Funding FOSS 2015-07-01
This past week I attended Open Source Bridge 2015 to speak about how the Internet works. A common topic both in the talks and the hallway track was how to improve the funding of open source software or, put more generally, how to ensure that the software ecosystem we enjoy today is sustainable. I wanted to summarize some of my feelings on the topic as this is something I care about deeply both personally and professionally.
What is a society?
Before I talk about FOSS, I want to take a bit of a detour. At its most basic, a society is many individuals working together for their collective benefit. As societies progressed from small family clans to agricultural collectives to city states and nations, we have seen an escalation of the Commons. Shared granaries meant no one farmer was at risk of dying from a random crop failure. Improvements in technology due to specialization allowed others to specialize themselves. In just a few hundred generations, collectivism has allowed us to go from nomadic apes to the vast civilization you see today.
What is the Commons?
The Commons is an abstraction of collectivism. There exists some pool of work, talent, and resources that everyone puts in to, and then everyone gets back out some benefit hopefully in excess of what they put in. In a modern context, an example would be taxes. Everyone (or almost everyone) pays in, and in return we all get roads to facilitate commerce, schools to ensure an educated population (and therefore workforce), etc.
Why does Open Source matter?
Open source software is a new, and very powerful, branch of the Commons. We have created an amazingly powerful collection of tools and techniques which now power the vast majority of new businesses. Even old holdouts like Microsoft have recently started to try to engage the open source community, to tap our willingness to build the Commons. I don’t think it is an overstatement to say that without our collective investment in this shared platform, we would not have nearly the same rate of technological development as we have today. Beyond the tools themselves, within a rounding error the Internet itself is a manifestation of all our communities. There are some places using Microsoft’s IIS here or Oracle’s database there, and many core implementations of things like BGP are proprietary, but the same community aesthetic powers groups like the IETF and W3C.
Where did it all go wrong?
In the early days of the open source movement there were relatively few projects and in general most people using a project were also contributing back to it in some way. Both of these have changed by likely uncountable orders of magnitude.
Contributing to a project like PostgreSQL or Python has a huge multiplier effect. So many people use these projects that only a small proportion of users need to contribute back to keep everything moving forward at an acceptable pace. Or to be more economical about it, the return on investment from a contribution was shockingly high. As we have moved to more and more niche tools, it becomes harder to justify the time investment to become a contributor. “Scratching your own itch” is still a powerful motivator, but that alone is difficult to build an ecosystem on.
The other problem is the growing imbalance between producers and consumers. In the past, these were roughly in balance. Everyone put time and effort in to the Commons and everyone reaped the benefits. These days, very few people put in that effort and the vast majority simply benefit from those that do. This imbalance has become so ingrained that for a company to re-pay (in either time or money) even a small fraction of the value they derive from the Commons is almost unthinkable.
Bonus Concern: Corporate communities and DevOps
Many of the projects we look to as examples of the best outcomes for open source were built as communities first. Even those that did come out of a company, like Django from the Lawrence Journal World, the community formed around the project and not the company. There is a long history of companies participating by either by donating resources or hiring contributors, especially in projects like the Linux kernel, but this has always been a relationship subordinate to the community as a whole. The operations software world has seen an explosion of single-company, often VC-backed projects where the company comes first and builds a community around itself. This has worked in some cases, but it is notable as a departure from the models we used to get here.
Where are we going
and why am I in this handbasket?
The tech sector and the glut of VC money that keeps it afloat is addicted to free (as in beer) software. It is flat-out assumed that a company can launch a product without paying a dime (again, either in time or money) in software costs. The biggest software expense for your average startup will be either buying SublimeText or their GitHub subscription. In and of itself, this reliance on everything being free at first is not a problem. The issue is that no one ever wants to pay the piper. Companies that succeed do not pay forward the “loan” they got from the Commons. We see lots of companies launching open source projects because they want to “give back”, but we’ve all felt that pang of dread when we realize the project is really a marketing stunt in disguise. Some of these projects are really beneficial to the community and I’m glad they exist, but even with that the balance between value derived from the Commons and value put in to it is woefully lopsided.
How do we make things better?
I have spent the last few years chasing the dragon of professional open source contribution, but it has been a hard battle. There are a few options available today for people that want to try them, and some that could work in the near-term future.
Kickstarter, Patreon, and other ad-hoc sponsorship
I did run a Kickstarter myself, and between that and an associated donation I’ve been able to fund the last six months of my work. That said, I really don’t recommend this option to others. Fund-raising (or advertising a Kickstarter campaign) is hugely emotionally draining, and the mental burden of “I’ve already sold these features to people” is considerable. Patreon is at least an ongoing donation, as opposed to a one-time cash infusion, but I’ve not yet seen anyone receive even close to a living wage on there or any similar service. Even if you do manage to raise a good amount of money, 10% immediately vanishes to fees. If you are far more outgoing and comfortable with self-promotion than I am, this might be a path for you.
Hiring contributors
Another common option is to try to find a company that works on open source tools and get a job there. If you don’t have ties to a specific community and just prefer working on open source code, this is probably viable given the number of companies with open source products either in part or in full. With specific communities this can be harder though, especially more niche tools. That said, some companies like Rackspace and HP have done great work hiring existing contributors for projects they want to see improved, and generally getting out of their way.
Patronage
This is very similar to hiring people, but I like the overall mental model of it better. When hiring an employee, there is an implicit assumption that you will do work as needed and directed by the company and the employee gives up most of their rights over the work in exchange for a paycheck. This is summarized by “work for hire”. I want to burn it down.
I think a more stable model for “I want to pay someone to make open source better” is that of Renaissance era patronage. It was considered socially “required” that a rich person or family (which is roughly comparable to a company in modern terms) would have one or more artists who they would support. These artists would make great art to show off to the patron’s friends and such. I would love to see a world where every successful tech company is simply expected to have a bunch of open source developers on staff not because it benefits the company’s products but just because it enriches the world. Stripe is a great example of this model, with their open source grant program. Finding others willing to put down the resources to be a patron has been difficult.
Foundations and grants
A solid model used in other arenas is big foundations collecting money and then handing out grants. We do have a lot of foundations in the open source world, and some have tried to run grant programs. The problem has generally been awareness and volume. The PSF grants program has gone mostly unknown, and most software foundations can’t fund more than a handful of full-time developers. This is a good step, but not enough to make a dent in the needs of the community.
This has been successful at small scales with mini-foundations like RubyTogether and Node Security. These mini-foundations generally find ongoing corporate sponsors and fund a small number of full time developers. Scaling up to either bigger teams or less niche tools is a stumbling block, but they are still a powerful story.
One avenue to improve this would be to hire professional fund-raisers. The non-profit and NGO world has lots of full time workers that focus on raising money for the foundation. As long as they bring in more than they cost, this is still a beneficial arrangement.
Grants by companies have also had some level of success. I am not a huge fan of Google’s Summer of Code or Highly Open Participation projects, but they do show the viability of corporate grants as a model. Bug bounties are also a small version of this, though rarely big enough to claim to actually support development overall.
National Endowment for Engineering
Many governments use tax money to support science and art. In the US we have the NEA, NSF, NIH, and more. These fund work to enrich society through grants and donations. I would love to see a similar structure in place for technology or engineering in general. Where these grant programs do exist, they are often impenetrable to non-university applicants so that would also be something to see improved.
This is probably the only option here that can make a difference in the long term. While I would love to see the others happen, I find it hard to imagine a world in which they are pervasive enough to continue the current rate of growth that has been powered by untold hours of unpaid labor. Unfortunately this solution is also the one I feel the least equipped to help move forward on.
Universal Basic Income
An end game solution: provide a basic standard of living so people that want to dedicate themselves to enriching society can do so without putting their own needs in jeopardy. UBI is, to put it mildly, a contentious issue. Small pilot programs in many places around the world have had promising results, but we don’t have any clear path to serious adoption right now. Beyond the logistical challenges, discussing any non-capitalist system is often politically unfeasible in most governments. My thoughts on a post-capitalism world are worth a whole post on their own, but this is something we should all keep in mind as we displace more and more jobs via automation.
What do we do now?
Unfortunately many of these options are unhelpful for one individual trying to find a way to work on software and communities they care about. Ad-hoc solutions like “spend one day a week on personal projects” do help, and management teams at companies successful enough to afford it can help to encourage contribution. In the end though, it is a bleak landscape right now. Open source burnout is being discussed more openly, which gives me hope that we are working towards reducing it, but for now the only real option is to try to take care of yourself and hope that some day all of this will be viable.
Many thanks to all those that reviewed drafts of this post: Donald Stufft, Steve Klabnik, Laurens Van Houtven, Jacob Kaplan-Moss, and Seth Vargo.