![](/static/253f0d9b/assets/icons/icon-96x96.png)
![](https://lemmy.ml/pictrs/image/d3d059e3-fa3d-45af-ac93-ac894beba378.png)
You’d probably want to segregate the hard-tech from the soft/bio-tech:
IF you require deterministic-performance, THEN stay away from meat-implimentation.
IF you require sentience-sensitivity, THEN stay away from hard-tech implimentation.
( same principle as requiring both people & systems, in order to project effective force, economically, militarily, in disaster-relief, in any domain,
same principle as using both a wordsless-wholistic metaphor-understanding sentience in one’s right-hemisphere, vs a reductionistic sentience in one’s left, as the split-brain patients showed.
2 orthogonal systems, collaborating … produces greater results than either-kind would on its own! )
Organic-sentience and machine-compute, organic doing the high-level-decisions, & machines doing the low-level-competencies, seems much more realistic.
As the real gurus of Agile point-out,
most of the nuveau methods/processes were ideological,
but agile had engineering-requirements, too ( test-1st develoopment, e.g. )
Which makes it much superior to all the ideology-but-no-engineering-hardening methods.
As they also pointed-out, you NEED disciplined individuals & teams to make it work.
IF you don’t have effective & driven-by-quality/integrity-of-work teams, agile isn’t the right method for you.
I’d go further:
I’d say that both waterfall & agile ( Wysocki’s book on project-management identifies that Traditional is the poorest match for reality, Agile’s best, & Extreme is research whereas Emertxe is where you’ve got a solution, but don’t know what it’s for, yet ( like Post-it notes glue, before sticky-notes were invented ) )
both waterfall & agile are mis-apprehensions of what’s required.
Until you understand the required architecture, you can’t make the right architecture-choices, right?
So, why not make a prototype agilely, until one has a proper domain-model, an executable toy-prototype which demonstrates all the key functions, & then when you’ve got the working, executable model, then you understand the architecture required, and only then do you switch from agile-prototyping to building-out the real, hardened thing…
Just seems sane, to me, but I’m just some idiot with a bit of thinking, not a working … anything, really.