What a Chain streak is, and how to keep one

A streak on this site is a small honest number: how many days in a row you have solved the daily puzzle. It is worth knowing exactly how it is counted, because most of the advice about streaks assumes rules that are not ours, and because the number measures less than people assume. This page covers the mechanics first, then how to build the habit, then the part nobody writes: what to do when the streak stops being the good bit.

What the streak actually counts here

It goes up when you solve today's board on the front page — the chain ends on the target, the result card opens, the counter moves. That is the only thing that increments it.

The specifics, from the code that does it:

• Solving is all that counts. Tries and time are recorded separately and shown on your stats page; they have no effect on the streak. A day solved on the seventh try counts exactly as much as one solved on the first.
• A day counts once. Solving is idempotent for a given date; there is nothing to farm by reloading.
• The run is consecutive days, and the display is generous by exactly one day. Your streak shows the recorded run if your last solved day is today or yesterday, and 0 otherwise. So on a day you have not played yet, yesterday's run is still on screen — it has not been lost, it is waiting. Miss two days and it reads 0.
• Your best never goes down. Once you have had a run of 19, that 19 stays as your best for good, whatever happens afterwards.
• The day is decided on the server, in your timezone. The date key that the streak is written against is rendered on the server for your timezone, not read from your device clock. Moving your clock forward does not manufacture a day.
• Archive plays never count. A board opened from the archive is the same puzzle with nothing attached: no streak, no stats, no recorded result. Solving five old days does not repair a broken run, and it is not meant to.
• The records are capped. The list of days marked solved keeps the most recent 60; the per-day history that the stats page reads keeps up to 365 entries. Older days fall off the end.

Where it is stored, and the ways you can lose it

Your streak, your history and your board state live in this browser's localStorage (see MDN's reference for what that is). Nothing about your play leaves the browser except an anonymous result row for the day's aggregate. That is a deliberate privacy choice, and it has a cost you should know about: a streak kept only in a browser is a fragile thing.

Ways people lose one, in roughly the order they happen: clearing site data or using a "clean up browsing data" tool; playing in a private window, where the storage is discarded at the end of the session; switching browsers; getting a new phone; a browser evicting storage under pressure. None of these are recoverable from our side, because we do not have a copy.

If you want the streak to survive that, an account is the only mechanism. You sign in with a link sent to your email address, or with Google; the account then holds your email plus, for each day you played, the date, how many tries it took and how long you took. Nothing else — no board states, no browsing.

The merge is worth one sentence because it determines what happens when you sign in on a second device: the two sides are unioned day by day, your best is the larger of the two, and the current run is rebuilt by replaying every solved day in order. That means a run assembled by syncing two devices is identical to the run you would have had if you had played all of it in one place. Signing in cannot lose a solved day, and it cannot invent one. Deleting the account removes the server copy; the browser copy stays where it is.

Anchor the puzzle to something you already do

The most useful finding we know of on everyday habit formation is Lally and colleagues, 2010, in the European Journal of Social Psychology. Ninety-six volunteers each chose an eating, drinking or activity behaviour to carry out daily in the same context — the example in the paper is "after breakfast" — and reported on it for twelve weeks. Automaticity rose along an asymptotic curve: fast at first, flattening later.

The practical part is the "same context" bit. Attaching the behaviour to a cue that already reliably happens does more work than deciding to be disciplined about it. So rather than "play the puzzle every day", pick the anchor: with the first coffee, on the train, after you close the laptop, while the kettle boils. The puzzle takes a minute or two on Monday and up to about ten on a hard Saturday, so it fits into most gaps that already exist in your day.

One anchor to avoid: last thing at night. The day rolls over at midnight in your timezone, so a late anchor gives you the smallest possible margin and turns a hard Friday into a race. Almost all the streak-related unpleasantness we can think of starts there.

The "21 days" number is not a real number

If you have heard that a habit takes 21 days to form, treat it as folklore. The Lally study measured the time to reach 95% of each person's automaticity ceiling and reported a range of 18 to 254 days across participants. That is more than a factor of ten, on simple behaviours, in a study designed to measure exactly this. Nothing in it supports a single figure you can put in your calendar.

The useful reading is not "it takes N days" but "the spread is enormous, so do not conclude anything from how it feels in week two."

A missed day is not a reason to quit

This is the finding we would most like people to take away. From the same paper's own summary of its results: missing one opportunity to perform the behaviour did not materially affect the habit formation process.

That is worth holding onto, because the streak mechanic pushes hard in the opposite direction. A counter resetting to 0 feels like losing the thing you built, and the honest read is that you did not: you lost a number. What you actually built — the anchor, the reflex of opening the site with the coffee — is not stored in the counter. On our side, too, less is lost than it looks: your best is untouched, every day you ever played is still counted on your stats page, and the day you missed is still sitting in the archive if you want to see what it was.

Start again at 1 tomorrow. The second run is easier than the first, because the anchor is already there.

What a streak measures, and what it does not

It measures one thing: consecutive days on which you turned up and solved the puzzle. It is worth being precise about everything it does not measure.

• Not skill. An easy three-move Monday and a hard five-move Saturday each add exactly 1. A player with a 40-day streak of one-try easy days and a player who grinds through every Friday have the same number.
• Not improvement. If you want a number that tracks getting better, use the tries distribution and median time on your stats page, not the streak.
• Not effort. A day where the answer fell out in twenty seconds counts the same as a day that took eleven minutes and five tries.
• Partly, just an uninterrupted life. Streaks end because of flights across the date line, hospital weeks, newborns, and days that were simply about something else. A streak is as much a record of nothing having happened to you as it is a record of anything you did.

Knowing that makes the number easier to hold lightly, which is the point.

When the streak stops being fun

There is a version of this that goes wrong, and it has recognisable signs. You are solving at 23:50 to save the number. You look up a solution rather than lose it. Landing on the target produces relief rather than pleasure. You stop opening the site at all because you already broke the run and the counter reads 0, which is the exact opposite of what a streak is supposed to do.

If any of that is true, the streak has stopped being a way to enjoy a puzzle and become a thing you owe. Some things that work:

• Change which number you look at. Your stats page shows days played and days solved as well as the run. Those two only ever go up — they cannot be broken by a holiday — and they are a better description of your relationship with the puzzle than a fragile consecutive count.
• Break it on purpose. Skip a Wednesday you could easily have solved. It is a surprisingly effective way to find out that nothing happens.
• Play from the archive for a week. Nothing there is recorded, so there is no run to protect and no result to send. Just the board.
• Stop. Nothing here is owed to anyone. The puzzle will still be there on a Monday, which is the easiest day of the week and a good one to come back on.

If you do want to keep a long one

A few practical things, all learned by losing streaks ourselves:

• Sign in on your account before you need it, not after your browser eats the storage.
• Play the hard days early. Friday and Saturday are five moves, and five moves at ten to midnight is how good runs die.
• Read the strategy guide properly once. Most long streaks are less about discipline than about hard days costing ten minutes instead of forty — last digits and working backwards from the target do more for a streak than any amount of resolve.
• Do not treat the clock as the score. The timer runs across tries and keeps counting if you close the tab, so a "slow" day is often just a day you walked away from. Tries are the more honest number, and neither one touches the streak.

If you are new, how to play has the rules and four worked puzzles — and if you want the argument for why the puzzle is daily at all, why one puzzle a day is the companion to this page.