What might software engineering look like in 10-15 years?

[flackjap]

Hey everyone!

I've been reflecting on how much software engineering has evolved over the past decade—from the rise of cloud-native development, CI/CD becoming the norm, and AI/ML starting to change the game. It’s got me wondering: what might the field look like in another 10 or even 15 years?

Of course, none of us have a crystal ball, and predicting the future is tricky. But I'm super curious to hear everyone’s thoughts and insights on where things could go, based on trends you’re seeing now or tech you’re excited about. What do you think might be the big shifts or breakthroughs that will define the next era of software engineering?

Whether it's specific advancements in languages, new paradigms, changes in how we work, or the role of AI (or something completely out of left field!)—I’d love to hear what you think and, more importantly, why you think that might happen.

Looking forward to some awesome ideas and perspectives!

[WheelsAtLarge]

We can look at the past to get a sense. Personally, I don't think there will be an extreme difference between now and then. A programmer or team of programmers will need to sit at a computer and put together a software app from start to finish. We won't be able to talk to a computer and have it build it for us. I do see a SWE having to have a stronger role in designing the software. Eventually, we will have tools that automate more of the process to write bigger programs with less effort. This is the way software has been advancing from the beginning. The idea that AI will take over is overblown at least in the near future.

What surprises me is that we don't have software that amateur programmers can program and customize at will. I think at some point when it's easier to program apps, professional programmers will be able to provide that functionality so we will be seeing more of that since the cost of a line of code will diminish. It will be more cost-effective to provide features that are too costly to provide now.

[GianFabien]

Perhaps more of wishful-thinking than prediction ...

Tooling that works top-down. Starting with UML and similar system of systems diagrams and requirements specs with semi-automated generation of working systems that can be fine-tuned as it runs.

We started down that path with Smalltalk, but as David Ungar remarked (from memory), "once you get past the system browser, it's like editing Pascal source code". The limiting factor was not the vision, but the meagre (compared to present day) hardware resources available for tooling. Some UML tooling has been written on Smalltalk systems, but Smalltalk run-times were never suitable for deployment of systems. And then Java happened.

[rhelz → GianFabien]

// Starting with UML and similar system of systems diagrams //

Have you tried asking using LLM's to generate your UMl? :-) Or to explain what kind of system a UML diagram is modeling?

Depending upon how fast they move, in 10 or 15 years software development may be unrecognizable and radically different. We might not even need UML (if we ever did).

[cheevly]

I do not think we’ll be writing code by the end of next year. AI will generate all code from natural language.

[d357r0y3r → cheevly]

Is there some betting market where I can put money on this other side of this prediction?