Watch the first video. The teacher counts the faces of Śiva: first this one, second that one, Aghora is third, Tatpuruṣa fourth, Īśāna fifth and on top.1
Four months later, same channel, same teacher, opening by telling you he's continuing the earlier episode.2 He counts again. Vāmadeva second — fine, that matches. Then Īśāna is "the fourth surupa."3 Then Aghora is "the fifth."4
Nobody mentions the discrepancy. Not in the second video, not anywhere. The faces are the same five, the elements attached to them are the same five, the functions are the same five. Only the numbering moved.
That's the interesting thing on this page, and it's worth getting straight before anything else: the correspondence holds and the sequence doesn't.
Strip the numbering out and a stable table is left standing. Every face keeps its element and its job across both recordings:
| Face | Direction | Element | Function |
|---|---|---|---|
| Sadyojāta | west | earth | creation |
| Vāmadeva | north | water | preservation |
| Aghora | south | fire | destruction, read as cleansing |
| Tatpuruṣa | east | air | concealment |
| Īśāna | upward | ether | revelation |
Vāmadeva is north and soft in both.5 Aghora is south and fiery in both.6 Īśāna looks up and takes in the whole cosmos in both — the second recording calls it subtle, vast, formless, "seeing the entire cosmos."7
So whatever slipped, it wasn't the content. A face is bound to its direction, its element and its function tightly enough to survive being retold months apart. What it isn't bound to is a position in a list.
Here's the thing about ordered lists in a teaching tradition: the order usually carries a claim. First means foundational, or earliest, or where you start. Last means highest, or hardest, or where you end up. When somebody hands you five things in order, you're being told something beyond the five things.
Watch what each ordering would mean if you took it seriously.
Take the first: creation, preservation, destruction, concealment, revelation. That's a story with a shape. The world gets made, held, unmade — and then, only after the unmaking, comes the veil, and only after the veil comes the lifting of it. Concealment is late. It happens to someone who has already been through the making and unmaking, which reads as a claim about spiritual maturity: you get hidden from after you've been around the block, not before.
Take the second: Īśāna fourth, Aghora fifth. Now revelation is not the endpoint. Something comes after being shown, and that something is fire. Read it straight and the claim inverts — you are shown the thing, and then you are burned. Revelation isn't the destination; it's what makes the burning possible.
Both are coherent. Both are teachable. They are not the same teaching, and no line in either recording chooses.
Push on the second ordering, because it's the stranger one and because the second recording is entirely about Aghora, so the source is not being careless about that face.
In that recording Aghora is bound to something the first one never says: Aghora is the guru of Bhairava, and that guru is Kāla.8 Time. So the fifth face, in the ordering where it's fifth, isn't just fire — it's the teaching function, and the teacher is time itself.
Now the ordering does work it couldn't do at position three. If the last thing after revelation is fire-as-teacher-as-time, then the scheme is saying: being shown is not the end, because you then have to live in time with what you were shown. That's a recognisable claim. Anyone who has had a genuine insight and then watched it fail to change their life knows the gap between seeing and being changed. Fire in the fifth position names that gap and says the burning is what closes it.
At position three, Aghora can't do any of that. It's the ordinary destroying-god slot, between preservation and concealment, doing what destroyers do.
So the two orderings aren't a slip and a truth. They're two different theologies of what fire is for. And the recording that thinks hardest about Aghora is the one that moves it to fifth — which is at least suggestive, though it is one instance and proves nothing on its own.
There's a moment in the second recording where the teacher stops mid-explanation to head off a misreading, and it tells you something about who he thinks is listening.
He gets to Vāmadeva, the north face, and immediately interrupts himself: Vāmadeva is not about vāmācāra.18 Then he goes back to describing the face as soft.
You have to know the landscape to see why he bothers. Vāma means left. Vāmācāra — the left-hand path — is the tantra of transgression, the one with the meat and the wine and the sexual rite, the one that outsiders find first and fixate on. Anyone who has read three articles about tantra has met vāmācāra and met almost nothing else.
So a teacher naming a face Vāmadeva to an internet audience knows exactly what half of them will hear. They'll hear the transgressive path. And what he actually needs them to hold is the opposite register — north-facing, soft, preservation, water.
The correction is a single sentence and it's not doctrinal. It's a repair on a predictable collision between a Sanskrit name and what his audience has already been taught to expect from that syllable. Which means the table on this page isn't only a transmission of a scheme; it's a scheme being defended, in real time, from the associations the medium brings with it.
Worth remembering when reading the rest of the corpus. Some of what looks like emphasis is actually damage control.
None of this is a new mess. The vault already holds a Purāṇic account where the five faces and the eight forms refuse to line up, where the mapping misaligns and a translator has to patch some seams and flag others.9
Which reframes what we're looking at. A single teacher contradicting himself across two videos looks like carelessness. A single teacher contradicting himself across two videos inside a tradition whose written sources already contradict each other on the same point looks like inheritance. He is passing on an instability that was handed to him.
That doesn't make the contradiction less real. It makes it less likely to be his fault, and much more likely to be load-bearing — because instabilities that survive centuries of transmission usually survive because something is holding them open.
You're at a desk, tired, and you've had one of those weeks where you can name exactly what's wrong with your life. Genuinely name it — not the vague dissatisfaction, the specific thing. That clarity arrived last Tuesday and it hasn't shifted anything.
That's revelation, and you've just discovered it isn't self-executing.
Sit with the second ordering for a minute. It says something comes after this. It says the fire comes after the seeing, and the fire's other name is time. Not a technique. Not a next step you can schedule. Just: you now have to carry this knowledge through months, and the carrying is the part that burns.
You notice you'd rather have the first ordering. In that one, the fire came earlier and the seeing was the reward at the top — you'd be finished by now.
Weeks later, you're in an argument and you feel the accurate cruel sentence arrive, fully formed. South face, fire. You don't say it. Nothing dramatic happens. You just notice that the not-saying cost something, and that the cost is the point, and that nobody handed you a technique for it.
Months later still, the Tuesday knowledge has changed something small in how you spend a morning. You couldn't say when.
That's the whole workflow the fifth position implies. It has no steps because time isn't a step.
Two video transcripts, one teacher, auto-generated captions, four months apart. Sadyojāta arrives as "sady jot," Tatpuruṣa as "tpurusha," Aghora as "agora," Īśāna as "ishana." Every name in the table above is reconstructed from context.10
And both recordings invoke purāṇas — the Liṅga Purāṇa, the Devī Purāṇa — as carrying this material, with no chapter and no verse.11 So the table cannot be checked upstream from inside this corpus. It is what this teacher teaches, twice, with the numbering different each time.
The evidence for the correspondence is as good as this corpus gets: two independent recordings, months apart, agreeing on every face's direction, element and function without agreeing on the order. That's the signature of something transmitted rather than improvised. If he were making it up, the details would drift and the sequence would hold, because sequence is the easy thing to remember. The opposite happened.
The tension, preserved and not resolved. Aghora is third in one recording and fifth in the other; Īśāna is fifth in one and fourth in the other. Neither acknowledges the other. Do not pick one. The two orderings encode different claims about whether revelation or fire is the terminal position, and this material does not settle it — see the source stubs, where the conflict is recorded as a fact about the sources.
A second tension, inherited. The written tradition this teaching claims descent from already fails to make the five faces line up with its own eight forms.12 So there are now two layers of disorder — the purāṇic one on the page and the oral one across two recordings — and it is genuinely unclear whether the second is a symptom of the first or an independent slip.
Open question. Is the ordering variable because it's unimportant, or variable because it's contested? Those look identical from outside and mean opposite things. A tradition that doesn't care about sequence and a tradition with two live factions would both produce exactly this record.
Come at the five faces through the vault's Śāktalineage page and the faces are doors — each one opening onto a goddess school, five faces resolving into four traditions, and the whole interest is in where the doors lead.13 Nobody there is counting. The faces are places you can go.
Come at them through the Purāṇic material and you're doing arithmetic, badly, and losing. Five into eight doesn't go, the exposition covers seven of them, the translator patches.14 The faces are entries in a system that won't balance.
Then come here, and the faces are neither doors nor entries. They're functions — jobs being performed right now, in you, on the world. And the moment they become jobs, the ordering question gets teeth, because jobs happen in sequence and sequence implies a claim about what has to happen before what.
That's the actual difference between these three accounts, and it isn't a difference about the faces at all. It's about what kind of thing a correspondence is for. A map of doors doesn't need an order. A ledger needs one and can't produce it. A process has one and disagrees with itself about what it is.
African spirituality — the lukasa, and what a memory board does and doesn't hold. The Luba lukasa is a hand-held board studded with beads and shells, carried by a trained keeper, read with the fingertips to recite genealogy and royal history.15 The vault's page on it lays out precisely the problem such a device solves: a genealogy that everyone recites from memory drifts, each person remembering it slightly differently, small variations accumulating until after fifty years there are three incompatible versions.16 The board stops the drift. What's fixed on it stays fixed.
And that page closes with an open question it can't answer from its own sources — is there a specific sequence an initiate learns, or is the encoding non-sequential?17
This corpus is a case where you can watch the answer. Two recordings, four months apart, one teacher. The bindings held perfectly: every face kept its direction, its element and its function, five for five. The sequence moved. Aghora third, then Aghora fifth, and no awareness that anything changed.
That is what a lukasa-shaped transmission looks like when the board is gone. The bindings behave like beads — fixed positions, fixed relations, unlosable. The narration order behaves like the keeper's performance — supplied fresh at each telling, because it was never on the board. A reciter working from a fixed encoding will produce a different route through it every time and feel no contradiction, since the route was always theirs and the encoding was always the tradition's.
The insight neither domain produces alone: the shape of what drifts tells you what kind of thing was encoded. Stable bindings plus unstable order is the fingerprint of a memorised correspondence being narrated. Stable order plus drifting content would be the fingerprint of a story being retold, which is the failure mode the lukasa exists to prevent. So the Bhairava material offers the lukasa page a data point it was missing — a tradition where the ordering demonstrably is not part of the encoding — and the lukasa page offers this one a mechanism that makes the contradiction stop looking like carelessness.
⚑ One caution, because the domains are not symmetrical here: the lukasa is a physical object with a documented ethnographic literature, and this is a caption file from a video channel. The parallel is structural and it is not evidence about Luba practice. Nothing on this page should be read back onto the lukasa.
Sharpest implication. The corpus preserves what each face does perfectly and loses the order they do it in — which means the five faces were transmitted as a table and are being retold as a process, and every ordering claim built on them is the teller's, not the tradition's.
Generative questions.