Reading Structure and Interpretation of Computer Programs in 2025

I had owned a copy of Structure and Interpretation of Computer Programs (SICP) for a couple of years, but never made it past the first few pages.

Last year in February, my friend Maud offered me a physical copy. Now I had two. It was time.

With about a decade of experience in web development, I wanted to challenge my thinking and build stronger foundations.

So I joined the Recurse Center SICP reading group and got started. It took me about a year to finish.

Read the "What to expect" sectionWhat to expect

SICP starts with the fundamentals of programming - procedures, recursion - then explores lazy evaluation, concurrency, and different programming paradigms. By the end, it develops a Scheme interpreter (written in Scheme¹) and a register-based virtual machine with a compiler.

I found the exercises very well-designed. I would recommend doing as many as possible.
Some took me an entire day to solve, but that was always due to a gap in my understanding - so I don't regret pushing through.

SICP feels like a reflection on how to design programs. The authors walk you through ideas and invite you to compare them. They give precise names to patterns you probably have seen before². Naming is powerful! It helps reason about the ideas and apply them intentionally.

I also thought the programs presented throughout the book (like the digital circuit simulator) were remarkably elegant. The implementation is so simple! I have written similar programs in the past and I know how easy it is to end up with a confusing pile of code. SICP shows how it's done.

Read the "Should you read it?" sectionShould you read it?

I would not recommend SICP to everyone.

• If you're just getting started and looking for a job, skip this. Come back later :)!
• If you want practical tips to ship software faster, there are better books.

However, if you want to reflect on programming and build stronger foundations, SICP might be worth the effort. Peter Norvig's review puts it better than me.

One caveat: The book assumes MIT undergraduate knowledge of math. If you are not math-inclined, some of the examples and exercises might be difficult. You can just skip some of them but it will make the overall experience a bit harder³.

If I had not committed to reading the whole book, I would probably have signed up for David Beazley's course based on SICP. I have taken a few of his other courses and can really recommend them.

Read the "Practical advice for reading SICP" sectionPractical advice for reading SICP

• A minimal setup is fine. I used MIT/GNU Scheme 12.1, rlwrap scheme, and a shortcut to run scheme < [current_file] from my editor.

• Learning to use the Scheme debugger early on saved me a lot of time. I'd say if you're in chapter 2 and intend to read the whole book, it makes sense to invest the time.

• Sometimes you need to read ahead to be able to test your code. Don't be afraid to do that.

• Build tools when needed. I added the ability to load files (like a file import) inside the custom interpreters of chapter 4 and it made testing much smoother.

Read the "Final Thoughts" sectionFinal Thoughts

Thanks to these people who discussed their solutions online. The ones I referred to the most:

• https://mk12.github.io/sicp/exercise/index.html, until half-way through chapter 4.
• https://eli.thegreenplace.net/
• https://www.inchmeal.io/sicp/review.html

In case all three of the websites above go down, my solutions are on GitHub.

Finally, a big thank you to the Recurse Center SICP Wizard group 🧙! Meeting weekly for many months helped me find motivation to finish the book. Thanks in particular to fellow wizards Banna, Charlie, Jeff, and Maud.

SICP did not change the way I write code overnight. But I made some progress, and picked up a few tools for the rest of the journey.

Happy wizarding 🧙 ✨