Syllabus
Fall 2026 Thursday 9-11 3 credits 300 Buell South
Adam Vosburgh adam (dot) vosburgh (at) columbia (dot) edu office hours by appointment, sign up at the link on the top of the page.
Description
Simulations, Models, Twins explores modes of simulative technologies and speculates on their potential for practitioners of the built environment. Digital twins, agent-based models, and other types of simulations serve a similar purpose to mock-ups and scale models in traditional architecture practices: there is utility in testing a design in a "real" environment, whether physical or simulated. At the same time, simulations are active agents in shaping the realities they depict, and the degree of "reality" they capture is a continuous source of debate. This course will emphasize hands-on experimentation with simulative technologies as tools for crafting arguments, building worlds, and testing how proposals might perform in the "real" world, while bearing in mind that these representations are partial, situated, and constructed.
The course is divided into two halves:
- In the first, participants work through a sequence of weekly tutorials and assignments, all of which are submitted to and shown on this website. Alongside this, we look closely at a set of sandboxes built for the course: playable simulations whose assumptions are exposed as controls.
- In the second, participants form groups and develop a final project on a topic of their own choosing, using the methods of the first half. This culminates in two "Show and Tell" sessions at the end of the semester, and a public-facing project page on this website.
No prior programming experience is required, but a willingness to work in code, is. The tutorials provide nearly all of the code you will need, and assignments will involve making tweaks on that code base. If you run in trouble along the way, you can find help either through the course-specific AI assistant, or through office hours.
This course sees the critical use of AI tools as a method of expanding the technical ambition of projects, and not as a substitute for having original ideas. From week four, the course teaches you to make software by writing a specification for it rather than by typing the code, otherwise referred to as "vibe-coding."
Course Overview
| Date | Week | Topic | Tutorial due | Assignment due |
|---|---|---|---|---|
| 9/10 | 1 | Introduction and Overview | ||
| 9/17 | 2 | Data and Representation | Tutorial 1 | Assignment 1 |
| 9/24 | 3 | Agents and Action | Tutorial 2 | Assignment 2 |
| 10/1 | 4 | Design Practice; Software as Medium | Tutorial 3 | Assignment 3 |
| 10/8 | 5 | Feedback, Sensing and Measuring | Tutorial 4 | Assignment 4 |
| 10/15 | 6 | Games, and Simulation as Play | ||
| 10/22 | 7 | Proposal Workshop | Draft Proposal | |
| 10/29 | 8 | "Spatial Intelligence" and "World Models" | ||
| 11/5 | 9 | Living in the Model | Tutorial 5, 6 or 7 | (Assignment 5, extra credit); Progress meeting, all groups |
| 11/12 | 10 | Show and Tell Day 1 | Day 1 groups present; Progress meeting, Day 2 groups | |
| 11/19 | 11 | Show and Tell Day 2 | Day 2 groups present | |
| 12/21 | Website Due | Final Project Page |
Weekly Schedule, Readings, and Assignments
All tutorials and assignments are on this website. Readings are listed under the week they are due and are in the course folder. Every assignment is submitted through its page here, and appears under Student Work. Each week also links to a short list of projects to look at before class, on the projects page.
9/10 | Week 1 | Introduction and Overview
An overview of the semester: the class ambitions, the kinds of output the course produces, and which parts of the course are technical and which are about representation and argument.
9/17 | Week 2 | Data and Representation
Data, representation, and the politics of each. In-class exercise: look at a set of projects and identify the datasets in them.
Readings due (in course folder or linked)
Kurgan, Laura. "Mapping Considered as a Problem of Theory and Practice." In Close Up at a Distance: Mapping, Technology, and Politics. Zone Books, 2013.
Optional: Couclelis, Helen. "People Manipulate Objects (but Cultivate Fields): Beyond the Raster-Vector Debate in GIS." 1992.
Tutorial 1 due – Set Up
This tutorial will get you up to speed with the course website.
Assignment 1 due – Hello
Tell us a bit about yourself, why you took this class, and what you want to get out of it. We will use responses to introduce the class to each other.
Projects to review
9/24 | Week 3 | Agents and Action
Forecasts, hindcasts, and simulations use data to make an argument about a time the data does not cover. This week covers what a rule is, what it means to run one, and how to read the output of a simulation someone else built. In-class exercise: intro to the course sandboxes, the ADU Forecast for Queens, the Sea Level Flood Map, Office to Residential Conversion, and Anthromes. For each: what is it trying to say, what is the data, and what is the mechanism?
Readings due
Levin, Noa and Sascha Roesler. "Editorial." e-flux Architecture, Simulation, 2026.
Kurgan, Laura, Dan Miller, and Adam Vosburgh. "Two Sides of the Same Coin." Check out the data-visualization, and click through to the original studies and data linked in the "info" button on either side
Tutorial 2 due – Mapping Where, in Python
Raster and vector spatial data in Python rather than in a desktop GIS, using the NYC street tree census.
Assignment 2 due – Two Datasets, a Site, a Map
A place you know, two datasets that were not collected to say anything about each other, and one 16:9 composition that makes an argument out of them.
Projects to review
10/1 | Week 4 | Design Practice; Software as Medium
What is design practice? In class: live demo of taking a sketch to a build doc, to an app.
Readings due (in course folder or linked)
Turkle, Sherry. Simulation and Its Discontents. MIT Press, 2009. [What Does Simulation Want (Introduction), New Ways of Knowing / New Ways of Forgetting]
Corner, James. "The Agency of Mapping." 2014. This is a classic reading in inter-disciplinary design courses. However, it is a bit long and meandering. Feel free to skim for bits of insight on why mapping is relevant to design practices.
Silver, Joshua. "Observing the Future: Ecologies of Architectural Simulation." e-flux Architecture, Simulation, 2026. https://www.e-flux.com/architecture/simulation/6783059/
optional: Rittel, Horst W. J., and Melvin M. Webber. 1973. “Dilemmas in a General Theory of Planning.” Policy Sciences 4(2): 155–169.
Tutorial 3 due – Simulating Trees
The same two datasets as last week, with rules applied.
Assignment 3 due – A Forecast, Sketched
A sketch of how you could apply rules to your Assignment 2 dataset to turn it into a model.
Projects to review
10/8 | Week 5 | Feedback, Sensing and Measuring
Pin-up of the Assignment 4 sandboxes. Second half: data layers on physical space, real-time feedback systems, and the idea of the smart city. What a sensor records, what it does not, and who it reports to. The final project is assigned.
Readings due (in course folder or linked)
House, Brian. "Stalking the Smart City." Urban Omnibus, 2019.
Halpern, Orit, Robert Mitchell, and Bernard Dionysius Geoghegan. "The Smartness Mandate: Notes toward a Critique." Grey Room 68 (2017).
Weckert, Simon. "Google Maps Hacks." 2020. Five minutes; browse the project.
Tutorial 4 due – From a Notebook to a Sandbox
The three documents (brief, build doc, code) worked through on the street tree sandbox.
Assignment 4 due – Sketch to Sandbox
The same three documents for your own Assignment 3 sketch, plus the sandbox the agent builds from them.
Final project assigned
Projects to review
10/15 | Week 6 | Games, and Simulation as Play
From Forrester's Urban Dynamics through SimCity to the present: the city simulated as a system of coefficients. The rules of a game are arguments about how a city works, and playing is one way of reading those arguments.
Readings due (in course folder or linked)
Baker, Kevin T. "Model Metropolis." Logic Magazine, 2019.
Vettese, Troy and Drew Pendergrass. Half-Earth Socialism. Verso, 2022. [Introduction] Half-Earth Socialism is paired with a simulation / deck-builder video game. In addition to the introduction, please read the overview on this page, and play the video game.
Optional: Chang, Alenda Y. Playing Nature: Ecology in Video Games. University of Minnesota Press, 2019. [Introduction]
Projects to review
10/22 | Week 7 | Proposal Workshop
We will go over draft proposals in groups and in desk crits, and assign presentation days.
Draft Proposal due
Three ideas, each with a title and thesis statement, an expected outcome, datasets and references, as a multi-page PDF.
10/29 | Week 8 | "Spatial Intelligence" and "World Models"
Generative models that claim to simulate the world, and the claim that "spatial intelligence" is the next stage of AI development. The curse of dimensionality, and what the world looks like from inside a model with a very large number of dimensions. In class: an activity with embeddings, and possibly a sandbox built on AlphaEarth, if it comes together.
Readings due (in course folder or linked)
Crawford, Kate. Atlas of AI. Yale University Press, 2021. [pp. 1–21 and selections]
Please watch: Kurgan, Laura, Adeline Chum, Michael Krisch, Adam Vosburgh, and Jia Zhang. "The Curse of Dimensionality." Venice Architecture Biennale, 2025.
Optional: Hao, Karen. Empire of AI. Penguin Press, 2025. [Selections]
Projects to review
11/5 | Week 9 | Living in the Model
Final lecture. Feedback loops between representations of the world. Look back on the semester and speculate on what it means.
Readings due
Steyerl, Hito. Medium Hot: Images in the Age of Heat. Verso, 2025. [Selections]
Wiener, Anna. "How Perfectly Can Reality Be Simulated?" The New Yorker, April 22, 2024.
Optional: Farocki, Harun. Parallel I–IV. 2012–2014. Course folder
Tutorial 5, 6 or 7 due
[TBD. Three short tutorials, of which you pick one. Likely candidates: preparing a Rhino model for the sunlight sandbox, reading data from a sensor into a space, and making spatial data out of plain text.]
Assignment 5 due - extra credit, optional
[TBD. Pick one of Tutorials 5, 6 or 7; submit the result]
Final project progress meeting – all groups
Asynchronous from class, any time in the week of 11/5. Questions about process, data, etc. Schedule an office hour with me (feel free to make them back-to-back, so one hour total).
Projects to review
11/12 | Week 10 | Show and Tell Day 1
Day 1 groups present
The expectation is that the mechanics of your project are mostly figured out. It's fine if you're still working through things.
Final project progress meeting – Day 2 groups
Asynchronous from class, any time in the week of 11/12. Final questions about process, data, etc. Schedule an office hour with me (feel free to make them back-to-back, so one hour total).
11/19 | Week 11 | Show and Tell Day 2 – Last Class
Day 2 groups present
The expectation is that the mechanics and data are completely figured out, and locked. From this point on, the work is representation.
12/21 | Website Due
Final project pages due on the course website, 9am
This is a hard deadline. Grades are due 12/23, and projects will be graded according to what is live on the website at that time.
Class Requirements and Grading
There are a total of 21 possible points. Attendance for each session (11 points), in-class participation (1 point), assignments 1 through 5 (5 points, 1 each), and the final project (4 points). 19 points and above is a HP, 15 points and above is a P, 12 points and above is a LP, and anything below that is a F.
Attendance (11 points) Attendance at all class sessions is required. Per GSAPP attendance policy, three unexcused absences is an automatic Unofficial Withdrawal (UW).
In-class participation (1 point) You are expected to give engaged and generous participation in class discussions and in critique workshops with your peers.
Assignments (5 points, 1 point each) Assignments are graded on completion and are intended to be limited in scope; experimental; and geared towards learning and creative engagement rather than demonstrating advanced skills. They must answer the prompts laid out in each assignment and use the methods taught in the relevant tutorials. All assignments are submitted to this website; polish of presentation is explicitly not a criterion.
Final Project + Presentation (4 points) Two points for the show and tell presentation, two points for the final submission on the course website.
Extra Credit (1 point) Extra credit will be given for the completion of Assignment 6, and additional extra credit can be assigned by the instructor in the event that a student has a LP or F.
Learning Objectives
After completing the course participants will:
- Be able to read a simulation, model, or digital twin critically: identify its data, its assumptions, its authors, and its intended and unintended uses
- Work with spatial data in Python: load it, join it, and map it
- Write a rule against a dataset and run it forward through time, and say what the run assumes
- Specify a small piece of software precisely enough for a machine to build it, and check the result against something outside itself
- Develop fluency with web-based visualization as a medium for argument
- Situate simulative technologies within the histories of representation in architecture, urbanism, and computation
AI Policy
This course makes use of AI tools as a method to expand the ambition of projects, and to make the technical portions of the course possible to those without a programming background.
The line this course draws is between assistance in understanding and substitution of authorship. Students are welcome to use AI tools, except for the purpose of:
- Producing original writing that the student turns in as their own work. Your gallery texts, descriptions, and briefs must be yours.
- Prompting a LLM to produce data. This will ALWAYS result in fabricated data. If the data you want doesn't exist, collect it, or scale the project to what does.
- Generating an artifact you cannot explain. If you cannot say what your sandbox assumes, what data it runs on, and what rule it steps by, it is not your project.
Work that uses AI in these ways, and in other ways that the instructor determines are counter to the objectives of the class, will not be awarded points.
Software
Simulation is not tied to any one particular software. The course privileges code over graphical user interface platforms (like Rhino,) because they are more flexible, more legible and editable to LLMs, and lend themselves to sharing work.
For the first three weeks we write Python in Google Colab, which needs nothing installed; instructions for running the same code on your own computer are in Resources.
From week four a coding agent (Claude Code, Codex, Gemini CLI) is recommended, but not required. Everything you make is a web page and lives on this site. Knowledge of 3D modeling software will be useful for some final projects but is not required.
Academic Integrity
The intellectual venture in which we are all engaged requires of faculty and students alike the highest level of personal and academic integrity. As members of an academic community, each one of us bears the responsibility to participate in scholarly discourse and research in a manner characterized by intellectual honesty and scholarly integrity.
Scholarship, by its very nature, is an iterative process, with ideas and insights building one upon the other. Collaborative scholarship requires the study of other scholars' work, the free discussion of such work, and the explicit acknowledgement of those ideas in any work that inform our own. This exchange of ideas relies upon a mutual trust that sources, opinions, facts, and insights will be properly noted and carefully credited.
In practical terms, this means that, as students, you must be responsible for the full citations of others' ideas in all of your research papers and projects; you must be scrupulously honest when taking your examinations; you must always submit your own work and not that of another student, scholar, or internet agent.
Any breach of this intellectual responsibility is a breach of faith with the rest of our academic community. It undermines our shared intellectual culture, and it cannot be tolerated. Students failing to meet these responsibilities should anticipate being asked to leave Columbia.
For more information on academic integrity at Columbia, students may refer to the Columbia University Undergraduate Guide to Academic Integrity as well as the GSAPP Honor System and Plagiarism Policy.
Community & Accessibility
This is a discussion and collaborative-critique based course. All students and the instructor must be respectful of others in the classroom. If you ever feel that the classroom environment is discouraging your participation or is problematic in any way please contact me.
GSAPP is committed to full inclusion of all students. Students needing any form of accommodation due to a disability should check in with Disability Services (DS) and speak with me at the beginning of the semester to provide the accommodations letter from DS. Alternatively you may ask your advisor to consult with me regarding your accommodations.
Email Policy
Students should not rely on or expect an immediate response to questions sent via email to the instructor. Please begin assignments with enough time to attend office hours or ask a question several days before the assignment is due.
Learning how to troubleshoot technical issues and locate relevant resources is crucial to your long-term success with computational methods. With this in mind, emails with technical questions must at a minimum contain the following:
- a clear description of what you are trying to do, and what the problem is
- a summary of the steps you have already taken to address the issue
- screenshots or error messages (where applicable)
Student Work Is Public
Everything submitted to this site is visible to anyone with the address. This is deliberate. The site is the class's pin-up wall, and the second half of the course produces public project pages. Write your gallery texts and descriptions with that in mind. If there is something you would rather not have public, talk to me before the deadline.