Blog #0188: On Knowing What Code to Throw Away
Source: Photo by liam ward on Unsplash
[ED: A short post on how agentic coding has changed the shape of the work. 1/5 hats.]
The Inversion
Something has shifted in how I write code, and I think it's worth naming.
The time cost of writing code is approaching zero. It may already be there. With agentic coding tools doing the heavy lifting, generating code is no longer the bottleneck. The bottleneck is judgement — specifically, knowing what code to throw away.
This is an inversion. For decades, the hard part was getting code written. You thought carefully before committing keystrokes because each line had a cost: time, cognitive load, review cycles, maintenance burden. Scarcity bred caution. Now the output is abundant and the scarce resource is taste.
Michelangelo's Block of Marble
I'm reminded of what is almost certainly an apocryphal story about Michelangelo describing how he created The David. He is reported to have said — and I'm paraphrasing for dramatic effect — words to the effect of:
"The best artist has no idea that a single block of marble does not already contain within itself — only the excess must be removed, and the hand obeys the intellect." [0]
The real quote is from one of his poems, and it's about the relationship between vision and material. The sculptor doesn't add — the sculptor removes. The form is already in the stone. The skill is in seeing it.
Sculpting, Not Building
This is how agentic coding feels to me now. Not to suggest that my output is anything like Michelangelo's, of course. But the shape of the work has changed. The agent generates a block of marble — rough, overbuilt, full of potential. The job is to see the form inside it and cut away everything that isn't that.
The skill set this demands is different from what we've historically valued in software engineering. It's less about construction and more about curation. Less about "can I build this?" and more about "should this exist?" You need architectural judgement, a sense of what's essential, and the discipline to delete code that works but doesn't belong.
Put differently: the hard problem used to be writing code. Now the hard problem is editing it.
What This Means
I don't think the industry has fully caught up with this inversion yet. Hiring processes still optimise for people who can write code under pressure. Engineering cultures still celebrate addition over subtraction. We still talk about "shipping" as though the act of producing output is the valuable part.
The valuable part is knowing what to keep.
[0]: "Four poems by Michelangelo by John Frederick Nims" — https://newcriterion.com/article/four-poems-by-michelangelo/
Regards,
M@
Originally posted on matthewsinclair.com and cross-posted on Medium.
_[ED: If you'd like to sign up for this content as an email, _click here_ to join the mailing list.]_