The Calculator Moment

In a time where every technical individual has an urge to share their experience with modern agentic coding tools (e.g. Claude Code, Codex), it is naturally difficult to cover new ground in the public discussion. However, I will try.

In my last post, I stated that the new agentic tools have, and will forever, changed how we code. Over the last three months, I have written less than 2K lines of code by hand, and prompted Claude Code to exhaustion.

In the past I have educated more than 200 students in Computer Science, and the experiences with agentic tools has made me seriously wonder how their curriculum could be adjusted for the future which lies ahead of them. If I pause the doomsday predictions of forever unemployment, I would argue that we have actually lived through a very similar technological breakthrough in a neighbouring scientific field in the past. Here, I refer to the introduction of the calculator into Mathematics. Suddenly the value of arithmetic turned into a trivial task, in a similar fashion to how now generating some kind of functioning code is trivial with agentic tools. This naturally struck the fear of no human being able to do basic arithmetic without a tool by their side, in a similar way that developers are currently afraid of becoming irrelevant and marginalized. Luckily, we can all agree that our brothers and sisters of Mathematics survived the introduction of this revolutionary tool, but how did it go about?

The revolution of the calculator mainly took place in the 1960’s, and was a deep concern for Mathematical educations during the 1970-1980 era. Sarah A. Banks (2011) wrote a great master thesis on the subject, covering the initial fears of this new tool would reshape Mathematics, and covers some interesting early research:

• Rudnick & Krulik (1976) mentions that, in the early 1970s, half of parents deliberately asked for their kids to be taught math without an calculator.
• In a survey (from Mathematics Teacher) 72% of teachers and mathematicians wanted children not to have calculators in their classrooms, in the fear of not being able to do basic arithmetic.
• When students used calculators teachers struggled to see where the students made the mistakes that caused incorrect results (Kiehl & Harper, 1979).
• Educators experiment students being less bogged down by computational problems and achieved higher motivation as a cause (Kiehl & Harper, 1979).
• In the mid-1970s, 84% of educators believed that calculators should be used in schools. (Wyatt et. al, 1979).
• Wyatt et. all (1979) also mentions that educators agreed on that students should master a skill or concept first, before being shown how to use a calculator to solve it.

After briefly covering some of this research, I realize the striking resemblance to modern perspectives on agentic coding. The company of Claude Code, Anthropic, seems to have only been increasing its desire to hire more software engineers as their tool has increased in capabilities.

Considering this, I think upcoming software engineers can remain calm as they are very unlikely to become replaced or undesired. I, however, think it is the right time to discuss how we should reshape our curricula to adapt to the rise of agentic tools in software engineering.