Strategy

Most people's first few dives at Krillion land in the Shallows, and almost none of that is a knowledge problem. It is a reflex problem. Seven habits separate a 300 m dive from a 1,700 m one, and every single one of them is about what you do in the first four seconds of a round.

1. Never type your first answer

This is the whole game and it is worth stating bluntly. The answer that arrives unbidden is, by construction, in the bottom 2% of the bank — the three to five entries filed as Plankton. It pays 10.

Your second thought is almost never in that group. On name a cheese, four of 169 answers are Plankton; the fifth cheese you can name is already worth 60 or 85. You do not have to be an expert to clear the floor. You have to wait about two seconds.

Answering with your second thought in all seven rounds is worth roughly 440 points — 1,760 m, the Thermocline band. That single habit is most of the distance between a bad dive and a good one.

2. Guess loosely — a miss costs time, not the round

A wrong answer takes a slice off the clock and leaves the round open. There is no attempt limit and no penalty tier. So a word you are only 60% sure is a real spice is clearly worth typing: you are risking a second or two against 85 points.

The corollary is that people vastly under-guess. If a half-remembered word is sitting there, send it. The matcher is forgiving about spelling, accents and plurals, so “asafetida” and “asafoetida” both land.

3. Beware the pick that feels clever

Each prompt can carry up to two traps, and they pay 15 — less than half of a plain Schooler answer. A trap is not an unusual answer. It is the answer with a reputation for being unusual, which is exactly why it arrives fast.

The tell is the feeling. If a word shows up with a small glow of self-congratulation attached, that glow is the reflex firing, and the reflex is what the tier is priced against. Something you had to genuinely dig for will not feel like that.

4. Move along an axis, not down a list

When a prompt stalls, people try to remember harder, which does not work. Changing the axis does. For name a dog breed: shift country (Shiba Inu, Xoloitzcuintli), shift size, shift job — herding, sledding, ratting. For name a gemstone: shift colour, then shift to the ones that are minerals first and jewellery second.

A prompt's bank is broad — most run past 150 accepted answers, and the banks total more than 21,000 entries across 140 prompts. Almost anything you can defend is in there. The constraint is retrieval, not validity.

5. Spend the clock, but not on precision

25 seconds is enough for three or four attempts, not for deliberation. A good rhythm is roughly: two seconds to let the reflex answer pass, five to find a real candidate, and the rest to type and, if needed, try again.

Do not spend the clock choosing between two good answers. They are almost certainly in the same tier — the Deep Cut band holds four fifths of the bank — so the choice is worth nothing and the seconds are worth a great deal.

6. Concede only when the well is actually dry

You cannot concede in the first 15 seconds, which is the game telling you something: the answers that pay arrive late. When you genuinely have nothing, conceding buys you a settled head for the next round rather than a blank one. A timeout and a concession score the same, so the only thing at stake is your composure.

7. Do not chase the gem

Every prompt has one designated answer worth 100, and it is never published. Hunting for it is a losing play: it is one entry out of 150-odd, and the search costs you the 85 you could have banked. Gems are something that happens to you, not something you aim at.

This is also why the Trench band is rare rather than difficult. It sits above 650 points, which no combination of Deep Cuts alone can reach — seven perfect Deep Cuts is 595. You need luck to enter the Trench, and no amount of skill substitutes.

What a realistic ceiling looks like

A dive over 1,700 m is a good day. Treat the bottom two rows as scenery.