OSHdata is an independent project that launched in 2020, starting by taking a look back and generating a static report about the state of the Open Source Hardware (OSH) ahead of the 10th Annual Open Hardware Summit in March 2020. OSHdata’s findings are for the community. For the founders, engineers, developers, artists, customers, suppliers, and all the other stakeholders who make this community what it is.
This report is authored by Harris Kenny and Steven Abadie, two members of the Open Source Hardware community who have contributed to the certification of dozens of products over the past five years. Today, we proudly work with OSH companies in our businesses. Combined, we have helped sell tens of thousands of Open Source Hardware products for tens of millions of dollars. We are passionate about this community and we are creating this report to help grow it.
We envision five different use cases for this report (in alphabetical order):
- Certified a product
- Considering certifying
- OSH customer
Based on the people who have expressed interest in OSHdata so far by signing up for the newsletter below, or following @OSHdata on Twitter, we have seen an even distribution across all five groups. Some of the largest businesses in the OSH community have raised their hands to express interest in this project. Remember, this report is the beginning of what we hope will be an ongoing resource for the community. Where things go from here depends on you and your feedback.
Sign up for the OSHdata email newsletter:
Table of Contents
- License Selection
- Growth and Adoption
- Next Steps
OSHdata findings show that the Open Source Hardware (OSH) community is dynamic, growing, and still in its early days as a formal movement. There are over 400 certified projects from 36 countries spanning five continents. The certification rate is increasing, too. Getting from 200 to 300 certified projects took nearly a year, but getting from 300 to 400 took a little over six months. Nearly 60% of the certified projects are available for sale at an average sale price of $211.47, though there is a big range here. The leading project categories include: electronics, 3D printing, tools, and education. Our research also finds opportunities for improvement in the certification program itself, particularly the application process and practices by creators.
There are ambiguities in the application process that are evident in the data over time. For example, creators are using the website and documentation URL fields in a variety of ways. This makes it difficult to know what to expect when looking between projects. The set of options presented for Project Type vary between methods of production, industry, and use case, making segmented analysis difficult. There is also not a standard established as it relates to version numbers, making it difficult to compare evolution or advancement between projects. This report will explore these shortcomings and also provide additional context thanks to research conducted by the OSHdata team around commercial availability, product pricing, and more. In the Road Map, we outline what comes next.
Before getting into the specifics, first a historical retrospective on all the work that made these insights possible in the first place. The Open Source Hardware Association (OSHWA) was founded in 2010 when Alicia Gibb and Ayah Bdeir hosted the first Open Hardware Summit. Shortly thereafter, they led the codification of the Open Source Hardware definition. The Open Source Hardware Certification was developed by OSHWA with extensive community input and introduced in September 2015. The first products and projects were officially certified in October 2016. You can find a full list of certifications here. Special thanks to OSHWA for creating the certification program and making the data publicly available. Thanks also to all the creators who have taken the effort to certify their products. Without these combined efforts, this project would be substantially more difficult, if not impossible. And finally, thanks to the peer reviewers who took time to read and share their feedback on this report.
The OSHWA certification data is well-organized, so we were able to create a relatively simple web scraper with Beautiful Soup to pull that data from the website into a spreadsheet. We scraped data on February 7th, 2020, note that there have been more recent projects certified since then. You can try this for yourself using this script in the OSHdata git repository.
To calculate pricing and commercial availability, the OSHdata team individually reviewed more than 400 certified projects and looked for product pages or other ways to purchase the certified items and standardized the currency at the time of accessing those pages. This analysis only shows what is available for sale, it does not show the revenue, profit, or unit sales for certified projects.
From there, we combed through the data and created a list of questions we wanted to explore, based on the different potential use cases that we identified. We have a list of questions that are answerable with current data, with a stretch effort, and then future goals. We then compiled tables and charts, converted them using Vega, and built the site using GatsbyJS and TailwindCSS.
This process was only between the co-authors (Steven Abadie and Harris Kenny). If others are interested, we could see having supporters of this project who help guide our research and suggest topics that they’d like us to look into. If you have ideas, let us know.
OSHdata 2020 Report is shared under a Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA 4.0 International). This license covers the entire report, including the text and graphics. OSHdata is a project of Kenny Consulting Group, LLC.
While there are many motivations for individuals and companies to participate in the Open Source Hardware community, it is clear that there is vibrant commercial market activity in this space. Of the over 400 certified projects, nearly sixty percent of all participating creators had their products available for sale. These products are available at an average sale price of $211.47. The average sale price does not tell the whole story, however.
Starting on the low-end, nearly half of the products listed have a base price of under $35. This includes the types of things you might typically associate with the Open Source Hardware community, from sensors to 3D printed parts and USB accessories.
A little more than one-fifth of the projects have a list price ranging from $35 to $100. These projects tend to be accessories, single board computers, and breakout boards.
There is also a healthy market for products ranging from $100 to $1,000, as these represent nearly one-fifth of the projects commercially available with a list price. These are typically sophisticated embedded devices and testing or diagnostic tools.
On the high-end of the price spectrum, nearly one in ten of the projects have a base list price of over $1,000, topping out at a maximum base price of $4,950. Several have configuration and accessory options that can bring the price of an Open Source Hardware product even higher. There are instances of configuration options that exceed $50,000. These projects show that high-end equipment and machinery manufacturers are participating in the Open Source Hardware community. These products are particularly oriented towards research and enterprise workflows.
There were a few dozen projects that did not have a list price available but did publish a way to get in touch with their team to discuss pricing and purchase their product. These unpriced products are not included in this analysis.
Licensing is a never-ending conversation in the open source community writ-large. Open Source Hardware is no different. This question identified what we believe to be one of the biggest shortcomings of the current certification application. It is our recommendation that OSHWA consider adjusting the options available or add additional explanation or suggestion to certifying individuals and organizations.
Hardware licenses are dominated by the Other selection, representing approximately three-quarters of the certified product licenses. The other two options available are CERN (nearly two-fifths) and finally TAPR with approximately two and a half percent. No other category had such a high percentage of Other selected.
Our direct experience certifying products has been that those organizations selected either a GPL or MIT license for hardware designs, thus selecting Other during the certification process. We heard from one engineer at a certifying organization that they opted for Creative Commons licenses. This question is complicated and it is unclear what’s going on here. As OSHWA articulates in their FAQ, CERN and TAPR licenses are specifically designed for hardware.
Are people mistakenly selecting other licenses, not knowing this distinction? Is there a disagreement over the scope of other licenses and belief that they may in fact apply to hardware? Are there other hardware-specific licenses (beyond CERN and TAPR) that are not captured? This is a separate conversation worth having. For the purposes of this report, we found it noteworthy. This uncertainty lessens as we move into software and documentation.
Software licenses are led by GPL (including LGPL), which constitutes a plurality of the dataset. Permissive licenses (Apache and MIT) are selected half as often as Copyleft. It is unclear what the approximately quarter of projects are that are using other licenses, including possibly proprietary software. With the abundance of software licenses out there, there are many options for certifying individuals and organizations to choose from. Interestingly, nearly a quarter of projects involve no software whatsoever.
Documentation options are streamlined to include three of the license options from Creative Commons: CC BY-SA, CC BY, and CC 0. However, with nearly one-third of the licenses being Other, it is interesting to think what is being chosen here. While GNU licenses are available in other domains, the GNU Free Documentation License is not included here. It is also possible that conventional copyright is applied to documentation for some projects or that perhaps there is no documentation whatsoever. Neither of these are presented as options, so those creators would presumably default to Other in this instance.
Additional Licensing Questions
There are deeper questions not covered here, like why are creators choosing these licenses? Creators are also taken at their word when submitting their license selection that the claimed license is indeed in-use. Because the OSHWA certification relies on self-reporting with an enforcement mechanism, it is possible that these licenses have changed over time. For more on this, read the certification License Agreement. Hardware also faces unique licensing challenges because of the upstream nature of components from various suppliers that hardware companies may or may not be able to make available–think a processor on a single board computer. While consensus is continuously evolving, we believe that conversations around licensing will continue to be central to this community. You can get updates from what we are observing by signing up for the newsletter below or following @OSHdata on Twitter.
Get email updates from OSHdata:
Growth and Adoption
One of the most exciting elements of this report is seeing the accelerated growth in the number of certified projects. In year one, there was a predictable surge where getting from 0 to 100 certifications took only 289 days. There is a policy-driver behind this, beyond the initial excitement about the creation of the program. Knowing that there would be a lot of interest in getting certification number one for a given country, OSHWA instituted a drawing system.
Every company interested in certifying a project submitted theirs ahead of time. Then a winner was drawn from that group. The more entries that a company submitted, the higher their odds of getting certification number one for their given country. This led to a one-time spike in participation at the outset. After this initial spike settled down, it took over one year (388 days) to get from certifications 100 to 200. Getting from 200 to 300 took under a year (339 days) and finally from 300 to 400 only 197 days.
Certifications generally come in three waves. First is the steady, incremental single project certifications. Next are multi-product releases, for example several versions of a project, or a single project with accessories and components. Finally, there are surges that come from larger efforts associated with existing products and/or projects. They are in many cases already open source, but deciding to formalize their participation in the certification program. All three of these are important and not captured in the overall certification numbers. The distinction becomes more clear when evaluating certifications individually.
Another way of looking at growth in the certification program is looking at the number of certified projects per year. When considering the growth from this perspective, the initial surge from the first year described earlier in this section of the report is more apparent. There is a dip in year two, but a strong bounceback in year three and continued growth in year four. Looking at certification data for 2020 year-to-date, the data indicates there will be continued year-over-year growth for the program. In other words, there should be more projects certified in 2020 than were certified in 2019.
What we don’t know, however, is why creators choose to certify and what they are hoping for in doing so. This is something we can begin to understand better by considering certifications by Creator.
Certification By Creator
A total of 169 creators have combined to certify over 400 products and projects. This calculates to a median of nearly 2.5 certifications per participating individual or organization, but that doesn’t tell the whole story.
There are certain companies that are more active than others, contributing multiple–if not dozens–of projects to the certification directory. A single certification can be anything from a simple one-part design, to a complex system with hundreds of components or more. A single complex machine could either be a single certification, or it could be hundreds of certifications. It may be helpful to think of this as “bundled versus unbundled.” It is up to creators to decide how they want to certify their products.
Creators also have an open field when submitting their certifications. We see discrepancies where a single creator uses different names between submissions. For example, submitting as a company name in one submission and submitting as a product brand in another submission — while they are in practice both submitted by the same organization. Or, submitting two slightly different versions of a legal entity’s name. This may be intentional or accidental.
When it comes to how to certify (bundled versus unbundled),which name to use, etc., these individual determinations do affect the overall numbers.
129 companies (~75%) have entered the program by certifying one product, equaling 32% of the total number of certifications. This lands us near the Pareto principle, with the remaining 25% of companies responsible for 68% of the certified products. For the high volume participating companies, many of them would benefit from more visibility and support in the community. Some names are generally recognized in the community (as measured by revenue, social media following, presence at events, etc.), however others are not and deserve credit for the work they are doing.
There is a large base of companies with a single certified product who represent potential growth for the community. More research is required to determine whether they have plans or current additional projects that could also be certified, or if they only have plans for a single project at this time and it is already certified. This group includes some high profile companies in the community. Do they not see value in the certification? Is the cost of participation too high? Is simply putting an open source license and the conventional “open gear” design sufficient signaling for their customers? Engaging these companies, understanding what’s working and what’s not, could increase their engagement in the community.
The leading certifying individuals and organizations are a diverse group, spanning numerous countries, product categories, and price points. The participants span multinational technology corporations, small- and mid-size open source oriented businesses, and even several individual hobbyists. This breadth of participation over the years demonstrates that many types of organizations see value in formally participating in the Open Source Hardware community. It’s not just the “maker” businesses that you might think of when you think of Open Source Hardware.
Editor’s Note: We had to establish a “stopping point” in order to conduct analysis when creating this report. There are more recent certifications that would affect the data in this section of the report (certifications submitted as recently as March 2020). In particular, it is worth noting that Field Ready has maintained a strong pace of certifying new projects in 2020. In between conducting analysis and finalizing this report, they have moved into the top slot as the single creator with the highest number of certifications.
Further, as mentioned earlier in this section, there are discrepancies in submissions that affect certification numbers by Creator. Because we don’t know whether these discrepancies are intentional or accidental, this static report does not reflect updates applied retroactively to the dataset that may affect any one organization’s overall numbers. These reporting lags are unavoidable until OSHdata is automated and generated in real-time.
Incidentally, dynamic reporting is something we would like to implement with financial support! If that’s something you’re interested in too, send us a note.
Certification By Country
One of the most exciting aspects of putting this report together was seeing the global scale of the Open Source Hardware community. There are participating individuals and organizations in 36 countries, spanning North and South America, Europe, Asia, and Australia.
While the United States is a leader by a long shot, Bulgaria has strong participation and Europe as a whole is worth considering as well. There are also countries that recently had their first certified project, like Turkey, which represent exciting growth and future potential for this community.
Certification By Project Type
Project Type is the area of certification data that has the greatest need for improvement. The options presented to certifying individuals and organizations are inconsistent, with categories spanning use cases, industries, and methods of production. Some insight can be gleaned from this data, for example the leading selected Project Types span Electronics, Tools, 3D Printing, and Education.
Overall, the current Project Type data makes it difficult to draw conclusions. With better data, segmented analysis could be done like considering the average sale price by category. Industry-specific research could be done about the price premium that Open Source Hardware products command, or reduced research and development costs versus proprietary alternatives. We believe there is a lot of opportunity and potential for improvement in this area.
Currently, project type is one category with pre-defined options. We suggest three categories be created to address the intention behind project type: Fabrication, Industry, and Application. To be clear, this data is not available at this time. In order for it to be made available moving forward, the certification application would need to be revised by OSHWA. Retroactively updating the dataset would depend on creators’ input, which may be difficult. Regardless, the following is our proposal for how OSHWA might consider this or how OSHdata might collect this type of data in the future. For a full outline of other ideas we are considering, visit the Road Map.
Signifying the method in which the creator fabricates the project as it is being certified. Can this be made at home with consumer-grade equipment, does it require the prosumer-level equipment found in a typical makerspace, or are industrial-grade machines and tools required? This will help show the level of investment required to replicate or clone a project, which will contextualize other aspects of the source code like the Bill of Materials (BOM).
Meaning the industry that the creator is operating within. This could be accomplished using conventional industry classification codes. These vary around the world, one that is well established in North America is the North American Industry Classification System (NAICS), which is used by governmental agencies in classifying business establishments for collecting, analyzing, and publishing statistical data related to the business economy.
This layer of classification would be specific to the application or use case for the end user. It may be the most difficult to come up with a set of these, however our initial thought is that it would cover the type of activity that this project is for. Outside of industry, the work that is being done. So this could be research and development, prototyping, testing, production, education, and other similar descriptors that span industry types.
Finally, the existing open tag system used for keywords allows creators to append additional classifications or distinctions as they see fit. So for example, a 3D printing-related project may want to specify that they are associated with the RepRap project. This enables additional layers of classification beyond what preset options permit. We believe it’s worthwhile to maintain this option to support differentiation and discoverability.
Where do we go from here? Well, that’s up to you. Read the OSHdata Road Map to see what we’re thinking about. If you already have feedback, use the Contact Form to get in touch. If you want to follow new updates, subscribe to the newsletter below. Let us know what you think on Twitter by tagging @OSHdata.
We know, writing a message might be outdated. We considered a survey, but we don’t want to limit which directions this might go. An open format seems like the best way to continue this conversation. So please let us know what you think. In the spirit of openness, don’t hold back.
Follow Open Hardware developments with OSHdata: