Hiring is not an incentives problem, it’s fundamentally an information problem
Finding top talent is one of the most important yet challenging tasks for companies.
As the way we work and learn changes, the traditional credentials that represent that work need to change to.
But when you think about it, little has changed in credential land when it comes to making hiring decisions. We still rely on degrees from XYZ University or experience at ABC Company, but using these outdated credentials practically guarantees quality candidates are slipping through the cracks.
Credentials need to be reinvented for the modern way we work, and we have a vision on how to use a network of peers to do that.
Why we need credentials in the first place
Any credential is just a heuristic — a simple shortcut for someone to make quick judgments from, without having to spend a ton of time doing research about a candidate, their work, skills and their overall ‘quality’ as a candidate.
The fundamental economic problem in hiring is one of matching with costly search and bilateral asymmetric information — Source: The science of smart hiring
This is why credentials have value and will continue to exist. What’s more likely to change is the form that these credentials take.
Previously — ‘Proof of Work’ used to mean the positions you had listed on your resume as a proxy for your experience and capabilities.
Today — that same proof-of-work can be found in many other places online, available for anyone to look at.
Github and Dribbble profiles are primary examples of this, but not the only ones. A Twitter account (proof of interestingness or unique point of view). A personal blog or newsletter (proof of clarity of communication) also hold a tremendous wealth of context about a person’s knowledge & skills.
While it’s presently difficult to review individual profiles at scale, years of experience is oftentimes the best alternative, resume-based signal for proof of skill — although this completely neglects high performers who grew quickly in their roles.
The problem is that we now know from over 100 years worth of data that years of job experience is one of the weakest predictors of job performance.
The reality is that uni-dimensional & education-based credentialing is already being devalued by employers in favor of more skill-based signaling methods.
Credentials are Broken, here’s Why
1. Proof of work has expanded, but credentials haven’t caught up
Part of the debate over current credentialing systems comes down to whether you view credentialing as primarily a responsibility of talent (to self-certify) via a trusted educational institution or as a human capital problem for employers handle by becoming better evaluators of talent.
To a large extent, the traditional credential, a four-year university degree, has already been un-bundled by savvy employers.
From a purely rational economic perspective, going to college makes no sense, yet it still ranks as the most powerful (and costly) form of credentialing/signaling in use today.
High productivity workers invest in education, but some less productive workers pay the cost too and thereby masquerade as high quality. — The Dynamics of Costly Signaling (link)
These are some of the current attempts to replace the college degree as a primary means of credentialing:
2. Education-based credentialing and the race to the bottom
Meet the new school, same as the old school.
New takes on credentialing via education such as Udacity or LambdaSchool offer an unbundled version of the college degree, that leaves out many of the constraints and barriers that exist with traditional educational institutions, including geography, cost, among others.
At the end of the day, while this approach can work well for offering previously ‘un-credentialed’ talent new opportunities to break through and prove themselves directly to employers, it is fundamentally limited by the same sorts of scalability bottlenecks that more traditional educational institution’s face, primarily time and cost.
Since Udacity & LamdaSchool are single institutions/accreditors themselves, they both suffer from the same adverse incentives universities face that come from scale, where paradoxically the more students that carry a Lambda or Udacity certification, the less signaling value it offers graduates since the value of the credential is directly tied to the educational brand.
2.1 Employer-based Credentialing
Meet the new boss, same as the old boss.
While it's not really a pure form of credentialing today, it's worth mentioning the various problems with employer-based credentialing: inherent biases, heterogeneous evaluation processes and more.
The reason these processes can’t form the basis of a universally accepted credential is that Interview scores, Take-home work assignments and other kinds of custom evaluations aren’t portable and are thus silo-ed away within individual organizational walls.
Not to mention the actual work that’s done while employed at a company.
3. Experience > Exams, Test just don’t cut it.
Meet the new exam, same as the old exam.
The premise behind outsourced certification providers like Triplebyte, Hackerrank, Karat (in coding) or Degreed (in other fields) sounds great in theory.
They make the process of evaluating a well-defined pool of talent, e.g: software engineers and systemize/standardize it, to the point where it closely resembles something like the SAT. Coupled with an open application process that is maximally inclusive at the top of the funnel and it sounds like a valuable and scalable solution on paper.
This problem is that this concept only makes sense for:
- An overly simplistic world — in which hard-skills are the only or primary means by which a potential hire is judged. Just like any test that becomes standardized, it can easily be gamed over time as potential takers study to the test.
- A static world — in which hard skills aren’t frequently acquired or lost. In other words, it ignores the current world we live in where most learning of new skills happens ‘just-in-time’, on the job.
Additionally, exams don’t make sense in a world where most ‘work product’ is already migrating to being captured and examined online.
An ideal digital credentialing system would instead allow for capturing broad sets of organic actions, not narrow sets of contrived, synthetic ones to translate into credentials.
It would lower costs tremendously and give everyone access to these credentials, without degrading the value of the credential itself.
A new credential
In a perfect world, we’d be able to dynamically issue and update credentials that factored in all past and current work (not just publicly visible work).
In the real (offline) world, we already have a few proxy’s for making trusted recommendations and hiring decisions, including:
The personal recommendation
The reason that recommendations/referrals still work so well for making hiring decisions, in spite of the fact that they are not very scalable, is that they are still costly to make and thus valuable forms of signaling as well as having personal reputation at stake.
This type of information can add color to a credential built around ‘soft skills’ like communication, honesty, and stick-to-itiveness.
In the digital world, we’ve already discussed how there are many artifacts e.g: Github commits logs, that can give us an indication of hard skills and technical knowledge.
“Proof of work” for hard skills
This information, if properly analyzed, can be a treasure trove of context that can help guide hiring decisions that are based around technical skills.
In this model: coming up with a digital ‘fingerprint’ of someone’s technical skills, can be a living and breathing record of their accomplishments, vs a static test that only captures a narrow subset of those skills at a place in time.