⚖️ One weighing = one comparison
A sorting algorithm never sees the numbers it sorts. It can only ask "is this one less than that one?" Your balance works the same way. The stones look alike; only weighing reveals the order.
In memory of Tim Peters · 1952–2026
Can you out-sort Timsort?
Every time your phone, your browser or a Python script sorts a list, it is probably running the algorithm Tim Peters wrote in 2002. Here's your chance to beat it. Weigh zen stones two at a time on a balance and find their order in fewer weighings than Python's list.sort() needs on exactly the same stones.
- 16 gardens
- 12 data shapes
- 1 very good algorithm
The gardens
Stars: 0 / 48 · ★ sorted · ★★ tied Tim · ★★★ beat Tim (or hit the unbeatable minimum)
Loading gardens…
Free play: any size, any shape
Level
Shape:
0your weighings
?Python's sort
The row (positions 1 → ; under each stone: its possible rank)
Pairs known 0 / 0
Lineup lightest → heaviest
Weighing log
Keyboard: press a stone's letter (A, B, C…) to put it on the scale, Esc to clear, ? for a hint.
Sorted!
0your weighings
vs
0Python's
list.sort()What Tim's algorithm knows
Timbot replay
How it works
🧠 Free deductions
If A is lighter than B and B is lighter than C, you never need to weigh A against C. The game tracks everything you can deduce. Each stone shows the ranks it could still have, and it settles once its place is certain.
🐍 The opponent is real
"Tim" is the exact comparison sequence CPython's classic list.sort() runs on your stones. For short lists that's a run check (count_run) plus binary insertion, as Tim Peters wrote it. The Lab shows the full algorithm with galloping merges.
📜 Read the story
Timsort is fast because real data has structure. Each garden tells you something about its stones. Python doesn't get to read the story, so you can beat it by using what it can't know.