docs: add shared variables and side effects

.github/workflows/joss-paper.yml

@@ -0,0 +1,25 @@
+name: Draft PDF
+on:
+  workflow_dispatch
+
+jobs:
+  paper:
+    runs-on: ubuntu-latest
+    name: Paper Draft
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Build draft PDF
+        uses: openjournals/openjournals-draft-action@master
+        with:
+          journal: joss
+          # This should be the path to the paper within your repo.
+          paper-path: paper/paper.md
+      - name: Upload
+        uses: actions/upload-artifact@v4
+        with:
+          name: paper
+          # This is the output path where Pandoc will write the compiled
+          # PDF. Note, this should be the same directory as the input
+          # paper.md
+          path: paper/paper.pdf

paper/paper.bib

@@ -0,0 +1,27 @@
+@article{de2015jspsych,
+  title={jsPsych: A JavaScript library for creating behavioral experiments in a Web browser},
+  author={De Leeuw, Joshua R},
+  journal={Behavior research methods},
+  volume={47},
+  pages={1--12},
+  year={2015},
+  publisher={Springer}
+}
+
+@inproceedings{
+demircan2024evaluating,
+title={Evaluating alignment between humans and neural network representations in image-based learning tasks},
+author={Can Demircan and Tankred Saanum and Leonardo Pettini and Marcel Binz and Blazej M Baczkowski and Christian F. Doeller and Mona M. Garvert and Eric Schulz},
+booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
+year={2024},
+url={https://openreview.net/forum?id=8i6px5W1Rf}
+}
+
+@article{musslick2020sweetpea,
+  title={SweetPea: A standard language for factorial experimental design},
+  author={Musslick, Sebastian and Cherkaev, Anastasia and Draut, Ben and Butt, Ahsan Sajjad and Darragh, Pierce and Srikumar, Vivek and Flatt, Matthew and Cohen, Jonathan D},
+  journal={Behavior Research Methods},
+  pages={1--25},
+  year={2020},
+  publisher={Springer}
+}

paper/paper.md

@@ -0,0 +1,43 @@
+---
+title: 'SweetBean: A declarative language for behavioral experiments with human and artificial participants'
+tags:
+  - Python
+  - online behavioral experiments
+  - large-language model experiments
+  - declarative language
+  - synthetic participants
+authors:
+  - name: Younes Strittmatter
+    orcid: 0000-0002-3414-2838
+    corresponding: true
+    affiliation: 1
+  - name: Sebastian Musslick
+    orcid: 0000-0002-8896-639X
+    affiliation: "2, 3"
+affiliations:
+  - name: Princeton University, USA
+    index: 1
+  - name: Osnabrück University, Germany
+    index: 2
+  - name: University of Chicago, USA
+    index: 3
+date: 13 August 2017
+bibliography: paper.bib
+
+---
+
+# Summary
+
+`SweetBean` is an open-source, domain-specific declarative programming language built in Python, designed for the  synthesis of behavioral experiments. It allows researchers to specify experiments once and seamlessly compile them into a `jsPsych` experiment [@de2015jspsych] for conducting studies with human participants or text-based simulations with synthetic participants using large-language models. 
+
+The `SweetBean` package seamlessly integrates with packages aimed at running online behavioral experiments, such as `AutoRA` [@AUTORA-MISSING] for automated research, participant recruitment, and data collection, and `SweetPea` [@musslick2020sweetpea] for experimental design. 
+
+# Statement of need
+
+The design and implementation of behavioral experiments often involve complex workflows, particularly when researchers aim to run the same experiments with both human participants and large-language models—a practice that is becoming increasingly common, for example, to train models to simulate human behavior [@demircan2024evaluating] or to support AI alignment research. While platforms like `jsPsych` offer powerful solutions for running online experiments, they require knowledge of *JavaScript*, which poses a significant barrier for researchers without programming expertise. Additionally, integrating synthetic participants, such as large-language models, often necessitates ad hoc solutions, leading to inefficiencies and errors due to the need for replicating the same experiment across different codebases.
+
+`SweetBean` offers a *Python*-based, declarative programming language specifically designed for behavioral experiment design. Python’s accessibility and widespread use in the behavioral sciences make `SweetBean` an ideal tool for researchers and educators. By enabling users to specify experiments in a single, concise format, SweetBean simplifies workflows and ensures compatibility with both human and synthetic participants.
+
+Beyond reducing technical barriers, `SweetBean` enhances reproducibility and collaboration by standardizing experiment specifications within a flexible and intuitive framework. This standardization not only saves time but also fosters more robust and scalable experimental workflows for the research community.
+
+# References