We analyzed 4,356 daily puzzle solves: Saturday's mystery takes 6.5 times longer than Sunday's

Last updated: October 22, 2026

Between 8 June and 21 October 2026, 902 players on bossfeed.fun posted 4,356 solves of our free daily puzzles to public leaderboards. We went through every one.

The short version

• Saturday's logic puzzle takes 6.5 times longer than Sunday's (7m 38s against 1m 10s), and about twice as long as a Wednesday or a Friday. That is the easy-to-hard ramp I built into the week, and the clock agrees with it.
• Not every day follows its name. "Fiendish Friday" (3m 21s) is the quickest weekday, and Thursday (7m 09s) almost matches Saturday.
• Regulars with 10 or more solves got about 28% faster by their eighth to tenth puzzle, with no drop in mistakes. Time of day barely matters.

Saturday takes 6.5 times longer than Sunday

Morning Mysteries is our daily logic-grid puzzle: read the clues, fill in the grid, name the culprit. I set its difficulty by weekday on purpose, from "Cozy Sunday" to "Grand Saturday".

The gap holds up. Sunday's plausible range for the median is 1m 07s to 1m 16s and Saturday's is 6m 09s to 8m 41s, so even the closest ends are nearly five times apart. Early and late in the season show the same shape.

Two grids are slow, whatever the day

"Gentle Monday" takes longer than Wednesday or Friday because Monday's featured puzzle uses two linked grids. The optional extra challenges show the same thing:

Why some days are harder (our best guesses)

How I tuned it. Each weekday gets a difficulty tier. Going up the week:

• Sunday is the smallest case: 5 guests, no liar (you just complete the grid), and no minimum length of reasoning.
• Monday and Tuesday seat 6 guests and still have no liar. Monday adds a second linked grid, and Tuesday is the one-true, one-false format.
• From Wednesday there is a liar: one guest lies and you must find the culprit. Thursday brings back the second grid and requires a chain of at least 7 reasoning steps, where each step needs an earlier one.
• Friday and Saturday require chains of at least 6 and 7 steps, let a suspect speak twice, add clues about traits and ranks, and make every clue necessary to the answer. Saturday seats 7 guests on a map instead of 6 on a grid.

In September I also switched off "either this or that" clues on every day. I wanted difficulty to come from how long the chain of deductions is, not from a coin flip between two options.

Now the guesses. These are guesses. The data shows what happened, not why.

• Why Saturday runs so long: it stacks the most hard things at once. A seventh guest, a map, a liar, a long required chain and no spare clues. Together they add minutes. Mistakes line up too: 57% of Saturday solves have no hints or wrong answers, against 97% on Sunday. Saturday is our smallest sample (106 solves).
• Why Sunday is so quick: a smaller table and no liar. But that cannot be the whole story, because Monday and Tuesday also have no liar and take four to five minutes. Or Sunday may draw more regulars. This data cannot tell those apart.
• Why "Fiendish Friday" is not the hardest: this one surprised me, because Friday is the second-highest tier I set. Friday's required chain (6 steps) is actually shorter than Thursday's (7), and Thursday has two grids. Maybe a long chain feels like a straight line once you see the first move, so it is hard to build but quick to solve. Or Friday's solvers may skew toward regulars. My own difficulty meter once rated four Fridays in a row Easy, Easy, Easy and Hard, so I now treat the tier as a plan, not a measurement. Friday is the day I most want to look at.

What I would like to test next. Line up each day's solve time against the real length of its reasoning chain, which I already measure when a puzzle is built. Compare new players with regulars for each weekday. And record the table size and puzzle type with each solve, so next time we do not have to guess.

Players get about 28% faster

To compare fairly, we measured each solve against that puzzle's own median, then followed the 35 players with at least 10 comparable solves. Their eighth to tenth solves ran a median of 28% faster than their first three, and 26 of the 35 got faster (plausible range 14% to 34%).

The gain is speed, not accuracy: 73% of their first three solves had no hints or wrong answers, and 74% of their eighth to tenth. Across everyone the gap looks bigger, but people who keep playing tend to be quick to begin with, so we trust the same-player comparison more.

When people solve barely matters

Using Pacific time, the clock our daily puzzles roll over on, solves were spread across the day, with about a fifth each from midnight to 5am, 5 to 9am and noon to 5pm. Adjusted for puzzle difficulty, every slot landed within about 2% of the typical time, except 9pm to midnight, about 12% quicker. That slot has only 165 solves and 47% come from the most frequent players, so treat it as a hint, not a finding.

Who plays, and how often

Of the 902 players, 719 (80%) appeared on only one day, 183 came back on another day, 39 played on seven or more days and 12 on thirty or more. Those 12 account for 42% of all solves, so every number above leans toward regulars. Growth is steady, from 138 solves in June to 1,581 in October so far.

Every daily game at a glance

Only Morning Mysteries has enough solves for a weekday breakdown. Dani's Diabolically Difficult Deduction is printed on paper, so its times are typed in by players and rest on trust.

How we measured

• What we counted: every completed solve on the public daily leaderboards, 8 June to 21 October 2026. 4,356 solves, 10 boards, 902 distinct anonymous player IDs.
• Players: one ID is one browser profile, so someone on two devices counts twice.
• Time: the time on the clock before penalty seconds. Medians, not averages, so a few very slow or very fast solves do not distort them. Days are Pacific days.
• Ranges: "plausible range" means a 95% bootstrap interval, which shows how much a number could move with a different sample of players.
• Limits: only players who post a time to the leaderboard are included, a small group of regulars supplies much of the data, and small boards are too thin to rank day by day. We report no names, and no individual players.

The leaderboards are public: play the Morning Mysteries and compare your Saturday with your Sunday. Journalists and bloggers are welcome to quote any figure here with a link back to this page.

Last updated: 22 October 2026. Source: Final Boss Games daily puzzle leaderboards.