Open-source software is one of the great success stories of the past few decades. The Apache HTTP Server is the world's most popular Web server, Linux has more than held its own against Unix and other proprietary operating systems, and Mozilla's Firefox browser has given Microsoft's Internet Explorer strong competition over the years.
Could the same philosophy -- the free and public dissemination of underlying code and specs, with multiple developers from disparate sources contributing to the design -- work for tech gadgets as well? Will we one day commonly use smartphones, netbooks or other gadgets that have been developed under an open-source model, maybe even preferring them over proprietary products like the iPhone?
After all, it's possible today to design a device -- including its electrical and mechanical architecture -- on a personal computer with CAD and schematic design software, order nearly all the components needed for it online, and then process the manufacturing of a prototype through a low-cost supplier. So the idea of organizing an open-source project online to build a device isn't far-fetched, nor is it one that requires millions in start-up funding.
But can such gadgets succeed against those developed by established commercial manufacturers with deep pockets? Mark Driver, a Gartner analyst who specializes in open source, thinks that open-source gadgets have the best chance in markets where the technology has matured to the point that it is commonplace.
"Open source is about commoditization," Driver says. "These products are taking a market where there really isn't a lot of concrete differentiation ... between what's out there and providing an alternative, which is exactly what open source does right. Linux got wildly popular not because it did something new; it's because it did what Unix did, but did it in a much more open fashion."
Defining open-source hardware
While there are numerous open-source computer and electronics components available today, only a handful of complete tech gadgets are being developed under an open-source philosophy. However, what exactly defines a hardware project as being open source remains ... well, open.
Generally, hardware that is "open sourced" means at least some of its plans have been made available to the public, thus allowing others to contribute to its development or, if permitted by its creator, to manufacture the device themselves or even modify the plans to create a new device.
Always Innovating Inc., for example, encourages outsiders to contribute to the development of its ARM-processor-based tablet/netbook hybrid, the Touch Book. Weighing 1.8 lbs., the device features a touch screen, a removable keyboard and a customized Linux operating system distribution. It can run for 10 hours on a single battery charge.
The schematics for the Touch Book are freely available on Always Innovating's Web site. "We also provide advanced support and consulting services for companies who want to build their own devices starting from our design," says Chief Operating Officer Alexandre Tisserant.
"This is the way we are following: Build reliable, innovative products, and by opening them, you will get the necessary feedback and contributions to improve them and design new ones faster and easier," Tisserant says.
That's the open-source ideal, anyway. On the flipside, "the worst-case scenario would be a project emerging using an open-source moniker, and it ends up being nothing more than a marketing gimmick," says Gartner's Driver. "If it's only from one vendor, or one source of support, those kind of things are the weakest forms of open source."
Who's in the market for open-source gadgets?
Unsurprisingly, the kind of user such gadgets are geared toward -- and appeal to -- the most is the tech hobbyist. The Touch Book has so far sold mainly to this crowd, says Tisserant, who says "several thousand" units have been sold. Yet his company is looking now to sell it to vertical markets. Because the Touch Book is highly customizable, it could easily be integrated into taxis or police cars, or connected to a hospital's private network as an "always on" portable device for medical staff, Tisserant says.
Then there's the Frankencamera, a Linux-based digital camera that can be programmed to control exposure, flash, focus settings and more. The camera is being developed by a team of graduate students at Stanford University and is meant for academic use.
"Specifically, we want to make this easy for graduate students doing research that could use a programmable camera, or undergraduate CS students doing courses in programming," says Andrew Adams, one of the lead developers of the Frankencamera. "We're graduate students ourselves, and this whole project is born out of our frustration with trying to program cameras to do what you want them to."
A consumer-oriented open-source project that has so far failed to catch on is the Neo FreeRunner smartphone and its supporting Linux-based platform, called Openmoko. The project was launched by Openmoko Inc., with both the operating system and the design plans for the internal electronics and housing available for others to use and improve on.
The company officially stopped supporting the project in April 2009, according to Product Manager William Lai. "As time and technology progressed, the funds involved in competing with the likes of Apple, RIM, Android, etc. were out of our scope, and we soon realized that the technology outpaced our ability to deliver on a timely basis," he says.
However, the Openmoko platform and FreeRunner phone are still being developed by a volunteer community.
Distributing and testing hardware is difficult
With software, anyone can download a copy of an open-source program and try it out practically instantly. It's equally easy to give feedback to its developers and contribute code to fix bugs or add features.
The open-source model in software development thrives on this constant distribute-and-test process: The more copies of the code you can get into the hands of other people, and the quicker you do so, the faster the project's developers can field feedback in order to fix and improve the software for its next release.
But applying the open-source model to hardware isn't as straightforward. Copies of prototypes can be expensive to produce and distribute to fellow developers for evaluating and testing, so development doesn't progress as quickly.
Tisserant calls this "the cost of the test": "When you get your first piece of homemade hardware, you can do some modifications. But you will have to order a new piece with your new design. This takes time -- a few weeks -- as well as money."
In order to seriously challenge the traditional proprietary model of developing hardware, a manufacturing time frame of less than one week would be ideal, says Tisserant: "The easier and faster you can test, the easier and faster you can learn."
Turnaround time could be lessened with the use of affordable rapid prototyping or fabrication machines. For example, the body of the Frankencamera is laser-cut acrylic. So anyone with access to a laser cutter can take the plans for the Frankencamera's body and make their own.
A device like the RepRap, a 3D printer for rapid prototyping, could play a significant part in open-source hardware development. The RepRap is itself open source. While commercially available 3D printers cost around $20,000 at the low end, the RepRap's design is freely available to anyone who wants to build one. (Its developers estimate that the materials cost around $480.) What's more, the RepRap can replicate many of its own parts, with the rest of its parts cheaply available, so you can build another one using the first.
"Someone with a RepRap or a laser cutter and a soldering iron can put together something. Open hardware designs combined with rapid fabrication gets at exactly the original intent of open-source software -- if the design is open, you can modify it to meet your needs, and freely share those modifications with others," says Adams.
Although the RepRap and other rapid-prototyping machines can speed up prototyping, they're not an end-all solution since their capabilities are meant for creating only the housing or external case for a gadget. Such a machine can help build prototypes of, say, a netbook's outer shell faster, but most of the device's internal electronics still need to be sourced out for manufacture.
Nevertheless, opening up a device to the public (especially during its early design phase) encourages the formation of a community that can propose and contribute improvements. This can help reduce the number of prototypes that need to be built, saving money and time.
The lack of open-source culture among component makers
A device that is open source does not necessarily mean every component within its design schematic is also open source -- in fact, it probably uses several proprietary parts.
Any consumer tech device is built with many smaller components. The makers of these parts are usually secretive about revealing their inner workings, unless it's to a paying client. This can be a challenge for anyone trying to develop open-source hardware if their device's design plans are to be released publicly.
"In the software world, there's a rich culture of providing basic open-source building blocks like compilers, editors, support libraries and operating systems," says Adams of the Frankencamera project. "Unfortunately, chip manufacturing is an inherently expensive business, and there's far less room for the kind of altruistic sharing that seems to be the major motivator behind a lot of open-source contributors. Having to sign [non-disclosure agreements] to even see how to use a part like an image sensor is common."
Although he and his fellow Frankencamera developers have encountered hesitation or refusals from companies they've approached to acquire information to help them build their digital camera, they have come across some willing to contribute -- in particular because of the open-source aspect of their project. (Most of the Frankencamera's electronics are commodity parts that anyone can buy. A few components, such as the camera's power circuitry, were specially designed by the project's team.)
"Companies that are hard to extract information or parts from don't care whether you're planning something open source or commercial -- they're equally reticent. People and companies that are willing to help are usually more willing to if it's going to be open source; they know they'll be able to benefit from any results too," says Adams.
The issue of intellectual property
A big question swirling around open-source hardware projects is the legal issue of intellectual property -- who owns what (including the whole and the individual parts) in an open-source device, especially if several people are contributing designs? Brendan Scott, a lawyer who specializes in IT law and runs the Web site Open Source Law, strongly advises the creators and lead developers of such projects to address this matter before anybody agrees to make anything.
As for how this should be handled, he says there is no one-size-fits-all answer. "In some cases, it will be better for individuals to retain intellectual property [in what they contribute]; in others, it will be better to transfer it to some holding entity. The main thing about intellectual property in a project is to turn your mind to the issue before you start -- or soon after you start -- rather than when you finish. By not addressing the issue, you may discover that the issue has been decided for you, perhaps in a way you are not happy with."
Michael Arrington, founder and co-editor of the TechCrunch blog, might agree. In July 2008 he announced plans to create a low-cost Web tablet, later dubbed the CrunchPad. While the hardware development process wouldn't be fully open, Arrington's idea was to "design it, build a few and then open source the specs so anyone can create them," as he wrote in the announcement.
The project got off to a promising start as TechCrunch partnered with Singapore-based Fusion Garage to develop and manufacture the CrunchPad. In late 2009, however, the agreement fell apart when Fusion Garage announced its intention to sell the CrunchPad without TechCrunch's involvement. Fusion Garage CEO Chandra Rathakrishnan claimed that his company had sole intellectual property rights to the device, while Arrington said both companies shared IP rights.
Fusion Garage plans to sell the device as the JooJoo tablet, and TechCrunch has filed a lawsuit against Fusion Garage. As of this writing, Fusion Garage has been taking pre-orders for the JooJoo, which the company's site says "will ship in 8 to 10 weeks."
Asked what legal steps or counsel he and his Frankencamera developer colleagues have taken to protect their hard work, Adams says, "In this regard, life is easier when there's no money to be made. Because everything we do is as students of Stanford University, we have pretty good legal avenues available to us if someone should try anything nefarious. So far, though, the vast majority of what we have heard from the general public is interest, encouragement and offers of help."
Applying current open-source licenses to hardware
Another legal matter is whether current open-source licenses apply to hardware, at least suitably enough. Most were drafted in the context of software, and this is evident in their wording: The commonly used GNU General Public License refers to "the Program."
Scott of Open Source Law postulates that with a generous reading, existing open-source licenses could be applied to hardware projects without modification. "It is not too far-fetched to think of hardware plans and component lists as the source from which an 'object' -- literally -- is 'compiled' or 'assembled,'" he says.
"The intellectual property landscape for hardware is a little different from that of software," Scott continues. "Copyright applies to copies of software, but does not typically apply directly to copies of hardware, particularly for items for which their form is functional -- although making a copy of hardware can result in an infringement for the plans from which the hardware is made."
Scott anticipates that, over time, licenses will be customized or amended in order to cover issues faced by hardware created under open source. For example, the TAPR Open Hardware License was specifically designed for hardware projects. And the Arduino project, an open-source electronics platform with both hardware and software components, uses a license for the designs of its hardware that is separate from the license for its firmware (the operating software that runs on it).
Making money (or not) from open-source hardware
Always Innovating's Tisserant acknowledges that hardware companies going the open-source route might have lower profit margins, but he says they can benefit from lower research-and-development costs and shorter development cycles. "The goal is not to keep your secrets and live on endless royalties, but to share the knowledge and grow upon fast innovation," he says.
Although Openmoko Inc. no longer supports the FreeRunner phone and Openmoko smartphone platform, Lai says the company isn't through with open-source: "For the last year, Openmoko as a company has been focused on bringing open source in front of an audience of mass appeal. We want to continue to design products using open-source elements," such as the WikiReader ($99), a pocket reader preloaded with Wikipedia content, says Lai.
As far as the developers of the Frankencamera are concerned, they have no business plan because their project isn't meant to sell an end product. Their goal is to get the schematics of Frankencameras into the hands of students at other academic institutions, so they can build their own at minimal cost to use in their coursework and research.
In turn, they hope their project will "convince camera manufacturers that letting end users program their cameras is something that actually adds value and makes people want their product more, because there's a community of enthusiasts constantly adding new features to it," says Adams.
"How successful would the iPhone have been without the app store? Now why can't you write and download apps to your camera? Our personal goals are to do interesting research, and give other people the tools to do interesting research, not to make money," Adams says.
Selling open-source gadgets beyond the techie crowd
Jeff Orr, a technology analyst with ABI Research, thinks for an open-source hardware project to succeed in the marketplace against proprietary, commercial products, it still needs "some ownership -- some individual, some entity -- that is providing the workforce to assemble and distribute these products ... Once I've bought it, what's the support like? Is there a warranty if something goes wrong?"
Still, he is cautiously optimistic about the potential of open source at gadgets' R&D stage: "Could [the open-source model] challenge the commercial research and development process? I think so ... because you create a larger pool of knowledge that any individual or organization could learn from."
But will an open-source gadget ever take off in the same way Firefox and Ubuntu have, becoming a household name among mainstream gadget users? Open-source gadgets will become more common, Gartner's Driver predicts, but he is unsure if we will see one that appeals to a wide user base and can challenge an equivalent proprietary product.
"Will we see the same kind of revolution in those kind of devices that we saw in software? That's probably a much less likely occurrence to happen, at least for the foreseeable future," Driver says.
Howard Wen is a frequent contributor to Computerworld.
This story, "Open-source hardware takes steps toward gadget mainstream" was originally published by Computerworld.