Chris Anderson, editor-in-chief of Wired Magazine and author of the books, The Long Tail and the soon-to-be-released book called Free: The Future of a Radical Price, used the expression, ‘atoms are the new bits’ in his Twitter stream. To clarify what he means, he’s referring to the phenomenon of Open Source Hardware (OSH), a relatively recent trend that follows similar principles to those of Open Source Software. Chris recently ran two open source hardware projects called the Ardupilot and the Blimpduino so he’s quite experienced with open source hardware. Chris also has a great writing style and if Free is as well written as The Long Tail, I expect it will meet with the same kind of success and fanfare as his previous book.
The term open source software can have different meanings depending on its particular open source licensing agreement, of which there are many. In its simplest form, open source software means openly sharing the underlying instructions on how to reproduce a software product. For software to be truly open source, not only should the source code be provided, but the tools needed to modify and re-compile the source should also be readily available. The requirements for providing the source code is often subject to interpretation, and many companies claim to abide by the the open source licensing agreement yet fail to deliver the ability to modify the source. This usually happens because they’ve left out some critical files that may not be bound by open source licensing requirements, and if those files are necessary for compiling, then without them the open source files are not very useful.
Source code for commercial software is usually kept secret to insure that customers will actually pay for the software and to prevent competitors from gaining access to a company’s intellectual property. To insure payment, a company usually needs to prevent others from simply copying the source, compiling it, and distributing it. Most companies selling closed source software will even lock down the compiled bits with digital rights management schemes that make sure the payment is made before supplying a unique passcode to unlock it. Otherwise, the compiled code could be copied and shared freely once the first copy of it was purchased. If a company locks down the compiled code, providing the source code would be like providing the keys to everyone.
There are many different open source licensing agreements and they usually fall into categories somewhere between “do anything you want with it” to “always include my name in the credits.” Some are more restrictive, for example, such as “if you make money with my source code, then you must pay me, otherwise, it’s free.”
Software is virtually free to reproduce and distribute. Hardware is a little different. Before you can do anything with hardware, you not only need the instructions to put it together, but you must also acquire the physical material, i.e., the atoms. And unlike a free software compiler, turning hardware into a usable product requires material and capital equipment that can be expensive. Providing instructions on how to make a hardware product has some value but it is much less of a proportion of the final product’s overall value than it is for software. People are often willing to give away information for free, but if a product includes atoms and if you give them away for free, you’ll go broke eventually. In other words, if I give you information, I can still keep my copy of it, but if I give you something made from atoms, then I no longer have them. This is a critical distinction between atoms and bits.
How much of the value of software is just information or bits? Software is primarily information so the instructions on how to create it would account to nearly 100% of its value. Some would argue this point and say that software always needs user support, basically education on how to use it, and that information also has value. In fact, this portion of the value is where open source software companies generally seek to make their money. For a fee, they will answer questions about the free software or customize it for you.
How about hardware? Where is its value stored? With products made of atoms, there is always some material cost, and the percentage of cost attributable to the atoms can vary considerably depending on the product. For basic commodity products, one can argue that the cost of the atoms is very close to 100% of their value. That’s why farmers have to remain competitive with the current market prices for commodities. Just a few percentage point change in the wrong direction and the farmer loses money. Energy commodities like oil, coal, and gas have similar economic constraints.
Manufactured goods may have a much wider range margins. Luxury goods, for example, have material costs that may be a very small percentage of the selling price. Perfume and jewelery come to mind. But the other costs of selling luxury goods such as demand generation and retailer margins usually make up for the low percentage of the actual material costs.
In the case of consumer electronics, it’s been my experience that the material costs are typically 30-70% of the selling price. There are a few exceptions, of course, such as integrated circuits where actual material costs are a very low percentage of their price. But integrated circuits have enormous capital costs for the manufacturing equipment along with a high development cost for the circuitry. There are even loss leaders, where retailers will sell certain products below cost just to get customers to visit their store. But for the most part, a retailer needs to maintain some margin to make money.
So far, open source hardware has been primarily confined to small circuit boards for hobbyists to build gadgets. The Arduino hardware platform is starting to give open source hardware some new momentum. I think that the Arduino has the potential to be a game-changer because its open source nature is all-encompassing. Not only is the software and hardware open sourced, but so are its development tools. Products based on the Arudino hardware platform are not your typical consumer electronic products. An Arduino-based product might remind you of a science project. Granted, it would be more sophisticated than what passed for a science project a few years ago, but still a far cry from a product that would sit on the shelf at a retailer like Best Buy. In addition to the software that goes inside these gadgets, the schematics, parts list, and artwork files to produce the circuit boards are included, thus helping them to earn a description of truly being ‘open source hardware’.
Providing schematics for electronic products is nothing new. Many electronics equipment manufacturers provided schematics up until the past few decades when they no longer had much value because electronic products became so highly integrated that is wasn’t possible to repair them in the field. Schematics were generally included primarily to help someone repair the product when it failed. The printed circuit files that are now part of open source hardware are a new twist and they have some value to the consumer, but only if one intended to produce multiple copies of a circuit board. Otherwise, it is more economical to purchase the board from someone who is already selling them, preferably with components already soldered to them, due to the economics of building multiple vs. one-off products. But supplying board files does have the effect of limiting the markup one might ask for raw board since with these files, anyone could have a board manufacturer produce the boards, and maybe even add some new features or improvements to it. This additional information, along with the Creative Commons license invites competitors to produce your design and that has been virtually unheard of in the world of manufacturing.
Soldering together a kit of parts was a common activity for electronics enthusiasts before surface mount technology largely replaced the easy-to-hand-solder electronic components. I took great delight in putting together many electronics kits from Heathkit in the 1970’s and 1980’s. Although it is possible with some practice to solder surface mount components to circuit boards by hand, most hobbyists tend to avoid it. It requires patience, steady hands, and some form of optical magnification to do it properly. In production environments, assembling surface mount boards can only be done economically by expensive robotic pick-and-place machines.
I’ve long thought that it would be great to have a manufacturing machine that could take basic raw materials and some downloaded information to assemble them into a finished product. There’s actually a name for this concept. It’s called a Santa Clause machine. I could imagine a machine that took in recycled aluminum, possibly empty beer cans, and produced car parts like the frame, body, wheels, and engine. After downloading the information, the rate at which you could produce a car would only be limited by your rate of beer consumption (or you could collect cans from along side the road to speed things up, of course). In reality, the machines needed to manufacture any product today are highly specialized and would not lend themselves to putting together products atom-by-atom. But this hypothetical machine is fun to contemplate and has been a topic of frequent conversation by those who have trouble differentiating between science fiction and the real world. Many people think that rapid prototyping technology like stereolithograpy has ushered in a Santa Claus machine era, but I’ve been having plastic parts made with stereolithography for more than 15 years and despite solid progress, it’s still nowhere close in terms of cost, speed, and functionality of plastic injected molded parts and doesn’t seem to be destined to ever close that gap. Laser cutters allow you to quickly make some interesting 2D parts, but it has limited applications when making 3D parts. And, of course, you almost always need some metal components and there are no Santa Claus machines for those parts. I am not holding my breath for the Santa Claus machine to arrive.
There’s also the issue of regulatory compliance requirements which is where you have to get the approvals from regulatory agencies like the UL, FCC, CSA, CE, etc.. which basically make sure you don’t:
1. burn down someone’s house
2. electrocute anyone
3. interfere with radio reception
You think that might be easy to pass these certifications, but it’s actually quite involved. That’s what is required for basic consumer electronics products. Some industries have many more additional regulatory requirements. Many open source hardware enthusiasts seem to be blissfully unaware of these testing requirements. Until you’ve actually shepherded a few products through these regulatory approval gauntlets, you haven’t really designed anything the general public could purchase in large quantities. Hobbiest products generally fall below the radar when it comes to regulatory agencies. And when you meet these requirements, it’s important that the end consumer does not make changes to a product that would jeopardize the test results that allowed the approvals to be granted. For those who want to sell products that can be modified by end customers, it’s a bit of a Catch-22.
So are atoms really the new bits? I think open source hardware has definite potential for growth, but like Linux, it will appeal primarily to individuals who thrive on knowing how things actually work, that is, learners and do-it-yourselfers who are technology enthusiasts, or, in other words, less than 10% of the population. The rest of the world just wants to purchase finished products, preferably those that are aesthetically pleasing and carry the brand of a well-recognized company and/or have the endorsement of some celebrity.
When you’re an engineer, designing products for other engineers is technically engaging and fun. You’re basically designing for people who think like you do. Designing for the masses can have an element of drudgery because you may have to intentionally limit products to make them acceptable to non-technical people so they will be mass marketable. Is it possible to do both? That is, can you make something that is technically elegant and capable of modification, yet acceptable as a mass-marketed product? I don’t know for sure, but it’s always fun to dream it is, and if a product like that does come to fruition, it will likely emerge from the open source hardware community. It should be interesting to see what innovations arise from this new trend.