Open Systems

Published in IEEE Spectrum Magazine, March 2011


Occasionally I indulge in a bit of nostalgia over those long ago days of Heathkit.  Of course, building a Heathkit was similar to doing a Rembrandt by paint-by-the-numbers, but I remember how nevertheless it resulted in a deep sense of pride of ownership.  What I’m considering today, however, is the openness of those old Heathkits.  Whatever might go wrong with the Heathkit, I knew I could fix.  I had easy access to all the parts and I had complete schematics and circuit descriptions.  It was truly mine.  The electronics world of today is profoundly different.  Like most houses, mine is littered with electronic gadgets, none of which could be considered to be open.  “No user serviceable parts inside” is the ubiquitous phrase of warning – like you’re going to electrocuted if you open the back of the gadget, or worse yet, void the warranty.  It has gotten so bad that a lot of very popular items won’t let you add memory or even change the battery!

These systems are closed, but they also differ in nature from the old Heathkits in some very fundamental ways.  For one thing, they very seldom break, so having the information needed to fix them is usually irrelevant.  Moreover, there usually is almost nothing inside anyway, since everything today is an embedded processor and some firmware.  But this is really important, because it means that whatever this gadget is and does, with a change in firmware it could be different, do different, and be more -- except, of course, that you’re not allowed to touch it.  In contrast, the old Heathkit, which you were allowed to touch, was what it was, and could never be more or different.

Understandably, manufacturers are reluctant to open their designs.  They fear someone else will make their same product more cheaply and undercut their market.  They also fear a loss of control and they worry about added support expense when people turn to them for replacement or help after they themselves have ruined their device.  They’ve put a lot of time, effort, and expense into creating a proprietary design; why give it away?

Given such considerations, there is only a short list of open products.  For experimenters examples include a camera, robot, and 3D printer.  Something between experiment and consumer product is an Internet clock/radio, and as a consumer product there is a wireless router which, perhaps by accident, is open by virtue of having used the Linux operating system.

Some devices get opened up by being jailbroken by hackers, and perhaps a small bit of anarchy to stimulate innovation might not be entirely a bad thing.  But there is a dilemma here, because in recent decades we have seen the tremendous potential value in opening innovation up to the public based upon a stable, underlying platform.  The Internet itself is the greatest example – a core platform consisting of a set of protocols – upon which millions upon millions of people have innovated and created enormous value for all humanity.

On a smaller scale we do have partially-open systems that have inspired a great deal of innovation.  Many systems have published APIs that allow third-party add-ons.  More recently, the advent of the app store has provided a wonderful model for allowing public innovation around a closed, proprietary platform.  For the platform manufacturer, this can increase their market many fold, and at little expense for themselves.

In maintaining a stable platform there is always the danger of being painted into a corner, where it becomes impossible to change the platform itself because of everything that has come to surround that platform.  Again, the Internet itself is a prime example.  A lot of people, now given hindsight, would like to change the underlying protocols of the Internet.  But the weight of legacy is so overwhelming that even the thought of such change is tantamount to heresy.

The whole question of how much and where a system should be open is fascinating.  With the PC, IBM opened the design to third parties, while Apple maintained a closed design.  The market chose the IBM design, but it was ultimately a market in which IBM itself had no inherent advantage.   On the other hand, Apple has been able to maintain its market – albeit with a smaller share – through the years, as well as its independence and design integrity.

As an engineer, I’m conflicted.  I don’t like these devices that lock me out.  But intellectually I understand that manufacturers have to make money or they couldn’t create these great products.  I also understand the value of maintaining stable platforms.  Nevertheless, I’m an engineer and yearn to tinker with these closed boxes.