Collaborative Product Development in an Open Source Hardware Environment: Thoughts …

This is a continuation of my previous blog post, responding to potential Handibotter Tim Deagan about the subject of how we are going to manage ongoing releases of Handibots and Handibot accessories. Here I reflect more generally on how and why open collaboration is attractive and practical.

When I asked Tim if it was okay to share his questions, he responded with some thoughtful and instructive observations that help frame the questions about open collaboration and kinda got me going on this. In Tim’s words:

The challenge of connecting a release cycle to purchasing is brutal. There are a few models in various industries, but ShopBot, especially with Handibot, is in a relatively new class. I’d love to suggest that your approach to Handibot offers some upside to the problem, but it might help if I enumerate what I think the problem/solution space is;

• Automobiles and similar large frequently updated hardware: Release schedules are known to all and usually hit. The industry mitigates this in two ways; 1) the old model year is known to be sold at a discount, 2) the timing of the release is predictable, so the buyers aren’t surprised when new models show up. Other large purchases (industrial equipment, etc.) follow a similar model in that they typically give warning about the new model availability and sell older stock at a discount to clear inventory.
• Software (my industry): Release schedules are a shot in the dark. They may be disclosed or undisclosed. They can be off by months or even years. They could be frequent or infrequent. The developers (hopefully) mitigate the impact on customers by ‘reverse compatibility’ if at all possible. This allows the customer to buy an older version, confident that the work they do in it will be digestible by the new product when they finally upgrade. Customers under a support contract are often offered free or reduced prices to upgrade.
• Closed Source hardware (especially electronic tools/gadgets.): Releases are frequently undisclosed or very soft.  New product is launched and old product is sometimes discounted (or the new product is sold at a markup.)  Buyers take the risk in the ‘grey zone’ and hope for the best.
• Open Source hardware: This is where I’d love to hope for something different. It can be the same as Closed Source. But, it has the potential to be philosophically different. Given the community involvement, there is potentially a stronger impetus to build upgrade paths for prior releases. This is a design constraint and not always worth it or even possible. The possibility increases the earlier the designs are opened to the public. If the next-gen design process has community eyes on it (with dev previews, etc.), then current users who have a stake in the upgrade path are motivated to contribute to exploring ways to implement the new features, either as submissions to the design in progress or as a forked path to create replacement components for the old model.

This can happen after the release, but the possible value proposition to new buyers is the confidence about what their position will be if they buy the old vs. new model. Certainly, this is a challenge to a commercial company. Open sourcing designs is an act of faith. IP issues aside, the pricing economics are challenging. If it’s known that the new model is going to cost more, there is incentive to buy old models. However, if the company has been clever and determined a way to add features while reducing, or even holding prices steady, the impetus to buy an old model is removed and the company is potentially caught holding old stock that no one wants unless they discount it (lousy RONA!)

This problem is becoming common. It might be that the economics rely on disgruntling a small set of edge case customers. But when release cycles are measured in months or quarters, that edge case occurs much more frequently, increasing the odds that customers will get pinched. As a maker, I’m very interested in exploring ways to both make companies profitable and give buyers maximum confidence (which will open their wallets and make companies profitable, repeat).

We hope with Handibot to addresses some of these challenges and suggest some solutions. The suggestions are in keeping with prospects for what has been called the “new industrial revolution”.

Upgradeability

For me, the rational approach to product evolution starts with a commitment to upgradeability — to create a path by which owners can add features to an existing product, some of the same features that are being introduced in newer models. Of course, in not emphasizing buying the next model, this approach precludes relying on new-purchase-replacement as the core revenue stream for a business. But, overly frequent replacement hardly seems a sustainable business practice for the 21st century. New business models promise to provide new options. The Phonebloks and Google Ara project are interesting examples. They start from the concept of upgradeability based on modular and interchangeable design principles (see below) and show how these can offer longer device usefulness.

Incremental Change

Getting away from using a highly-touted, new-model release to drive business actually eases the pressure on the development process and allows for the regular introduction of small and incremental improvements into a product. With some thoughtfulness, most of these can be offered to existing customers as upgrades or accessories. Occasionally, a major change in a product becomes warranted, but this can happen in a way in which most existing units can either have been brought along with successive updates; or alternatively, certain critical components of the new model might be retrofitted. In many product areas, an upgrading approach could help move us beyond our captivity to the next-model-hype cycle … but, I appreciate, it’s not going to work for everything.

Here’s the 15-year-old ShopBot CNC in my barn. This tool has received many regular ShopBot upgrades over the years, including: getting an industrial spindle, a Z-axis upgrade called a Retro-Z, and a new Control Box and several of the Control Card updates noted below. My wife Sallye and I believe this makes it about the equivalent of today’s shipping ShopBots. Sallye uses the tool most days for signage, contract cutting, and educational projects for ShopBot (note the totally abused spoilboard). You can see her test-cut materials on the deck for things she is doing at SolidWorks World this week.

Design for Rational Products

Upgradeability through incremental change is fostered by several design principles:

• Modularity. Modularity of the sort that allows interchangeability of new components for old is the basis for projects like the Phoneblok mentioned above. It is a practical way to support incremental change in products while also insuring serviceability and usefully long product lifespans. An example in our own design practice is to maintain modular compatibility across models as much as possible. The Control Card on ShopBots is the same for every tool, whether a 10’x30’ CNC for cutting large panels for home construction, a 5-Axis Mill, or a Handibot. Over the years, we have maintained the connector compatibility of the Control Card so that a current Card can simply be plugged-in to upgrade the capabilities on an 18-year-old ShopBot. We’ve attempted to use the modularity principle for many hardware and software components of our tools. It also creates opportunities for quantity purchases and ease of manufacture.
• Use of Standard or Common Components. Many electronic and hardware items can be sourced as standard products. By sticking with items that are readily available and interchangeable, we make products that are more maintainable and serviceable and that are also good candidates for open and shared development. Motor-drivers and linear-rails in Handibots are a good example of standardized components.
• Easy Hackability. It goes without saying that for incremental enhancement to take place, you need to be able to take something apart and work with it. And, you will need the documentation to make sense of it. If there is also a community available that provides additional resource help … all the better.

Alicia Gibb with others from the Open Source Hardware Association have released a new book, “Building Open Source Hardware,” that summarizes many of these principles, considers business aspects of open hardware, and provides nice checklists for design and manufacturing. There is also a useful chapter on going from making to manufacturing by Matt Bolton, who has himself done it for Sparkfun.

Design for Digital Fabrication

The above design principles are important, but the game-changer for the “next industrial revolution” is digital fabrication. Whether it is additive-fab or subtractive-fab, the enabler here, and the essence of digital fab, is the precision in the work of the tools and their fidelity to the digital model. This capability allows complexity of form and function to be built-in at little cost. The oft-heard expression for this virtue in 3D printing is that “complexity comes free.” The same is true for subtractive-fab tools (CNC, laser cutters, etc), these tools provide cutting, drilling, and machining that few individuals could do manually. It is just as easy for a CNC tool to cut a curve as a straight line, to make a complex joint or fitting as to cut something square. The tool does not care. Sure, it might take a few seconds longer, but the point is that now engineering can be designed into parts in ways that was either impossible or prohibitively expensive to embed into production in the past. It would have involved machining that is too labor intensive. Now, we can make “smart parts;” parts that help assemble themselves because of their designed-in features. Look no further than the digital fab joinery in your Handibot for a good example of such parts in practice. Indeed, at a conceptual macro-scale, smart parts made with today’s subtractive digital fab tools anticipate the smart nano-materials in our future.

Digital-fab joinery throughout Handibot as well as extensive detailed machining. (Plastic components for Handibots are all machined on ShopBot Desktop CNC’s.)

In “Makers: The New Industrial Revolution“, Chris Anderson diagrammed the efficiency and competitiveness of digital fab compared to centralized mass production processes such as injection molding. Chris’s example is for 3D printing, but it holds true for digital fab tools in general. We can appreciate that there will always be situations where centralized, mass-production makes the most sense. But for other products, there is a volume range in which digital fab can be competitive and thus be supportive of alternative manufacturing strategies.

The “duck” idea here is that until one gets to thousands of ducks, it is cheaper to make the duck with digital fab (3D printing, in this case) than with injection molding. Note that the cost of production is relatively low and constant for small runs of digitally-fabricated parts.

Being able to competitively employ digital fab in production opens the door for incremental upgradeability. That’s because there is little or no cost to evolving a design and doing it frequently. Compared to needing to create an expensive new mold or modify a highly capitalized production process in traditional mass production, a design change can be readily incorporated into ongoing digital fab production – the files change; some fixtures may change.

For Handibot, using digital-fab design principles in developing the tool and digital fab in production allows manufacture of plastic and aluminum parts to be done efficiently and competitively in small or medium volume. We are not locked into large batches of injection molded components.

Of course, part of the charm here is that Handibots themselves are digital fab tools. We are expecting that they will be able to make a lot of their own upgrade parts, at least for those makers who really want to DIY.

And, there is the “Democratization” Thing

The machining capability of digital fab tools is at the heart of their potential contribution to our productivity. But their increasing accessibility and affordability makes adoption of these tools possible for small scale production. Computers, software, and the web make digital fab ever more accessible: accessible as something people can understand and engage in; affordable because the costs for software and hardware keep decreasing. These tools were once only used in industrial mass production because their costs were too high for anything else (and they did not have the usability provided by today’s PC’s). Now, a small digital fab manufacturing operation can be created with little capital.

So the technology is available — and its agility will help make it competitive. It’s about new opportunities for small manufacturing in communities everywhere. It’s a prospect that has been described as providing a “democratization of the tools of production.” … Powerful stuff, and I wouldn’t want to get too carried away.

In the Handibot scenario, we look forward to growing production to levels that allow distributed manufacturing, that support spreading out the production of Handibots to the communities where they will be used. One can envision local, enterprising individuals or groups in small manufacturing facilities who make the parts and assemble the tools, who re-manufacture and upgrade old tools, and who also provide support and training to new users while contributing to the collaborative development of the product.

Who knows whether any of this will work? But digital fabrication does create the opportunity to grow this project and the production of Handibots in small steps. With such bootstrapping, we basically test the business viability at each level of scaling as we go along. My bias is that the game-changing nature of digital fab — the continuum from design through prototyping to production, and supported by digital resources and infrastructures — will make it possible. We will also see how digital technologies afford a growth medium for a product like Handibot and thus an example of attractive new forms of distributed production, something envisioned by many who have forecast a next industrial revolution.

Evolution of the pre-Version 1.0 Handibot.