# The Art of Doing Science & Engineering

Stripe Press recently reprinted Richard Hamming's "The Art of Doing Science and Engineering": Learning to Learn. It's been one of my favourite books, but I've only ever read it in electronic form – the printed version is a joy to hold and read.

Instead of using a pen to mark up the book, I'll use this note instead to write my take on the book. In general, I recommend getting a copy: online or otherwise.

As an exercise, I'll summarize each chapter in my own words; and perhaps in a few years come back and compare if I still read it the same way.

## Chapter 1

The whole point of life is to aim for excellence in whatever direction I choose. Alternatively, the unexamined life is not worth living.

The most important take away from this book would be to form a vision of the future and be willing to tailor my work and career towards it. Understanding what's possible, what's likely to happen, and if it's something that should happen are critical.

This is something that I've also read in Peter Drucker's books, and is one of those sound pieces of advice I've consistently managed to forget.

Knowledge doubles every 17 years; what I'll be working on at the height of my career is something that I will inevitable learn after school. It's hard to keep up with what's going on in the world and balancing learning and working, but it's a balance I must learn to strike. Learning the fundamentals (Lindy effect) and keeping up with the latest is one tactic.

I should balance between education and training: education lets me prioritize and choose the right thing to do, training lets me actually execute on it.

Asides: It can be very valuable to build a small mathematical model, then parametrize and play with it to gain a deeper understanding of a system.

Another example is the random walk that forms the cover of the book.

## Chapter 2

My main take-aways from chapter 2 were more on the ways of thinking displayed instead of the actual contents: back-of-the napking modeling to predict the future, constantly figuring out what the future looks like.

The quote I appreciated the most was to assume responsibility for what I believe in – which is one of the few pieces of advice I feel comfortable giving others (with far less eloquence).

Quickly modeling the S-curve with assumptions on the rate of change was a fascinating exercise, and one I'm more excited about than the rest of the book at the moment. I might take a detour to build my technical skills more deeply before coming back to this book.

Asides: It's funny to see so many predictions aimed at 2020 – 80% of the US workforce is in the service sector, startlingly close to Hamming's prediction of 75%.

On the other hand, a prediction that most large companies would be replaced by many smaller companies doesn't quite ring true given FAANG and other companies – but I'm not quite sure to validate this for real.

Another aside: I think there is a small error in the description of the S-curve equation. On page 26, to simplify

$$\frac{dy}{dt} = ky(L-y)$$

the book recommends replacing

$$y = Lz$$ $$x = \frac{t}{kL^2}$$

to get

$$\frac{dz}{dx} = z(1 - z)$$

I think it should be

$$x = \frac{t}{kL}$$

instead to get the expected result.

## Chapter 3

It's strange to read about the history of computer hardware in a book that is now more than 25 years old. If I had to extract a pattern from this chapter, it would be to never underestimate the uses of computers given how we use current technology.

The other one would be to appreciate the S-curve, and perhaps to better understand where my lifetime fits on different S-curve trends in the world. I can only imagine what the next 20 years of hardware will look like, with GPUs, TPUs – and Quantum computing arriving.