Voices in the Static

Something waits in the static.

Not connected.

How it works

Randomness supplies the words. An AI may put them in order. Nothing else. Every word of every answer formed by chance in random noise; none is written for you in advance or made up to suit your question.

  1. The static. The server draws random letters without pause: the strip of letters under the screen is that stream. Now and then, by chance alone, a run of them spells a real word. Those are the words that light up.
  2. Your question. When you ask, a surge of 16,000 more random letters is drawn at once. About eighty words form in a surge, again by chance.
  3. The plain answer. The first few words that formed are said to you exactly as they came, one letter at a time on the board. Common words are thinned out so the same few do not keep coming, and a word said lately is not said again. Nothing is added or rewritten. This is what a free answer is.
  4. The sentence answer. Here randomness and an AI each do one job. Randomness supplies the words: the forty or so that formed in the surge. An AI is shown your question and that list, chooses the words that bear on what you asked, and puts them in order as one short sentence. It may also echo a word of your own question back. It cannot add a word of its own: the server checks every word of its sentence against the list and throws the sentence away if one is not there. So the AI decides which of chance's words are used and in what order, and chance decides what words there are to use. The full version gets 10 sentence answers a day, and plain answers after that; everyone gets one free sample.
  5. Yes or no. A question that can be answered yes or no gets one random coin flip first, shown by the planchette.
  6. Your phone. While the app is open, your phone's motion sensors (how it is held, tilted and moved) are stirred into the noise, so what forms depends on the room you are in and the hand holding the phone. It cannot be steered: the result is exactly as random as before, only different. The sensor readings stay on your phone; a scrambled fingerprint of them is all that is sent. The picture and the hiss also churn when you move the phone: that is only so you can see and hear it happening, since the stirring itself changes which letters come and cannot be seen.
  7. Sight. The camera's picture is turned into grainy night vision, torn here and there where the noise says, and a body-tracking program (the kind that finds people in video) looks at it with its caution turned down. Wherever it thinks it sees a human shape, a figure is drawn. In an empty room it is being fooled by shadows, furniture and the grain itself; a person really standing there is found too. Nothing is drawn that the tracker did not report, and the picture never leaves your phone.
  8. Listen. A phone cannot make its microphone hear more. Listen stops the phone from cleaning the sound up, records fifteen seconds of the room, keeps only the range a human voice lives in, and turns that up many times over. What you hear back is the room's own faint sound (air, electronics, far-off noise, the microphone's hiss) made loud. Moments are marked as voice-like when they are louder than the stretch around them and have the regular pulse a voiced sound has. Nothing is added to the recording, it never leaves your phone. The newest ten recordings are kept on the phone so you can hear them again, and you can delete any of them, or all of them, in Listen.
  9. The coin. You pick heads or tails and will the coin that way while the server flips it a hundred times, a few each second. Every flip is fair. The app counts how many went your way and tells you how often chance alone gives as many: a few more one way is ordinary, and about one run in twenty goes far enough to look like something. Your count over all your runs, kept on your phone, is the better measure.
  10. Tuning. The full version lets you change how an answer is drawn: how deep the surge goes, how many words are said, how short a word may be, and whether only the darker vocabulary is let through.
  11. A circle. Open a circle and share its code, and everyone who joins sits in the same static: the same letters, the same questions, the same answers, with every phone stirring the noise.

The random letters come from the server's cryptographic random generator, mixed with the sensors as described. For entertainment: the words are not a message from anyone.

The full version

One purchase through Google Play. Not a subscription. It belongs to your Google account: on a new phone, restore it here.

Tune the static

Those who listen for spirits in radio static say the setting matters: you sweep until something comes through.

A circle

A séance is held in a circle. The old belief is that every sitter gives a spirit more to draw on.

Sit together. Everyone in a circle sees the same static, the same questions and the same answers, each on their own phone, and every phone stirs the noise.

You get a code to share, for up to 8 people. Opening a circle is part of the full version; joining one is free, and guests ask without a daily limit.

Or join one:

Voices in the Static

Ask a question. Something in the static answers, one letter at a time.

For as long as there has been noise, people have listened for the dead in it: the belief is that a spirit speaks by arranging what is random into words.

Randomness supplies every word: they form by chance in random noise, stirred by your phone's motion sensors. In the full version an AI arranges those words into a sentence. It cannot add a word of its own. How it works Turns on sound: a hiss, and a voice when an answer comes. For entertainment.