The polite feedback trap
The team takes Severo to Juan's old university for its first in-person user tests, about eight strangers approached at study tables. The usability findings are concrete: people get lost, ignoring the big Adventure button and drifting sideways into the pile of minigames until they're overwhelmed; Google login silently fails for a couple of them; the audio-record button breaks, partly because the first tap triggers the mic-permission prompt and eats the recording, partly because very short clips come back empty and the API can't read them. But the lesson that outlasts the bug list is about the feedback itself: almost everyone said the app was cool, because in a hallway, pitched by a stranger, people are polite. Only a few gave hard feedback. So the qualitative signal is inflated the same way vanity metrics inflate the quantitative one, and 'it's cool' from a stranger is worth almost nothing.
// trace: where this idea came from
- ↳ video diary @ 3:42 (la gente se pierde, ignora el botón grande de aventura)
- ↳ video diary @ 4:36 (el botón de audio se rompe, permiso y audios muy cortos)
- ↳ video diary @ 8:04 (le haces un pitch a cualquiera y dice suena chévere)
- ↳ Entry 243-1: A manufactured audience (la señal inflada que aquí aparece del lado cualitativo)
They took Severo to Juan’s old university for its first real user tests, approaching about eight people at study tables and letting them log in and poke around before saying much, just to watch. The usability findings came fast. People get lost: instead of tapping the big Adventure button, they drift sideways and start opening every exercise, then get overwhelmed by the sheer pile of minigames ▸ 3:42. For two or three of them Google login just didn’t work, cause unknown ▸ 4:29. And the audio-record button broke in the wild: the first tap fires the mic-permission prompt and swallows that recording, and very short clips come back empty because the API can’t read them, leaving the exercise stuck ▸ 4:36. The fix is to simplify: one tap to start recording, drop the clever WhatsApp-style hold gesture.
en la universidad casi todos dicen “está chévere” →
The finding that outlasts the bug list is about the feedback itself. Most people said the app was great, but that number is suspect ▸ 7:52. In a university hallway, pitched by a stranger they don’t want to offend, almost anyone will say “sounds cool”, and it is not the truth ▸ 8:04. Only a handful gave feedback with any edge to it ▸ 8:13. It is the qualitative twin of a manufactured audience: a signal that looks like validation and carries none. The praise you can watch, someone getting lost, someone giving up on the mic, is worth more than the praise you’re told…