One aspect might be that a developer who engages in "business" effectively stops being "subordinate". Management decisions need to be justified on a different level to maintain legitimacy.
It's one of the reasons I went back for a business degree and then re-entered tech. No, of course nobody in Silicon Valley cares about the "MBA" title (HN sees it as a negative), but everywhere I've interviewed/worked they've appreciated that we could talk about the economic and business impact of the software, and not just the algorithms and data structures.
Chicken/egg imho.
These roles require wildly different skills and knowledge.
Usually the outcomes are better if you combine two people who are good at their jobs rather than hoping one person can do it all.
Why it makes sense for them to be a single person? Often, "changing requirements" really comes from an engineer learning new things (this framework does not provide this, this external dep is going to be late, I'd need to learn 2 new things so will need more time...), and really, an engineer is the first one who'll know of some of the challenges and what's even feasible!
Now, the skills an engineer needs to develop to be a good PM is good communication and ability to document things at the right level, and lots of empathy for a customer and a business person (so they can "walk in their shoes"). Arguably, all things that will make a great engineer even better.
I've been in teams where we've had a very senior, experienced PM tell us that he's looking for another position in the company because our team does not need them: we already did the stuff they were hired to do. That was a sign of a great PM who did not try to actively wrestle control out of our hands when the team was chugging along just fine.
Scoping tickets is more of a project management skill. Again, not a dev skill.
Estimating effect on user experience - requires empathy, again not a dev skill.
If you redefine the dev job as including PM skills then sure, PM skills are dev skills.
But theyre not.
>Why it makes sense for them to be a single person? Often, "changing requirements" really comes from an engineer learning new things
So? Happens to me too. I can tell the PM these things i learned. Thats a hell of a lot easier than managing all stakeholder interactions, empathizing and balancing their demands.
It only really makes sense to combine the two roles if the project is inherently very straightforward, a salary can be saved and the person doing both roles is suffiently qualified for both roles.
If you are not doing that, you are being micromanaged and I feel for you in your engineering job.
And trust me, non-technical PMs are ill-equipped to figure out an incremental path to that North Star product or feature you want to ship — how you split branches and deliver value incrementally is something only a good engineer can do (well).
If you do not consider how an implementation will affect the user, you might just satisfy the requirement with an actually terrible experience (but the ticket never said it needs to load in under 30s and with no visible redraws and jumping elements): a good engineer will implicitly consider all of these, even if unspecified in a task (and many more, I only used an outrageous example to make a point).
Breaking down problems is certainly a life skill, but engineers are inherently good at it: it's the very definition of an engineer, and you can't be one without it. I have however seen PMs who mostly channel and aggregate customer experiences and stakeholder requests without an ability to consider (broken down, stepwise, incremental) paths to completion.
If you are good at all of these, you'd likely be a good engineer too: this does not mean that one can't be good at PM without being an engineer, just that a great engineer is very close to being a great PM too.
I am not against the division of labour and different motivations driving where each person invests their time, but if we are talking about a great either PM or engineer, they are pretty much of the same mindset with focus on different parts of the job that needs to be done — 90/10 split vs 10/90 split (and anything in between).
And finally, whether you are a great craftsman at engineering (or PMing), it is slightly different from a great engineer.
I've found it possible to migrate to a less top-down Desert style just by finding executives who are frustrated by those problems and saying, "I have an idea I've seen help" and then getting the team together and saying, "hey, it turns out the executives would like us to write software well. What should we try first?"
Product has plenty of work remaining: they should be handling whatever subset of strategy, prioritization, analytics, BI, QA, facilitation, design and contracts that they have the skills for. But it requires engineers to actually collaborate with them as a peer, rather than engage in power struggles, and that requires everyone on the team to understand what we are building, for whom, and why.
This is manifest in management methodologies: developers are largely interchangeable cells in a spreadsheet. I'm not saying this is a good thing.
The reasons for this are complex, but generally, business people want us to solve the technical problems they can't handle themselves, they don't want us to "relieve" them of product management, customer relationships, and industry knowledge. Why would they? It would devalue them.