FactOTD
How Linus Torvalds Built Git in Two Weeks — and Why It Conquered Software Development
technology

How Linus Torvalds Built Git in Two Weeks — and Why It Conquered Software Development

March 28, 2026 · 4 min read

The Fact

Git, the most widely used version control system, was created by Linus Torvalds in 2005 to manage Linux kernel development.

Version control is one of those infrastructure technologies invisible to everyone outside software development but absolutely fundamental to how the modern world's software gets built. The basic idea is simple: when many people work on the same codebase, you need a system that tracks every change, records who made it and when, allows experimentation without breaking the main code, and enables merging of work from many contributors. Before Git, this was done with systems like CVS and Subversion that worked reasonably well for smaller teams but struggled at the scale of the Linux kernel project, which involved thousands of contributors worldwide.

The Trigger: A License Dispute

The Linux kernel had been using a commercial distributed version control system called BitKeeper, which was provided free of charge to the open-source community under certain conditions. In 2005, a dispute arose between the BitKeeper vendor and the open-source community over reverse-engineering attempts. BitKeeper withdrew free access, and the Linux development community — handling thousands of patches per month from developers worldwide — suddenly needed a new solution.

Torvalds surveyed the available alternatives and found none of them acceptable for the kernel's specific requirements: the ability to support extremely distributed development, fast branch and merge operations, strong guarantees of data integrity, and performance at large scale. Rather than compromise on these requirements, he decided to write something new.

He announced his intentions in April 2005 with characteristic bluntness: he needed the system to be fast, distributed, and protect against data corruption, either accidental or malicious. He had the initial design working within days. Git — the name was Torvalds's self-deprecating humor, British slang for an obnoxious person — was operational within two weeks. Torvalds handed off maintenance to Junio Hamano in June 2005, and Hamano has served as the primary maintainer since then.

What Made Git Different

The key design principle that distinguished Git from its predecessors was genuine distributed operation. Systems like SVN used a central server: all version history lived on one machine, and developers checked out copies to work on. Git gave every developer a complete copy of the entire repository, including the full history of every change ever made. Work could be done entirely offline, branches could be created and merged locally, and no central server was required for the fundamental workflow.

This design had profound consequences for how collaborative software development worked. Because branching was fast and cheap in Git — creating a branch takes milliseconds regardless of repository size — developers could create experimental branches freely, work on features in isolation, and merge them when ready. The "feature branch" workflow became standard practice, transforming how teams organized their work.

Git's data model was also novel: rather than storing differences between files (as many earlier systems did), Git stored snapshots of the entire repository at each commit, using a content-addressable object store where each piece of data is identified by the cryptographic hash (SHA-1, later SHA-256) of its contents. This design made it impossible to change historical data without being detected and gave Git extraordinary data integrity guarantees.

GitHub and the Social Coding Revolution

Git itself was a command-line tool with a steep learning curve. GitHub, founded in 2008, built a web interface on top of Git that made repository hosting, code review, and collaboration far more accessible. The combination of Git's technical model with GitHub's social interface created a new paradigm for open-source software development. Forking a project, making changes, and proposing them back as a "pull request" became the standard workflow for contributing to any open-source project.

GitHub grew to host over 100 million repositories by the 2020s, essentially becoming the world's largest source code archive and the central hub of open-source software development. Microsoft acquired it in 2018 for $7.5 billion, an acknowledgment of its centrality to the software development ecosystem.

As of 2026, Git is used in essentially all professional software development contexts. The problem Torvalds set out to solve in April 2005 — how to manage a large, widely distributed software project — turned out to be everyone's problem, and his two-week solution has endured as the universal answer.

F

FactOTD Editorial Team

Published March 28, 2026 · 4 min read

The FactOTD editorial team researches and verifies every fact before publication. Our mission is to make learning effortless and accurate. Learn about our process →

Related Articles

technologyLinux: How a Hobby Project Became the Software Powering the WorldIn August 1991, a 21-year-old Finnish computer science student posted a message to an internet newsgroup saying he was writing a free operating system 'just a hobby, won't be big and professional.' That casual announcement introduced Linux to the world — a kernel that now runs more than 90% of the world's servers, all Android smartphones, and every supercomputer on the TOP500 list.technologyThe Black Box Is Actually Orange: Why Aviation's Most Important Recorder Is MisnamedEvery commercial aircraft carries devices universally known as 'black boxes,' yet they are not black — they are a vivid, high-visibility orange. The misnomer persists, but the color choice is deliberate and potentially life-saving: these devices need to be found in the wreckage of catastrophic accidents.technology347 Billion Emails Per Day: Inside the World's Most Trafficked Communication SystemEvery day, roughly 347 billion emails travel across the global internet — more than 4 million per second, 24 hours a day. Of those, nearly half are unsolicited junk. Understanding how email infrastructure handles this volume, and how spam filters manage to keep inboxes functional, reveals one of the internet's most important and least visible engineering achievements.technologyThe First Internet Message Was 'LO' — Because the System Crashed After Two LettersThe first message ever sent over the network that would become the internet was supposed to be 'LOGIN.' Instead it was 'LO' — because the receiving computer crashed after two characters. The accidental poetry of that truncated greeting, inadvertently echoing 'hello' or 'lo and behold,' seems fitting for the birth of the technology that would eventually connect most of humanity.