I generally agree with your point on ease of experimentation, but if we insist on calling it software engineering, then maybe the field needs to adhere to engineering principles, as the GP highlighted.

I believe part of engineering isn’t over-engineering for the task at hand as well. If the costs of a “failure” are low/zero then the right thing can be to move quickly expecting some problems.

I think the field could get better at knowing when costs are low (eg sometimes scalability, cheaper to change a database choice than rebuild a bridge) and where the costs are sometimes very high (eg security).

Notice OP says "software design". Design is still an art, even in mechanical engineering.

Engineering applies to looking at a design and proving something about it.