College Is (Almost) Obsolete

A few days ago a neighbor in her eighties offered me a piece of advice regarding my six-month old son: “start a college fund.” For some reason it sounded to me as if she’d told me there’s a great future in plastics. I have no intention of starting a college fund for my son, and this post is an attempt to explain why.

When I entered the workforce in the early nineties I had only one tool to increase my chances of getting the job I wanted: my resume. Looking for a job worked like this:

  • scan the job ad section of the newspaper
  • choose a few interesting positions
  • make copies of my resume and try to get them to the right people
  • wait for a call

Some time later when I was on the recruiting side, finding the right person was a costly task. I started with a pile of resumes but only had time for two or three interviews per week; I had to pick promising candidates based on the little signal I had. This was a time before Facebook, LinkedIn or Google; all the information I could reasonably obtain from applicants was in those pieces of paper. Most of us believed that resumes had a high chance of containing lies, because most claims were unverifiable. Probably the easiest ones to verify were someone’s degree and GPA, so we believed that people would have less of an incentive to make those up. For that reason we prioritized candidates from the best schools and with the best grades (assuming their resumes didn’t contain negative signals such as horrendous grammar).

Over the course of the next decade (and especially in the software industry) this started to change. First, people became searchable. The best software developeres often had a web page or some online trace of their work. LinkedIn slowly started making resumes irrelevant: the chances of being found out are much higher if you lie on your LinkedIn page. GitHub made it much easier to know if someone can write code.

When I look for software developers today and want to decide whether someone is worth meeting, the college signal has a tiny influence compared to everything else. As I said four years ago, I probably won’t interview someone who doesn’t have a LinkedIn profile and a GitHub account with some activity. I do not think the college signal is worthless though: to me, being able to stick with at least four years of structured coursework and tests is an indicator of consistency, discipline, and a certain type of intelligence.

If someone asks me the question “I want to become a professional software developer. With that objective, is college worth the investment?” my answer is no. This is one of the reasons I became involved with Platzi: it that’s your goal, it gives you a much better ROI.

Of course there are other fields for which a formal education and certification is still important (medicine is the most obvious one). Also, there are other reasons to go to college besides increasing your value as a worker: it’s a life experience, a good way to meet like-minded people, and it provides a good place to learn from others (of course not the only place or even the best, as was the case before the internet). Unfortunately I do not believe those reasons justify the cost of college in the US for someone who’s not quite wealthy. I’m sure the deans of Ivy League universities would entertain a different opinion, given that the future of their industry depends on the perceived value of their services. What they won’t say is that the commoditization of degrees caused by the expansion of their business means that having one is no longer the differentiator it was one or two generations ago.

In conclusion, I don’t think it will be indispensable for my son to pay hundreds of thousands of dollars to become a successful professional. He may choose to do so if he wants to enter a field requiring access to expensive laboratory equipment, or one requiring legal certification monopolized by universities. However, there’s a growing list of professions that will employ smart and resourceful people regardless of where or how they acquired their skills.