Somebody unsubscribed. Somebody else left a comment that made it clear they'd misread the whole thing and were annoyed about it. The retention graph on the video has a cliff at ninety seconds.
Every instinct says these are problems.
Pashupati's position is that at least some of them are the mechanism working:
If you're a creator with high energy, you'll attract viewers who are high energy. You'll repel viewers who are low energy — they can't take the energy of your channel, they'll run.1
And then the part that's easy to miss:
They won't understand a word that you are saying. And that's a good thing. It's like not only that they'll get repelled and run away — they feel now this is not for me, and they just go. So it's like a filter.2
The mechanism is more specific than dislike, and the specificity matters.
He doesn't say they disagree. He says they won't understand a word — and that they conclude, on their own, that it isn't for them.
Which is a different event from being repelled by content. Disagreement means the signal arrived and was rejected. What he's describing is a signal that doesn't resolve — the person can't get purchase on it, experiences that as irrelevance rather than as difficulty, and leaves without any hard feelings or any memory.
That's why he can call it a filter rather than a cost. Nobody was driven off. They failed to receive, correctly identified that as a mismatch, and went somewhere their band was being broadcast on.
The worked example is his own and it's the best evidence for the mechanism, because it shows the sorting propagating.
When he worked in an art gallery, people would come in and ask which piece to buy. His answer was always the same: go and stand in front of it and feel the energy the artist put in.3
Then the curation:
When I was curating, I would only take artists with high energy. I started attracting clients with high energy — because I took artists with high energy.4
Notice the causal direction. He didn't find high-energy clients and then source work for them. He set the contents of the space, and the space selected its own visitors.
That's an ordinary and observable dynamic that needs no metaphysics: what a venue contains determines who is comfortable in it, which determines who returns, which determines what the venue becomes. Bookshops, bars, forums, comment sections and Discord servers all visibly do this.
The uncomfortable corollary he doesn't state: it also runs downward. A gallery that takes one easy show attracts the audience for it, and that audience shapes what the next show has to be. The filter has no preferred direction — it just amplifies whatever you put in.
The reason this can't be gamed is that the filter reads the whole signature rather than the stated positioning.
The mass-produced aesthetic page and the bear's cave page both make the argument: an audience receives convergence across channels, most of which you aren't supervising. So the filter isn't sorting on your stated niche or your about page. It's sorting on reply speed, paragraph length, room tone, what you refuse to cover — the unsupervised channels.
Which explains a failure that's otherwise puzzling: creators who have correctly identified their audience, positioned precisely, and still attract the wrong people. The positioning is one channel. The filter is reading all of them, and the others are saying something different.
You cannot filter for an audience you aren't. The mechanism is downstream of what you are, which is why the only reliable lever on it is the slow one.
The mechanism is invisible while it's operating, and everything it produces looks like a problem in the dashboard.
A filter working correctly generates: a lower view count than the same effort would otherwise get, a retention cliff early in anything demanding, unsubscribes after your strongest pieces, and a growth curve that looks flat next to peers making broader work.
Every one of those is a red metric. There is no dashboard anywhere that reports successfully repelled 4,000 people who would never have acted on this.
So a creator running the filter well and a creator whose work is simply not good enough see identical numbers, and the framework offers no way to tell them apart from the inside. Which is precisely why the doctrine is dangerous as a comfort and useful only as a discipline: it's available at any moment to explain away genuine failure, and the availability is what makes it seductive.
The only check that helps is the composition count from the sibling page — not how many left but what are the ones who stayed doing. A working filter produces a small audience that acts. A bad piece of work produces a small audience that does nothing, and the total is the same either way.
The comment says the piece was pretentious and needlessly long, and you've read it four times.
Old you does one of two things. Either you're wounded and spend the afternoon on it, or you're defensive and draft something. Both treat it as feedback.
So instead you do a two-minute check on whether this person is inside your filter.
You look at what else they engage with. Not to judge them — to establish the band. And it's immediately clear: everything else is short-form, quick-take, high-frequency. This person consumes at a completely different tempo from the one your work runs at.
Then the honest question, which is the one that stops this from being a rationalisation: is there a version of the complaint that's true?
Sometimes there is. Long isn't a virtue and pretentious is a real failure mode, and if three people who are in your band said the same thing you'd have a problem.
Today it's one person, and it's one person whose tempo is structurally incompatible with what you make.
So you do nothing. You don't reply, you don't adjust, and — this is the part that takes discipline — you don't file it as a data point either, because a data point from outside the filter isn't data about your work.
And you notice, because it's worth noticing: the same piece got two replies from people who read the whole thing and wanted to argue with the third section. Those are the readings that count, and they're easy to under-weight because they arrived without friction.
The most instructive case is one where he didn't let the filter operate, and it shows the doctrine's limit.
A man complained that the retreat was too expensive — three lakhs, and it seemed unfair.5
By the filter logic, that's a clean exit. Wrong band, not for him, let him go.
Pashupati's actual response was the opposite. He reframed it: it's not fair that you're not being taught to see three lakhs as not expensive. That's the unfair part — not that I'm charging. And he pointed him at a free video.6 The man, by his account, said he'd never looked at it that way and left with hope rather than resentment.7
[UNVERIFIED], and 🚩 this is a sales interaction described by the seller.
But the structural point survives, and it complicates the filter doctrine usefully. He treated the man's band as changeable rather than as a verdict — which is precisely what the filter model, taken strictly, says it isn't.
So his own practice contains an exception he never theorises: the filter sorts, and he intervenes on people he thinks are sortable upward. Which means the operative question isn't whether someone is currently in your band. It's whether their being outside it is a disposition or a circumstance — and the source has no way to tell those apart except by trying.
One property the source doesn't discuss and that follows from the gallery: the filter compounds, and it does so on a delay.
Each piece of work sorts an audience. That audience then shapes what the next piece can be — through what gets engagement, what gets replies, what gets asked for, and the entirely ordinary human pull toward the people who responded well.
So the composition you have now was selected by work you made a year ago, and the work you make now is being shaped by that composition. Two loops running at different speeds, with the slow one always a year behind.
Which explains a pattern creators experience as being trapped by their own audience. It isn't a betrayal by the audience — it's the filter operating with a lag. The people who arrived for the accessible piece you made in a thin month are now the majority voice on everything you publish, and they're asking, reasonably, for more of what brought them.
The practical consequence is that changing composition takes about as long as accumulating it did, and the interval is uncomfortable: you produce work the current audience doesn't want, before the audience it would sort for has arrived. There's a trough, and nothing in the dashboard distinguishes the trough from a mistake.
His own fifteen unmonetised years are, read this way, exactly that interval.
Evidence. The gallery case is strong because the mechanism is observable in any curated space and requires nothing exotic to explain.
Tension — unfalsifiable as a defence. They couldn't handle the energy absorbs every piece of negative feedback, including the accurate kind. The framework needs an ordering — check whether the complaint is true first — and doesn't supply one.
Tension — his own exception. The three-lakh man should have been filtered out and wasn't. The doctrine has no account of when to intervene.
Open question. The filter amplifies whatever you put in, including downward. What does a creator do who has already accumulated an audience selected by their weakest work — where the filter is now actively holding them at that level?
Set this against The Peacock and the Blind Man — his own — and the two describe the same sorting from opposite ends, with different action implied.
The peacock page is about the creator's discretion: don't display where it can't be seen, have the judgement not to work with those people. The move is active and it's yours.
This page is about the audience's self-selection: they arrive, fail to resolve you, and leave on their own. Nothing is done.
Which is a genuinely useful pairing, because it identifies two different jobs. The peacock governs what you accept — commissions, collaborations, clients. The filter governs what happens to everyone else, and the correct posture there is to leave it alone.
The failure mode is running the wrong one in each situation: exercising discretion over your audience, which is contempt, or passively accepting whatever client arrives, which is the designer's ruined work.
Against Rasa Weaver vs Content Monk — The Two Paths the tension is worth naming. The Content Monk's edge-dwelling and the filter's repulsion look similar and aren't. The Monk withdraws from the audience by disposition. The filter is a property of work that's fully present and simply not receivable by everyone — you're broadcasting at full power and some receivers can't tune it.
To Branding Hub — business
The premium corpus documents deliberate repulsion as strategy: the brand that signals not for you, the price that excludes, the deliberate refusal of mass legibility. Its warrant is that a signal legible to everyone carries no information.
The energy filter reaches similar outcomes without the deliberation, and the difference is where the sorting criterion lives.
Designed exclusion sorts on an attribute the brand selected — income, taste, membership, whatever the boundary was drawn around. The filter sorts on capacity to receive, which nobody chose and which cuts across every demographic. A wealthy person and a broke one can be equally unable to resolve a piece of work, and equally able.
Which predicts something the branding literature would find awkward: an energy-filtered audience should be demographically incoherent in a way a designed-exclusion audience never is. If your audience looks like a segment, you built a boundary. If it looks like a strange assortment of people with nothing in common but how they respond to things, the filter is doing the work.
To Somatic Trauma Theory Hub — psychology
The somatic literature describes what happens when material arrives that a person cannot currently metabolise: not rejection, but a kind of sliding-off — the content doesn't land, attention moves, and the person often can't report that anything happened. Clinicians treat it as a capacity signal rather than as disinterest.
That is almost exactly they won't understand a word and they'll just go.
What the pairing produces is a caution about the word low-energy, which does a lot of unexamined work in this corpus. The clinical account says non-reception is frequently protective and state-dependent — the same person, differently resourced, receives the same material fine. Which means the filter isn't sorting people into permanent categories at all; it's sorting moments. Somebody who slid off your work in March may be exactly your reader in November, and the framework's language — cowards, low-energy people — encodes a stable trait where the evidence supports a state. That's the difference between a filter and a judgement, and Pashupati's own three-lakh exception suggests he knows it in practice while his vocabulary doesn't.
Sharpest implication. The filter reads every channel, not the positioning — which means you cannot filter for an audience you aren't, and a creator who has correctly identified their ideal reader while broadcasting a different signature will keep attracting people they can't serve. The lever on composition isn't targeting. It's the slow one.
Generative questions.