3D printing: what you (probably) haven't considered (yet)

^01c March 22, 2014 -- (asphalt)

A couple of days ago I was reading an article called "The Hardware Hacker Manifesto", in which the author laments on the sorry state of user freedoms in regard to hacking devices. He makes a fine point, but he only barely scratches the surface of the problem with the tip of his finger; because let's face it, "hardware hacking" wouldn't be possible without the help of actual open hardware, as well as "software hacking" wouldn't be possible without actual "open software", be it "open source" or "free software".

In a fashion quite similar to software, hardware offers various degrees of freedom. The main difference between the two lies in the nature of the restrictions which get to be applied upon users or hackers: while software artificially imposes a so-called "license" which allows or disallows the user to modify, redistribute or sell a given program, hardware restrictions are, well... hard. Vendors could in theory come up with a piece of silicon on a PCB along with a schematic of said PCB, and market it as "open"; this already happens in practice with the Raspberry Pi, Arduino, Galileo and other wonderful pieces of hardware. There's only one tiny problem about them: they're not open, or rather, they're not fully open.

So at this point you find yourself in the rather nasty situation of trusting the chip producers¹. Even more, you find yourself in the even nastier situation of trusting the chip designers². So what's all this nonsense about "trust" anyway?

Well, for example Intel implement this neat instruction called RdRand, which is supposed to provide users with a "trustable" source of random numbers. There is however a non-zero, and most probably non-negligible, probability that Intel, being an American company, are politically influenced by parties which are anything but trustworthy, namely the NSA, nowadays hated by most people who have an interest in security. The same goes for any other hardware manufacturer, which begs the question: are we going to implement our own pieces of hardware anytime soon?

This isn't news; open hardware enthusiasts have been thinking about this for more than a year now. The goal is attainable not only with a CPU, which can be easily designed by a second-year Computer Engineering undergrad, but also with GPUs and other high-performance hardware that would catch highly secretive hardware makers such as Nvidia off-balance. This is doable, in fact MIPS cores have provided a starting point for years, while OpenRISC has the true potential of becoming the Linux of hardware. Still, there's only one tiny problem about that: we don't have the technology to brew our custom pieces of silicon at home; not yet.

Surely, FPGAs are a fairly good solution, although their cost can go at least one order of magnitude higher than the Raspberry Pi. Still, FPGAs are more trustable than a Pi, while ASICs are even more trustable and expensive as shit. Otherwise if you're looking for security at the expense of large-scale integration, then you might as well go solder your own stuff, just like dad used to thirty years ago.

Now consider the following idea: at the time of writing, the costs of 3D printers vary about the same as FPGA costs. Printing electronics is still a hot research topic, so by the time this goes mainstream, FPGAs will probably cost less than the Raspberry Pi does now. Single-board computers will probably cost less than $5. A couple of decades later everyone and their dog will be able to print custom phones, or even better, hire robots to design their favourite tech junk according to some informal specification.

I'll admit that I'm sounding like an overly-optimistic prick right now. But as a science and technology enthusiast and a person with an educated view on the subject, I predict without even the slightest hint of optimism that this will certainly happen. Mark my words, there's no way around it; it's either this or the dark ages.