Extraverted intuition takes whatever is currently in front of you — an object, a sentence, a problem, a stranger’s throwaway remark — and returns the set of things it connects to. Not the deepest connection. All of the shallow ones, at once. It is a breadth-first search: expand every adjacent node before walking down any of them. That single mechanical property explains everything Ne is famous for, including the parts nobody enjoys: the tangents, the six-tab brain, the abandoned projects, and the specific way a Ne user texts.
Divergence, stated mechanically
Ne is a perceiving function, which means it does not evaluate. It has no opinion about whether a branch is true, useful, kind or achievable — those are jobs for the judging function sitting next to it in the stack. Ne’s only operation is expand. Give it one node, get back five or six neighbours, each of which is itself a node with five or six neighbours. The tree grows faster than any human can walk it, which is why Ne feels like abundance from the inside and like interruption from the outside.
Watch it run on something deliberately boring. A hotel key card, the flat white plastic kind, sitting on a desk. Here is one pass — six branches, generated in maybe four seconds, none of them checked for truth:
Branch 1 — the unit
Thirty cents of plastic, reissued every time someone loses one. How many does a chain burn a year, and is the replacement flow a business somebody already owns?
Branch 2 — the protocol
The card knows nothing. The lock holds the schedule and the card is just a bearer token with an expiry. So a hotel is an authorisation system that happens to have beds attached.
Branch 3 — the desk moment
They slide two cards across and never ask who the second person is. There is a whole social contract compressed into that non-question, and everyone has agreed not to look at it.
Branch 4 — the story
Someone who keeps the key card from every hotel they have been fired in. Sixteen cards, a drawer, and one of them is from a place that no longer exists.
Branch 5 — the folklore
Everyone says a phone demagnetises the card. That has probably not been true for years, so why did the belief outlive the technology, and what else are we still explaining with a dead mechanism?
Branch 6 — the category
A key used to physically do the work. This one only requests permission. When did objects stop acting and start asking — fobs, barriers, elevator buttons, the whole quiet handover?
Notice what did not happen. Ne did not rank those, did not test whether the clock claim in branch two is even correct, and did not ask whether anyone wants to hear branch four. It also did not go down. A depth-first search would have taken branch two and spent the four seconds pushing it toward one conclusion. Introverted intuition does that, which is why Ni users produce a single unexplained certainty and Ne users produce six explainable maybes.
The same six branches, two different survivors
This is the part most Ne explainers skip, and it is the part that actually determines what a Ne user is like to know. The generation step is basically identical across Ne users. What differs is which branches get a second step — and that is decided entirely by whichever judging function sits next in the stack.
Ti-paired Ne — ENTP (Ne-Ti-Fe-Si) and INTP (Ti-Ne-Si-Fe) — keeps the branches that make a checkable claim. Branch two survives, because “a hotel is an authorisation system with beds” is a structural statement that could be wrong, and being able to be wrong is what makes it interesting. Branch six survives, because it proposes a category. Branch three gets dropped in under a second, not out of coldness, but because “that moment at the desk is strange” does not generalise into anything you can push on.
Fi-paired Ne — ENFP (Ne-Fi-Te-Si) and INFP (Fi-Ne-Si-Te) — keeps the branches with a person inside them or a value under pressure. Branch three survives and turns into a whole scene. Branch four survives and gets a name and a reason for the firings. Branch one survives, but only its ethical wing — the waste, not the unit economics. Branch two gets politely agreed with and never mentioned again.
Ne in writing: the structural signature
Ne is more visible in text than any other function, because writing is sequential and Ne is not. A branching process forced through a one-dimensional channel leaves marks: the sentence that answers and then keeps going, the parenthesis that opens a second thread rather than clarifying the first, the clause abandoned mid-word when a better branch arrives. Three real shapes:
“yeah 7 works. although if it is 7 we could take the back room, which means the projector is actually possible — or we skip the projector entirely and do the whole thing outside, weather permitting, which it will not be”
“finished the deck (mostly — the middle section is still the version you hated, I just moved it) and I think the argument holds. it is only that the argument turned out to be three arguments”
“the reason it broke is the retry logic — actually no, wait. the retry logic is fine. nothing clears the queue on restart, which also explains that thing from March nobody could reproduce”
Sample one is the branch. The question was answered in three characters. Everything after “yeah” is expansion, and the “or” is not offering a choice — it is generating another node. A Te user answers and stops. A Si user answers and anchors: 7 worked last time.
Sample two is the parenthetical. Look at what is inside the brackets: not a clarification of the sentence, but a second, more honest sentence running underneath it. Ne users nest because two things became true at once and the format only allows one. The tell is that the aside is often the load-bearing part of the message.
Sample three is the self-interrupt. This is the highest-confidence marker of the three, because it shows the search updating live. A branch was walked, a better one arrived, and rather than delete and retype, the writer left the wreckage visible and kept going. Then it closed with a connection to an unrelated event from March — the fan-out reaching sideways into old material, which is Ne doing the thing it is actually good at.
This matters practically because it is detectable. Structure survives topic: you can strip every noun out of those messages and the shape still says Ne. It is also close to impossible to self-report — almost nobody experiences their own tangents as tangents, because from the inside each branch arrived as the obviously relevant next thing. If you want to check yourself, look at conversations you already sent rather than asking yourself whether you are open-minded.
Ne in all four slots
Eight of the sixteen types carry Ne, and position changes it more than presence does. The stack logic explains how the order is built; this is what the Ne slot specifically does in each position.
Dominant
The search runs out loud, before filtering. Talks over you because branches decay in seconds. Argues a position into existence and only then discovers whether it is held.
Auxiliary
Same six branches, generated behind the judging function. You see one survivor with a lot of structure already inside it — and mistake the filter output for a smaller generator.
Tertiary
Sits under Si, so it activates at the edge of the known process. Genuinely inventive, on a leash: every branch serves a plan that already exists rather than proposing a new one.
Inferior
Not an absence of imagination — a generator pointed entirely at downside. Six vivid futures, all bad, all unweighted, experienced as realism rather than as intuition.
Dominant — ENTP (Ne-Ti-Fe-Si) and ENFP (Ne-Fi-Te-Si)
The search runs out loud and in public, before filtering. Dominant Ne users talk over you not because they are rude but because a branch has arrived and branches decay — wait ninety seconds and it is gone. The most reliable behavioural tell is that they will argue a point into existence and only then find out whether they hold it. ENTPs do this with claims; ENFPs do it with enthusiasms, which is why an ENFP can be genuinely, wholly committed to a plan on Tuesday and unable to locate the feeling on Thursday. Neither is insincere. The commitment was real; it was just attached to a node, and the search moved.
Auxiliary — INTP (Ti-Ne-Si-Fe) and INFP (Fi-Ne-Si-Te)
In second position the branching happens behind the judging function, so what reaches the room is pre-pruned. An INTP generates the same six branches as an ENTP and says one of them, usually the one that survived contact with Ti. This is why auxiliary-Ne types read as quiet and then produce a sentence with an unreasonable amount of structure already inside it. It is also why they are so often mistaken for having fewer ideas. They have the same number. You are seeing the output of the filter, not the generator — and the unspoken five are still consuming attention, which is why an apparently idle INTP is not idle.
Tertiary — ESTJ (Te-Si-Ne-Fi) and ESFJ (Fe-Si-Ne-Ti)
Third slot Ne sits directly under Si, and the pairing is the whole story. These types run on precedent first: the thing that worked, the way we do it here. Ne shows up as a late, bounded pass — the ESTJ who has run the standard process, hit the edge of it, and then generates three genuinely inventive workarounds in ten minutes, all of them concerned with getting the existing thing to finish. Tertiary Ne is real creative capacity with a leash: it activates in service of a plan that already exists, rather than proposing plans that do not. When an ESFJ reorganises an event on the fly after the venue falls through, that is tertiary Ne, and it is often better than the dominant version precisely because every branch is already load-tested.
Inferior — ISTJ (Si-Te-Fi-Ne) and ISFJ (Si-Fe-Ti-Ne)
Last position, opposite end of the Si-Ne axis, and here is the thing the guides get wrong: inferior Ne does not show up as poor creativity. It shows up as catastrophising, because the generator is running perfectly well and is pointed exclusively at ways this could go wrong. An ISTJ at 2am does not fail to imagine possibilities. They produce six of them at speed, all downside, each more vivid than the last, and they experience the whole cascade as realism rather than as intuition.
The structural giveaway is that the branches are unweighted. Dominant Ne generates six futures and knows they are six guesses; inferior Ne generates six and treats each as a live threat requiring a response, because there is no practice at holding possibilities loosely. An ISFJ who has decided a friend’s short reply means the friendship is ending has run exactly the same operation as an ENFP inventing a business over lunch. Same function, no calibration, and pointed at threat instead of opportunity.
Ne vs Se: the room test
Both are extraverted perceiving functions, both look high-energy, and both make their user conspicuously alert. They are opposites in what they take as input. Se registers what is physically present, in detail, right now. Ne registers what the present thing could connect to. Ask two people to describe a room they have just walked into and the answers separate immediately.
“Long room, maybe twenty by twelve. Two windows on the left, the near one painted shut. Radiator under it, ticking. Someone's coat on the third chair and it is still wet. The table is off level — there is a folded envelope under the leg.”
“This was two rooms, look where the cornice stops. Whoever knocked it through did not care about the fireplace, so probably a rental by the eighties. It smells like a school library, which is either the varnish or the radiator. And nobody puts the projector on that wall unless the meeting is not the point of the meeting.”
The Se answer survives a photograph. Take a picture, check it line by line, and every item is in the frame — the wet coat, the ticking radiator, the folded envelope under the table leg. It is dense, it is accurate, and it is entirely about now. Se also notices change: move one chair between visits and an ESTP will clock it without trying.
The Ne answer barely survives the photograph at all. Half of it is inference — the room used to be two rooms, someone did not care about the fireplace, the projector is on the wrong wall for a reason. Ne users routinely cannot tell you the colour of a room they spent an hour in, because the wall was never the object; the wall was a launch point. The clean decision rule: if the description would be settled by a photo, it is Se; if it would be settled by an interview with the landlord, it is Ne.
The energy reads differently once you know what to look for, too. Se energy is responsive and bodily — it moves toward the thing, picks it up, does something with it. Ne energy is verbal fan-out; it talks fast because branches expire. An ESFP and an ENFP can look identical at a party for an hour. Ask each what happened and the ESFP tells you what people did while the ENFP tells you what people are probably like.
The cost: breadth-first search with no terminator
Here is the mechanism behind the single most common Ne complaint, and it is not laziness. Breadth-first search has no natural stopping point. Depth-first search terminates when it hits the bottom of a branch; breadth-first only terminates when the frontier is exhausted, and in an open-ended possibility space the frontier never is. Every node you expand adds more nodes than you consumed. The queue grows forever.
Which produces a very specific pattern: the beginning of a project is the richest part of it, because that is where branch density is highest. Everything is still undecided, so everything is still generating. Then you make choices, the tree gets pruned, and by the middle of the work there is exactly one path with no branching left on it. To a Ne user the middle of a project does not feel hard. It feels like the function has been switched off — and the cure for that feeling is always visible, because there is a brand-new project one thought away with full branch density.
That is how you end up with seventeen beginnings and no endings. It is also why the standard advice — focus, pick one thing, stop having so many ideas — fails completely. You cannot ask a perceiving function to stop perceiving. What works is importing a terminator from outside the stack, because Ne will never generate one on its own.
Four mistypes Ne causes
Ne is over-claimed on questionnaires, because every item about being curious, open-minded or imaginative reads as obviously true to nearly everyone. These are the four confusions worth knowing, each with the tell that actually separates them.
ENTP mistyped as ENFP (and the reverse)
Identical generation, different survivors. The probe is a branch that is both funny and quietly unkind about someone you both know: Ti-Ne follows it for the structure and is surprised later that it landed badly, Fi-Ne stops mid-sentence. Second check: ENTPs argue positions they do not hold to see if the position stands. ENFPs will do it briefly and then cannot, because arguing for something they find wrong costs them more than winning is worth.
ENFP mistyped as ESFP
Both are warm, fast and physically present in a room, and both are Fi-paired — ENFP is Ne-Fi-Te-Si, ESFP is Se-Fi-Te-Ni. Use the room test. Ask what happened at the party: the ESFP tells you what people did, in order, in detail; the ENFP tells you what people are probably like and which two of them should meet. Se recalls the frame, Ne recalls the inference.
INTP mistyped as ENTP
The letter is not about how much you talk, it is about whether the branching happens before or after the filter. Look at first drafts. An ENTP sends the raw fan-out and edits in public across four messages. An INTP sends one message that already contains the objection to itself, having deleted the other five branches unsent. If you routinely think of the good version an hour after the conversation, Ne is your auxiliary.
INFP mistyped as INFJ
The highest-value one, and it is Ne versus Ni. Count the endings. INFP (Fi-Ne-Si-Te) produces five plausible futures and cannot rank them; INFJ (Ni-Fe-Ti-Se) produces one and cannot explain where it came from. Ask about a decision six months out: branching uncertainty is Ne, a single unargued conviction that turns out to have been held for weeks is Ni.
Checking this in your own messages
Self-report is weak here for a structural reason: Ne is a perceiving function, so its outputs arrive feeling like observations rather than like generation. Nobody experiences their own tangent as a tangent. The evidence that discriminates is already written down — branch density per message, whether your parentheses clarify or fork, whether your descriptions are settled by photographs or by interviews.
Our typing study is worth reading before you trust any single result, including ours. Across 1,179 de-duplicated chat-based readings, 42.3% came back INTP and 23.3% INFP — which means roughly two-thirds of the sample were auxiliary-Ne types, with INFJ at 1.2%. That is a fact about who volunteers their chat logs to a typing engine, not about any population, and it is a good reason to be suspicious of a result that agrees with the crowd. Two other numbers from it are directly useful here: MBTI and Big Five agreed on introversion 80.2% of the time, so the E or I in front of your Ne is the letter you can trust most — and it is precisely the letter that decides whether Ne leads or supports. And INTP readings mapped to Enneagram 5 in 56.1% of cases, which is worth holding lightly: 5 is a motive, Ne is a channel, and the overlap is real without being identity.