# “Crouching Auto, Hidden Piano,” or Probability Theory in Action

For a long time I wanted to participate in some small but entertaining web-experiment, tied to a full-screen interactive video broadcast, and now it happened: a client appeared who was ready to sponsor this.

So we started making http://carvspiano.ru - a project of only a week length, but with the bold intention to test the Butterfly Effect in action.

Given:
- Beautiful black car - 1 pc.,
- No less beautiful black piano - 1 pc.,
- Camcorder - 3 pc.,
- Projector - 1 pc.,
- Power Mac G5 to which it is connected - 1 pc.,
- Ropes with fasteners - 9 pcs.,
- Conditions under which the ropes break off - 12 pairs,
- A platform with the Internet and a relatively high ceiling - 1 pc.,
- “Probability theory and mathematical statistics” (Kremer's textbook) - 1 pc.

Required:
- Find out how many conditions out of 12 will play,
- Understand whether the piano will fall on the car in the allotted 6 days (from Monday to Saturday).

All conditions are verifiable. We check them twice a day, at 11:00 and 17:00.

To date, the following has already happened:

1. Monday, morning. If an eagle falls on a coin thrown in front of the camera, cut the rope.
Result: tails fell out, not cut.

2. Monday evening. If the number of likes for this video at the time of verification is even, we cut the rope.
Result: it was even, cut.

3. Tuesday, morning. If the traffic jams at the time of verification will be more than 6 points (according to Yandex.Traffic estimates), we cut the rope.
Result: traffic jams were 7 points, cut.

4. Tuesday evening. If the number of messages on Twitter with the word "piano" for the last hour at the time of verification is even, we cut the rope.
Result: there were 26 tweets, cut.

5. Wednesday, morning. (This one we all love very much) If Barcelona reaches the final, we cut the rope.
Result: Chelsea came out, not cut.

At present, 3 ropes have been shot, 6 more are left.
We do not know how many of them must remain in order for the piano to fall. In addition, when shooting each rope, additional stress is created - the lid may simply not withstand.

The next condition check is today at 17:00 Moscow time.

Since theoretically the piano can fall not only when we shoot the ropes, but generally at any moment (the thing, in general, is not new and not very adapted for acrobatic somersaults), we record video from each of the three cameras around the clock.

Considering that a 1-hour video with a three-megabit bitrate weighs approximately 1.3 GB, by the end of the project we will have accumulated about 3 x 24 x 7 x 1.3 = 655.2 GB of video material. Now it is being written directly to the server (thanks to Scalaxy for this!)
Maybe, post factum we will mount a small video from it with the most important points.

PS In many ways, this project is a test of the pen and checking how realistic it is to make a simple but technologically advanced site in a short time and with a limited budget.
Comments, questions and constructive criticism are welcome!

upd .: Many thanks to the Habrausers for stress testing. Programmers cursed, but more or less restrained the blow. At the peak, there were 1,700 video connections, but the servers survived.