Business and free software have been intertwined for years, but the two often misunderstand one another. That's not surprising -- what is just a business to one is way of life for the other. But the misunderstanding can be painful, which is why debunking it is a worth the effort.
An increasingly common case in point: the growing attempts at open hardware, whether from Canonical, Jolla, MakePlayLive, or any of half a dozen others. Whether pundit or end-user, the average free software user reacts with exaggerated enthusiasm when a new piece of hardware is announced, then retreats into disillusionment as delay follows delay, often ending in the cancellation of the entire product.
It's a cycle that does no one any good, and often breeds distrust – and all because the average Linux user has no idea what's happening behind the news.
My own experience with bringing products to market is long behind me. However, nothing I have heard suggests that anything has changed. Bringing open hardware or any other product to market remains not just a brutal business, but one heavily stacked against newcomers.
Searching for Partners
Both the manufacturing and distribution of digital products is controlled by a relatively small number of companies, whose time can sometimes be booked months in advance. Profit margins can be tight, so like movie studios that buy the rights to an ancient sit-com, the manufacturers usually hope to clone the success of the latest hot product. As Aaron Seigo told me when talking about his efforts to develop the Vivaldi tablet, the manufacturers would much rather prefer someone else take the risk of doing anything new.
Not only that, but they would prefer to deal with someone with an existing sales record who is likely to bring repeat business.
Besides, the average newcomer is looking at a product run of a few thousand units. A chip manufacturer would much rather deal with Apple or Samsung, whose order is more likely in the hundreds of thousands.
Faced with this situation, the makers of open hardware are likely to find themselves cascading down into the list of manufacturers until they can find a second or third tier manufacturer that is willing to take a chance on a small run of something new.
They might be reduced to buying off-the-shelf components and assembling units themselves, as Seigo tried with Vivaldi. Alternatively, they might do as Canonical did, and find established partners that encourage the industry to take a gamble. Even if they succeed, they have usually taken months longer than they expected in their initial naivety.
Staggering to Market
However, finding a manufacturer is only the first obstacle. As Raspberry Pi found out, even if the open hardware producers want only free software in their product, the manufacturers will probably insist that firmware or drivers stay proprietary in the name of protecting trade secrets.
This situation is guaranteed to set off criticism from potential users, but the open hardware producers have no choice except to compromise their vision. Looking for another manufacturer is not a solution, partly because to do so means more delays, but largely because completely free-licensed hardware does not exist. The industry giants like Samsung have no interest in free hardware, and, being new, the open hardware producers have no clout to demand any.
Besides, even if free hardware was available, manufacturers could probably not guarantee that it would be used in the next production run. The producers might easily find themselves re-fighting the same battle every time they needed more units.
As if all this is not enough, at this point the open hardware producer has probably spent 6-12 months haggling. The chances are, the industry standards have shifted, and they may have to start from the beginning again by upgrading specs.
A Short and Brutal Shelf Life
Despite these obstacles, hardware with some degree of openness does sometimes get released. But remember the challenges of finding a manufacturer? They have to be repeated all over again with the distributors -- and not just once, but region by region.
Typically, the distributors are just as conservative as the manufacturers, and just as cautious about dealing with newcomers and new ideas. Even if they agree to add a product to their catalog, the distributors can easily decide not to encourage their representatives to promote it, which means that in a few months they have effectively removed it from the shelves.
Of course, online sales are a possibility. But meanwhile, the hardware has to be stored somewhere, adding to the cost. Production runs on demand are expensive even in the unlikely event that they are available, and even unassembled units need storage.
Weighing the Odds
I have been generalizing wildly here, but anyone who has ever been involved in producing anything will recognize what I am describing as the norm. And just to make matters worse, open hardware producers typically discover the situation as they are going through it. Inevitably, they make mistakes, which adds still more delays.
But the point is, if you have any sense of the process at all, your knowledge is going to change how you react to news of another attempt at hardware. The process means that, unless a company has been in serious stealth mode, an announcement that a product will be out in six months will rapidly prove to be an outdate guestimate. 12-18 months is more likely, and the obstacles I describe may mean that the product will never actually be released.
For example, as I write, people are waiting for the emergence of the first Steam Machines, the Linux-based gaming consoles. They are convinced that the Steam Machines will utterly transform both Linux and gaming.
As a market category, Steam Machines may do better than other new products, because those who are developing them at least have experience developing software products. However, none of the dozen or so Steam Machines in development have produced more than a prototype after almost a year, and none are likely to be available for buying until halfway through 2015. Given the realities of hardware manufacturing, we will be lucky if half of them see daylight. In fact, a release of 2-4 might be more realistic.
I make that prediction with next to no knowledge of any of the individual efforts. But, having some sense of how hardware manufacturing works, I suspect that it is likely to be closer to what happens next year than all the predictions of a new Golden Age for Linux and gaming. I would be entirely happy being wrong, but the fact remains: what is surprising is not that so many Linux-associated hardware products fail, but that any succeed even briefly.