Veritas Silentii Conversations Across Time
Conversations Across Time
Conversation No. 6 · September 2026
John Keats and the Vanishing Room
Before every answer, there is a room we call uncertainty where the answer does not exist yet. Predictive AI is very good at making sure you never have to enter it.
The boss offers you a promotion, but it's across the country. Or a doctor lays out two treatments and asks which one you'd like to try first.
What do you say?
Almost certainly, you say some version of the same thing.
Give me a little time to think about it.
I need to sleep on it.
Let me mull it over.
Can I run this by someone first?
We treat these sentences as throat-clearing…a polite way of buying a few hours or days before the real business of deciding begins. Something to get through on the way to an answer. The thinking itself, we assume, is just friction.
But what if it isn't friction at all?
What if it's actually a place?
John Keats and the Room He Refused to Leave Early
In December 1817, a twenty-two-year-old poet named John Keats wrote a letter to his brothers describing a quality he had just observed in a great writer at work, and that he believed was essential to any serious achievement, in literature or otherwise.
He called it Negative Capability…when man is capable of living with uncertainty, mystery, and doubts, without any irritable reaching after fact and reason.
Keats wasn't describing indecision. He was describing a discipline…the discipline of staying inside a question longer than is comfortable and resisting the urge to grab the first plausible answer just to make the discomfort stop.
He thought Shakespeare possessed this quality "so enormously." Not because Shakespeare was slow to understand things. Because he was willing to remain unresolved for as long as the work required it, rather than rushing toward a tidy conclusion that would betray the complexity of what he was actually looking at.
Even if we've never named it, most of us have felt a small version of this ourselves. The unfinished thought that isn't ready to be finished. The problem that gets worse the moment you try to force a premature solution onto it. The conversation that needs to stay open a while longer, because closing it too soon just feels wrong.
Keats gave that discomfort a name and, more importantly, gave it dignity. He didn't treat it as a defect to be cured. He treated it as a capability…something a person could get better at, the way you get better at anything you practice.
Two hundred years later, an entirely different kind of thinker arrived at almost the same place, from almost the opposite direction.
Graham Wallas and the Stage Nobody Wants to Admit They Need
In 1926, the social psychologist Graham Wallas published a short book called The Art of Thought, based on his own observations and the accounts of inventors, scientists, and writers describing how their best ideas actually arrived.
He proposed that creative thought moves through four stages.
- Preparation. You gather information. You define the problem as clearly as you can.
- Incubation. You stop working on it directly. The problem sits, unattended, often while you're doing something else entirely.
- Illumination. The idea arrives, often suddenly, and often when you least expect it.
- Verification. You test the idea to see if it actually holds up.
Almost everyone recognizes Preparation, Illumination and Verification. They look like work. They produce visible output. You can put them on a timeline and report them in a status meeting.
Incubation is the outlier. It often looks like nothing is happening. No one can see it, measure it, or schedule it with any confidence. And yet Wallas found, again and again, that the genuine insight and innovation almost never arrived during Preparation, no matter how hard the thinker pushed. It arrived after the pushing stopped…on a walk, in the shower, in the gap between two unrelated conversations, or while lying awake at four in the morning.
Incubation isn't a delay before the real thinking happens.
It is the real thinking.
It just doesn't look like it from the outside.
Keats Meets Wallas
Put our Romantic poet and our Fabian social psychologist, who lived a century apart, in a room together, and you might not expect them to agree on much.
But imagine them looking at our computer screen full of AI agents promising instant answers to hard questions, and Keats might say it first, because he already had the words for it:
You have built machines that cannot tolerate uncertainty for even a moment. And you are starting to build people who can't either.
Wallas would nod, and add the mechanism:
You have engineered away the incubation stage. And you are calling the result efficiency.
But I don't think either of them would be protecting slowness. They would likely be talking about protecting the specific kind of unfinished, undirected, unmonetizable time in which a mind is allowed to keep working on a problem without yet being required to produce an answer.
Call it a place between not-knowing and knowing.
Up until now, we have treated that place as a hallway from a room we call the past. Somewhere you pass through, as briefly as possible, on your way to a future room. The one that actually matters. The one with the decision in it.
Keats and Wallas both, in their own vocabulary, would be making the same argument.
The hallway is actually not a hallway at all…it is a room, an important room that matters.
The Agentic World Doesn't Believe in Hallways
We've spent two previous Conversations describing what happens when AI agents become the dominant filter between us and the world's possibilities.
Herbert Simon told us that a wealth of information creates a poverty of attention, and that agents earn their keep by deciding what we don't need to see.
Claude Shannon told us that predictable messages carry little information, and that a system built to eliminate noise risks eliminating the surprises most worth having…including, at the level of a whole business, the improbable strategy that never gets attempted because nothing in the data recommended it.
Both of those Conversations were fundamentally about selection. What gets shown to us. What gets excluded.
This Conversation is about something that happens earlier than selection.
Before an agent can show you an option, it has to have already resolved the uncertainty about which options exist at all. It has already been through stages something like Preparation, Incubation, Illumination, and Verification…but compressed, instantaneously, into a single returned answer.
You never see any of that happen. You only see the output.
Which means, increasingly, with perhaps the exception of Preparation, you never have to go through the stages.
Ask an agent what job you should take, what strategy you should pursue, what to say in a difficult conversation, and it doesn't say: let's sit with that for a while.
It doesn't suggest we stop and incubate. It answers. Immediately.
And the terrible, seductive thing about that fast answer is that it is often good. Sometimes very good. Because AI can produce in seconds an answer that may look remarkably like the result of prolonged human thought.
But it didn't incubate the problem in Wallas's sense.
And more importantly:
Neither did you.
Where the Room Used to Be
Consider how much of ordinary life used to be organized around protecting that place where uncertainty and unfinishedness get resolved.
- The night before a big decision, when nothing needed to happen except your own mind quietly working on it in the background.
- The colleague you called specifically to talk something through, not because they had the answer, but because saying it out loud to another uncertain person helped the shape of the problem become clearer.
- The meeting that ended with "let's sleep on it and reconvene tomorrow," which everyone understood was not a delay tactic but part of the actual decision-making process.
- The blank page you stared at before writing the first sentence, during which, if you were honest, most of the real work was already happening.
None of these looked like productivity. They would not show up on a dashboard. And that is precisely why they are the first casualties of a world optimized for instant, defensible, data-backed answers.
And that can be dangerous. An agent that can give you a recommendation in four seconds doesn't just save you time. It quietly removes your social permission to ask for it in the first place.
Why would you tell your team you need to sleep on a decision when the system in front of you has already modeled the outcome? Why would you stare at a blank page when the page can be filled with possible answers even before taking the time to incubate the question you meant to ask it?
Do that often, and over time that room in between the past and future gets bypassed. Ignored. And then never entered again…eventually dismantled and never rebuilt.
Uncertainty Is Not Just the Interval In Between
Here is the reframe that Keats and Wallas are both, in their own centuries, pointing toward.
We have almost always treated the past as a place…somewhere with a shape, a record, a history you could return to and examine. And we have always treated the future, once it becomes the present, as a place too…a decision made, an outcome you now inhabit.
But the interval between them, the time when you no longer fully belong to the past and don't yet belong to the future, we have treated as a problem to be overcome. A gap. An absence of knowledge, waiting to be filled.
What if that was never correct…what if that interval is instead structurally equal to the other two? Not a hallway between two rooms, but a third room, with its own furniture, its own weather, its own rules. Residing there, not just passing through, offers equal authority to living in the past or the future.
I call it Inhabiting Uncertainty, and it requires the discipline to remain, on purpose, in the state where a problem has been genuinely engaged but not yet resolved…neither returning to the safety of what you already knew, nor reaching prematurely for a solution that only resembles an answer.
Keats gave us a name for the human capacity to remain there. Wallas showed us something that can happen while we do.
But predictive AI introduces something neither could have anticipated.
It can dramatically shorten our stay.
I call that Agentic Compression: the condition in which a system moves us from problem to answer so quickly and smoothly that we may never notice that human thought and creativity were bypassed along the way.
And, even more concerning, that we have quietly stopped inhabiting the place where a genuinely new and creative answer, one that wasn't already sitting somewhere in the training data, might have had time to form.
Now Let's Turn Our Attention to Business
A 2026 study in the California Management Review, looking at AI's growing role in mergers and acquisitions, found something that should sound familiar by now: AI compresses analytical preparation time while leaving judgment, authority and execution work largely unchanged. Deals move faster. Decisions arrive sooner.
But the natural pauses that used to sit between analysis and judgment…the incubation stage, in Wallas's terms…are quietly disappearing from the process.
That should concern any business leader, for the same reason it should concern any individual asking an agent what to do next.
AI can accelerate research and analysis, but judgment requires something AI cannot yet mass produce on our behalf: the lived experience of having struggled, wrestling with a hard problem long enough to feel its edges. A senior executive's value has never been simply processing information quickly. It comes from dealing with thousands of unresolved problems over a career and learning, in some largely unconscious way, what it feels like when a plausible-sounding answer is actually wrong.
Compress that incubation stage across an entire organization long enough, and you don't just get faster decisions.
You get judgment that has never matured in the first place.
The associate who is handed an AI-generated strategy memo instead of a blank page and a deadline may produce something that appears perfectly competent. But they probably didn't spend a sleepless night with the problem, and never built the internal, hard-won sense of when a data-backed answer is quietly wrong…the judgment predecessors built by having no choice but to sit in that room of uncertainty, the one that has since been abandoned.
An organization can optimize away that creative incubation room for years before anyone notices that it has also, without quite deciding to, optimized away its capacity to produce the kind of judgment that only comes from having stood in them.
The Keats Echo
Keats never met an AI agent, and Wallas never watched a large language model return an answer in under a second. But between them, they left us a habit worth adopting deliberately, now that the room they were describing has become so easy to skip.
When an agent hands you a fast, confident answer to something that matters, before accepting it, ask:
- What would this look like if I had actually incubated it myself?
- Am I skipping the room, or did I genuinely not need it this time?
- What did the system arrive at in seconds that a person would once have taken a night, a walk, or a difficult conversation to reach?
- If I gave this problem more unstructured time, not more data, what might change?
And occasionally, on the decisions that actually matter, do something almost defiantly old-fashioned.
Tell no one, ask no agent, and simply sit with it.
Not because uncertainty feels good. It rarely does. But because, as Keats understood two centuries before any of this existed, the capacity to inhabit uncertainty, to actually live there for a while, without impatience, may be the thing that was always doing the real work.
Don't abandon the room between not-knowing and knowing. It's not empty.
It is where your best work happens.
What Inspired This Conversation
This Conversation draws on John Keats's December 1817 letter to his brothers George and Thomas, in which he coined the term "Negative Capability"; Graham Wallas's 1926 book The Art of Thought, which proposed the four-stage model of Preparation, Incubation, Illumination and Verification; and 2026 research published in the California Management Review on how AI is compressing analytical preparation time in mergers and acquisitions while leaving human judgment and authority largely unchanged.