DailyVox builds a private model of you — entirely on your iPhone, nothing collected, nothing transmitted. That architecture has a consequence we refuse to dodge: nobody can check the model works by watching a server, because there isn't one. So we measure it the hard way, in the open, and we publish what we find — including the failures. The next measurement needs a handful of real people. Maybe you.
One honest question, precisely stated: how well do the Twin's estimates of your traits agree with how you see yourself? You journal by voice; the model estimates seven things about your style of expression; you answer seven short questions about yourself; we compute the agreement — and compare it against a baseline that guesses blindly. If the model can't beat the blind guess, we say so in public. We've already published one negative result on the developer's own twenty-year diary; the instrument is real.
Full documents: participant sheet · consent note · the public research kit. This is an independent, unpaid, consented study run by the developer as an independent researcher; it is not a university study.
The first wave is deliberately small — a handful of people, measured carefully, reported honestly. If you journal in English on an iPhone and are 18 or older, you're eligible.
Saw the invitation card in the app? This is that study — the card appears once to journalers with a few weeks of entries, and everything beyond it happens here and over email, never inside the app. Two ways to take part: answer the seven-question self-report and compare it with your Twin's estimates (numbers only, two minutes), or additionally share an encrypted export of your diary for model validation — your choice, spelled out in the consent note before anything moves.
Email hello@getdailyvox.comQuestions first? Same address — ask anything before sending your materials.