sensation of leaving the crowd. Learn to trust it.
Principle 3: Stock second-page knowledge, on purpose
Depth is inventory. The divers who bank Rares daily are not smarter in the moment; they arrived carrying stock: the capitals of small states (Vaduz, Andorra la Vella, San Marino), the losing finalists of famous finals, the pre-Roman empires (Akkad, Elam), the world instruments (erhu, duduk, theremin), the lanthanide elements, the regional dish one province deeper than the famous one. Build the inventory deliberately:
- Read yesterday's bank. The published answers are a daily lesson in where the crowd stops — study which answers were banked Rare and Deep Cut in categories you play.
- Specialise in two categories. Depth in geography and food will carry a dive while your movies catch up. The packs exist for exactly this drilling.
- Collect runners-up. Second voyages, second inventions, second capitals, losing nominees — the mirror loves a runner-up, because the crowd only remembers winners.
- Learn three per category. Three reliable deep answers per category transforms your floor: a dive that can't fall below Schooler-with-Rares is a dive that lives in Mirror Deep.
Principle 4: Spend the clock like money
Twenty-five seconds is a budget, and most players overspend it on prompt one. A working split: 5 seconds to name the crowd's answers (out loud in your head: "everyone says X, the clever ones say Y"), 12 seconds to search your own second page, 5 seconds to commit and type, and a 3-second reserve for the "did you mean" lifeline or a spelling fix. If the search comes up empty at second 15, take the Schooler you have rather than gambling the whole prompt on a guess that isn't banked — 30 points banked beats a heroic zero, and the dive's total is arithmetic, not honour. Skip early on hopeless prompts; a fast zero buys calm for the six prompts you can actually win.
Principle 5: Play the week, not the dive
One dive is variance; a week is signal. Track which tier you land most, per category, across a week of dailies — the results screen's breakdown makes this easy. Patterns emerge fast: most players discover a personal Plankton category (often the one they know best — expertise makes first instincts stronger, not rarer) and a surprise strength. Then train the weakness in Unlimited for ten minutes, where mistakes cost nothing, and bring the new depth back to the daily where it counts. Warm up with two Unlimited dives before the daily if the daily matters to you today: the backwards reflex takes a prompt or two to wake up, and you'd rather it woke up off the record.
Principle 6: Every category hides its depth in a different place
Principles travel; inventories don't. Each of the eight categories has its own crowd, and its own blind spot:
- Geography — the crowd stops at the famous map. Depth lives in landlocked states, second cities, small capitals (Vaduz, San Marino) and any country whose name you must picture before you can spell.
- Movies — the crowd names winners and blockbusters. Depth lives in losing nominees, directors' early films, and the co-star the poster left off.
- Sports — the crowd names champions of the big four leagues. Depth lives in forgotten finalists, niche events, and the second-most-famous player on a famous team.
- Food — the crowd names the national dish. Depth lives one province deeper: the regional variant, the baker's cut of pasta (orecchiette, bucatini), the fermented thing a grandmother makes and the internet barely lists.
- Science — the crowd names planets and Einstein. Depth lives in the lanthanides, the second-discovered, the units and constants that never got a documentary.
- Animals — the crowd names the zoo's front row. Depth lives below and behind: siphonophores, axolotls, the creature named in the zone labels on your way down.
- History — the crowd names empires at their peak. Depth lives in the pre-famous and the runners-up: Akkad before Babylon, the voyage before the famous voyage.
- Music — the crowd names the single. Depth lives in the album cut — the deep cut the tier is named for — plus world instruments (erhu, duduk, theremin) the charts never meet.
Drill the two categories where your own crowd — you — stops earliest. The packs are sorted for exactly this; the tiers page shows what each shelf is worth when you get there.
The 25-second routine: a drill you can actually practise
Strategy becomes points only when it fits inside 25 seconds. Rehearse this exact routine in Unlimited until it is reflex:
- Seconds 0–3 — name the crowd. Silently list the first answer ("everyone says Zambia"). This is not your answer; it is your map of where the surface is.
- Seconds 3–6 — name the trap. Say the clever one too ("the quiz crowd says Czechia"). Two shelves mapped, neither of them yours.
- Seconds 6–16 — search the second page. Run your stocked inventory for the category: the small capital, the losing finalist, the regional dish. Take the rarest bankable thing you are sure is valid.
- Seconds 16–21 — commit and type. No more browsing your own brain. Type it, watch for the spelling lifeline, confirm if offered.
- Seconds 21–25 — reserve. If nothing bankable surfaced by second 16, this reserve is your escape: take the Schooler you passed on the way down, or skip cleanly. A planned 30 beats a panicked 0.
Run the routine on three Unlimited dives a day for a week. By day four the first two steps stop costing time — the crowd and the trap start arriving as one picture — and the search window quietly grows from ten seconds to fifteen. That growth, not any single gem, is what a rising weekly average is made of.
The weekly practice loop (how players actually get better)
"How do I get better at this?" has a boring, reliable answer: a loop, run weekly. Study one archive day — read yesterday's bank and mark the answers you would have given; the gap between your marks and the gem shelf is your syllabus. Drill two weak packs — twenty minutes across the week in the categories where your marks were shallowest. Run the 25-second routine in Unlimited until the crowd-scan is reflex, then protect the streak: one honest daily, every day, reviewed against its own published answers afterwards. Players who run this loop don't feel themselves improve dive by dive — they notice it a fortnight later, in the rank band that quietly changed.
The mirror mindset, in one paragraph
The crowd is not stupid; the crowd is fast. Plankton answers are good answers delivered at the speed of recognition. Noillirk simply pays for the second thought — the one that arrives after you've asked what the first thought was worth. Slow down one beat, reverse the question, distrust the clever glow, spend your seconds like money and your memory like stock. Do that for a month of midnights and the trench stops being a rumour.
Strategy is cheap. Go prove it.
One dive per day, seven prompts, 25 seconds each. The rarer your answer, the deeper you sink — 700 points is a perfect dive.