NetLogo

NetLogo is a free, open-source programming language and integrated modeling environment built for one purpose above all others: making it easy to build and explore agent-based models of complex systems. A direct descendant of the Logo family, NetLogo lets a modeler give simple, local instructions to thousands of independent agents — “turtles” moving across a grid of stationary “patches” — and then watch global, often surprising, patterns emerge from their interactions. Created by Uri Wilensky in 1999, it has become the most widely used agent-based modeling platform in both education and research, prized for being approachable enough for a middle-school classroom yet capable enough for published science.

History & Origins

NetLogo’s lineage runs straight back to Logo, the educational language developed in the late 1960s by Seymour Papert and colleagues, famous for its turtle graphics. In the late 1980s, Mitchel Resnick at the MIT Media Lab extended Logo into StarLogo (around 1989), a “massively parallel” version in which the single turtle became thousands of turtles all executing instructions at once — a natural fit for modeling decentralized systems like ant colonies, traffic jams, and markets.

Uri Wilensky, working in that tradition, built StarLogoT in 1997 at the Center for Connected Learning and Computer-Based Modeling (CCL), then based at Tufts University. StarLogoT was the immediate predecessor of NetLogo. In 1999, Wilensky first created NetLogo itself, redesigning the environment and adding the influential notion of patches — a grid of stationary agents that hold local state — alongside the mobile turtles. In 2000, the CCL relocated to Northwestern University in the Chicago area, where NetLogo has been developed and maintained ever since.

From there the version history is steady and long-lived: NetLogo 1.0 arrived in 2002, 4.0 in 2007, 5.0 in February 2012, 6.0 in December 2016, the final 6.x release 6.4 in November 2023, and the major 7.0 release in September 2025.

Design Philosophy

NetLogo is built around a famous educational maxim associated with the Logo tradition: a “low threshold and no ceiling.” Beginners should be able to write a working model in minutes, while experts should never hit a wall that prevents serious work.

• Agents, not algorithms first. Instead of thinking in terms of central loops over data, modelers think in terms of what each agent does. Behavior is described locally; global structure emerges.
• Four agent types. NetLogo’s world is made of turtles (mobile agents), patches (the stationary grid cells they move over), links (connections between turtles, for networks), and the observer (a global vantage point that issues commands to everyone).
• Readable, near-English syntax. Commands like ask turtles [ forward 1 ] read almost as plain instructions, lowering the barrier for non-programmers in the sciences and classrooms.
• Exploration over performance. The environment is optimized for rapid model building, visualization, and experimentation rather than raw execution speed, reflecting its roots as a tool for understanding rather than production computation.

Key Features

The agent model

A NetLogo program is organized around giving instructions to agents. The ask primitive runs a block of commands for each agent in a set:

to setup
  clear-all
  create-turtles 100 [
    setxy random-xcor random-ycor
    set color green
  ]
  reset-ticks
end

to go
  ask turtles [
    right random 50
    left random 50
    forward 1
  ]
  tick
end

Here setup scatters 100 turtles across the world, and go makes each one wander — a minimal model that already shows the language’s local-instruction style.

Turtles, patches, links, and the observer

Every model is a world of patches (a coordinate grid), turtles that move across them, and optional links for representing networks and relationships. The observer sits above the world and can address any agent set. This small, fixed vocabulary is what makes NetLogo so quick to learn while remaining expressive enough for elaborate simulations.

BehaviorSpace

BehaviorSpace is a built-in tool for running a model many times across systematically varied parameter settings — a parameter sweep — and recording the results. It can distribute these runs in parallel across multiple processor cores, turning a single model into a controlled computational experiment.

HubNet

HubNet turns a NetLogo model into a multiuser, participatory simulation: each participant controls one or more agents from a networked client device, while the model runs centrally. It is widely used in classrooms to let students become agents in a shared system, from trading economies to epidemics.

Extensions and the Models Library

NetLogo bundles an extensive Models Library of hundreds of curated, runnable models, and supports an extensions mechanism (for GIS, networks, Python interoperability, sound, and more) so the core language can reach beyond its built-in capabilities.

Implementation

NetLogo is written mostly in Scala, with some parts in Java, with the Scala code compiling to Java bytecode and running on the Java Virtual Machine. According to the project’s documentation, recent versions require Java 11 or later, and the official downloads bundle their own Java runtime so users do not need to install one separately. The environment runs on Windows, macOS, and Linux; the documentation notes it does not run on mobile operating systems such as iOS or Android.

NetLogo Web

For browser-based use, the project maintains NetLogo Web, an implementation in JavaScript/HTML5 that runs models in a web browser without a desktop install. The project’s documentation describes NetLogo Web as an experimental implementation aimed at lighter-weight and tablet/phone use; the desktop application remains the primary, fully featured environment.

Evolution

Over more than two decades, NetLogo has evolved while preserving its accessible character. Major language milestones include the substantial engine and language revisions of 4.0 (2007), the plotting and language refinements of 5.0 (2012), and the introduction of first-class anonymous procedures (closures) in 6.0 (2016), which modernized how reusable behavior is passed around in models.

The 7.0 release in September 2025 was the largest update in years on the presentation side: it introduced a modernized graphical interface with light, dark, and classic themes, redesigned widget tooling, and a new XML-based .nlogox model file format replacing the older .nlogo format, intended to make models easier to extend and to process with other tools. Maintenance releases have continued into 2026.

Current Relevance

NetLogo remains actively developed at Northwestern’s CCL and is, by wide consensus in the field, the most popular agent-based modeling environment in both education and research. It anchors a large community: university courses, the long-running NetLogo users group and forum, the curated Models Library, third-party extensions, and a substantial body of published research that cites it as the modeling platform. The MIT Press textbook An Introduction to Agent-Based Modeling (2015) by Uri Wilensky and William Rand uses NetLogo throughout, further establishing it as a standard reference tool.

Why It Matters

NetLogo’s significance lies in how it democratized agent-based modeling. Before tools like it, simulating a complex system of many interacting parts typically meant writing substantial custom code in a general-purpose language. NetLogo replaced that with a friendly environment where a biologist, economist, social scientist, or schoolchild could express “what each agent does” in near-English and immediately see the emergent consequences. In doing so it carried the Logo tradition of learning-by-building into the era of complexity science, and gave an entire research community — and countless classrooms — a shared language for studying how local rules give rise to global patterns.