Follow the Huh
- Rich Washburn
- 1 day ago
- 8 min read
Updated: 1 day ago
I've spent a lot of time over the years building systems to organize information, which is funny because the way I actually think has never been particularly interested in being organized. My brain is less filing cabinet, more cognitive landfill. Not landfill in the useless sense. More like the place where everything gets dumped because I'm not always qualified, in the moment, to decide what is trash and what is going to turn out to be important six months from now.
A weird article. Something somebody said. A diagram. A half-baked business idea. Some piece of hardware I probably don't need. A question about physics. Something AI did that it probably wasn't supposed to be able to do yet. A sentence that just randomly showed up in my head while I was driving.
Most of it has absolutely nothing to do with whatever I'm supposed to be working on.
Until suddenly it does.
That's the part I've started paying more attention to. I came across a video about how people like Leonardo da Vinci used notebooks. The interesting part wasn't really that Da Vinci took notes. Everybody knows that. It was how little concern there seemed to be for keeping the notes obediently separated. Anatomy could be sitting next to engineering, shopping lists next to scientific questions, random observations beside whatever else happened to be occupying his attention.
That looked suspiciously familiar.
Not because I'm comparing myself to Leonardo da Vinci. Settle down. Because the behavior made sense. The page wasn't necessarily a place where finished knowledge was being stored. It was a place where cognition was happening. That sent me down another rabbit hole, which is generally how these things go.
The Huh Before the Idea
I realized I'd been thinking too far downstream whenever I talked about ideas.
An idea is already fairly developed. Before the idea there's usually something smaller. You're reading something and stop for a second.
You hear somebody say something. Two completely unrelated concepts touch in your head. Or nothing happens externally at all. Your internal dialogue just coughs up a sentence that sounds interesting and some other part of your brain goes: Huh.
That's it. That little moment is the beginning. It might be profound.
It might be stupid. It might be completely wrong. It might disappear fifteen seconds later. The problem is that most systems ask that little Huh to explain itself before we're willing to save it.
What project does this belong to?
What category?
What's the next action?
Is this important?
What are you going to do with it?
I don't know. It's eleven seconds old. Maybe give it a minute. The phrase we ended up with while talking through all of this was: Attention is enough to earn a first move.
Not enough to become a project. Not enough to become a commitment. Just enough to preserve it. That feels different.
Why I Think That Matters More Now
There was a time when ruthlessly filtering ideas made a lot of sense.
Following a random curiosity could get expensive quickly. Research might consume an afternoon. Prototyping something might take a week. Building software required actually building software. Turning an idea into something you could examine often involved enough friction that deciding whether it was "worth it" was a useful survival mechanism.
AI has changed that equation.
The cost of curiosity has collapsed.
I can take a strange thought and ask questions about it immediately. I can challenge it, compare it against something else, research the background, sketch an architecture, build a crude prototype, argue against it, or turn it into something tangible before the initial enthusiasm has even worn off.
That doesn't mean every idea is suddenly good.
It means bad ideas have become much cheaper to kill and potentially good ones have become much cheaper to explore.
That moves the threshold. Something no longer has to be interesting enough to justify three days. Maybe it only has to be interesting enough to justify three minutes. That's a pretty meaningful change.
Structure Follows Gravity
The other thing I've realized is that I don't actually want everything organized. I want things to become organized when they earn it. That sounds subtle, but I think it's a pretty big difference.
If I have some random thought about AI memory while I'm also sketching a circuit and making a note about something a client said, I don't necessarily want those three things immediately separated into AI, Electronics and Client Projects. They happened next to each other for a reason, even if I don't understand the reason yet.
Sometimes adjacency is information. Sometimes the thing that looks like disorder is the part you want to preserve. That is probably why my cognitive landfill works as well as it does. The landfill allows things that aren't supposed to touch to touch.
Eventually something starts showing up repeatedly. I mention it again. I notice another connection. I find myself researching it when I'm supposed to be doing something else.
Now it has gravity. That's when it deserves structure.
So the system I've been putting together works more like:
Huh → Capture → Chase → Connect → Test → Learn → Reframe
And if something keeps pulling:
Develop → Brief → Build
That's very different from immediately turning every thought into a task.
Sometimes curiosity is the work.
This Became the "Huh" Protocol
Once I started pulling on this, it turned into something I'm calling the Huh Protocol, which is an appropriately serious name for a system whose primary entry command is basically a confused noise.
The idea is simple enough. Give spontaneous thought somewhere to exist before forcing it to justify itself. I started designing a small family of reMarkable pages around different states that I already seem to move through naturally.
There's an Everything page, which is the cognitive landfill in its purest form. Almost nothing on the page. Just somewhere to dump whatever showed up. There's a Huh page for when one thing catches enough attention that I want to spend a little time with it. A Thread page is for when something starts developing gravity. A Collision page exists specifically for the moment when two things that shouldn't necessarily have anything to do with one another suddenly appear connected. There's a Poke page for giving reality a chance to disagree with me. A Brief page comes much later, once the ambiguity collapses enough that I actually know what I'm trying to make. And then there's Return, because nothing really ends. The thing you build teaches you something, which becomes another input, which creates another question, which sends you around again.
It isn't really a loop. It's more of a spiral.
You come back to the same places, but hopefully not as quite the same person.
Then I Put Rails Around the Landfill
The templates ended up getting one more layer that I think may be the most interesting part.
I call it the Huh Rail.
The middle of the page remains mostly open because I don't want the system interfering with the thought. But around the edges are small marks I can use to leave context for whatever intelligence deals with the page later.
Things like:
research this
connect this
verify this
challenge this
hold this
this is a question
this is a hunch
this came up before
this reminds me of something
this might branch somewhere else
The point isn't to turn handwriting into a form. Quite the opposite. The handwriting stays authoritative. The little marks are basically breadcrumbs. Human meaning in the middle. Machine hints at the edge.
The phrase we eventually landed on was: Capture freely. Annotate lightly. Organize computationally.
That feels much closer to what I want than another elaborate note-management system.
My Version Is, Admittedly, Stupid
This is where I should disclose that I have taken this substantially further than anyone needs to.
My reMarkable is already effectively networked paper. I wrote about that in a piece called Networked Paper, because I realized how useful the device became once the page could exist somewhere other than directly under my hand. I can stream the reMarkable screen live, so something I'm drawing can be visible elsewhere while I'm drawing it. The weirdly useful part is that paper remains paper. I can talk, listen, draw boxes, circle things and connect ideas without constantly operating a computer interface.
I've now taken that a step further. My bot can see that stream when I ask it to. I can trigger that from a Siri command or send a text message and essentially say, go look at what's on the paper. At that point the paper isn't disconnected anymore.
I can scribble something by hand, mark parts of it with the Huh Rail, then ask the AI to look at it, understand the page, research something, connect it to previous material, or help move it forward. That is, objectively, an unreasonable amount of plumbing behind what still feels like writing on a yellow pad.
Which is extremely on brand. But none of that plumbing is required.
If somebody wants 90 percent of the idea without building Washburn NORAD for a notebook, take a picture of the page and drop it into ChatGPT, Claude, or whatever multimodal system you're using.
"Read this."
"Research the thing I circled."
"Does this remind you of anything we've discussed?"
"Challenge this hunch."
"Turn this into a brief."
That's already enough to make the paper computational. You don't need the live stream. The stream just removes another little piece of friction, and I have an unhealthy relationship with friction.
The Human Part Should Stay Human
That may actually be the larger thing I'm trying to preserve here.
We have a tendency, whenever computers get involved, to redesign human behavior around what the computer wants. Databases want fields. Software wants categories. Project-management systems want status. AI systems want clean inputs.
I'm becoming increasingly interested in doing the reverse. Let human cognition remain weird. Let it be spatial. Let it be incomplete. Let it jump around. Let a thought exist beside another thought for no reason we can explain yet. Let someone write a question diagonally across the page because that's where it fit. Then make the machine adapt to that.
AI is particularly good at this because it doesn't require the rigid schema that earlier software did. It can read context. It can infer relationships. It can retrieve semantically instead of needing you to remember exactly which folder you put something in. That changes what we can ask the human to do. Or, maybe more importantly, what we no longer need to ask the human to do.
The machine can carry more of the organizational tax. The human can spend more time noticing.
Who This Is Probably For
I suspect this is particularly useful for people who have spent their lives being told their thinking is a little scattered. People with too many interests. People who get excited about something and disappear down a rabbit hole. People who notice connections between unrelated domains and then struggle to explain how they got there. People who generate ideas faster than they can organize them. People who have opened a note app to write down one thought, discovered forty-seven other notes they forgot existed, and somehow ended up researching something else entirely.
There are plenty of labels around that kind of cognition. ADHD, neurodivergent, creative, multidisciplinary, distracted, obsessive, pain in the ass. I'm less interested in the label than I am in the design problem.
Maybe the answer isn't always making that person think more linearly.
Maybe sometimes you build a better system around the way the person already thinks.
That's what I'm trying here.
Follow the Huh
I'm giving the pages away. PDFs and PNGs. No course. No certification. No 47-step productivity doctrine explaining the approved way to have a thought.
Use the whole thing. Use one page. Steal the parts you like. Ignore the rails. Change the symbols. Print them. Put them on a reMarkable. Take pictures and give them to an AI. Build an insane streaming pipeline because you also have poor impulse control. Whatever.
The protocol isn't really the templates anyway. It's the underlying permission. Something caught your attention. Before you dismiss it because you don't know where it belongs or what it's worth, give it somewhere to land.
Maybe it goes nowhere. Maybe it becomes something. Maybe six months from now it collides with some other piece of junk sitting in the cognitive landfill and suddenly you understand why you kept it.
That happens to me often enough that I've stopped assuming the landfill is the problem. Maybe I just needed better recycling equipment.
Follow the Huh.
Rich Washburn is a technologist, strategist, and Founder & Chief AI Architect of ARIA AI Labs, working at the intersection of AI, infrastructure, communications, and capital. He also serves as Managing Partner and Chief AI Officer at Eliakim Capital.
Comments