https://assertible.com

Users have the option of including experiment configuration parameters in one of two locations: the experiment's config.txt or a global .dallingerconfig file. This separation has two benefits: (1) it allows keys for Heroku and other web services to be stored outside of the files that are packaged up in a zip file each time an experiment is run; and (2) it allows for default settings, such as the organization name, to be overwritten once instead of needing to rewrite it for each new experiment. However, .dallingerconfig is not saved. It should be, because the file can contain information relevant to the experiment and how it was run.

From @suchow on March 26, 2015 20:59

Copied from original issue: berkeley-cocosci/Wallace#136

From @thomasmorgan on September 3, 2014 22:11

We should make demos:

1 - two co-evolving species (think host parasite)
2 - one where participants fill out the behavior of the environment (participatory fitness function). This is also quite like Stephan's game where other participants acted as filters.

Copied from original issue: berkeley-cocosci/Wallace#30

Expected Behavior

The README on PyPi should appear as rendered rst.

Current Behavior

Markup is visible.

Steps to Reproduce (for bugs)

1. Go to https://pypi.python.org/pypi/dallinger.

From @michaelpacer on November 11, 2014 0:17

This is a reason to have vector types as an extension of the vector class so as to allow multigraphs over the same sets of nodes that obey different constraints on the kinds of information transmission that can occur.

Possible examples:

Evolutionary Graph theory: Breaking the symmetry between interaction and replacement

Gender based sexual reproduction restrictions, no gender based behavioral restrictions

Species level reproduction restrictions, alarm calls and communication between species

Copied from original issue: berkeley-cocosci/Wallace#72

From @suchow on July 20, 2016 21:55

Copied from original issue: berkeley-cocosci/Wallace#391

From @suchow on August 10, 2016 19:13

Copied from original issue: berkeley-cocosci/Wallace#440

From @michaelpacer on November 11, 2014 0:17

This is a reason to have vector types as an extension of the vector class so as to allow multigraphs over the same sets of nodes that obey different constraints on the kinds of information transmission that can occur.

Possible examples:

Evolutionary Graph theory: Breaking the symmetry between interaction and replacement

Gender based sexual reproduction restrictions, no gender based behavioral restrictions

Species level reproduction restrictions, alarm calls and communication between species

Copied from original issue: berkeley-cocosci/Wallace#72

From @suchow on September 1, 2014 0:26

Once the simulation is ready for them, the waiting room sends the participant to the experiment. This isn't needed yet, but will become quite useful when we need sets of people to be simultaneously present.

Copied from original issue: berkeley-cocosci/Wallace#27

From @suchow on May 3, 2016 23:17

https://en.wikipedia.org/wiki/Nomic

Copied from original issue: berkeley-cocosci/Wallace#326

From @jhamrick on April 20, 2015 16:58

Python 3 is the future! (And is much nicer than python 2). At some point we should add python 3 support.

Copied from original issue: berkeley-cocosci/Wallace#155

From @suchow on May 7, 2016 3:48

Copied from original issue: berkeley-cocosci/Wallace#329

The configuration parser should be as similar as possible to what's there, but should not depend on psiTurk. This may also be a good opportunity to address #73, an issue where user-modified settings in the .dallingerconfig file are not recorded in the experiment package.

Expected Behavior

JavaScript libraries and CSS stylesheets, such as Bootstrap's, can be downloaded by the client directly from a CDN:

http://getbootstrap.com/getting-started/

Current Behavior

These assets are served from the Dallinger server and are stored in the code base multiple times (e.g., once per demo).

e.g., measure a user's Weber fraction.

From @thomasmorgan on November 3, 2015 20:5

here's how to do it:

http://flask.pocoo.org/snippets/8/

Copied from original issue: berkeley-cocosci/Wallace#251

From @suchow on February 14, 2015 0:27

Copied from original issue: berkeley-cocosci/Wallace#129

From @suchow on April 25, 2015 22:13

Copied from original issue: berkeley-cocosci/Wallace#167

From @suchow on August 22, 2016 17:1

Copied from original issue: berkeley-cocosci/Wallace#470

https://en.wikipedia.org/wiki/Turing_test

From @suchow on July 26, 2016 18:34

Consider the nested HTTP requests in the following snippet from questionnaire.js:

submit_responses = function() {
    if (lock===false) {
        lock=true;
        reqwest({
            url: "/question/" + participant_id,
            method: 'post',
            type: 'json',
            data: {
                question: "engagement",
                number: 1,
                response: $("#engagement").val()
            },
            success: function (resp) {
                reqwest({
                    url: "/question/" + participant_id,
                    method: 'post',
                    type: 'json',
                    data: {
                        question: "difficulty",
                        number: 2,
                        response: $("#difficulty").val()
                    },
                    success: function(resp) {
                        submit_assignment();
                    },
                    error: function (err) {
                        err_response = JSON.parse(err.response);
                        if (err_response.hasOwnProperty('html')) {
                            $('body').html(err_response.html);
                        }
                    }
                });
            },
            error: function (err) {
                console.log(err);
                err_response = JSON.parse(err.response);
                if (err_response.hasOwnProperty('html')) {
                    $('body').html(err_response.html);
                }
            }
        });
    }
};

Because the second HTTP request does not depend on the first, it is unnecessary to nest them in this way. It would be simpler to write a for loop over the questions and post them one by one.

This appears in several places elsewhere.

Copied from original issue: berkeley-cocosci/Wallace#395

From @thomasmorgan on November 3, 2015 20:5

here's how to do it:

http://flask.pocoo.org/snippets/8/

Copied from original issue: berkeley-cocosci/Wallace#251

From @suchow on September 5, 2016 17:43

http://www.nature.com/articles/srep26889

Copied from original issue: berkeley-cocosci/Wallace#486

https://en.wikipedia.org/wiki/List_of_games_in_game_theory

• Hedonic games #189
• Schelling spatial segregation #188
• Voronoi game #137
• Keynesian beauty contest #42
• Iterated prisoners dilemma tournament
• Battle of the sexes
• Blotto games
• Fair division
• Centipede game
• Chicken
• Cournot competition
• Deadlock
• Dictator game
• Diner's dilemma
• Dollar auction
• El Farol Bar problem
• Game without a value
• Guess two thirds of the average
• Kuhn poker
• Matching pennies
• Minority game
• Kolkata Paise Restaurant Problem
• Nash bargaining game
• Pirate game
• Ultimate game
• Pigs
• Princess and monster game
• Prisoner's dilemma
• Iterated prisoner's dilemma
• Public goods game
• Rock, paper, scissors
• Screening game
• Signaling game
• Stag hunt
• Traveler's dilemma
• War of Attrition
• Ultimatum game
• Nim
• Samaritan's dilemma
• Pareto coordination (see pg. 29 of the Joel Watson game theory text)

• Avenir body text
• Smaller h2 headings
• Solid underline for links

From @suchow on July 20, 2016 22:3

(See email between Tom Griffiths and Ming Hsu.)

Copied from original issue: berkeley-cocosci/Wallace#392

(This is what you'd want if participants arrived one by one and were arranged into pairs. Then you can generalize to the case of triads, small groups, etc.)

From @suchow on April 14, 2015 8:14

http://www.replicatedtypo.com/iterated-learning-using-youtube-videos-and-speech-synthesis/6123.html

Copied from original issue: berkeley-cocosci/Wallace#147

From @thomasmorgan on October 15, 2015 0:43

Wallace uses notifications from AWS to advance a particular experiment. For example, when recruiting participants in batches of 40, Wallace waits for a submitted notification from AWS as a prompt to update the status of the participant in the database to 101 (meaning completed) and it then uses this to determine whether it needs to put up 40 more slots or not.

Because processing the notifications can take a long time they are dealt with by worker dynos.

During the most recent large scale study, in some instances Wallace paused, seemingly awaiting a notification from AWS. However, the logs show that the notification is sent, but that it is not processed successfully. At first it appeared that it wasn't being added to the queue successfully and so was never being picked up by the worker dynos, however, in some cases the worker dynos appeared to start working on it, but did not finish successfully.

Copied from original issue: berkeley-cocosci/Wallace#237

From @thomasmorgan on June 12, 2015 16:32

we can allow the databases to include workerIDs whilst running, however, to protect identities we can replace them all with indexes upon export. This way Wallace can use worker ids while running

Copied from original issue: berkeley-cocosci/Wallace#187

From @suchow on April 18, 2016 22:47

Copied from original issue: berkeley-cocosci/Wallace#307

From @suchow on September 3, 2016 3:7

Copied from original issue: berkeley-cocosci/Wallace#483

• Cache apt packages
• Don't install Heroku toolbelt

From @suchow on September 5, 2016 19:43

https://devcenter.heroku.com/articles/heroku-button

Here's how you get user-set configuration:
https://github.com/cobyism/ghost-on-heroku/blob/master/app.json

Copied from original issue: berkeley-cocosci/Wallace#487

https://en.wikipedia.org/wiki/Interactive_evolutionary_computation

(Modify MCMCP demo.)

From @thomasmorgan on October 19, 2015 18:39

Copied from original issue: berkeley-cocosci/Wallace#240

From @suchow on April 23, 2015 0:24

Copied from original issue: berkeley-cocosci/Wallace#165

The goal is to allow Dallinger to run without psiTurk as a dependency. This will require a good bit of research into how exactly we depend on psiTurk.

• Use a different configuration manager (#72)
• Flask app
• psiTurk shell
• psiTurk access keys
• Documentation
• Overwritten routes (compute_bonus/, quitter/)
• scoped_session_decorator in db.py
• Ad server

http://www.aaronkoblin.com/

One click replication + writeup of a behavioral experiment.

http://www.automaticstatistician.com/

https://en.wikipedia.org/wiki/Keynesian_beauty_contest

From @michaelpacer on November 11, 2014 1:9

Given an info identity, create a complete history of any particular piece of info by tracing the source infos (in a many to one transformation) recursively until you arrive at a source or a human participant who created the info.

A secondary convenience method might be to create a graph of this information path through the nodes, transformations and vectors it went along and through.

Copied from original issue: berkeley-cocosci/Wallace#74