What a useful certification usually actually says isn't "this person can use this tool", it's "this person has been educated in a certain way and paid attention during that education, leading to the ability to use this tool".
For the vast, vast majority of certs out there where you don't have to take a course to take the exam, you can knowledge-cram for a comparatively short period of time, then take the exam without actually having much experience, even if "only" in a lab, then forget most of what you've learned shortly afterwards. As a result, few employers care about them.
If you want to do certs separate from education paths, you have to figure out a better way of testing people - possibly real-world (or at least lab-based, scenario-oriented) testing over a period of time.
Where I disagree is the tacit implication that, therefore university education is somehow better at being an indicator of those skills and qualities. I've seen a lot of terrible graduates who I wouldn't employ if they paid me: graduates who couldn't reason their way out of a paper bag. So I'd respectfully disagree with that implication.
It does tend to be a better indicator - just because it's not perfect doesn't make it the same.
In the world of IT, university degrees mean very little, as universities almost universally fail to teach anything useful for that industry.
There's still very few certs that mean anything. At the entry level, all of the certs are dreadful, and past the entry level, experience usually matters more than what certs you have.
Most state-administered professional certification exams have similar eligibility requirements.
You could theoretically get certified with only high school and work experience, although I don't think any employers would hire someone without a bachelor's. The degree can really be in anything though as long as you've passed an exam or two, as this shows an employer that you're serious and have learned at least something.
I believe the system works in our case because there is an ecosystem of employers, employees, recruiters, and government agencies that all buy into the exam process. It's sort of a throwback guild system that's not very similar to IT certifications such as CCNA.
Neither MOOCs nor certification programs offer that key function. When Coursera only accepts people with a certain SAT score, then issues "degrees" ranking candidates by class performance, I think we'll see employers take them more seriously.
Filtering alone doesn't explain why those companies go 50% deep into the pool of Harvard graduates before considering the top 5% of UCLA grads.
But I partially agree: certification programs need to indicate degree of attainment, not just pass or fail. I disagree that SAT entry requirements is particularly important to an employer. If I wanted to make sure a candidate had a certain SAT score, I could ask them for it, couldn't I?
I see something similar happen on a smaller scale when a business tries to re-design a successful legacy product: The conditions that made it possible to design the original product no longer exist. Nobody really understands why the legacy product was successful, and the original designers are long gone -- they were "acqui-fired."
The legacy product might have a number of glaring shortcomings, but the re-design team ends up creating something that's worse.
I realise that degree-certification-bodies isn't what you meant by "certification" -- that you're thinking more of the "industry cert" model, where a training centre teaches you material, and then you pay to take an exam from the certifier. The point though is that's not what society (or at least as represented by government) wants university to be. It is, at least currently, supposed to be education not just training and testing.
That's not so surprising to me, in that a degree tends to be an entry to a career. And that career can be long enough that competencies are likely to change in unpredictable ways over your career. So, it makes some sense for graduate employers to be more interested in an education they believe means you are likely adapt well to those changes, and perhaps lead those changes. Rather than a competency certification that shows you know the current state-of-play.
Whether that will change over time is an interesting question. If the market does move towards just certifying competencies, I have an awkward suspicion that it will be because of general trends towards workforce casualisation. If you see employees as disposable (hire the skills of today, and tomorrow fire them to hire the skills of tomorrow) then education is less important than today's competency.
This isn't intended as a criticism of MOOCs, nor in any sense to diminish them. I'm not only interested in them, but trying to develop some -- I have something of a vested interest towards MOOCs... It's just that I'm personally more interested in how we can improve MOOCs to offer an education, rather than how we can get the market to stop demanding an education and just want a competency-certification.
By competency, I don't mean some narrow technical expertise, but whatever skills it is that an employer needs their workforce to have, that universities claim to impart.
There's actually one or two other concerns, too: networking and fun. Some of the value I and many other people got from college is meeting others with similar interests.
Regarding fun (and parties), I get the sense that most HN readers probably aren't or weren't party animals, but that's actually a pretty common reason for people to go to college, per Sperber's Beer and Circus (http://www.amazon.com/Beer-Circus-Crippling-Undergraduate-Ed...) and Armstrong and Hamilton's Paying for the Party (http://jakeseliger.com/2014/04/27/paying-for-the-party-eliza...).
I don't really have a sense of how the university bundle really breaks down into core value.
Greatly complicating things is the fact that you often don't need to know anything in particular to succeed. There is no shortage of people who learned on the job and became successful. Filtering out such people with an exam may badly damage your hiring.
* General Computer Science Cert
* Software Architecture Cert
* C Lang Cert or a Java Cert
from an accredited institution and be instantly employable at the entry level. This is kind of how it works with Network Engineering, I wonder why it can't with Programming/SoftDev.
The certification is what you need. But to get it, the university can require you to purchase a whole bunch of very expensive education you may not need.
Only by separating the two will this problem be solved: having certification authorities that are independent of educational providers.
In that utopia, even a regular university can't grant you the certification. They can market themselves as the best and most cost effective way for you to reach the competency needed to achieve your certificate. But, perhaps MOOCs are better, or intensive library study, or perhaps you're already that competent, and don't need any education to get there.
It is only that which can solve the scandal of indebting vast swathes of young people by tens of thousands of dollars.
So I'm suspicious of this article, because the solutions described still make the same mistakes. The body that provides you with the education, certifies you as having had that education. Okay there may be more evidence as metadata, but still, the incentive is there for providers to bias their certification towards the educational products they want to sell.