The point of the CLA signing process is to have a credible record of a developer stating that they really intend to contribute to an open source project. Technically speaking, the thing we are aiming for is a non-repudiable statement from a contributor, that is, a statement that would be a blatant falsehood to later deny.
This is an important step in assuring that an open source project — any open source project — truly is open source. More specifically, it helps guard against bad actors who contribute to a project only apparently in good faith, and then later make trouble by claiming they weren't really contributing under the project's open source license. For some projects "trouble" has historically come, for example, in the form of patent lawsuits.
Mark J. Crane wants to do his part to guard against this potential trouble, and believes so should you. We aren't innovating here by asking you to sign an agreement, but we are trying to innovate by making the agreement and process nearly-transparent, natural to do for folks already active in open source, and (we hope) extremely understandable.
For comparison, here are a few other open source projects and organizations that use contributor license agreements or have similar processes:
- ANTLR: http://www.antlr.org/doc/ANTLR-contributor-agreement.pdf
- Apache: http://www.apache.org/licenses/icla.txt
- GNU: http://www.gnu.org/prep/maintain/html_node/Copyright-Papers.html
- Google (Android, Chrome / ChromeOS, and more): http://code.google.com/legal/individual-cla-v1.0.html
- Linux kernel: http://elinux.org/Developer_Certificate_Of_Origin
- Node: http://nodejs.org/cla.html
- 10Gen (MongoDB): http://www.10gen.com/contributor
In addition, there is unsurprisingly a Wikipedia page about CLAs. It's brief but has some good information and useful links.
In an older time, this sort of agreement might have been collected in paper form. You might have been asked to sign a piece of paper and then send it into an organization (perhaps physically or as a fax), which would in turn keep it in a real filing cabinet.
We no longer live in that time.
One wonderful thing about the world we live in today is that we can achieve the necessary non-reputiability without having to have a physical document. In the case of the CLA, we bootstrap this ability off of the infrastructure provided by GitHub: More specifically, we treat GitHub as a neutral third party to witness the transactions between a would-be contributor and Mark J Crane. GitHub ends up acting sort of like a notary, in that its records of the actions — such as in particular the pull requests — of people using it can be taken as authoritative and unbiased.
So, when a contributor forks this project, commits a change indicating agreement to the CLA, and files a pull request back with this project, GitHub knows that all that happened, knows when it happened, and knows the identity of the entity taking all that action. Should there ever be a dispute about whether any of that took place, it's not just he-said she-said, because GitHub has unbiased information which it can use to shed light on the situation.
In addition to what's in the pull request, we need to have a little bit of extra information, which can help verify that you are who you say you are, should the need arise. In particular, we need to have some information that can link you to your contributions, even if you later delete your GitHub account. Since people rightly desire privacy about their addresses and phone numbers, we don't ask for this information to be made public in the pull request, instead going for a traditional email. We promise never to use this information for any purpose other than resolving authorship disputes.
The upshot is that filing a pull request containing a statement of agreement to the CLA, along with the supplementary email, is close enough to having submitted a signed physical document saying the same things. That is, this tactic is a workable solution to the problem. And since fits in naturally with how actual contributions get made, what we do here ends up being the least-intrusive way of getting the assurance that we all need to operate in safety.