Aviation vs. the computer industry
We’re back after some time off. This year I actually got to take a week of vacation, a rare occurrence here at the Advosys Consulting world headquarters. Of course, there were the usual small projects for clients and taking a couple online courses so it wasn’t entirely idle time.
The Ottawa Air Show was held in June this year. The show reminded me of all the differences between the aviation and computing industries, especially how far computing has yet to come.
I earned my private pilot license in 1998. Though I don’t fly right now, two things always struck me as I was going though the training: first, a lot of computer people become pilots, and second, computing could learn a lot from the way things are done in aviation.
There’s little mystery why many so computer people are attracted to flying: like computing, aviation is a highly technical and multi-disciplinary field: flying requires knowledge of aerodynamics, engines, navigation, meteorology and regulations. Also like computing, flying is a relatively rare skill that tends to impress people when you talk about it at parties and in blogs.
Those factors also appealed to me, but what was also interesting was the conservative nature of aviation professionals and the over-engineering of every aircraft and aviation system.
“New” equals “bad” in the minds of pilots and aircraft engineers. For example, fuel injection is considered a radical unproved new concept in small aircraft engine, even though automobiles have used it since the 1970s. I trained on the venerable Cessna 172 which has been nearly unchanged since it was introduced in 1955. When Cessna added fuel injection to recent models, many pilots refused to fly them. Even the new piston aircraft with fuel injection still use a manual fuel mixture… as you gain altitude you have to lean out the fuel/air ratio using a knob, unlike cars which do that automatically using oxygen sensors.
Likewise, GPS navigation has been widely available for over a decade. Yet it is still mostly illegal to use for aircraft navigation… pilots are still required to use the aging system of radio nav aids or visual reference to the ground as the primary way getting around.
That’s how it is in aviation: every new idea large or small is viewed with skepticism, thought about, hotly debated, tested, and re-tested. Even after being adopted, no change is trusted by most pilots until there are a few decades of proof behind it.
In computing, ideas and technologies only a couple years old are considered antiquated. Most IT professionals scramble to be the first to adopt new technology, often for no better reason that “it’s new”. The factors of “geewiz” and lust fueled by rabid over-marketing rule the IT profession. In computing, systems don’t have time to be perfected before being replaced with a radically “better” idea, with the consequences that best practices never mature and staff experience is changed from an asset into a liability.
The aviation industry tends to learn from mistakes. When an incident happens there is usually a formal investigation to determine the cause and how to prevent it from happening again. Errors are investigated, documented publicly, and corrected. Most of the time, aviation incidents are a result of human error, not mechanical failure. Even so, changes in procedures, regulations, and design are proposed and tested to reduce the risk of those human factors.
For example, early retractable gear aircraft used a normal switch to raise and lower the wheels. Even though extending the gear was part of the written landing procedure, pilots kept hitting the wrong switch, resulting in many belly landings. Now the gear switch in all aircraft is shaped like an actual wheel, even in small private planes. Pilots can tell instantly by touch they have the right switch. Belly landings have become rare events.
Compare that to computing, where it is still standard practice to admonish users to never to open e-mail from an unrecognized sender, lest an attachment or HTML unleash a worm that destroys the corporate network. Obviously e-mail software and networks that frail should never have been designed or used, but less obviously such a policy ignores the human factor that people will always open new mail, even it if the subject line says “Click me to destroy your network”.
Disasters in computing, such as massive outages, financial loss and security breaches, are frequently covered up. Bugs and design flaws are rarely tracked down and reported. Experiences are almost never shared openly or learned from among industry players. And anyway, there’s usually little point in learning from mistakes since the hardware and software is likely due to be replaced with something completely different in a few months.
Obviously the stakes in aviation are much higher than in computing: when aviation technology fails, people tend to die. Most computing failures aren’t yet killing large numbers of people (failures in computerized control, navigation and medical systems have killed, but low numbers of people). Computing failures are causing massive economic losses, sometimes destroying companies, but that doesn’t seem to be enough yet to make industry practices more prudent and thoughtful.
Aviation has been around just over 100 years. It was wild in the early years with many truly stupid aircraft designs, reckless pilots, and companies more interested in money than public safety. But the aviation industry matured. Now failures are rare: thousands of flights are completed each day without incident. Training, procedures, regulations and the careful, responsible people who do the engineering, operating and maintenance have turned an inherently dangerous activity into the safest form of transportation there is.
Not only that, but massive projects like the Boeing 777 and Airbus A380 can be designed and built with confidence that they will perform as designed and won’t crash on the first test flight. Most medium to large IT projects, such as the infamous FBI Sentinel boondoggle, either fail completely or come in late, seriously flawed and vastly over budget.
The extreme conservatism in aviation has some downsides. For example it has slowed adoption of GPS for navigation, use of composites for airframes and stall-proof canard wing designs.
Computing has been around maybe 65 years, microcomputers maybe 25 and widespread use of the Internet only a decade or so. Disasters are still the rule, not the exception. Fortunately, progress is happening… there seem to be fewer cowboys and Rube Goldberg architectures around these days than ten years ago. Methodologies like ITIL are gaining acceptance and starting to bring order and thought to the change process.
As an industry we have a lot of maturing to do, and many practices in aviation are good models for the IT industry to learn from.
Related posts: