The release of the Commons Clause in August 2018 created a flurry of debate and media attention. Now that the dust has settled, here is a recap of how Commons Clause served as a bellwether of a licensing trend. The Commons Clause was an early iteration in a new wave of licensing called “source available” — an awkward term, but probably the most accurate and broadly used term at this time.
State of the Art
Here are examples of source available licenses being used today.(FN1)
In an effort to help standardize this category of software, in 2019 we started a project called the PolyForm Project, which offers a menu of peer-reviewed, publicly developed, plain language forms for the most popular kinds of license scopes, with limitations such as non-commercial, free trial, and non-competition. The first tranche of licenses was released over the course of the last year, and projects have begun to adopt them.
So in sum, this category — its awkward name notwithstanding — is alive and growing as we start into the 2020s.
Is this Open Source? No.
Much has been made about putative confusion created by the “source available” category. But source available licenses have been around for a long time; there just wasn’t a recognized term for them. People mainly just called them source code licenses. Any confusion with open source, such as it is, goes beyond the category of source available, and comes from the gulf between a layman’s view of open source, and a specialist’s view. That gulf has been around for decades.
For those who specialize in software licensing, “open source” has a specific meaning: meeting the Open Source Definition. That definition, as demonstrated by months of controversy over MongoDB’s SSPL,(FN1) is becoming ever more like a holy text than a working definition. It has not been updated in over 20 years, and the advent of new trends in computing, like cloud computing, have made it harder for even the most ardent pundits to interpret. (FN3) But even so, it is still concrete enough to distinguish between open source and source available.
The gulf in meaning is driven mainly by two qualities shared by both categories: whether the license is generally available at no royalty, and whether it is deployed in a frictionless manner.
What is a Scope Restriction?
Commons Clause is an example of a license scope restriction. It was created as a “portmanteau” approach to apply a restriction side by side with an underlying open source software license grant (of the licensor’s choice). The Open Source Definition does not allow scope restrictions (FN4), so two sets of terms taken together in this way are not open source. Similarly, licenses like the Confluent Community License and the PolyForm licenses — and Creative Commons Non-Commercial — all have license scope restrictions. But unlike Commons Clause, they all were written from scratch, without incorporating the terms of an existing open source license.
The easiest way to understand a scope restriction is to look at the sister Free Software Definition, promulgated by the Free Software Foundation. It says, in an elegant and charmingly-numbered fashion.
A program is free software if the program’s users have the four essential freedoms: The freedom to run the program as you wish, for any purpose….“Freedom Zero”
How could a license violate this freedom? Let me count the ways: field of use restrictions, user restrictions, server restrictions, enterprise restrictions, CPU restrictions, site restrictions, and all the other metrics that commercial software vendors use to tot up their license fees. Open source is blissfully free of not only these restrictions, but the compliance tooling necessary to avoid violating them.
Engineers and businesspersons frequently ask if there is an open source license that only allows non-commercial use, or doesn’t allow SaaS use, and the answers is no. Those are license restrictions, so they can’t be open source.
Free as in Free Beer
But the lay use of open source is sometimes broadened to mean any license that is free of charge. Anyone who has worked in the field of software licensing has seen this at work for many years. These days, in most financing, M&A, or even commercial software licensing deals, companies must disclose which open source software they are using. Almost every company answers this disclosure request by including everything it downloaded from the web for free. In fact, there is a lot of free-of-charge software licensed only in binary form that is not open source, or even source available. That category of software is often called freeware, and includes products like the Adobe PDF reader. But as Richard Stallman famously said, free software means free as in free speech, not as in free beer.
Sometime lay persons call software “open source” for a reason that has little to do with the actual license terms applied to the software. For example, I have heard many engineers call anything with a click-to-accept mechanism a “EULA.” That’s not accurate either, but it’s instructive to consider the root of both these misconceptions. Open source and proprietary software licenses are not only different in content, they use different delivery mechanisms.
One aspect of open source licensing that is often given short shrift is its lack of a click-to-accept mechanism. The Open Source Definition requires:
Distribution of License. The rights attached to the program must apply to all to whom the program is redistributed without the need for execution of an additional license by those parties.OSD Section 7.
In contrast, most end user licenses require each licensee to execute an assent mechanism, which is a way for the licensee to indicate its agreement with the terms of the license and form a license contract.
The reason limited licenses and open source licenses have historically worked differently is rooted in some slightly arcane software licensing law. As GPL2 says,
You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it.GPL2 , Section 5
And this is the nutmeat of the legal issue. A license is a take-or-or-leave-it proposition. Either you accept the license, or you don’t. If you don’t, you can’t exercise the rights. If you do, you have to accept all the terms of the license. Note that the implicit acceptance of the license is triggered by “modifying or distributing” but not use. GPL2 section zero says that “Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted.” It is no coincidence that under US law, mere use of a program does not require a license. US copyright law says:
[I]t is not an infringement for the owner of a copy of a computer program to make or authorize the making of another copy or adaptation of that computer program provided:17 USC 117(a)
(1)that such a new copy or adaptation is created as an essential step in the utilization of the computer program in conjunction with a machine and that it is used in no other manner…
So, mere use does not require a license. Any attempt to limit use of a program, as opposed to copying, modification or distribution, would require the licensee to bargain away the right to take actions that do not require a license. That must be done by agreeing to a contract. Therefore, it has long been considered a best practice for commercial software vendors to restrict use with a click-to-accept mechanism. And that is why lay persons equate the click-to-accept mechanism with an end user license agreement, or EULA. EULAs are the most common kind of limited license, and those with which every user is familiar.
But today, deploying software is no longer a question of making one copy for use on one machine, which is what 17 USC 117 allows absent a license. Using software at scale in our cloud and container world requires making many copies. Section 117 was written to enable single-machine use. It it not carte blanche to deploy thousands of containers. And that’s why source available licensing works in “frictionless” style, when traditional EULAs don’t. Source available licenses are usually not mere end user licenses. They often grant rights to modify, redistribute, and copy at scale.
The State of Play Today
Source available is an emerging category of limited licenses, but it is probably stealing more territory from binary licensing than open source licensing. Open source and limited licenses have co-existed for many years. To understand how this works, we need to understand a few things about how the software business works.
- Platform software. The trend toward source available is most prevalent among the kinds of software sometimes called platform software or middleware. It is no accident that so many of the splashy adoptions of source available licenses in 2019 took place in the database sector. Applications have always been dominated by proprietary licensing, and infrastructure has become dominated by open source. That’s because software in ubiquitous use, like infrastructure software, is usually most valuable when is it standardized, and the cost to develop it is shared by the industry. For infrastructure like the Linux kernel or Kubernetes, plenty of industry players are willing to collaborate to maintain this infrastructure, so they can leverage it for their businesses. Specialized applications at the top of the stack, in contrast, usually have a smaller audience, and need private funding to develop. Platform software is somewhere in between: more like applications in monetization, and more like open source in the need to deliver source code to enable compatibility and security.
- Source code delivery expectations. One side effect of the success of open source software over the last 30 years is that users now strongly prefer to receive source code from vendors, and exercise their bargaining power to demand it. For a long time, there was a tension between a vendor’s unwillingness to provide source code, and a user’s desire to get it. During the 1990s, commercial vendors tended to deliver binary only software, backed up by a mechanism called a software escrow, from which the source code was released if the supplier failed to maintain the software: for example, if the vendor went bankrupt or retired the product. This was always an awkward model, and escrows were difficult to negotiate and even more difficult to use if the code was actually released. Today, source available licensing acknowledges the need to deliver source code, even for proprietary products.
- Software security. Security of software has become a huge issue, and most technologists agree that security by obscurity doesn’t work. Transparent software is more secure, and most source-available licenses grant the right to modify, so licensees can fix security flaws quickly on their own, if necessary. So many customers care more about transparency of software than the right to modify or use it free of charge.
- The true landscape is a mix. Source available licenses are often used as part of an overall model that includes some mix of open source, source-available, binary-only and SaaS parts. SaaS is not a license at all, and users get no access to the software. As computing moves to the cloud, source available elements provide what customers need in source code form, even if they are commercial products.
More About the Commons Clause
The Commons Clause was adopted by Redis Labs, Dgraph, Neo4j, and others in 2018 and early 2019. Today, it is not used much in the wild, but it did serve as a catalyst for discussion and adoption of source-available licenses. So, this section is mostly for historical interest.
An FAQ about the clause is here. The Clause has been mis-characterized as a “non-commercial restriction” — in fact, it is a more lenient restriction, unlike CC-NC, or PolyForm Non-Commercial, which are in fact non-commercial restrictions.
In the highly political world of open source, the media responses to the Commons Clause were polarized. Here are some examples to give you a flavor of the controversy.
Note: I coordinated the drafting of Commons Clause and advised several clients on adopting it, and drafting forms of source available licenses, so I have kept the information here as neutral as possible while communicating the facts of this trend. Any opinions expressed in this blog post or in the articles cited here should not be attributed to my clients, and I have not written this article on behalf of any client.
(FN1) Many people discussing this trend put MongoDB in this category, but technically, it is not. It adopted the Server Side Public License, a modified form of Affero GPL3 with enhanced source code sharing requirements, which was the subject of a controversial submission to the Open Source Initiative. The SSPL does not contain license scope restrictions.
(FN2) Creative Commons Licenses are not software licenses; they are general copyright licenses, mostly applied to content like images, music, and video. But the lack of a suitable non-commercial software license caused many companies to adopt Creative Commons Non-Commerial for software. Let’s hope PolyForm Non-Commercial gives them a better option.
(FN3) At this point, the OSD runs the risk of devolving to an “I know it when I see it” standard; indeed, OSI has added a requirement for approval. The license must not only meet the OSD but also “guarantee software freedom” — without guidance on what that might mean.
(FN4) The planks of the Open Source Definition that prevent license scope restrictions are less pithy than Freedom Zero, but they make the point clear nonetheless. They are:
- 5. No Discrimination Against Persons or Groups
- 6. No Discrimination Against Fields of Endeavor
- 8. License Must Not Be Specific to a Product
- 10. License Must Be Technology-Neutral