{{ title }}

Ihave been circling the same thought for months now, and writing it down is the only way I know to find out whether it holds.

It started with a particular kind of person. They sit inside a regional development agency in Norway, a næringshage, and their job is to make the small companies around them a little stronger. A campsite owner. A timber framer. A cheesemaker who wants to sell beyond the valley. The advisor is supposed to know a bit about everything, stay current on everything, and still say something specific when a founder calls on a Tuesday afternoon with a real problem and twenty minutes.

I kept watching that job and feeling that something was off about how we talk about it. We assume the advisor’s difficulty is a lack of knowledge: that if they only read more, subscribed to more, attended one more course, they would be ready. But that is not what I saw. What I saw was someone drowning in availability and starving for relevance. The problem was not too little. It was far, far too much.

Abundance is a tax on judgment

We inherited a story from the last century that says information is scarce and valuable, and that more of it is always better. For most of history that was true. It is not true anymore, and it is least true for the people whose entire job is to exercise judgment.

The research here is older and blunter than I expected when I went looking. Studies of decision support systems show that past a certain threshold, more information actively degrades the quality of decisions, because the mind starts spending its limited budget on filtering instead of deciding (Monash University). One older line of work names the condition plainly: information overload is the state where the inputs exceed the decision maker’s capacity to take them in and act (ERIC). So the advisor opens another newsletter about artificial intelligence, skims it, feels vaguely behind, and closes it. Nothing changed. The newsletter told them what happened in the world. It said nothing about what to do with it, on Tuesday, for the cheesemaker.

That gap, between what happened and what to do about it for this person, is not a content problem. It is a judgment problem wearing a content costume.

The same difficulty, all the way up

What unsettled me, the more I looked, was that this is not a small or a local problem. It is the exact difficulty carried by some of the most expensive people in the world.

Management consulting, stripped to its bones, is the business of selling judgment: the knowledge, the frameworks, the read on a situation that a client could not produce alone, or could not produce in time (Dewx). The economics are enviable, and a partner at one of the famous firms bills somewhere between three and eight hundred dollars an hour, with the whole enterprise floated on reputation and thought leadership (TacticalVC). But the engine under the slide deck is not charisma. It is a knowledge system. BCG has said this about its own clients almost without embarrassment, arguing that a genuine knowledge capability creates a measurable advantage worth tens of millions a year to a large fund (BCG). McKinsey has written that the real value comes less from storing knowledge and far more from creating and exchanging it (McKinsey Quarterly). The famous firms, seen from a certain angle, are intelligence layers that happen to wear good suits.

The venture world has its own version. A studio or an accelerator spreads scarce operational expertise across a portfolio of young companies, so that each one can borrow knowledge in product and finance and growth that none of them could afford to own (LinkedIn). The operating partner is the valley advisor again, only with carried interest and better coffee. They sit above a portfolio, they are expected to be current across all of it, and the thing they are actually short of is not effort. It is attention.

So the difficulty is the same all the way up the ladder. The only thing that changes is whether you can afford to build the intelligence layer in house. McKinsey can. A studio with three partners cannot. An agency in a Norwegian valley certainly cannot. I find that asymmetry quietly fascinating, because it means the most ordinary advisor and the most prestigious one are wrestling with the same beast, and only one of them gets to hire help.

What a good week of advice actually looks like

When I try to describe what would genuinely help any of these people, I keep arriving at the same shape. Not more information. A particular way of cutting it.

Take any single thing that happened in the world this week and ask four questions of it, in order.

I notice that the third question is where the meaning lives, and the second question is where the difficulty lives. Because before you can say what something means for a portfolio, you have to decide whether it belongs to that portfolio at all. And that, it turns out, is a piece of mathematics.

A small equation I cannot put down

This is the part I find genuinely beautiful, and I want to lay it out as a thing I am still working through rather than a thing I have solved.

The oldest idea in information retrieval is to turn a question and a document into vectors and measure the angle between them. Two pieces of text are alike to the degree their vectors point the same way, which we write as cosine similarity:

The classic way to build those vectors, TF–IDF, says a word counts for more when it shows up often in one document but rarely across the whole library, so common words are cheap and distinctive ones are dear (Information Retrieval Facility). Search engines have spent decades refining this. The library underneath most of the search you touch every day deliberately dampens these terms, because relevance does not actually rise in a straight line, and modern scoring bends the curve so that the tenth mention of a word matters far less than the first (fred.glass).

That much is borrowed and well understood. Here is where my own thinking begins, and where I am least certain.

A portfolio is not a search query. It is a living thing with weights and soft spots. So when I imagine scoring a single incoming signal s against one advisor’s portfolio P, I find myself reaching for something like:

Read slowly, every symbol is a small claim about what attention is worth. The cᵢ are the segments of the portfolio: the tourism people and the builders and the food producers. The sim(s, cᵢ) is how near the signal sits to each of them, the cosine idea from before. The wᵢ is how much that segment actually matters to this advisor, because a region that lives on tourism should not hear as much about heavy industry. The τ(s) is a decay over time, because intelligence spoils, and a signal that demanded action three weeks ago is just history now.

And then there is σᵢ, the exposure term, and it is the one I keep turning over in my hands. It is my attempt to capture not how related a segment is to a change, but how vulnerable it is to it. A small operator with a thin web presence is far more exposed to a shift in how people search than a national chain that will be fine either way.

Anyone can compute the similarity. The vectors are a commodity. The part that would be hard to copy was never the model: it was the weights, and the weights are just a precise name for local knowledge.

The whole quality of that third question, the political and structural read, is bounded by how honestly you can set σᵢ, and exposure is not a technical quantity at all. It is a claim about power and geography and who absorbs the cost when something changes. You cannot scrape it from anywhere. You have to actually understand the place.

Noticing as a discipline

The other half of what I keep circling comes from a field called strategic foresight, which I came to late and wish I had found earlier. Its central practice, horizon scanning, is an organized search for weak signals: the faint early signs that something is shifting in ways that could later matter enormously (Policy Horizons Canada). Done with any consistency, scanning is how an organization learns whether it is ready for what is coming before it arrives (National Academies Press). Serious institutions now build their anticipation of risk on exactly this, beginning with weak signals from unglamorous sources well before a trend is obvious to everyone (OECD).

What I find quietly radical is that foresight treats noticing itself as a craft. Not predicting, which is mostly vanity dressed as confidence, but noticing. A weekly rhythm of scanning and filtering and translating is not glamorous work. It is closer to keeping a workshop swept and the tools where you can find them. But it is the thing that lets an ordinary advisor sound, to the people who depend on them, like they can see a little way around the corner.

Why this is not simply a cheaper version of the famous firms

I should be honest about a comparison that hangs over all of this, because when I describe it to people they reach for it immediately. Is this just a budget McKinsey.

The more I sit with the question, the more I think the differences run in the opposite direction to almost everything those firms are built on, and that the contrast is more interesting than flattering to either side.

The famous firm runs on leverage, a pyramid of junior people beneath a partner, and to serve twice the clients you need roughly twice the bodies. The thing I am describing is closer to a product than a pyramid. The scanning happens once for everyone, and most of the effort in the next client is the patient tuning of their particular weights. Even the newer firms feel this current pulling at them, which is why some are now built around senior judgment sitting on top of capable software rather than an army of analysts (FourWeekMBA).

But the deeper difference is about credit, and it took me a while to see how total it is. A consulting firm’s product is its name. You pay, in part, so that you can say the name out loud in a boardroom. Their authority is the asset, and an asset like that has to be visible to be worth anything. What I am describing is the precise inverse. Its entire value depends on the client’s name being on the work and mine being nowhere at all. We are not fighting over the same authority. The point is to make authority that someone else gets to wear.

And from that one fact, a strange kind of durability follows, almost by accident rather than by cleverness. There is a concept from Hamilton Helmer’s writing on competitive advantage called counter-positioning, which describes a newcomer adopting a model that the established player declines to copy, not because they are blind to it, but because copying it would damage the business they already have (Commoncog). I did not set out to build that. But it is sitting right there. A famous firm could, in principle, offer cheap and invisible work tomorrow. They will not, because invisibility is poison to a business whose whole margin rests on its name being seen, and Helmer’s quiet irony is that the higher those margins are, the more they have to lose by changing (Gavel). Their greatest strength is the very thing that holds them in place.

I notice the same logic in the math. The famous firms sell general brilliance, people who can drop into nearly any problem. What I keep arriving at is the opposite, a reading of exposure that only makes sense for one client’s particular companies in one particular place. General brilliance travels anywhere. The exposure term refuses to travel at all, and the part that refuses to travel is the part that has to be earned slowly, on the ground, over time. I have come to think the things that are hard to copy are usually just the things that were hard to earn.

What I think I am actually doing

Strip away the equations and the citations and it gets simple, almost humble. I am trying to build a quiet system that reads widely, judges narrowly, and hands a busy advisor a very short list of things worth their attention, in their own voice, ready to act on.

There is a line I keep written where I can see it.

This happened. Here is what it means for your people. Here is the weather around it. Here is what you might do about it now.

Everything else, the vectors and the decay curves and the stubborn exposure weights, is just an attempt to earn the right to say those four short things honestly, week after week, to someone who has too much to read and a founder already waiting on the line.

One temptation I want to name, because it is real and I feel it. Once you have an equation and a capable model, you want to let the machine write the thing and send it. I do not believe in building it that way, and not only for safety. A message generated and sent with no one reading it is no longer judgment. It is content again, the very thing I was trying to escape. The machine can propose. A person has to decide. That is most true in the third question, where a careless sentence about regulation or about children’s data could do real harm, and it is true in the fourth, where the words go out under someone else’s name and someone else’s reputation. The point was never to replace the judgment that made an advisor worth listening to. The point was to give that judgment more room to breathe.

I still do not know whether the exposure term can be learned by a machine or whether it will always need a human hand on it. That is the open question, and it is the one I am building toward rather than away from. But I have started to believe something that runs against the grain of the age. The systems that help us think well are not the ones that produce the most. They are the ones with the discipline to leave almost everything out.