🏁 Relativistic Compute Race β›°οΈπŸ“‘

Three programs start at the same moment β€” Cincinnati, Mt. Everest, and a GEO satellite. Which one delivers its answer back to Cincinnati first?

LOSE
GEO ONLY
BOTH WIN
πŸ“‘ GEO
7 yr
⛰️ Everest
1,736 yr
both lose GEO only both win
1,000 years
Run time (N):
β€”
⛰️ Mt. Everest
Time gain (NΒ·Ξ΅)
β€”
Signal back
βˆ’40.0 ms
Net margin
β€”
πŸ“‘ GEO satellite
Time gain (NΒ·Ξ΅)
β€”
Signal back
βˆ’120.0 ms
Net margin
β€”
Press Run the race to watch the answers race back to Cincinnati.
The math: Each location's clock runs at a slightly different rate vs. Cincinnati. Over a calculation of length N, each racer finishes NΒ·Ξ΅ seconds earlier (from Cincinnati's frame), then loses time sending the answer back. The race animation exaggerates each lead for visibility; the numbers above are real. Each racer's signal duration is scaled so it exactly reaches Cincinnati's finish line at that racer's own break-even.
r
Assume a spherical cow.

πŸ„ The fine print: simplifying assumptions

Like every back-of-envelope physics problem, this one assumes a few cows are spherical. Things this viz pretends don't exist:

  • Vacuum signal paths. The 40 ms Everest figure is the theoretical chord through Earth β€” impossible with photons. Real fiber is ~60 ms; via satellite relay, longer.
  • Idealized clocks. Each location is one perfect clock at one elevation. Real atomic clocks drift, vibrate, and care about temperature far more than they care about relativity.
  • Identical compute. Same CPU, same workload, no thermal throttling, no GC pauses, no cosmic-ray bit flips.
  • No protocol overhead. Tcomm is bare light-time. The real world has TCP handshakes, error correction, and queuing latency that swamp the relativistic effect.
  • Frozen Earth. Tides flex the crust, the geoid wobbles, the atmosphere refracts β€” none of which the viz models.

The math is sound to ~3 sig figs. But it's still a cow-shaped sphere.