The Loop  ·  Issue 024

The Loop

A field journal of the AI frontier — for engineers who ship.

  Lab bench

Experiment №015
filed May 18, 2026

game

Filed under

  • #game
  • #reflex
  • #physics

Echo

You have one cursor. The game has two. The second runs your moves 1.5 seconds late.

  Primer

Skip if you already know the theory; the interactive is right below.

You have one cursor. The game has two. The second is a translucent copy of yours, replaying your motion 1.5 seconds behind. Catch yellow coins with the live cursor for one point. Catch purple coins with both cursors at the same instant for five. Avoid red hazards — they kill on either cursor.

Three lives, spawn rates ramp with time. The longer you play, the more crowded the field gets and the less margin you have for darting moves your future-shadow can't afford.

▶  Try it

Score

0

Echoes

0 ×5

Lives

3 / 3

Best

0

runs · 0

Your cursor has a shadow that replays your motion 1.5 seconds behind. Touch yellow coins with the live cursor for one point. Touch purple coins with both cursors at once for five. Avoid red dots — either cursor hitting one costs a life. Three lives, density ramps with time.

saved locally · sign in to sync

▤  Leaderboard · top 25

Loading…

  Notes from the bench

What to watch for, why it matters, and the one thing that usually surprises people.

The twist

Your past is in the playfield. Every move you make replays itself, verbatim, 1.5 seconds later. Most arcade games let you treat movement as free — you decide, you dart, you forget about it. Echo makes every dart a debt. The shadow will run that same path in a second and a half, no matter what you do, and the hazards you swept past will still be there.

What the two scoring tracks are for

Yellow coins exist so there's always something obvious to do. Purple coins are the reason I built this. To collect one, you need your live cursor on it and your shadow on it, simultaneously. Which means you had to stand on that exact spot 1.5 seconds earlier — knowing then that you'd want to return now. That's a planning loop tucked inside a reflex game, and it's where the run-to-run skill gap lives. Best round, best echo count both get tracked.

Hazards do the inverse work: they punish darting through space carelessly. A red dot you brushed past at second 3 is sitting in wait for your shadow at second 4.5. Some of my deaths in playtest were entirely self-inflicted that way — me dying to a hazard I'd already avoided.

What I'm not sure about

Whether 1.5 seconds is the right delay. At one second you can almost react to your shadow live, which kills the planning layer. At two, you've forgotten the path entirely and it feels like a separate agent. 1.5 was the gap where I started thinking ahead instead of reacting backward. Could be tuned per-player; for now it's fixed.

In a line

Canvas reflex game with a single mouse input and a delayed shadow cursor that replays your motion. Yellow coins score on the live cursor; purple coins score only when both cursors overlap them at once — which means standing on a future spot 1.5s in advance. Three lives, density ramps with run length.

Other experiments

15
  1. Exp 001

    How a sentence becomes tokens

  2. Exp 002

    Temperature and top-p, visibly

  3. Exp 003

    What does this prompt actually cost?

  4. Exp 004

    Tokens per second

  5. Exp 005

    How far should the model think?

  6. Exp 006

    Neural language vs a Markov chain

  7. Exp 007

    What each token looks at

  8. Exp 008

    Words in space

  9. Exp 009

    The injection arena

  10. Exp 010

    AI or human?

  11. Exp 011

    Context Tetris

  12. Exp 012

    Magnet flip

  13. Exp 013

    Ricochet

  14. Exp 014

    Wick

  15. Exp 016

    Orbit