Find the Odd Disk

Original Article ↗ Hacker News Discussion ↗

Perceived Difficulty and Scoring

  • Many report it starts very easy and becomes noticeably harder around rounds 10–15; late rounds often feel like pure guessing.
  • Reported scores range widely (roughly 7–20/20), with most self‑described non‑colorblind users clustering in the mid–high teens or 19–20.
  • Several note specific trouble with certain hues, especially blues/purples and sometimes reds or pinks.
  • Some users improve markedly on a second run by changing strategy (looking at each disk in sequence, blinking, looking away briefly).

Desire for Feedback and Data

  • Strong demand for richer feedback: comparison to others, possible color‑blindness indicators, per‑color error breakdown, and an explanation of what the test is measuring.
  • People are curious why more data is requested and whether aggregated statistics will be published.

Display Quality, Calibration, and Environment

  • Major thread on whether results measure vision or display quality:
    • Arguments that you “can’t take it seriously” without a calibrated, high‑gamut display in good lighting.
    • Counter‑arguments that calibration doesn’t necessarily affect relative distinguishability on the same device except near gamut limits.
  • Device differences (cheap phones/tablets vs OLEDs, high‑end calibrated monitors, blue‑light filters, “night mode,” brightness level) clearly change scores for some.
  • Suggestions that the experiment should record device type and maybe test display capability.

Color Vision and Accessibility

  • Color‑blind participants generally score lower and describe the test as frustrating or “torture.”
  • People wish for an “I can’t tell” or “all the same” option to avoid forced random clicks that skew data.

Perceptual Effects and Visual Phenomena

  • Several note afterimages and adaptation: the disk they stare at seems to change brightness/color, making discrimination harder.
  • Strategies like looking at the triangle center or using peripheral vision help some.
  • Discussion branches into related visual phenomena: averted vision for dim stars, flicker sensitivity in peripheral vision, visual/“eye” migraines and scintillating scotoma.

Test Design, Implementation, and Cheating

  • One commenter inspects the code: difficulty ramps in discrete steps over 20 rounds; a blacklist avoids repeats; every answer is sent to the server.
  • Some think control trials with identical disks would help detect positional bias.
  • Using browser dev tools to read RGB values is mentioned and immediately labeled as cheating.