Differences between revisions 2 and 3
Revision 2 as of 2016-03-30 22:17:27
Size: 2474
Editor: MattJohnson
Comment:
Revision 3 as of 2016-03-30 22:17:51
Size: 2474
Editor: MattJohnson
Comment: Spoonerized permissions -- now fixed.
Deletions are marked like this. Additions are marked like this.
Line 11: Line 11:
 * Get the 'active' version of the experiment into the right directory on the web server, make sure the Apache configuration is set up correctly to run Python scripts in that directory, and check all permissions -- Python scripts should be 644, everything else should be 755 (assuming owner is root, which it should be).  * Get the 'active' version of the experiment into the right directory on the web server, make sure the Apache configuration is set up correctly to run Python scripts in that directory, and check all permissions -- Python scripts should be 755, everything else should be 644 (assuming owner is root, which it should be).

Mechanical Turk (M-Turk) experiments will be hosted on the same server as Matt's website / the lab website (matthewrjohnson.net or raplab.rocks, depending on how you get in), but via the domain mturkstudi.es.

For now, Matt knows the intimate details of how the web server is set up and is the only one with root-level access to the server, so making a study "go live" will have to go through him at some point.

Here are a few notes on what we have to do in order to get a study ready to run on M-Turk (roughly in chronological order)... will refine as we go later.

  • Set up the task. There will be a 'debug' area of the server where folks can more freely upload files to test, which will be separate from the 'active' area where studies linked from M-Turk will be hosted. For now, we'll set up access to the 'debug' area on a per-experiment basis. Experiments in the 'debug' area should be set up to write files to a 'debug' directory, so that the output is not confused with the output from 'active' studies. We should remember to delete and deactivate 'debug' versions of each experiment (both removing them from public_html and potentially also turning off CGI within the Apache configuration for the 'debug' directory) when the experiment is ready to go into 'active' status.
  • Load money into the lab M-Turk account -- for now, we'll probably do that with Matt's personal credit card and get reimbursed after the data have been collected.
  • Set up the experiment description (including generating the PDF file that contains detailed instructions and making sure it is hosted properly on the mturkstudi.es site) on the M-Turk website.
  • Double-check the 'active' version of the task -- make sure it is set up to write output to the correct places, run through a test subject to make sure everything works (and delete their output), etc. This includes checking the various points that are on the old Yale per-study checklist. (That sentence is just a placeholder for now -- we'll track down the old Yale checklist and add the various items on it to this page.)
  • Get the 'active' version of the experiment into the right directory on the web server, make sure the Apache configuration is set up correctly to run Python scripts in that directory, and check all permissions -- Python scripts should be 755, everything else should be 644 (assuming owner is root, which it should be).
  • Configure TurkPrime to run the experiment.

  • Unleash the hounds!

rapwiki: RunningMechanicalTurkExperiments (last edited 2016-06-13 17:50:37 by MattJohnson)