Learn Architecture

And do the stuff AI can't

Jan 17, 2024

Poll Results
- Almost everyone is worried about AI over some time horizon
Learn Architecture
- You should be doing this anyway
- Tech has always been moving in this direction
- Concentrate on the what and why, not the how
Poll of the Day
- What newsletter content would you like more of?
Bonus Link
- How to design software architecture: Top tips and best practices

Would a friend like this newsletter? It would help me out if you could share it with a friend.

Poll Results

The results of the poll seem clear, and certainly not positive, apart from the one person sitting pretty saying No! We don’t know why, but I imagine that one person is already in a job they feel AI can’t replicate (e.g. management).

Learn Architecture

Introduction

Recently I was asked by a mid-level software engineer whether they should move towards architecture. They were concerned, as a lot of people are, about AI coming for their job.

Whilst I’m measured about AI taking jobs short-term, it pays to be prepared. However you look at it, the farmer planning crop cycles still has a job, the person wielding the scythe does not.

My instant answer to the question was ‘yes’, but the more I got to think about it over subsequent days, the more this engineer should have been thinking like that even without AI.

Why it matters

Today, I’m guessing all of you use high-level programming languages, leveraging powerful, highly abstracted libraries, using cloud infrastructure that can be deployed with a few clicks.

Almost every aspect of the engineering process is higher-level than it has ever been. We have been on this trajectory since I started my career. In fact, since the introduction of computing.

So that being the case, it makes sense to concern ourselves more with higher-level considerations than nuts-and-bolts implementation. If you have ever used a third-party library, you already do.

How to apply it

Think of the WHAT and WHY, more so than the HOW. It might sound counter-intuitive in a profession where your job is implementation, but this is where AI is less capable.

WHAT database technology should we use?
WHAT library offers best performance?
WHAT stack fits best with where the business and industry are heading?
WHY should we use a queue not a database?
WHY should we put business logic on the client not the server?

The answer to all of these questions is ‘it depends’. These are sprawling questions laced with context and opinion. If you ask AI to write code, it will happily make assumptions on these things. The danger is that without skilled direction, humans or AI can easily create a bad implementation.

I hope writing code manually never goes away, but the trajectory is obvious. I’m not saying throw your coding books away, but at least look beyond them. Thinking more broadly is a quality of a good software engineer regardless of AI.

Do you enjoy this newsletter? It would help me out if you could share it with a friend.