Experience level

Learner

Session Track

Humans

In free software circles, a disturbing number of conversations treat

it as given that fewer new projects are choosing copyleft licenses,

and more are going the route of lax permissive. Though this has been

repeated in news articles and blog posts, when we look deeper for

evidence of the claim, we find either anecdotes (often from the field

of corporate-backed project license choices), or highly questionable

and unscientific data sets.

Sometimes the claim is repeated by people just trying to set the table

for other discussions about trends in free software or open source;

sometimes it is promoted by people specifically arguing against

copyleft (especially in business); sometimes it is raised as a concern

by people defending copyleft. All three situations are worrisome in

their own ways.

I'll discuss:

1) recent "data" and articles published, including my own analysis;

2) whether the contexts in which permissive license use does appear to

be increasing mean what they are purported to mean, and whether those

contexts are actually relying purely on permissive licenses or on

other sorts of supplementary legal structures;

3) prominent instances of copyleft usage, especially the AGPL and GPL;

and

4) different ways to study and understand the trends people are trying

to get at when they talk about licensing choices in quantitative

terms.

In a community with so many computer scientists, lawyers, and talented

researchers, it's important that we habitually put popular assumptions

up for critique. Assumptions and bits of conventional wisdom need to

be audited just like code, to ensure that the business plans and

movement-organizing decisions which flow from them aren't based on

rotten foundations.