Arnt Gulbrandsen
About meAbout this blog
2012-04-23

Choosing an opensource licence

Last night I read that use of the GPL is declining, ever faster. Makes sense. And makes me want to braindump another minor aspect of Trolltech's history, with a bit of a comment.

Trolltech started using 100% closed licensing, then, when I suggested it a few days after being hired, gradually released Qt under ever more permissive licenses. Finally, years later, Qt was released under the GPL, which may or may not be viewed as more permissive than the previous license. My opinion is that users had a choice: The previous license or the GPL, so at least it wasn't necessarily a bad move. Confusing, perhaps. But confusion can be a feature, e.g. if it makes advokids quarrel among themselves.

Every time we changed Qt's licensing terms, we did so after much deliberation, and we chose such terms as we felt would benefit the general audience the most. We kept in mind (I did in particular, since I never had enough savings to pay even a single month's rent) that the audience for a programming toolkit needs maintainers. If we took one too-big risk, lost, and had to close Trolltech, Qt's users would be shot.

There was a popular theory at the time that someone would eventually step up and resume development of any open source abandonware. I personally didn't trust that. Besides I wanted to work on Qt, not run Trolltech into the ground and get a random job, then hope that someone else would take over Qt.

However. After that digression, to the real matter at hand. I think we were wrong.

I now believe that we should have picked a much, much more permissive license. I believe we should have picked the BSD, MIT or Apache licenses for Qt, early.

We had good arguments against. For instance, we believed (and our customers confirmed) that we had to give our advocates within potential customer companies a reason to buy Qt. When the manager asked can't we use the open source version? we thought the engineer needed a reason to say no, such as No, that doesn't allow us to do what we do.

For the past few years I've shifted to think that was wrong. It may have been right. Maybe a series of small risks, 3-400% yearly growth and precarious cashflow was right, but I now believe we should've taken the big risk. Just a feeling.

Someone at MySQL who should know told me that people didn't buy MySQL because the license terms forced them to. They bought to be friends with MySQL-the-company, or because the MySQL salespeople said it was the right thing (and it was much cheaper than Oracle anyway). They didn't buy support because they needed it, but rather because they wanted the option of support.

Something like that would've happened to Trolltech too.