SAKSIN · THE WITNESS
Twelve essays on the seam between a person and what they are doing — beginning with the one that has a laboratory record.
Somebody is about to do something small and entirely familiar. Sign their name. Walk downstairs. Say a sentence they have said a thousand times.
And for the first time in years, they are present as an observer of it. Not distracted — the opposite. Attending closely, to the thing itself, from the outside.
It goes wrong. Not dramatically; it comes out stiff, or a half-beat late, or with a small hitch in it that nobody else notices and they feel from the inside like a stumble.
The strange part is that the attention was well-intentioned and the skill was not in question. What broke it was the arrival of somebody watching.
This collection is called Not-Two, and the phrase needs handling carefully.
Advaita is a technical term in a specific school of Vedāntic philosophy. Wu wei is Daoist. Anattā is Buddhist. Fanā is Sufi. Over the twentieth century these were gathered into a single English category called “non-duality”, largely by perennialist writers and later by teachers in the West, and the gathering has been so successful that the category now feels older than the thing it describes.
It is not old, and the traditions it covers are not versions of one another. The sharpest case: Advaita Vedānta holds that there is one reality and that your deepest self is identical with it. Buddhist anattā denies that there is any such self to be identical with anything. It is a claim about an enduring, unchanging essence, not a denial that there are persons or that ordinary self-reference works. These are not two accents on one insight. They are the opposite answers to the same question, and Indian philosophers argued about it for something like a thousand years.
So this collection does not claim the traditions agree. It takes an observation that several of them happen to have made about the same seam, uses each in its own language, and says at the end where they part company. That refusal is the spine of the collection, and the reason for it is in the closing essay.
One failure installs the observer and leaves it running. The other tries to abolish it.
Attention turned inward as a standing policy, because it feels like diligence. It degrades anything already learned, and the person cannot tell, because from inside it feels like care.
“Stop thinking about it.” This does not work, cannot be done on command, and adds a second process whose job is to check whether the first one has stopped. Nothing below is an attempt to stop thinking.
Sākṣin is an Advaitic technical term, and inside that system it carries metaphysical weight this collection is not borrowing.
What is borrowed is narrow: the observer and the performer are not the same function, and running both at once on a practised action costs something. That is a claim about attention, it is checkable, and the next section is where it gets checked.
The best-known version of this belongs to no tradition at all. It is a piece of Victorian light verse about a centipede who is asked which leg moves after which, falls into a ditch considering the question, and cannot walk again.
It is usually attributed to Katharine Craster and dated to 1871, though the attribution is traditional rather than certain and the rhyme circulated anonymously.
It is a better teaching text than most of the scriptural material, because it makes no metaphysical claim whatever. It just describes the mechanism and stops.
Explicit monitoring of a well-learned skill degrades it. Sian Beilock and Thomas Carr demonstrated this directly: skilled performers made worse by attention directed at the components of what they were doing, and — the detail that matters — novices in the same conditions were not harmed the same way.4
Rich Masters’ reinvestment theory gives the mechanism a name: under pressure a performer reverts to consciously controlling a process that had become automatic, and the conscious control is slower and clumsier than the thing it replaced.5
Two limits on what this supports. It concerns skills that are already automatic, which is why the novice finding matters — for something you are still learning, attention to components is how you learn it. And it says nothing about awareness in general, only about attention directed at the mechanics of an action while performing it. Extending it into a claim about consciousness would be exactly the overreach this collection is trying not to make.
The strongest version of both is worth having before the practice.
The case against
Nothing in this collection generates undividedness. Every practice here removes something that prevents it, which is the only honest thing on offer, and a reader who treats these as techniques for achieving a state will be checking whether it has arrived — which is the supervisor, wearing a different hat.
The second objection is about who this advice suits. Somebody still acquiring a skill needs exactly the explicit attention this essay warns against. The warning applies to what is already learned, and applying it earlier will slow the learning down.
There is also a version of this that becomes an excuse. “Do not analyse it” is excellent advice about execution and poor advice about review, and the difference is a matter of when rather than whether.
The reliable move is not to reduce attention but to point it outward, at something specific and external, so that the observer has a job that is not you.
Pick one external cue for the duration of the action. A sound, a point, an edge, a rhythm in the environment. Specific enough to hold and dull enough not to become interesting.
Separate execution from review completely, in time. Analysis afterwards, and as much of it as you like. None during.
And notice when the supervisor arrives rather than fighting it. Naming the arrival — two words, silently — costs less than arguing with it, and arguing is the second process.
The point is not to acquire something. It is to tell two kinds of hour apart.
Attention turned on a practised action from outside it costs something, reliably, in a way that has been measured. That is the whole load-bearing claim of this essay and it needs no metaphysics at all.
The traditions that noticed it noticed it separately, in different centuries, meaning different things by it. They are not confirming each other and the centipede is not confirming any of them.
What they share is an observation about one seam. The rest of this collection is eleven more, and the last essay is about where they stop sharing anything.
Give attention one specific external object for the duration of the action. Move all analysis to afterwards.
And when the observer arrives, name it rather than arguing with it — the argument is a second process.
Terms are given in the language of the tradition they belong to — Sanskrit for Vedāntic and Buddhist material, Chinese for Daoist, Arabic for Sufi — and are not treated as synonyms for one another, because they are not. Where a rendering is condensed it is marked as such. Modern research is cited by author and publication, with links where a stable public page exists.