Solutions to the upcoming crisis of human obsolescence.

Solving the biggest issues in the future of work.

A recent breakfast discussion with Dean Nitin Nohria, Dean of the Harvard Business School, inspired this article.

The harsh truth is that if you are an employee, you are either already obsolete — and won’t get a coveted role — or you are going to become obsolete in that role within 5 years. I’m not kidding.

• For example, in just 10 years, jobs in the legal industry have been transformed largely due to the increased application of legal software systems. The result is fewer highly-paid legal jobs, reduced demand for law school graduates, and the emergence of a higher volume of low-paying jobs to complete legal tasks aided by software systems (such as e-Discovery review at ~$30/hr).

The half-life of talent is scary and has to be faced. Artificial intelligence, software, automation and robots are changing every job, blue and white collar alike.

The pace of technology is such that newer ways of working, newer literacies, newer competencies are showing up every day. If solutions are not created, as a society we will be faced with mass unemployment, decreased quality of life and increased poverty.

Good work is being done on many fronts: in policy, on the front-lines managing a business’s talent pool, operating workforce development agencies, delivering higher education, building talent recruiting systems, and so on.

But which of these solutions is really going to make a dent in this problem? Which initiatives are the ones to support and invest in?

In this article learn about structural solutions that have the potential to massively change outcomes in reducing human obsolescence.

What is “Structural” Change?

Let’s start with a case study that demonstrates the jaw-dropping power of structural change from the manufacturing industry.

Until the 1980s, the US manufacturing industry was leading the world. US manufacturers believed that there is a firm tradeoff between Quality and Cost. The manufacturing department and the quality department were two separate entities, and if higher quality products were to be shipped to customers, more products found their way into the rejects pile. Supporting this world view was a wealth of related business management techniques.

Higher quality products meant incurring higher costs. Producing lower cost products meant lower quality. It felt like a universal truth.

Japanese manufacturers were known for shoddy, low cost manufacturing. In the 1940s they sought to change this. Using little-used research that had, in fact, originated in the US, they eliminated the quality department. Instead they imbued quality across the entire manufacturing process, applying Statistical Quality Control methods to identify and remediate the sources of defects along the entire manufacturing chain. They invented the business practices of Total Quality Management, lean manufacturing, and kaizen. These methods worked incredibly well and by the 1970s Japanese companies were eating the lunch of US manufacturers who lamented If the Japanese Can… Why Can’t We?.

Japanese companies showed manufacturing could be done with higher quality AND lower cost. This way of managing manufacturing has become standard operating practice for manufacturing everywhere. The cost and quality tradeoff was broken.

Structural Change Ideas for Labor & Work Markets

Big problems require structural change.

The question is, what are the structural changes that could stop human obsolescence? What are the breakthrough ideas that apply to labor?

Here’s an initial list of 5 areas to look at — with initial ideas described— and I’m looking for more.

1. Blur the line between learning and working

There has always been a tradeoff being learning and working: invest in learning and THEN work; invest in a college degree and then you’ll get a better paying job; give employees time off from work to complete training.

We could simply accept this tradeoff: let people try harder to manage a part-time job while studying, or provide online learning that can be juggled with a job.

The structural question asks, though: how might we actually break the tradeoff between learning and working?

Apprenticeships for white collar work.

Learning is defined as acquiring knowledge and being able to subsequently apply it in new situations. Last year Accenture spent $950M in class-based training of 440,000 employees. When they reviewed the impact — using a rubric of 8 elements of durable knowledge — they discovered that their programs were failing: only 25% of their programs made the cut, and none excelled. Neuroscience tells us several pertinent ideas:

• (1) The brain is plastic throughout life, not just when young; (2) Learning is not measured in training hours but in attentive moments; in those attentive moments learning happens very fast; (3) Learning requires oxytocin, a primary modulator of brain rewiring, which increases when the learner feels trust and falls when the learner feels fear; (4) learning is boosted through constructive criticism, or corrective feedback, especially if the learner has a growth mindset. (My thanks to Dean Dan Schwartz, Dean of the Stanford Graduate School of Education, whose lecture I attended).

These are hall-marks of apprenticeship programs where a learner is working on actual work, in a trusted environment, to learn a particular skill or set of skills from someone who already has mastery. While common in trades, the apprenticeship model could be applied to white collar work. Rather than going off to “do training”, completing an apprenticeship within the workplace might have a higher impact in achieving learning outcomes.

10 year degrees.

Bill Gates and others recommend we spend 5 hours per week to learn something new. Professor Tyler Cowan (author of Average is Over) urges us to retrain ourselves every 3- 5 years. “Becoming life long learners” is a well-used mantra in the media and amongst teaching professionals.

Achieving this requires good intentions to follow through — and only a small percentage of people can actually do this.

My MBA at Harvard Business School took almost 2 years. What if Harvard Business School made its MBA not a 2-year degree but a 10 year one? 1 year on campus coupled with additional learning over the next 9 years. Employers employing a Harvard MBA would know their new hire is going to stay current. I wonder if large employers of business management personnel (such as General Electric, McKinsey, Bain Consulting and Accenture, to mention just a few) would support this?

While I use the example of the MBA, this could be applied to all kinds of degrees. Rather than individuals stringing together “life long learning” in an ad hoc way, life long learning would become a systematic focused effort. Rather than life long learning relying on “good intentions” this could systematically entrench “life long learning” into the fabric of work.

2. “Learning is expensive”

We believe that learning is expensive, due to the tuition and time spent, and we are unsure of the returns. Embarking on a course of learning is fraught with friction.

We know however that being able to afford to learn is orthogonal to the ROI of learning. Having financial commitments (like supporting yourself, supporting a family, or commitment to a mortgage) is a circumstantial fact. People with these circumstances are not less capable or less motivated to become more relevant in the workforce.

Taking financing out of the decision to re-train makes huge sense.

Financial arrangements tied to future earning.

Boot camps focused on training people for technology sector jobs are leading the way in de-coupling the cost of tuition from the desire to learn. Income-sharing agreements eliminate upfront tuition and replaces it with future payments calculated as a percentage of future earnings.

• For example, Flatiron’s deferred tuition program allows students to pay $2,000 upfront then 10% of their income once they start to earn at least $40k, for a period of 4 years with a cap on the total paid.

Some even offer a living stipend in return for a share of future earnings (see details from Flatiron, Lambda School, and others). These kinds of financial arrangements for training in other sectors would systematically increase the flow of motivated students into learning endeavors.

3. Learning benefits are “unique” to each learner

There is a belief that the value of training is hard to measure and benefits are hard to quantify. You’ll hear comments that it depends a lot on the individual concerned, and of successful people who gained success despite their lack of specific training. This creates an overall lack of accountability between training and impact. According to this thinking, learners are such individual snowflakes that the impact of a program of training can’t be predicted. This leads to a general environment in which learners invest in learning activities as a leap of faith, having to believe they will get a sufficient ROI. Any kind of investment falls under circumstances of uncertainty. Structural changes would work to reduce the environment of uncertainty.

Mandatory impact/outcome data for training programs.

Many consumer products like cars, fitness supplements or wrinkle-creams exist in a world where decisions are made in the midst of massive marketing expenditures, and buyers muddle through claims and endorsements made by vendors to make their ultimate decisions.

Learning is however a massively high stakes decision. Buyers can’t see through advertising claims. Vendors that offer training people, must talk about the success and failure modes of their offering. This should be information that is freely and cleanly available in all industries, regardless of the institution type, for profit or non profit, or length of the training.

• An example of how this could be done: Coursereport.com covers one sliver of the technology labor market: entry level jobs in web development, mobile applications, data science, UX design, digital marketing, product management, and cybersecurity. The site covers for-profit bootcamps that operate in this field. The site provides easy to read information about graduation rate, employment rate, salary, and unstructured information in the form of Yelp-like user reviews. Coursereport.com gets about 250,000 visitors/month who spend an average of 2 minutes on the site, reading 2.5 pages. Others like Switchup.org provide rankings as well.
• Another example: YearUp uses the gold-standard — a Random Controlled Test. It’s clear which kinds of students are the “inputs” to the program, what the training is, and what impact was created. While the results need to be boiled down for the lay-person to understand and posted more prominently on the website, kudos to YearUp on measuring and reporting the impact of the training program.

4. Stop treating hiring as an art: tighten the link between skills and jobs

The college admission process in the US looks very different from other countries. There’s a spectrum of how predictable the process is for high school students seeking admission to college. The US process is at one end of the spectrum — more subjective and less predictable — while other countries at the other other end, with a process that is less subjective and more predictable. It’s actually a structural choice.

The hiring process can be thought of on the same spectrum. Currently it is skewed towards being more subjective and less predictable. We might want to make it less subjective and more predictable. Doing so would reduce the uncertainty around undertaking training efforts to stay relevant for new tasks and new jobs.

Here are two ideas around reducing subjectivity, and I’m curious about others:

Demonstrate or prove skills as part of job applications.

Skills can be gained in any number of settings: at a school (well ranked or unknown), through employment, or in a non-traditional setting such as volunteering. Many of these ways of acquiring skills tend to be overlooked in the hiring process. Job applications could require a micro-task to be completed, and the candidate’s performance on the assessment task heavily weighted in the filtering process.

• Several technology companies work on this kind of candidate assessment process — Pymetrics, Aspiring Minds and many other providers.

I wondered, would I mind being asked to prove my own abilities in, say product design? Would I be affronted being asked to demonstrate something I’ve done for years? Or would I go ahead and take the test? I think it would take some getting used to the first time, but in the end I’d be ok with it.

Provide Applicant Tracking Systems with more recent, diverse data.

Due to the internet, the volume of job applications received by employers exponentially increased. The initial wave of Applicant Tracking Systems (ATS) managed workflows and were based on simplifying specific HR tasks, for instance, narrowing down a high volume of applicants to a short-list of candidates to screen using Boolean (key word) search tools.

The latest ATS systems use machine learning to determine a subset of candidates that would match a particular role, but without being tied to explicit Boolean search terms. Instead, data about existing employees is used to detect similar characteristics in the applicants.

However, newly acquired skills don’t make it into the process that machine learning is operating on. The machine learning algorithms are looking at past data about current employees. Someone completing a new bootcamp might have acquired excellent skills for a particular role, but the training data for the machine learning algorithm has never seen this in its dataset before and therefore can not weight it.

Can we train machine learning algorithms to be more “present” rather than backward looking? We could, using the outcomes from candidate assessments as training data. If current employees and prospective applicants both take the same assessments, the patterns in this data would allow better detection of strong candidates in the applicant pool by the machine learning algorithm.

5. Re-think what makes an employer a good place to work

Employers strive to be a good place to work. I have two structural changes around how changes in the employer environment can overall improve the ability for individuals to remain productive and less likely to become obsolete:

Forced Brutal Honesty.

Information dissemination isn’t as clear as you’d think. At Siebel Systems I led the creation of Employee Relationship Management, an enterprise software product which aligned employees around company goals (similar to OKRs) and engaged managers and employees in quick, useful dialogs around performance. This was in response to almost all employees in large companies agreeing with the statement “I don’t know how what I do fits into the company’s direction”, despite immense efforts by management to communicate with employees.

Making information crystal clear when it comes to individual productivity could be a place to start. Imagine receiving your W2 each year, along with a brutally honest answer to a single question:

No matter how nice the manager, the work environment, the current work or the current pay, the answer to this question matters. Anything less than “probably” is going trigger the individual into thinking about and (possibly) taking action on how to become more valuable.

• AT&T and IBM have started down this path, announcing to large numbers of employees that they are in positions that are going to become obsolete, while launching paths for those employees to re-train to the jobs that are emerging.

Remote first.

Remote work has emerged in response to talent shortages in some locations, because relocating talent isn’t that easy. Yet remote employees (or remote contractors) are usually not as productive as those in the office. Many top-notch employees would rather be onsite at a company than work remote for a different company because it doesn’t feel impactful to be remote. Promotions come more slowly. Knowledge transfer is lack luster. Taking a remote job can feel like a compromise.

Could we structurally accelerate the emergence of remote-first companies, that are centered around the productivity of remote employees? What if remote-first companies — say, those with 90%+ employees remote — got a significant tax break? Companies would invest in systems to produce productive outcomes with remote employees, and the tax break would allow them to be more competitive than others without remote workforces.

Where Next?

I believe the way forward is to be very clear that we’re looking for structural change. Structural changes make a huge difference, yet don’t rely on the good intentions of individuals or the power of marketing. The problem of impending human obsolescence requires it.

With a good set of candidate solutions in hand, the next step is to run experiments to test the ideas. This is analogous to the first factories in Japan where US researchers (Deming & company) applied their statistical control theories to improve manufacturing quality.

With promising experimental results, we can amplify that knowledge to speed implementation by others.

The Total Quality Management story is one of 40 years. But we can go faster. I’m excited about the path ahead. Please let me know what you think.

Continue the discussion (Twitter)

Sunita Parbhu is a seasoned startup product executive with expertise in labor markets. Currently undertaking an independent study on the opportunities for improving labor economics in the field of The Future of Work. Sunita is a proponent of thinking from first principles. She has worked closely with tech leaders such as Tom Siebel (Siebel Systems, C3.ai) and James Currier (NfX Fund) to identify and create new markets from first principles. Sunita is an economist, a Fulbright Scholar, and held leadership roles in multiple technology startups with exits exceeding $6B. Her education includes two degrees from Victoria University of Wellington in New Zealand and an MBA from the Harvard Business School.