May 13, 2018

Commons Club open source as accessory, copyleft as network

For fellow software developers, today’s public hand-wringing about privacy, security, and user manipulation comes comically, pitiably late. The tech-savvy have long taken extra steps when using the big online platforms, from privacy settings to browser extensions to Tor. Those steps tell what insiders know about their industry.

But the tech-savvy also know that it’s a losing fight. We’re tracked, profiled, and hacked all the same. We know technology, but we might as well be plumbers, insurance brokers, or grandma as far as other devs’ apps are concerned. They’re about as opaque to us as to anyone else. Quite despite the rise of open source.

Open source software is winning. Software freedom—the power to be both developer and user of the applications that matter to you—is not. If there were ever a short case for a difference, there it is.

Commons

Many programmers believe that open source doesn’t belong in conversations about privacy, transparency, or end-user choice. They’re right. Open source chooses not to belong in those conversations. Free software takes a stand on user-facing issues like transparency, access to code, and modification. Apart from the right to be an end-user, open source takes a stand on user-facing issues if and only if the users happen to be programmers.

That makes open source something of a club. Within the club, there’s camaraderie, a more refined code of conduct, and benefits—the mythified software commons. Common practices inside the club, like forking, happen only rarely outside it. Common practices outside the club, like invasive ads, earn scorn and protest within it. Membership in the club is an exclusive privilege of programmers. Non-programmers benefit only by invitation, which usually costs: time, money, privacy, or the lot.

Accessory

That’s why open source wins today. It wins, dilutely, as accessory to proprietary software, which is winning bigger. It’s more efficient to build proprietary code—social networks, marketplaces, locked-in user communities—on open source freebies. But it’s more profitable, prevailing wisdom says, to deny users of those networks meaningful rights to set policy, guide development, or defect. As standard operating procedure, a profitable app or service creates its own laity to conspire against, for cash.

But isn’t all software these days really open source, or heading that way? After all, proprietary software is ever more made of open source, and even made like open source, as innersource. When we look at proprietary apps and services as a whole, the layer of closed, proprietary integration, application, and design code on top of the open, common stack often measures relatively shallow. At least technically speaking.

But that’s a distinction falling conveniently short of a user difference. I can improve Facebook’s front-end framework, making it easier to develop facebook.com, and drawing in more developers Facebook might like to hire. I can optimize LinkedIn’s database, making linkedin.com cheaper to run, and fewer tickets for LinkedIn staff to handle. But Facebook isn’t a front-end framework, and LinkedIn isn’t a database.

Not really. Not even close. Even if sloughing away all the open source leaves but a thin residue of proprietary code, that code implements the choices that meaningfully affect users, that define user’ relationships to the owner and each other. Patching that implementation, the network itself, is network-owner prerogative.

Network

Network-owner prerogative is a pervasive kind of power. Networks are value propositions, but the one making the proposition isn’t the one making most of the value. Rather, lots of valuable users make a beneficial network.

That collective benefit can overwhelm the cost of even an irksome or incompetent facilitator. So we’ll abide a creepy social network to socialize with the right people. And we’ll abide a shifty marketplace to buy the right product at the right price. We’ll even abide social networks and market makers who get creepier, shiftier, or simply bungle over time, as long as network effects stack up faster than the ambiance takes a dive.

Network effect is Valley Speak, not a hacker term. But software freedom advocates bid consciously for network effect, explicitly to benefit end-user rights, in 1989. Quoth GPLv1:

The license agreements of most software companies try to keep users at the mercy of those companies. By contrast, our General Public License is intended to guarantee your freedom to share and change free software—to make sure the software is free for all its users.

In specific domains, newer GPLs still reign. It’s easier to port GCC to your new architecture than to write a compiler from scratch, even if you think RMS is nuts. It’s easier to tweak WordPress than to seed you own blog-platform community, even if copyleft gives you migraines. The freedom-indifferent abide GPL politics and GPL legal for free code and documentation all the time. Even as old hands get nuttier and license combinatorics explode. Even though giving users code and a license meaningfully shifts the balance power.

That is exactly the promise of network power. A free software network affects not just the software that freedom advocates write or choose to use, but software that anyone writes. The cost, if any, is a set of structural constraints that enforce transparency, adaptability, and the ability to secede from a network and start one’s own if the mad monarch of your current network becomes insufferable. But we’ll suffer a lot to avoid disturbance to our networks.

Past

Copyleft-based software freedom was a sound network play, and it achieved significant network-level results. It’s also succumbing to the signature crumbling decadence of networks: backward-looking hubris. When network participants mire down in reverence for past accomplishments, secure in the belief that their network is too big now not to matter forever, we dub them incumbents. And we wait.

It’s famously easy to topple unwary incumbents in software. Opportunity costs are relatively low, and with patents again in decline, it’s often possible to clone, reimplement, and otherwise run with what you find…or sit around until someone else does. That dynamic bites software networks of all kinds, both open and proprietary. Techniques, standards, conventions, and usability lessons from GCC or Linux end up not just in proprietary compilers and kernels, but also permissively licensed, freedom-indifferent alternatives like LLVM and Fuschia. Entire toolchains get rebuilt likewise, as with Go.

Figuring out what to code is damn hard. Rebuilding what’s been done and proven, slightly better this time, is the norm.

Present

That dynamic partially explains why pillars of the copyleft network feel ever more like legacy systems. That’s not to say stagnant or doomed. Far from it: GPL software remains much of the most mature and battle-tested, with the broadest supporting constituencies. But the content of LICENSE no longer seems to define those projects or their successes, if it ever did. It dates them.

A combination of factors conspire against vanguard spirit in the free software movement. For one, the licenses don’t seem to evolve with the times. The application service provider loophole, which allowed creators of proprietary software services to avoid the requirement to share source code under copyleft licenses, especially GPLv2, represented a practical vulnerability. FSF patched it in AGPL, but failed to merge into mainline GPL, in part due to threats to fork development of existing GPLv2 projects. We’re left with GPLv3, which suffers from a known, unpatchable software freedom vulnerability, and AGPLv3, a separate license under a distinct name that can be identified, isolated, scorned, and written off as an extremist excursion, rather than a necessary adaptation.

Meanwhile, the ASP business model, rebranded software as a service, exploded again. As did proprietary use of software development tools, which slide through a loophole no GPL has ever tried to close. Software developers see these dynamics, but lack a public license tool with which to respond.

Licenses without teeth—and a lot of unhelpful complexity—don’t make or keep strong reputations. In time, that perception ferments into a widely held, ahistorical belief that licensing can’t effectively express or protect software freedom values. Many active developers feel not just that open source doesn’t belong in conversations about software freedom, but that copyleft or even licensing in general don’t really belong. No wonder. They’ve never seen them coupled with efficacy.

Future

Over time, we should expect a repeating pattern of permissively licensed software, funded or blessed by industry, cloning or replacing software first proven by developers more devoted to software freedom. Copyleft licenses impose some additional friction over permissive licenses, by their nature. When a concept is proven, demand for similar functionality under lower-friction terms increases, and lower-friction terms themselves bestow marginal competitive advantage.

The free software movement was born cloning proprietary software, and refining its clones. When permissive software can as easily clone copyleft software, the software freedom network can’t expect to survive with vitality in a permanent defensive position. Rather, software freedom has to stay at the front lines, pushing the state of the art, and offering developers a compelling proposition: enjoy the state of the art, on an ongoing basis, but respect user rights as you do.

Part of that project is making sure software freedom’s tools remain state of the art. That means reassessing licenses and norms about sharing and consuming code with others. The License Zero Reciprocal Public License closes loopholes, reduces legalese, and reflects what we’ve learned about public licensing in the last ten years. Dual licensing through licensezero.com with L0‑R replaces prevailing, permissive norms with infrastructure for sustainable development.

Copyleft, permissively licensed, and proprietary software overlap, and therefore compete with one another. It’s equally important to acknowledge and account for that natural competition. By encouraging those advancing the state of the art to stand up for user rights in their licensing choices. By building out infrastructure, support networks, and methods to make free software programs and services viable competitors to both permissive and proprietary alternatives.

Free software values and objectives haven’t changed. With time, experience, and changes in the industry, we’ve come to understand them better. The methods of expressing those values, and supporting those objectives, have to evolve with our understanding.

Software freedom is a movement. Movements have to move.