Homepage Eckhard Etzold - Cyty - Menschen

Control runs

First tests (with beta version 0 and Javascript experimental field) supplied different results: obviously users of the Netscape navigator are psychokinetic more talented ;-) than users of the Microsoft Internet Explorer. But it would be surely still too hasty to draw conclusions from it about the Outgoing of the war of browsers.
Unsupervised control test with 8000 runs (seed number = 5-digit hexadecimal random numbers from the Fourmilab server, Java version 2.0) had a standard deviation of +-14.172666. Platform was Netscape 4.05.
If you like to make control runs with system time as a seed number on your computer (you to have to do it, if you wish to take part seriously in the experiment to get clear results), you have the choice between

  1. 25 runs - approx. 25 minutes of run time
  2. 50 runs - approx. 50 minutes of run time
  3. 100 runs - approx. 1 hour, 40 minutes of run time
  4. 200 runs - approx. 3 hours, 20 minutes of run time
  5. 500 runs - approx. 8 hours, 20 minutes of run time
  6. 1000 runs - approx. 16 hours, 40 minutes of run time

For scientific purposes it is more necessary to execute to each run a control run with likewise a random seed number (from another "pool") without mentally influencing this run. In addition you create a text file, which contains e.g. a list of 50 or 100 seed numbers. You can run Telemech in an accelerated version without graphic feedback display. If you input under file name "zahlen.txt", then Telemech processes within a short time of some seconds this run list, and you obtain the results displayed. For own analyses of test runs or control attempts you must indicate the URL of your own txt-file.

In order of low online costs, wait until the Java window is completely loaded with all graphs, let the Browser opened and go offline. Mark the results with the mouse pointer and copy it into the clipboard for et cet. Note: Switch your monitor off with long-term runs. Keep however your computer under observation, so that you can intervene, if electrical damage occurs.
As a check it is also helpful to view the cumulative deviation graph on the RetroPsychokinesisProject in summary.html. This full-scale test had meanwhile (November 1998) more than 14 ,000 runs. If you take a view of the analysis for Visual-Feedback, you find clearly the different cumulative scores in the different sections, which the more runs in the individual category is the higher was completed. In the Fourmilab log file you find a user who has a particularly high and constant (negative) cumulative score (Subject 623), the single experiment data you can download in ZIP-Format to get an impression of the result of a "successful" user in another experimental situation.

© Eckhard Etzold, 1998-7-31

Zur Homepage

[Zurück] [Menschen]
Impressum, http://bs.cyty.com/menschen/archiv/telemech/, Stand: 6. März 2002, ee