“Code was never the hard part” is an insult to all programmers

The percentages are different for different types of SWEs, not all SWEs spend 80% of their time on XFN comms. I agree with your larger point that a non-trivial amount of a SWE's time is spent on XFN comms and team alignment. My argument is that if an individual contributor is not spending a majority of their time on problem solving and implementation (including maintenance of legacy code), they are not maximizing their potential. If they are spending 80% of their time on non-coding activity they are better suited for an Manager role (Engineering or Product). At the end of the day, coding is not hard only if you are a good coder to begin with. If you are a good EM/PM then people issues will not be hard (which coders often complain about).

There’s too many dimensions to make broad claims like this and it always makes me wince a little.

In my career code was the hard part for the first few years. Then I got over the hump and everything else about my job was harder. Today code is the easiest and least interesting part. But I’ve also experienced in 13 years maybe… I’m going to say 2% of the world of professional coding. I bet if tomorrow I was asked to do some kernel optimization or make Postgres better or reverse engineer an emulator for some PLC something something, I would be deep into a land where code is the hard part.

The truth of that statement depends on what you consider as "coding".

The way I would define it, coding is to software development as cutting is to open heart surgery. It is the final part of the process after 99% of the decision making and application of experience has already been done. There is still some craft or "surgical technique" to it, but the hard part was surely everything that came before (requirements, architecture, design, etc, etc) that got you to the point where all you had left was a bunch of classes and fully specced out modules (and implied test cases) to code up.

I feel that some people, maybe including a lot of the people working at the AI companies, think that the software development job mostly consists of "coding", and perhaps in machine learning (not much code in an LLM!) it mostly does, but if you are a developer and define coding as the final "sit down and implement it" phase, then surely that is the easy part.

Maybe the future has those as two different jobs re: hard problems.

- The sorcerer: Have meetings with others until you know just how valuable a solution to this hard problem actually is, characterize it well, pool resources use cases and documentation, and then work with whatever wizard (or university thereof) is known to be able to solve that kind of problem. Have them find a solution to the hard problem and publish it. Use that publication as context, and have your LLMs integrate the solution.

- The wizard: Find hard problems with adequate funding behind them. Solve them. Don't worry about stakeholders or integrations--the problems are hard enough on their own.

It used to be we found ourselves jumping back and forth between sorcerer and wizard. But there are so many hard problems with solutions that are now in the training data for these models. A relevant skill for the sorcerer, besides the skills that are relevant in those meetings, is not solving hard problems head on, but being a sort of remixer of existing solutions to hard problems.

I think this would actually be better, because more hard problems would get solved in the open where they can benefit everybody, rather than ending up as IP-shaped ammo for zero sum games.

Those who know, do. Those that understand, teach.