The difficulty curve and what each tier actually trains

If you've worked through every difficulty tier on this site at least once, you've probably noticed something the difficulty label doesn't quite capture: an extreme puzzle isn't just a hard puzzle that takes longer. It's a different kind of effort. Easy goes by in a kind of light forward-momentum. Master sits you down and asks you to hold an entire chain of pencil-marks in your head at the same time. Calling both of those things "Sudoku" obscures more than it reveals.

This isn't a piece about which tier is best for you. It's a piece about what's happening cognitively at each tier, so that the choice you make is informed by something other than ego. The short version is that the cognitive flavours change as you climb the curve, and the right tier on a given day is the one that matches the kind of attention you actually have available.

Easy — pattern recognition, mostly automatic

The easy tier rewards eye-discipline more than reasoning. The empty cells have enough constraint already given that the next move usually lives in plain sight: a row with one missing digit, a box where eight of the nine candidates have been narrowed to one cell, a column where a digit has nowhere else to land. You scan, you place, you scan again. Pencil marks are mostly unnecessary.

Cognitively, this is the most automatic of the tiers. The pattern-finding part of your visual system is doing most of the heavy lifting; the deliberate-reasoning part is taking notes and getting out of the way. That's not a knock — there's a reason easy puzzles aren't just for beginners. On a tired afternoon or a distracted morning, an easy is the tier that meets you where you are.

Medium — candidate-tracking enters the picture

Medium puzzles introduce the moments where scanning runs out and you have to start writing things down. Not on every cell, and not yet for chains of reasoning — just for the handful of cells where two or three candidates need to be tracked while you work elsewhere on the grid. Your first medium puzzle goes deeper on what changes between easy and medium; the through-line is that medium begins the small handover from automatic pattern recognition to deliberate working memory.

The cognitive load is still modest. You're not yet building long deductive chains; you're keeping a small handful of options open in two or three places, and using the resolution of one to unlock the next. This is the tier where most casual solvers settle in for life, and there's nothing wrong with that.

Hard and expert — small chains, longer eye-time

Hard puzzles are where pencil-marking stops being a stylistic choice. There are now patches of the grid where a move requires you to notice a configuration across two or three cells that aren't obvious one-at-a-time. The X-wing, the pointing pair, the locked-candidate move — these are the patterns hard and expert tiers tend to demand, and we've written separately about why the X-wing keeps tripping people up and pointing pairs and the snake.

Cognitively, this is where working memory load gets real. You're holding two or three pieces of information in your head and turning them sideways to see the move. Sustained attention matters more than at the lower tiers — distraction is more expensive, because losing track means re-laying-down the chain. Most readers find this is the tier where puzzling shifts from "background activity" to "needs the room to be a bit quiet."

Master and extreme — the chain tier

At master and extreme, the puzzle is rarely going to surrender to a single observation. The moves are chains: a candidate elimination two cells over only matters because of a deduction three cells before that, which only worked because of the fork structure you noticed at the start of the chain. The half-finished-grid problem lives here — you're 60% of the way through, none of the obvious moves are working, and the next move requires you to plan three steps ahead, hold the plan, and execute it without losing the thread.

This is the tier where Sudoku starts looking less like pattern matching and more like a small piece of long-form reasoning. It rewards a particular kind of patient, slow attention that most of life doesn't ask for, which is part of why the people who love it really love it. It's also the tier most exposed to having a bad day. If you sit down distracted, extreme will tell you so within five minutes.

What this means honestly

It would be tidy if the tiers also formed a cognitive ladder — easy trains you for medium, medium trains you for hard, all the way up to a notional version of you that solves extreme in eight minutes flat. The picture is messier than that. Practising at any tier makes you better at that tier and modestly better at the tiers immediately around it. The transfer doesn't extend cleanly to unrelated cognitive tasks — that's the narrow-transfer point we've written about elsewhere — and it doesn't extend cleanly even to other puzzle types you haven't practised. People who train on cognitive tasks get better at those tasks and not at much else¹.

The most useful framing is that the tiers are different shapes of attention. Easy trains scan fluency. Medium trains light working memory under modest interference. Hard trains pattern recognition across multiple cells under sustained attention. Master and extreme train chained reasoning with planning depth. None of these are general cognitive abilities you should expect to transfer to your job, your conversations, or your taxes. They're the local cognitive flavours that Sudoku happens to exercise.

Picking a tier on a given day

The pragmatic move is to think of the tiers as a menu, not a ladder. On a tired evening, an easy is genuinely the right pick — there's no virtue in choosing a tier you're not going to enjoy. On a quiet Saturday morning with coffee and an hour, an extreme that takes you forty-five minutes is a small, well-shaped use of the morning. On a focused Wednesday lunch break, medium or hard is the sweet spot for most people.

If you've fallen into the habit of always picking the same tier, try one a notch up and one a notch down across a couple of weeks. You'll probably find that the variety is the most cognitively interesting thing of all — not because it's training a wider general ability, but because it keeps the puzzle from settling into a single rut, and a puzzle in a rut stops being the small daily thing it's good at being.