Stop Fighting Git. Start Using It the Way Linus Intended

The Latency Gambler

Follow

4 min read

·

Jul 27, 2025

706

29

Listen

Share

Most developers treat Git like a glorified backup system. They create linear histories, fight merge conflicts, and complain about complexity. But Git wasn’t designed for this workflow. Linus Torvalds built Git as a distributed content-addressable filesystem optimized for tracking changes in large codebases with hundreds of contributors.

Understanding Git’s original design philosophy will transform how you work with version control.

Press enter or click to view image in full size

The Content-Addressable Storage Model

Git doesn’t store files. It stores content addressed by SHA-1 hashes. Every object (blob, tree, commit, tag) has a unique identifier based on its content.

Git Object Storage Architecture:

Working Directory
           |
           v
    +-------------+
    |   Staging   |  <-- git add
    |    Area     |
    +-------------+
           |
           v
    +-------------+
    |  Local Repo |  <-- git commit
    | (Objects DB)|
    +-------------+
           |
           v
    +-------------+
    |  Remote     |  <-- git push
    |   Repo      |
    +-------------+

When you commit, Git creates immutable snapshots:

# Each commit points to a tree object
git cat-file -p HEAD
# tree 4b825dc642cb6eb9a060e54bf8d69288fbee4904
# author John Doe…