Farewell open source

A year ago today we found ourselves in San Francisco proudly accepting an Elastic Search Award.

A year ago today, COVID-19 had not yet been declared a global pandemic, and the Elastic Stack was open source. Today, the Elastic Stack is no longer open source, and we’re in the second wave of a global pandemic. Reflecting on the past year we ask:

How is it that we were able to double down on open source (software and hardware!) during a global crisis while some organisations instead changed course?

Organisations?   As in more than one?   Yes.   Here we are referring to Oxford University reversing its decision to open source their coronavirus vaccine. And Elastic’s license change which was announced in January of this year.

What’s so special about open source, you may ask?

In a word, community. At Elastic{ON} 2020 we were impressed by many things, but none more than the strong sense of community among Elastic’s diverse and distributed workforce and network of contributors, partners and investors. A project on the scale of the Elastic Stack inspires purposeful contribution, not just to the code base, but to the project as a whole, from the community and, critically, for the community.

Open source is about distribution terms (i.e. implications for the community), not just access to code. Hence the OSD: Open Source Definition. And that’s where Elastic’s recent license change establishes terms outside of this definition, with resulting implications for the community, even if the code remains as accessible as before.

Imagine now the Oxford vaccine being distributed as open source, as originally intended. Imagine the strong sense of global community this would have instilled, let alone the purposeful contributions it would have received to facilitate both distribution (130 countries still had not administered a single vaccination around the time of writing, according to UNICEF!) and continuous improvement as new challenges and strains of the virus emerge.

Why the departure from open source, you may ask?

In a word, pressure. We invite the reader to inform themselves and draw their own conclusions as to the source(s) of pressure for both Oxford and Elastic to reverse course on open source. And, critically, to identify what prevented each organisation’s leadership from successfully resisting that pressure?

Indeed we too received pressure against open sourcing Pareto Anywhere and its precursors. This past year, our decisions to open source DirAct and our ca-va-bracelet open hardware collaboration were strongly contested. How were we able to successfully resist? The ownership structure of our purpose-driven organisation affords us—and only us—the final say.

Will we continue to use the Elastic Stack?   Yes, of course. It remains an outstanding example of open source collaboration and we remain proud of our award and contributions! Let’s look back again in a year to see what impact the recent license change will have had for the community.

Will people still get vaccinated?   Yes, of course. The global vaccine effort has been unprecedented even if the precedent of proprietariness remains unbroken. Let’s look back again in a year to see what impact this approach will have had on immunisation rates for the global community.

Finally, were we ourselves A Fool to Open Source? We mused tongue-in-cheek about the subject in our traditional April Fool’s blog post five years ago. As we hope we’ve shown, there is clearly no consensus—even in the software industry where the open source movement began decades ago. For those who bade farewell to open source this past year, we trust that their communities—of which we are a part—will continue to fare well. We nevertheless maintain our philosophy that open source is what’s best for the community.

Open hardware: a closed case?

If you’re familiar with our blog posts, you’ll know we never get bored populating them with puns (oof, there they are already). In case it isn’t obvious, the title is playing on the word case: we’ll be talking about enclosures here. Specifically,

why is open hardware so seldom designed to take advantage of an off-the-shelf enclosure?

In other words, why invest the time and energy to develop a hardware board that doesn’t fit in any standard box? For those who are less familiar with hardware development, one of the first steps in designing the physical board (PCB) on which the circuits and components reside is to establish its dimensions, shape and the location of the holes with which it can be mounted into an enclosure. At this point, it would be straightforward to choose a commercial off-the-shelf (COTS) enclosure that closely matches the target dimensions, and adjust these and the locations of the mounting holes in consequence.

Why does the board need to fit in a standard enclosure? Again, for those less familiar with hardware development, developing and producing a custom enclosure typically runs in the tens or hundreds of thousands of dollars. A COTS enclosure might cost $5 and will be available in small quantities. The economics speak for themselves, especially when many projects based on open hardware require only tens or hundreds of completed units.

Case in point (pun intended, of course), the first generation of hardware we built at reelyActive in 2012: both our active RFID tag and reelceiver were designed around off-the-shelf enclosures from OKW Enclosures (MINITEC) and New Age Enclosures (3 inch Dongle) respectively. These were economically sound decisions as much then as they are today with all the units we’ve produced to date. And our customers enjoy the benefit and freedom of customising the enclosures (colour, form factor, etc.) either with our help or directly with the enclosure vendor.

And therein lies the philosophical issue with open hardware that lacks a standard enclosure: if one of the principal objectives of the “open” movement is freedom from lock-in, this practice doesn’t sit well (pun very much intended). One may argue that open hardware offers developers the freedom to adapt the design to a standard enclosure, however this requires specialised skills and iterations that significantly extend the timeline. And, if there are any radios onboard, as is typically the case with IoT designs, any such adaptation may engender costly, risky and time consuming re-certification.

Designing open hardware around COTS enclosures is both economically and philosophically sound.

The motivation for this post stems from our disappointment in a recent survey of hardware platforms (both open and closed) on which to develop our own next-generation hardware. While there are plenty of options available, we were unable to identify a single one that streamlined rapid prototyping out-of-the-box, or “into-the-box” as is surely more apt. So, eight years after developing our first “into-the-box” prototypes it seems this task is still up to us—open hardware is far from a closed case. We promise to make our design available to the community and hope it encourages wider adoption of “into-the-box” best practices.

A Fool to Open Source

This week, while paying a visit to one of our Fortune 500 clients, they asked us why we open sourced our software. “Anyone could just copy your work” they said. “Aren’t you afraid that someone steals your business from you?” Wow. We hadn’t thought of that. We just figured we were following the Lean Startup guidelines by sticking to the free tier of GitHub. It costs $25 per month to hide your code in private repositories you know! But then we looked into the gravity of our mistake and here’s what we found:

Other companies can create value on top of our platform. For themselves! How selfish!

We were feeling quite confident that we had the $19 trillion Internet of Things market opportunity all to ourselves given our early mover status. After all, we announced our IoT pivot a week before Cisco, the company responsible for that claim. But it turns out that other companies could leverage our open source platform to “create” additional value. For instance, a third party could focus on a specific vertical outside of our expertise, and develop a useful targeted product with our code base as a foundation. That supposed additional value would in effect be a stolen slice of our $19T pie! Unless, of course, they’re naïve enough to open source too, allowing us to reciprocate and rightfully reclaim our slice!

Collaborators could contaminate our code base under the auspices of free contributions!

We also felt confident that, in our conquest of the global IoT market, we could maintain a clean, pure code base dignified of such an endeavour. But then, to our chagrin, individuals outside our organisation insisted on collaborating and contributing to our open source code. Sure, they claimed to be “fixing bugs” or “adding features” but we all know that can’t be true: why would anyone work without pay? Their intentions could only be malicious. Surely they must be saboteurs seeking to steal from our pie! After all, look at what open collaboration has done to the Linux operating system. Yikes! Is anybody still using that?

The community could continue to use our software even should our organisation perish!

Okay, we’re still having a hard time wrapping our heads around this one. It’s bad enough that we have to share our pie with selfish third-parties and “collabo-raiders”. Now imagine that they eat the whole $19T, causing us to starve and perish as an organisation! Apparently our open source code base would continue to live on indefinitely! How unjust! This surely explains why so many clients and partners have been keen to adopt our platform: we’d have absolutely no recourse to childishly revoke our work out of spite! In fact, we are relegated to subsist but from the meager rations of pie that they dole out each month in exchange for our open source software as-a-service. What kind of pathetic business model is that?!?

We learned a tough lesson this week. By accidentally open sourcing our software, it has become nothing more than a platform bastardised by open collaboration and trampled by an influx of clients and partners. If only we had the prescience to patent a proprietary code base. Imagine how much further ahead we, and the IoT, would be today.