← More writing

On Vibal Knowledge

04-23-2026

Vibal knowledge. The undocumented "tribal knowledge" decisions that a large language model makes based on confidence levels. It might be documented clearly or vaguely within the code.

What do we do with the steady, quiet increase of undocumented decisions? Un-fought over decisions? Decisions without toil that are tied to how our coworkers think of us, tied to the readability and reliability of our designs?

Developers think of a desired outcome: a real one with business impacts and systems that impact people. Then we meditate on them. Often brainstorms happen and we put our thoughts into words, our feelings into moodboards, or our architectures into charts. And we want to validate that our dreamed-up outcome is a desirable change!

Then somewhere in there we slowly start asking questions of how cloud services or user interfaces will support this in reality. So maybe we start pseudocoding or building small prototypes. We begin to try to speak like the machine, the server, or the service.

But when we get AI-assisted, the temptation is always there:

"You want that prototype? Just ask. Maybe I'll come up with a system that works better than your system. I work better if you don't give me constraints, trust me. Why wait? I have the explicit knowledge of millions of you weighing confidences in your favor (or my best guess of what favors you)."

What do we do here? When do we yield to it, and when do we push back? What do we do with the anchoring effects of letting it have the first go?

Not a new problem

Peter Naur wrote in 1985 on software development that necessitates ongoing access to the original author or at least extended apprenticeship with the original author to comprehend the author's understanding of the world for which the system is aiming to optimize.

From Programming as Theory Building by Peter Naur (discovered thru Chris Krycho's talk):

The dependence of a theory on a grasp of certain kinds of similarity between situations and events of the real world gives the reason why the knowledge held by someone who has the theory could not, in principle, be expressed in terms of rules. In fact, the similarities in question are not, and cannot be, expressed in terms of criteria, no more than the similarities of many other kinds of objects, such as human faces, tunes, or tastes of wine, can be thus expressed.

Written down rules and guidelines don't ever seem to be enough. There is both tacit knowledge and explicit knowledge. Software developers are really good at designing explicit knowledge to make it more readable (whether human-readable or machine-readable for performance optimization). But tacit knowledge: what's contextual to a specific person or group of people.

Technology literally means "stored knowledge", and software development allows syntactic, dynamic, and automated rearrangement of explicit knowledge. And now we're automating this rearranging, "world-making" process left up to probabilities.

Explicit Knowledge with a long tail

In the beginning was the Word, and the Word was with God, and the Word was God. — John 1:1

Jesus Christ himself is described as explicit knowledge. The gospel according to John is saying that Christ and his life-example are themselves, "the Word", or explicit knowledge. The Word and a message for us. If we follow along with the telling of his life (i.e., explicit knowledge told in narratives and prescriptive parables) in the gospels according to Matthew, Mark, Luke, and John. Throughout the gospels, Jesus asks his disciples to "Follow him." He doesn't ask them to take a pamphlet, or follow his X feed (or blog!). He asks them to follow along with how he is. To see how he treats food; how he looks at people that don't know him; which words he emphasizes when he speaks; what he brings with him everywhere he goes. So they got the tacit practices and the explicit stories and sermons on mounts and seashores. And then someone wrote down as much as they were able.

And so shouldn't we be concerned? We didn't get to do that! Jesus' life is now legacy software. According to Programming as Theory Building, it's a liability if we're to try to live it out! Except that Jesus tells us a Helper has been sent. (Agent-assisted apprenticeship to Jesus? No, this is 0% hallucination, much more edifying, and doesn't mind telling you, "Not yet.")

I will put my Spirit within you and cause you to follow my statutes. — Ezekiel 36:27 (CSB)

So we can follow explicit knowledge (i.e., "statutes"), because we will have a helper. We will be able to pick up the tacit knowledge day-by-day of what it feels like to apprentice to Jesus. We have the Helper. Day by day we ask:

When I know that God is my master and not money, what does it mean when I'm deciding where to live or which job I take?

If it is good to love my enemies, how does it really feel when a coworker gives me busy work? Do I do the busy work out of love or graciously recommend another solution?

As time goes on, we slowly increase alignment with how God has designed us to live. But the explicit knowledge, the Law that God gives us, wasn't meant to be the way. We are to live and walk in the Spirit, and he'll show us what is good. And honestly, when he works all things for his good, it reminds me of vibing, but we know there's nothing probabilistic about it. This is not personalized: it's personal and it's three-personal And that's our hope. 👈

← More writing