From Reid Hoffmann, during a podcast with Russ Roberts.
The whole notion of separating the academy from the work life is insane.
That is a radical statement. It is even more radical than the view that I associate with the Montessori movement that separating students by age is insane.
The radical view goes something like this: in ordinary life, people learn naturally. The academy is an artificial environment, which tries to stimulate learning in an alternate reality, a sim if you will.
If so, what does that make educational software? A sim of a sim?
I think that there are some deep issues here.
I think given how successful specialization has been in many other things, the broad claim is approaching the kind of extraordinariness that requires extraordinary evidence. Some kinds of activities are easy to separate, some are not, quite a few vary enough depending on details that generalities are tricky. (Does it work to separate computer software expertise from the rest of a given business? Beware of answers that are less than a two-page memo.)
What he means in practice, though, may be something close to “once you separate education from practice, you are flying on instruments, with no easy guides to limit how badly you can screw up.” I would agree with that. Further, I suspect he and I agree that the state of the art in our social science instruments and in our social science institutions and government institutions is not very impressive, and even further agree that performance of many educational policies looks far too much like “controlled flight into terrain”. Therefore, as far as policy consequences go, I think he might be arriving at a roughly-correct conclusion. But I think if so it’s by the interaction of the factors above, not the first-principles reasons that you’d expect from the quoted generalization.
Everything I need to learn I get from scouring the internet and then my resume becomes my verification of projects worked on or delivered.
Can this system be improved or bridged and integrated with formal education?
So I have made some videos of my stuff. Is there value in publishing super specialized training videos and then certifying that someone watched them with a quiz?
It is not insane.
Once you attach learning to work, you are obliged to make that relationship work as a market transaction. That is not necessarily a bug in many circumstances, but it seriously limits the bounds of education.
Perhaps he should have said “having so little of our education connected to our work life is insane”.
Higher education started out as an extremely precious and specialized endeavor for a tiny monastic elite, and grew gradually to support a larger body of people desiring training for professions like law, medicine, and engineering, and even those were arguably cases of the academy capturing training that had been done mostly by apprenticeship. Applying the academy model to more than ten percent of the population makes as much sense as requiring a Computer Science degree to be able to purchase a computer.
Isn’t this like saying you should only learn to box in the boxing ring or learn to fly by getting in a plane? I’d argue for training and learning in simulated situations before trying the real thing. The question is how close to the real thing are you making the training so the lessons can be carried over.
As a pilot who boxes regularly, I can tell you that the number of hours spent flying or throwing punches were far greater than those spent reading or being lectured on either. Flight simulators are useful precisely because they feel so similar to flying a plane. Classroom education as we do it today actively disavows trying to achieve that level of fidelity.
“Classroom education as we do it today actively disavows trying to achieve that level of fidelity.”
True, and we do that too much, and a considerable amount of the teaching is incompetent, and a considerable amount of the target material is irrelevant. But I think there are nonetheless some important cases where it is appropriate to do that, teaching fundamentals which are too distant from the goal activity to be learned naturally by practicing the goal activity. One that seems to have worked well for me, and that a number of other people have also remarked worked well (search for “Andreessen assault rifle”:-), is the core STEM mathematics curriculum. Our society fetishizes it too much, teaching cargo cult calculus and/or statistics to too many people who are pursuing careers without many opportunities to use even the non-cargo-cult versions, and giving too much respect to misguided technical work that is dolled up with misguided mathematical formalism. Nonetheless it can be done right, and learning and using the tools for systematically estimating and approximating and simulating and optimizing is rather powerful. Opportunities to use them formally are not scarce, especially when computers make some opportunities orders of magnitude cheaper. And learning to use them formally seems to be a significant advantage even in situations where computing everything formally on the fly isn’t practical, like poker or various kinds of basic finance, where people who haven’t learned to think straight about uncertainty and optimization and estimation seem to operate at a disadvantage. Against that, various legal trends are introducing a lot of friction, too, but not enough to actually eliminate all the incentives for doing things better. At the high end, there is lot of money in understanding tricky things like speech recognition or the distribution of pennies and steamrollers in markets; at the low end, there is a lot of money even in straightforward things like simple A/B testing of websites. And within specialties like “computer networking” and “computer security” unexploited opportunities range from very tricky to facepalmingly obvious.
Also, even though I am unusually good at learning things for myself through self-study (and have successfully learned a lot in nontrivial fields like programming and electronics), in high school I found it particularly slow going teaching myself this math even with guidance from my father (very knowledgeable and helpful, but thousands of miles away), and I mostly ended up learning it later from classes in college. (partial exception: Careful self-study of the little precalc-fundamentals book _A Concept of Limits_ worked well and served me well in classes later; recommended.) And it seems my experience is pretty typical: I know other people who have mastered hard subjects for themselves, sometimes on demand when they needed it for work, but things close to calculus seem to be underrepresented in the subjects which are mastered that way.
“The object of education is the generation of power. But to generate and store up power whether mental or physical or both is a waste of effort unless the power is to be exerted. Why generate steam if there is no engine to be operated? Steam may be likened to an idea which finds expression through the engine, a thing? Why store the mind with facts, historical, philosophical, or mathematical, which are useless until applied to things, if they are not to be applied to things? And if they are to be applied to things, why not teach the art of so applying them? As a matter of fact the system of education which does not do this is one-sided, incomplete, unscientific.”
–Charles H. Ham, Mind and Hand: manual training, the chief factor in education (1900)
I would suggest that this is not radical in the sense of new or progressive. This is an old idea, a throwback to the way things were before acadamies were widely available.
Clearly, the assertion that things learned while being applied are better learned resonates with the intuition of personal experience. While our intuition is not alway correct, that resonance certainly bears attention. Inspecting that personal experience, I believe most of us would also recognize that what we have learned on the job tends to be only the portion of the content needed for the job. For instance, I learned C++ as a tool to solve problems, but I have large gaps in my knowledge of the subject of programming because of a lack of formal, systematic computer science training.
So, learning in a context of application appears to be superior to the academy for regarding the absorption and practical use of information. However, it also appears to be inferior to the academy regarding broad mastery of a topic in preparation for tackling larger, more complex applications of the field of study.