Why free hardware fails

Off the Beat: Bruce Byfield's Blog

Free-Licensed hardware has yet to have even a moderate success. Detractors take this fact as proof of the unviability of free hardware; supporters as a reason to despair.

Both conclusions, however, are premature. Most new products fail, and, although free hard has some unique processes, enough free hardware has not been released for any accurate estimations of its chances of success.

I first learned this basic lesson around the start of the millennium, but nothing I have heard since convinces me that anything has changed. Crowdfunding is helping would-be producers of free hardware, but, as any product manager can tell you, bring any product to market is a slow and difficult process. Producers of free hardware are no exception, except insofar as they may have more technical than marketing experience.

The rocky road to market

The first problem in building free hardware is that it is typically built in numbers that are irrelevant to manufacturers used to shipping hundreds of thousands of units. Manufacturers often refuse an order of 2500, let alone one of 250. When free hardware orders are accepted, its makers rarely have access to the highest level of executives, and their orders are taken less than seriously. The blog of Keyboardio, a company producing a keyboard whose keys are mounted on slabs of maple, mentions receiving samples that have been improperly cut, unevenly varnished, and obviously unfit for any professional use.  Often, after weeks of discussion, negotiations may be broken off without any word of explanation or demands will be made for additional money. None of these things suggest that that efforts at free hardware manufacture are treated with any respect by many manufacturers -- although perhaps they do suggest a reluctance to refuse outrightor an attempt to gain bargaining power.

Such problems are difficult to handle by themselves. But in addition, any production run has fixed costs that, as Len Leighton, who is currently fundraising for the EOMA68 laptop, explained to me in a private email, remain very much the same regardless of how many units are built. As a result, units in a small run are more expensive, and have a hard time matching the specifications of leading manufacturers. This is a serious problem because, while idealists will buy free hardware for ethical reasons, others potential users will not consider free hardware unless its specs are competitive with the proprietary hardware sold in computer stores.

Another problem is that the ethics of producers of free hardware are often unfathomable to the manufacturers. Keyboardio's partners intend their product to be the ultimate keyboard, with dozens of specially sculpted keys and both character assignments and backlighting completely customizable. These standards of excellence were hard to explain to the sales representatives of manufacturers, whose response was apparently something like:

“Wow! Your keyboard is beautiful. The wood is just for the prototype, right? What would you think of switching to a nice white plastic? LEDs on every key? How about just a few LEDs and some light piping? Wait. You want to do custom keycaps? You know that costs money, right?”

Similarly, to many manufacturers, free software and hardware are equally foreign concepts. Aaron Seigo, who attempted to build a free tablet in 2012-14, recalls one manufacturer who unilaterally changed free-licensed components in the interest of reducing costs. Similarly, the Keyboardio blog recalls that, when one manufacturer warned the company that it should safeguard Keyboardio's intellectual property, co-founder Jesse Vincent had to explain what open source software meant. "You have no idea how much fun it was too explain," he comments.

Reaching an agreement

And these are only the innate challenges in bringing free hardware to market. The process of reaching an agreement is laborious in itself. Manufacturing in North America or Europe is rarely cost effective, which means either frequent trips to Asia, or, else as Pi-Top, which produces computers built with the Raspberry Pi found, having a partner move to Asia to oversee negotiations and manufacturing. Meanwhile, the producers may be still refining their software or any hardware they are producing themselves.

Some producers benefit from the services of HWTrek, a company that exists to help bring producers and manufacturers together, but the process remains a series of stops and starts. Delays such as the Lunar New Year may also crop up unexpectedly for those unfamiliar with them.

The process begins by deciding which manufacturers to approach and arranging meetings and factory tours. If the meetings are successful, they may lead to a Statement of Work, a summary of the work the manufacturer will do and the tentative cost and delivery dates, followed by a request for bids that spells out the work in more detail. Terms are negotiated at every step at the way, and false beginnings and dead ends are common.

When formal bids are received, the producer may be lucky to find one that meets its preferences. Just as often, though, choosing a bid will be a matter of trading off advantages and disadvantages. The producer responds to bids by proposing a Letter of Understanding which, in theory, should help prevent any surprises. Only then is the contract itself signed, opening the way to manufacturing -- although, depending on the manufacturer's other contracts, there may still be delays. If the producer has been miscalculated, the manufacturer might bump them in favor of a larger or more important contract. or try to insist on a change in terms.

Nor is that all. After the hardware leaves the factory, the producer has to begin a similar round of negotiations to convince distributors to carry and promote the product. Should the distributor not be enthused about the product (as I believe happened with the Aquaris M10 Ubuntu Edition), the hardware may received half-hearted promotion and be kept alive only so that the distributor can claim diversity or prestige.

Limited vision
Once the producer has one successful product, they may find their next products easier to bring to market. However, for the first-time producer, or the producer whose past products have had lackluster sales, a doublebind exists: without a past record of success, finding a partner can be next to impossible, yet, unless a partner can be found, no record of any sort is possible.

So far as I can see, nothing is wrong with free hardware. Rather, the problem is with a conservative industry that makes bringing any product to market next to impossible for a small and inexperienced company.

When several thousand free hardware products fail, then we can begin to draw some conclusions. Until then, we can only watch as effort after effort develops more slowly than the producer expected, and hope for the best, aware that we are only learning the small part of the story that escapes to the public.