Is source code protected by copyright?

Yes, automatically, as a literary work — both source and object code. The protection covers the code as written. It does not cover the functionality: 17 USC 102(b) excludes ideas, procedures, processes and methods of operation, so an independent implementation of the same behaviour is generally not infringement.

Can I register copyright in code without publishing it?

Yes. Registration requires depositing identifying portions of the code, typically the first and last 25 pages, and the Copyright Office permits redacting trade secret material. That mechanism exists specifically so that registration and confidentiality can coexist.

Are git commit dates good enough as proof?

For internal purposes, yes. As evidence against a motivated opponent, they are weak, because commit dates are metadata your own machine writes and can be rewritten with a rebase. Anchoring a release hash to a public ledger produces a date that is not yours to change.

Does open sourcing my code affect my rights?

It ends any trade secret protection permanently, which for server-side software is often the protection doing the real work. Copyright survives and is what the licence operates on, and trademark rights in the name survive — most open source licences reserve them explicitly, so a fork may be free to use the code and not the name.

Trademark vs Copyright for Source Code

Key takeaways

Copyright protects the code as written, automatically, as a literary work.
It does not protect functionality, algorithms or methods of operation — 17 USC 102(b).
Trademark covers only the product name, not the software.
Trade secret is often the stronger practical protection, and it depends on the code staying confidential.

The short answer

Copyright protects your expression of the code. It does not protect what the code does.

Someone who copies your repository infringes. Someone who reads your documentation, understands the approach, and writes their own implementation generally does not.

What copyright covers

Source code and object code are literary works under US law. Protection is automatic on creation and covers:

The code as written, including its structure, sequence and organisation to some extent
Accompanying documentation
Non-literal elements in limited circumstances, though courts filter out anything dictated by efficiency, external constraints or the public domain

What is excluded is substantial: 17 USC 102(b) puts ideas, procedures, processes, systems, methods of operation and concepts outside copyright entirely. Algorithms, protocols and the functional behaviour of your program fall on that side of the line.

Registration is roughly $45–$65 and required before suing in the US. Registering software means depositing identifying portions of the code — typically the first and last 25 pages — and the Copyright Office allows redacting trade secret material, which is the mechanism that makes registration compatible with keeping the code confidential.

What trademark covers

Only the name, and the logo. Your product's name is registrable at $350 per class; software usually falls in class 9, with software-as-a-service in class 42.

This does nothing for the code itself. It stops a competitor selling a confusingly similar product under a confusingly similar name.

Note that open source licences frequently reserve trademark rights explicitly — you may be free to fork the code and unfree to keep the name.

Where trade secret comes in

For most commercial server-side software, trade secret is the protection doing the real work. It covers exactly what copyright does not — the methods, the architecture, the tuning, the accumulated know-how — for as long as it stays secret and you take reasonable steps to keep it that way.

The catch is in the definition. Publish the code and the trade secret is gone permanently. This is why the decision to open source is an IP decision, not only a distribution one.

Side by side

	Copyright	Trademark	Trade secret	Patent
The code as written	Yes	No	If kept secret	No
Algorithms and methods	No	No	Yes	Possibly
Product name	No	Yes	No	No
Arises	Automatically	On use	On secrecy	On grant
Survives publication	Yes	Yes	No	Yes
Duration	Life + 70, or 95 years	Indefinite with use	While secret	~20 years
Cost	~$45–$65	$350 per class	Internal discipline	Thousands

Where a timestamp fits

Version control already gives you a detailed history — but git timestamps are metadata written by your own machine, and they can be rewritten. git commit --date and a rebase will produce any history you like. That is fine internally and weak as evidence against a motivated opponent.

Timestamping the hash of a release archive, or of a repository bundle, anchors that exact state to a public ledger at a moment nobody controls. Two specific uses:

Release records. A dated, tamper-evident record of exactly what shipped, useful in an infringement claim and equally useful in defending one.
Pre-disclosure records. Before code goes to a client, an auditor, an acquirer or an escrow agent, without the code being uploaded anywhere — the hash is computed in your browser and only the hash is transmitted, which is the point when the code is the trade secret.

It records existence and integrity, not authorship. It is not a patent and not a registration.

US law. This is general information, not legal advice.

Trademark vs copyright for source code