Why Teachers Won’t Be Replaced By Software

Marc Andreesen believes that software is eating the world. It’s a very visceral image, and in one sense it’s absolutely true. Software is spreading into every industry, changing how established players must play and even what the rules of the game are. But while many in Silicon Valley and Educational Technology think that software will “eat” teachers, replacing many of them, at trinket we believe software’s role is to create openness, making teachers better and more connected. Far from there being less teachers in the future, we think openness will enable and encourage more people than ever to teach.

Godawful Teachers?

In the midst of a longer Twitter conversation I was having with him and others (which I will likely blog about separately), Andreesen made an interesting comment:

My suggestion was that increasing openness into what teachers are doing and what the results were was the solution to bad teaching. Sunlight, disinfectant, etc:


Andreesen agreed but thinks there needs to be some sort of culling of the worst teachers. He thinks of education as a government monopoly that has been too long shielded from adaptive pressures. So, logically, he thinks that it’s a natural thing for Software to eat.

Software as Archetype instead of Omnivore

But, backing up, it seems that Andreesen’s assertion that software could replace the worst performing teachers isn’t the only possibility we should consider. Another possibility is suggested by the trajectory of the profession of programming itself. In this view, software won’t replace teaching so much as model its future as an occupation.

There was a time when programmers were regarded as mostly “godawful”, insulated from competence by structure and size. Those of us who have had to endure the dictates and systematic negligence of large IT departments can see where the term ‘Godawful’ might apply.

Yet software has gotten better in the past two decades. Why? How? Can we replicate this success for teaching?

Openness and Teaching’s Future

We can anticipate what’s happening with teaching by looking at how the software industry matured: it became friendler, more open, and more accessible. It did this despite more junior, inexperienced programmers flooding the job market. And, importantly, without someone having to fire the “bottom 10%” of programmers. By connecting people (rather than separating them), transparency gave a better account of who was ‘good’, helped to improve the skills of those who weren’t and has led to the craft of coding to flourish. The craft of teaching is beginning to undergo the same overhaul. In 140 characters, that is:

Coursera, Udacity and other massive platforms are delivering content to students but they’re also opening up these instructors’ methods and content to other instructors, for critique, reuse, and inspiration. The Open Courseware Initiative, spearheaded by MIT’s forward-thinking leadership, has made a default of openness a reality for a growing number of universities. And, all along, the humble course page has remained the most prevalent form of open teaching.

Professors have been sharing course materials online for almost as long as the Web has been around, often via hand-written HTML. Inspired and encouraged by this, we’re building the easiest way to make an interactive course page to support classroom teaching. While the purpose of most online course materials is to let students access them, we’re also building direct support for instructor-to-instructor interaction around materials. This is harder than it looks, but we think we’ve cracked the code. More in a future blog post. For now, let’s sum this up.

Why Andreesen and other VCs are Wrong about Software Eating Teaching

I’ll admit it’s somewhat unfair to write a blog post around a few tweets, inferring deeper thoughts on complex issues from 140 character snippets. So I may very well have misrepresented Andreesen’s thoughts, though I’m confident that I’m pretty close to the mark. Like I said before, though, I don’t think that Andreesen or most venture capitalists have malevolent intentions. Far from that: they’re seeking business opportunities that do real good for the world. In that way we’re on the same mission.

But I think they’ve made an error in logic when they assume that teachers will be replaced by technology. Encouraged by software’s staggering proliferation into every corner of the modern economy, they’ve been blinded to the parallels between the professions of programming and of teaching. If software ‘eats’ teaching, it will look like how software has ‘eaten’ itself: more tools to make humans more productive, effective, and connected. We will not see the rise and triumph of Teaching Machines that replace teachers any more than we’ve seen Coding Machines make coders obsolete. Rather, the need for teachers will increase apace of human innovation more broadly, and the most innovative companies in the ed tech space will augment, connect, and amplify these professionals.

In our industry, companies like Bloc, General Assembly, and DevBootCamp understand that the human element is central to teaching and are, similarly, building technology that augments good teachers rather than seeking to replace them. That’s also the approach we’re taking here at trinket.

We don’t know what the future of education looks like but, if we’re reading these trends right they point to more openness, more teachers, and software firmly ensconced as a tool for open teaching.

Thanks to Dave Paola from Bloc.io, Jenn Marks, and Brian, Ben, Julia, and Pardees from the trinket team for reviewing earlier versions of this post. And, of course, to Marc Andreesen for helping spark the discussion on Twitter.

14 thoughts on “Why Teachers Won’t Be Replaced By Software

  1. I think you’re right that teachers will not be replaced by software, and I like the analogy with software that through openness and transparency, better tools and methods have been developed that developers (and teachers someday) themselves can use more effectively.

    Yet I still think that even this process, and any one that follows it, might significantly diminish the industries of software or teaching as professions.

    For instance, is it not possible that we will yet see “coding machines” eat the software as profession industry? Or at least, software might become commoditized through a host of better tools and visual/interactive techniques that make programming and design manageable for anyone? In effect, rather than software developers being replaced by machines, they may be replaced as an industry if software can largely become commoditized. Even in this case there would likely still be software developers, but certainly not as many.

    The two reasons I don’t think this can happen with teaching though, are that as you say, the human element is central to teaching. I don’t think that is because individuals cannot learn things on their own, but rather because teaching itself is perhaps the highest and most effective form of learning. Teaching is also not just a profession, It is also an enjoyable and irreducible basic aesthetic activity of humans. We enjoy it, it is fulfilling, we all do it at some point, and it helps us learn and grow.

    The other reason is that even if teaching could be automated, in a world of abundance where people could choose what they did, is there any doubt that there would still be teachers?

    1. Let’s take and example from software: WordPress. Using this software, anyone can make a blog, buy a cheap custom theme, and be off and rolling. Many people can hack at/slightly customize their WordPress sites. Yet the web design business is flourishing. How can this be?

      It’s an expanding market. Room at the top and room at the bottom. You can have both an increase in supply and an increase in demand, supporting both higher volumes and higher prices.

      For teaching I think we’ll see 1) more people engaging in it 2) more often 3) at a lower average cost but 4) with plenty of room at the top end of the market for high-cost services. So we’ll have both lots more informal MeetUp workshop AND increasing competitiveness at the Harvard and Dukes of the world, assuming they continue to evolve as they are doing.

      Admittedly, this view requires an interestingly contrarian view: Education as an industry will dramatically expand in the coming years. But that’s exactly what I see happening. Similar to Andreesen’s comments on the future of Journalism (expanding, high volume and high quality).

  2. “We will not see the rise and triumph of Teaching Machines that replace teachers any more than we’ve seen Coding Machines make coders obsolete.”

    There’s a saying “Those who can’t do, teach.” Coding is often creating, manipulating something new; teaching involves engaging with a finite, pre-existing, fairly objective body of knowledge. Computers excel at repetition and following rules, perfect for teaching masses of students. Humans excel at pattern recognition, possessing an advantage over “code generating” machines.

    As an aside, I’ve met fewer teachers who I believe were good at teaching and were appropriately emotionally/psychologically mature enough for the job than not. I’d welcome computers replacing teachers so fast in the classroom it would make you dizzy.

    Quite possibly the optimal blend would be to have teachers encouraging students to pursue whatever truly interests them and to pursue research on computers. The computers would deal with the bulk of time involved with studying, while the teacher may be available to supplement learning with whatever human answers and suggestions that they can contribute.

    1. Thanks for the considered response.

      As an instructor, I can tell you that the act of teaching is much more creative than people who don’t do it often realize. This is another parallel with coding. You seem to be familiar with coding and understand how creative it is. BUt many who are outside the field might say it’s just a rote activity- there are well-defined tasks, and you make programs that accomplish them. As we know it’s not that simple. So with teaching.

      Regarding teaching coding specifically, I personally think that code is much more a cultural product than many recognize. As a cultural product, it’s best learned in cultural immersion, with a guide familiar with the culture you’re coming from.

      One of the root causes of a lack of diversity in tech is that we have too few teachers who are willing/able to meet students where they are and connect coding to their lives. But this failing again doesn’t mean teachers should be replaced- it means we need to equip a more diverse set of instructors to teach.

  3. Hi Elliott, excellent column, and interesting tweets by you and Marc. I think he is the sharpest VC in Silicon Valley right now, and I value his opinion on this subject. Actually, I think you two are in total agreement on the direction, but he is pointing out a systemic problem that can’t be corrected just by reinventing the teaching profession with better technology. He went straight to the solution, which is about all you can do in 140 characters, which is to get rid of poor teachers. A better way to say it is that there are some teachers that will never adapt to the new paradigm, and because of the system of government schools and unions, that means that some percentage of students will never get the benefit. He suggested the only way to fix that is to force the current system to be able to fire the non-adaptive teachers. That is not the only way, but it is one way to improve opportunity for students.

    Another way to say it: your analogy to code development over the last 20 years is right on target. But it assumes that people who want to code have the freedom and opportunity to find and learn from the teachers they want on the subjects they want, and in a way that allows all of us to benefit from the collective advances of the art. That is the target model that we are going for. But that is not the model that students in government education or even higher education are participating in. Yet. Changes need to be made, and it all revolves around student choice. Perhaps non-adaptive teachers don’t need to be fired, but when students have a choice, they probably won’t have any students either.

    For me, I realized this was starting to happen on a vast open community social scale back in 2006 when YouTube first came on the scene. I posted a guitar lesson, and a few weeks later over 10,000 people had viewed it and were asking for more. Suddenly, at least for a little while back then, more people had probably learned to play “Tears in Heaven” from me than any teacher ever. Today that has evolved to a virtually unlimited number of teachers teaching a virtually unlimited number of topics to anybody who wants to search for it on YouTube. And that all happened without any kind of “Future of Education” strategy, plan or debate.

    There is no stopping this rocket to the future, and better tools like yours will help make it happen.

    Mark Easley Sr.

    1. Mark,

      Student choice is something I hadn’t fully considered, but you’re right that it’s an important part of letting the best rise to the top- provided that students choose the best. Frank Noschese has written extensively on pseudoteaching:


      This is a phenomenon where students think and feel like they’ve learned a lot, and self-report confidence. But this is not borne out in measures of understanding. Addressing pseudoteaching will be a key component to introducing student choice into the equation, especially in elementary or middle schools.

      Peer instruction and peer assessment might be an approach toward this that can engage students regardless of instructor skill, and avoid the perils of pseudoteaching. I have yet to experiment much with this in my own class but am intrigued by Cathy Davidson’s work:


      Thanks again for your comments. As your experience with guitar lessons on YouTube shows, the Web can drastically alter who is considered an instructor, which is for the better. Our challenge is to weave the personal interactions back in to teaching technology.

  4. You raise some good points, but I can speak as someone who was a programming tutor in college for several years, a teacher in a very poor and very low performing school for a year, and is now a Senior Developer. And my thoughts, basically, are that you’re *both* correct. Software and other technology is now to the point where we can provide transparency to teaching… while replacing the bad teachers with electronic interaction with the good teachers, combined with (cheaper) parapro presence in the classroom to handle disciplinary interaction. Thus, there will still be a teacher in the “classroom”, but possibly/ probably not in the same physical space as the student – while still having *some* adult in the same physical space as the student.

  5. I think that a teacher is necessary to create a solid programming structure for the individual. Once that is done, teachers are no longer necessary for that particular individual, he can learn any programming language he wants.

  6. Andreesan is a fool. During my years of association with educational institutions, there have been repeated predictions that teachers were replaceable by automation and software. Each of these great trends has flopped. And they will continue to do so. Quality teachers are people with a great deal of skill. Treat them like crap and they will question their career choices and do other stuff. The half-life of new teachers seems to be about five years.

    The invective Andreesan spews indicates he has an interest in spreading heat and not light. We need to get serous about recruiting, retaining and supporting quality teachers. But that might entail work and expense. And you know America will never stand for that.

  7. My humble opinion is like Alan Turing’s in the film”The Imitation Game”: humans think like humans and machines think like machines. We are proud to be very different one from another and would not say, that someone who thinks differently does not think at all. Our brains / minds consist of biological material, whereas the brains of machines have wire, aluminium and steel. Hence we need human teachers because our students are humans. We would be bloody awful teachers for machines. Vice versa: machines could be good teachers for machines, but would be bloody awful teachers for humans.
    Excuse my poor English.

  8. The root of the problem (teaching in general, not just programming) is that teacher salaries are not competitive with industry jobs. This is why “those who can’t do, teach”.

  9. Elliot, I just found out about trinket.io today, as I am starting an edX course on Python programming. This course is using trinket.io. I really enjoyed this post.

    I am an IT guy (“programmer”) since 1977. I have been working in the field ever since. I will be teaching a Python course this fall at local community college.

    How did I learn to program (back in the days before the Internet)? I learned BASIC programming from my Math Teacher in 11th grade. I learned COBOL programming from a community college instructor (teacher). She learned how to program as the first woman COBOL programmer at Chrysler years before that. I’m sure they had an in-house or hired trainer to teach them how to program a brand new language (at that time).

    Another teacher, my first college IT instructor (Business Data Processing 101) was the person who gave me a “tip” on where to apply for a computer operator job. In the days before online job boards, this was a valuable tip. I had gotten an A in her course. She really helped me out. My Teacher helped me get my first IT related job! I hope to pay it forward and help out my students as I can (at least with job references), for students who do well in course.

    Even though there is so much info and training materials and platforms available online today, someone has to create the content. Even though you might call them tech writers, designers or programmers, those who provide this training, I would also consider them as “teachers”, in a sense (even if not stand up classroom teachers).

    As the programming profession and related tools and techniques have improved over time, so will the teaching profession.

    How many Teachers today can program computers, in comparison with Venture Capitalists? Perhaps a friendly programming competition between the today’s Teachers and today’s VCs would be an interesting event.

Comments are closed.