[License-review] Approval: Server Side Public License, Version 1 (SSPL v1)

[Kyle Mitchell]

Copyleft has always been dual-use technology.  Any
sufficiently general copyleft license that activists write
to advance software freedom can be repurposed for getting
patches back, excluding competitors, or dual licensing, as
long as the default operating mode of the end-user software
industry remains nonfree.  Which it decidedly does, despite
three decades of software freedom activism.

I don't support bullying with unwelcome nicknames.  Thirty
years of software freedom activism show that doesn't work,
either.  But I've written elsewhere that SSPL's mix of
permissiveness and strong copyleft appeals exclusively to
the business or competitive constituencies for permissive
and copyleft terms.  SSPLv1 doesn't align with any activist
position, either anti-IP or pro-software freedom.

https://writing.kemitchell.com/2018/11/04/Copyleft-Bust-Up.html

That's why I think Eliot's question about how many and which
communities have to want a license was key.  If business
users on all sides want this, but activists on all sides
don't, is that still enough?

When software freedom activists and upstart businesses want
a stronger copyleft license, but anti-IP activists and
established businesses favoring permissive licenses don't,
that is evidently enough.  See AGPLv3.  When anti-IP
activists and established businesses approve of an even more
stripped-down permissive license, but software freedom
activists and upstart businesses don't, that is evidently
enough.  See FPL/0BSD.

We've seen business-oriented licenses without activist
support.  They've been approved.  Consider the business
reciprocal licenses, with stronger reaches and triggers,
circa 2002.  Consider the business hybrids of permissive
with intentionally weak copyleft, in MPL and CPL/EPL.  I
don't think we've seen a business hybrid of selective
permissive with _strengthened_ copyleft until SSPLv1.

Without reopening the advisability of L0-R, or if you
prefer, assuming away its controversial dev-tool trigger,
I'd offer it for contrast.  L0-R's strengthened copyleft
wasn't hybridized with permissive terms for any particular
development use cases.  It was consistently strong.  As a
result, L0-R was reusable for hardline activist purposes,
for those who reject AGPLv3's loopholes and compromises, and
not just to exclude competitors, or for dual licensing.  In
other words, it could make perfect sense as a license choice
for a solo developer without any business plan or
affiliation.  Activists can repurpose sufficiently general
maximaleft licenses from business, just like business
repurposes sufficiently general activist copyleft licenses.

By contrast, an inconsistent or hybrid copyleft license
that's actually permissive for many use cases, and draws the
line at serviceization specifically, doesn't map to any
activist agenda I'm aware of.  Even if we take as a given
that some software freedom activists oppose SaaS on ethical
grounds---"service as a software substitute" didn't stop
that train, either---and want to exclude its providers from
their work, they'd rather use copyleft terms that apply to
distributed and network-service applications, too.

As for freedom, to quote What is free software?:

  [C]opyleft (very simply stated) is the rule that when
  redistributing the program, you cannot add restrictions to
  deny other people the central freedoms.  This rule does
  not conflict with the central freedoms; rather it protects
  them.

Copyleft has always limited freedom 3, by requiring exercise
of that freedom be consistent with the freedom of others.

The Free Software Definition says more about copyleft design
than OSD does.  It implies that copyleft should trigger as
the GPLs do, on exercise of freedom 3.  But the reason it
gives---that the limit protects software freedom
overall---isn't specific to freedom 3.  I can only make
sense of AGPLv3 by reading the rationale as invited, to
extend to other freedoms, too.

That's why I question FSF's position on private changes.
That seems like compromise, not principle.  We can do
copyleft on freedom 3.  That's GPL.  We can do copyleft on
freedom 0.  That's AGPLv3, whether we want to admit it or
not.  Why can't we do copyleft on freedom 1?

For those who want all the rabbit holes, I've blogged my
thoughts and questions on the strength of copyleft in the
frame of FSF writings here:

https://blog.licensezero.com/2018/09/14/free-to-take-freedom.html

-- 
Kyle Mitchell, attorney // Oakland // (510) 712 - 0933