> When might this future arrive? It's uncertain. A good indicator, however, might be when engineers confidently say "yes" to the question: "Would you happily go on-call for a system of fully AI-generated services?".
I dare say there are already people saying this, but I don't know of any SWEs (or anyone who has maintained a significant codebase) who would be willing to go on-call for such a system. But I would hesitate to go on-call for even an entirely human-built system unless I'd done some in depth review of its architecture and playbook.
Maybe there's a middle ground here, like willingness to update the codebase of a vibe-coded application. Do that enough times and maybe the confidence in the system increases enough to be willing to on-call for it.
I agree with the position that the value of an "engineering" role in this context should be judged by those who have held that position, moreso than the opinions of management or media in that regard, anyway.
I think being "on call" is a sign you're not doing engineering in your role, but that's not absolute and kind of a deeper discussion about what "software engineering" even means.
I swear I'm not trying to take your bait, but I did want to point out that I want to live in a world where developers are responsible for running the products that they produce, otherwise the incentives are not aligned and then -- relevant to this vibe discussion -- tossing slop over the wall becomes Situation Normal
I am always wary of "should" sentences. As best I can tell, it's only ridiculous if every human in the whole chain executes flawlessly. I am cognizant that very few space shuttles blew up, but the amount of development and testing time that went into them is similarly impressive
> but the amount of development and testing time that went into them is similarly impressive
Absolutely. As before, there are tradeoffs to be made. Not all software needs engineering, and it may even be detrimental to it. But if you have proclaimed engineers working on your product...
One of last month fun bugs I got from a 10 year old ecosystem was due to some backend calling an API which (aside from fetching info from an API provided by the backend) needed data from some third system generating it using a CRON run mon-thu at 13h. The original input in this third system was done on thursday around 17h so it was not available on friday (date of the ticket) nor monday morning (which led to this treasure hunt) but things were automagically resolved around 13h15. Gotta love the world of in-house legacy applications.
I dare say there are already people saying this, but I don't know of any SWEs (or anyone who has maintained a significant codebase) who would be willing to go on-call for such a system. But I would hesitate to go on-call for even an entirely human-built system unless I'd done some in depth review of its architecture and playbook.
Maybe there's a middle ground here, like willingness to update the codebase of a vibe-coded application. Do that enough times and maybe the confidence in the system increases enough to be willing to on-call for it.
I agree with the position that the value of an "engineering" role in this context should be judged by those who have held that position, moreso than the opinions of management or media in that regard, anyway.