In recent posts, I’ve been thinking about the situation behind the current grounding of Boeing 737 Max due to two crashes linked to the MCAS. I’ve been thinking more about the fact that MCAS was a new element added to this particular type of aircraft in order to get the aircraft’s behavior to match that of earlier models of the 737. Without MCAS, the aircraft would have become a different type, requiring a distinct rating that qualified pilots must train and test for. As some point in the process, there was a choice: either offer the plane as a new type that requires separate pilot rating, or offer a plane with additional automation that adjusts the behavior to match what current pilots are trained and experienced at flying. The automation was introduced to eliminate the need for training. This was a sales feature because training is expensive to customers, both for the cost of the training itself, and for the fact that there would be a smaller pool of pilots who would be available to keep schedules running.
The MCAS introduction illustrates a broader problem with automation.
I have always thought of automation in terms of enhancing productivity: needing less human labor to accomplish more. There are criticisms or concerns about use of automation in this manner because it inevitably eliminates jobs, or reduces the pay of the worker by needing fewer of his hours to accomplish the same task. Recent examples are with autonomous cars, and autonomous freight in particular. Such automation will inevitably reduce the size of the workforce for that industry. While I share the concerns about automation to replace jobs, in my personal career I embraced automation with the motivation of putting myself out of a job. Somehow, I get into jobs where I find a niche that is hard to fill so that the only way out is to automate it, somehow.
I have overlooked another impact of automation, one that troubles me more than replacing jobs: using automation with the primary motivation being the avoidance of training instead of increasing productivity. There are many jobs that have established coverage hours that require people to be present to do the tasks. We want humans to be available at their designated hours so we can rely on their being available when we need them. We also need a reliable pool of qualified candidates to fill in vacancies, where existing staff can transfer any job-specific skills by showing the skills the staff knows themselves. In this environment, there is a desire to perpetuate the relevance of skills that have been relatively constant over a long period of time.
I am describing this problem from my own perspective in a technical field that is rapidly changing in terms of technologies and the best ways to use them. We deal with this aspect of change with specialized training (frequently a week-long course), or with more extensive certification training and testing that can be obtained reasonably quickly and affordably. The distinction, though is that this training is to get familiar with some new technology so that the worker can incorporate it into his existing practice that is necessary to continue the organization’s internal processes. This specialized training enables the worker to perpetuate his previous skills while using the newer innovations. The training does not address those core skills of corporate practices.
In the case of technical training, automation built into the new technologies will simplify and quicken the training process for that technology. A frequent way to describe this type of automation is reducing the learning curve. Once learned, this automation offers no real productivity. In fact, such automation can hamper the achievable automation by eliminating or hiding the more specialized operations that advanced users could otherwise exploit when they are ready.
This type of automation minimizes the cost of training and enables the organization to preserve existing staff, or to maintain the ease of filling vacancies. The downside for this approach is that it perpetuates established practices necessary to retain the qualifications of the existing staff. The existing staff go about their work as if they are still using older technologies operating within a simulated environment that matches the one they knew in the past but is no longer real.
This is the analogy of the MCAS on the 737 Max. The automation simulates the behavior of an airplane the pilots are familiar with despite the fact that they are flying a plane with significantly different characteristics. The organizations benefit by preserving their pools of pilots who can be assigned to both older or new models depending only on scheduling availability and organizational convenience. Meanwhile, these pilots lack the opportunity to personally master the real characteristics of the newer aircraft, possibly resulting in changes in procedures that would make sense only with the newer aircraft.
I am at all familiar with any of the details of piloting, let alone airline piloting. What little I know, I learned from following the news of catastrophes involving this particular aircraft. I’m sure there are many other examples in other aircraft models, and certainly in different industries.
I am struck by the similarity with a complaint I have had within my own experience. I started my career at the end of an earlier era that emphasized training on the job. Without asserting much authority on the subject, but it was my impression that the earlier period embraced a corporate culture of explicit coaching and training within the organization. One of the consequences of this older model was that all promotions or advanced job opportunities were filled from within the existing workforce. In other words, the term entry-level had a very concrete meaning where nearly everyone had an entry level experience, and that experience was very similar for everyone. You would start with little expectation of skills, and your elevation in rank would be through skills demonstrated in your actual performance on the job.
Another consequence of this older model is that of a long term commitment mutually between employer and employee. The employer needed the employee’s longevity because that was the employer’s primary pool for future advanced openings. The employee’s commitment to the employer was that it was his sole opportunity to fill those future opportunities.
This may be partially fantasy on my part, but that was my impression when entering older organizations at a period of transformation. My first job was not entry level in the same sense as earlier generations. True, my job was lowest in the hierarchy, but the expectations were that I would come to the job already prepared to do the work, or to do the necessary homework on my own to catch up wherever I lacked. In addition, newer opportunities were more frequently than not filled with new hires instead of internally. Of course, I could compete for those openings but my candidacy did not benefit much (if at all) by the fact that I was already inside the organization. It is likely that my established position disqualified me in the sense that the competing applicant had opportunities unavailable to me to obtain or demonstrate his skills, typically through his college education earned while I was working in my previously assigned role.
I started my career in the early 1980s, a time when affordable computing transformed the workplace. I sometimes think that that particular time may have been the worst time to start a job because it was at the cusp of an automation revolution. I entered a workforce where I was encouraged to follow the lead and advice of the more senior staff was fully invested in the older culture (before automation). At the same time, automation was aggressively making its entry, demanding new skills unknown to my seniors, and even demanding mastery of even newer skills unknown to myself.
Somewhere along the line there was a shift in corporate culture. The earlier culture was of gradual career skill progression from entry-level to advanced lead level. The newer culture was career skill preservation in order to perpetuate the occupancy of a particular role.
While corporations need to adopt new technologies, and fill roles to manage those innovations, they also needed to preserve older technologies that they continue to rely on. The problem is that there is no outside source of new labor skilled in obsolete technologies. There is no incentive for younger people to pursue old technologies. Even if they were curious, there are very few educational opportunities to learn obsolete technologies. The only feasible opportunity would be in the old model of entry-level employment into an industry that still used that technology. That type of entry level opportunity no longer exists.
Entry-level positions are where a new hire was expected to learn all of the skills on the job. Such positions may have been lost accidentally through a broader process of adapting to more rapid technology changes. This environment forced employers to fill new positions (involving innovations) with outside staff, and to retain existing staff in their roles with the otherwise technologies that are more or less obsolete everywhere else but inside one organization.
In the past, if you were to meet an older worker, you may reasonably expect him to have some higher level of respect or authority. Today, when you see an older worker, it is likely someone who is still doing what he was hired to do decades in the past. In both cases, these older workers are essential to their organizations, and both are more or less irreplaceable, but for completely different reasons. In the past, the older worker benefited from the progression of responsibility and authority. Now the older worker is preservation of skills required to operate something at closer to an entry level, but in a field no young person would be interested in getting trained to qualify for the hire.
An example that occurred to me was the Y2K bug panic of the late 1990s, where it was discovered that older mainframe software encoded years with the last 2 digits to save precious memory and storage: once the new millennium started, the software would think the current year would be 100 years in the past. To fix this problem, there was a need for burst of employment to review and revise old software, software often written in the late 1960s and early 1970s, using languages popular at that time, and more importantly using software practices at the time. Current colleges were lucky to offer courses on the old languages, and even if they did, few would bother to take them. No colleges at the time would train for the old software practices. The primary way to fill these positions was to delay retirement plans of existing staff, attempt to rehire previous workers who had either retired or moved on to other jobs.
Y2K labor shortage made big news at the time, but I think it was illustrative of something that continues to this day: the creation of an essential worker class consisting primarily of older workers who learned how to do things at the time when they were cutting edge, but those things are otherwise obsolete. Younger workers would not have the necessary context of why things are done the way they continue to need to be done, even in the unlikely event that they would have the basic know-how of the technologies.
Compounding that is the modern hiring practices that eliminate the concept of on-the-job training for entry level positions. When a vacancy opens for some old-timer doing a near entry-level position from decades ago, preference would be given to someone who is postponing retirement (or returning from it) over a younger person. There is an implicit notion of stasis for these positions: the jobs may be essential indefinitely, but at their core they will never change.
As an aside, I think this partly explains the modern anomaly of higher employment for older workers at the expense of younger workers. Organizations have essential jobs in obsolete technologies and practices even while they eagerly attempt to modernize their environment. Given the hopeful replacement of these jobs, and the lack of within-organization preference for new opportunities, these types of jobs will appeal mostly to older workers who have prior relevant experience. Eventually, that workforce will disappear. Just saying.
Back to the topic of this post, there is a need to extend the work life of older workers who are skilled on older technologies and practices that remain essential for the organization even if these are obsolete anywhere else. Meanwhile, the physical technology must be replaced with modern technology that if used properly would require a completely different set of skills and more importantly a different work culture. To preserve the not yet eliminated but essential business practices, we introduce automation to simulate the old world despite the fact that little (if anything) of that old world still exists.
We still use qwerty keyboards originally optimally designed for mechanical typewriters, but thankfully this particular skill became a standard for everyone. Think of what the computing workforce would look like if only older workers knew how to use qwerty keyboards, while newer workers used some other arrangement. Basically the same condition occurs in job categories but for different reasons. Older people are more fluent on older practices, and they require special accommodations to preserve the “look and feel” of what they did in the past. Meanwhile, no such accommodation would be available in newer fields.
Some jobs will be dominated by older workers. Others will be dominated by younger workers. Unfortunately, the jobs dominated by older workers are only that way because the involve practices or investments that the organization has not yet been able to retire or may never be able to do so. Meanwhile, it is easier to replace the more recent technologies, filling roles with outside hires, and letting go of those recently hired for the now obsolete tech.
There is a built-in bias to perpetuate old-person jobs, jobs that lack any entry-level training trajectory yet will remain essential for the organization. At the same time, there is a bias to make young-person jobs into short-term engagements of so-called hot jobs that last only a couple years.
Concerning those old-person jobs, there is a need to preserve the workers in those jobs. Automation accomplishes this through simulating an older world for the worker to interact with despite the fact that the underlying technologies are different. The simulation translated newer more exotic concepts into older more familiar but less specific language or presentation.
In the data-science realm, we have automation to give us SQL interface despite the fact that the underlying data stores have no resemblance to structured relational databases. There remains some appeal to use SQL even for newcomers. The difference is the the older workers will use SQL exclusively while the newcomers would use more direct techniques when those make more sense. The result of which is that both groups will eventually create constructs that the other group would not comprehend or at least not understand why anyone would do it that way.
Much attention is spent on the job-loss implications of introduction of automation to improve productivity. Meanwhile, automation is also used for job-preservation of older workers in outdated yet still essential practices, and this too has some unfortunate implications for the future. Eventually the simulation of an earlier age will fail catastrophically in the fact that that age no longer exists. Alternatively, eventually we will run out of older workers who can work in that simulation.