RituStudio · Phoenix, Arizona · The Instrument
The ArtGraph
Fingerprint
A complete reading of a painting, in a single mark. Five dimensions become a shape. The sixth becomes its color.
I. A Language, Not an Interpretation
What I am obsessed with is giving language to abstraction.
Not interpretation. Language. The difference is the whole project.
A person can stand in front of a representational painting and talk. They have words — subject, likeness, composition, period, the fall of light. The words are not always good ones and the conversation is not always a good conversation, but it exists. There is somewhere to stand.
Stand the same person in front of an abstract painting and the vocabulary collapses into two positions. I like it. I don't. Everything in between — everything that would constitute an actual reading — has no words to be said in.
So abstraction sits in a different seat from representational art. Not a lower one, exactly. Often a more reverent one, spoken about in hushed tones by people unsure what they are permitted to say. But a seat outside the conversation. Admired, or dismissed, and unknown in both cases.
I want it in the same seat. That is what the taxonomy is for.
What Hindi taught me
I was raised learning Hindi, and I avoided Sanskrit like the plague — too structured, too many rules, too much rigour for a boy who wanted to be somewhere else. I have since spent years building a taxonomy with nine dimensions, a scoring scale and a notation system, which is either a late apology or an education that took its time arriving.
What Hindi taught me, without my noticing at the time, is that a script is not its letters.
English separates its words with space. Devanagari binds them with a line — the shirorekha, ruled straight across the top, from which the letters of a word are suspended. Opposite mechanics, identical job: telling the eye where one word ends and the next begins. Remove either and you have letters in a row with no way of knowing what belongs to what.
For years the dimensions were letters. Texture, Form, Color Range, Palette, Mood, the Abstraction Quotient — and the three that read the encounter. Every one of them true. Every one useless alone. I could say Texture 5 and I had made a sound, not a word.
The fingerprint is the line. It binds them, and the moment they were bound they stopped being measurements and became a word — something with a shape, something that occupies volume, something you can recognise across a room without reading it.
The objection
It arrives quickly and it deserves a serious answer. Abstraction's power, someone will say, is precisely that it escaped the language representational painting was trapped in. It refused subject. It refused likeness. It refused the whole apparatus of decoding. Give it a grammar and you have put it back in the cage it broke out of.
I would agree, if the grammar described meaning. It does not. Not one of the nine dimensions asks what a painting is about. They ask what it does — how the surface behaves under light, how far the work has travelled from the visible world, what happens in the body of the person standing in front of it before their mind catches up. A language for what a painting does is not a decoding. It is close to the opposite: a way of talking about the encounter without ever claiming to have solved it.
II. A Reading Is Not a Recognition
Here is a complete reading of a painting called Sindoor Field. Texture 5. Abstraction +5. Form 3. Mood 5. Color Range 2. Palette: vermilion, crimson, rose madder, oxblood, near black.
Every one of those is accurate. Every one is checkable against the canvas. And together they are almost impossible to hold in your head.
Read six values for one painting and you have learned something real. Read them for ten paintings and you have learned less, because the numbers begin to interfere with each other — the fours blur into the threes, and you can no longer say which painting was which. Read them for two hundred and sixty-three and you no longer have knowledge at all. You have a spreadsheet. A spreadsheet is a thing you query. It is not a thing you know.
This is not a failure of the taxonomy. The six dimensions do exactly what they were built to do: they describe the object precisely, in terms that can be argued with. The failure is one of format. A list is built for retrieval — you look things up in it. Recognition needs something else entirely.
Consider how you recognize a face. You do not enumerate features — eye spacing 3, jaw angle 4, brow height 2 — and then compute a match. You perceive a whole, instantly, and the perception is far faster and far more reliable than any list you could assemble about the same face. The features are all there. They simply arrive as one thing rather than six.
That is what the fingerprint is for. Not to replace the reading. To make it perceivable in the way a face is perceivable — as a single object, recognized rather than parsed.
III. Five Dimensions Become a Shape
Five of the six dimensions are scalar. Mood, Texture, Form and Color Range each run from 1 to 5. The Abstraction Quotient runs from −5 to +5. Each has a floor, a ceiling, and a meaningful middle, which means each can be a distance along an axis.
Give each dimension a spoke, space the spokes evenly around a circle, plot each value as a distance from the centre, and join the points. The result is a polygon — and the polygon carries every one of those five numbers at once, in a form the eye takes in without counting.
A fingerprint reads outward from the centre. Farther is more.
A tall narrow shape and a wide flat one are different paintings before you have read a single label. A shape that reaches the outer ring on one axis and hugs the centre on another is a painting with an extreme commitment and an extreme refusal, and you can see both facts in the same glance.
This is the whole mechanical claim, and it is not a large one. Radar charts are old. What matters is not that the five numbers can be drawn — it is what happens to the sixth.
IV. The Sixth Dimension Will Not Plot
Palette is the sixth dimension of the Taxonomy, and it does not behave like the others.
Palette is not more or less of anything. Sindoor Field is not a 4 out of 5 in palette. It is vermilion, crimson, rose madder, oxblood and near black — a sequence of named pigments, not a position on a scale. There is no axis it could travel along, because there is no direction in which a palette increases.
Every earlier version of this diagram handled that the same way: five spokes, and a note underneath explaining that the sixth was an exception. The sixth reads as color, not score. A footnote apologizing for the framework's own shape.
The resolution came from noticing that the five and the sixth are not answering the same kind of question. The five say how much. Palette says what of. Those are different grammatical roles, and forcing them into the same notation was the error.
So the five make the shape. And the sixth fills it.
The polygon is divided into equal wedges — one for each pigment in the painting's palette — and each wedge is painted in that pigment. Not a representative color. Not a mood color. The actual hex values recorded in the catalog for that specific canvas.
Two paintings, as far apart in temperature as the catalog goes. Identical on Abstraction, Texture and Form. The shapes differ only in Mood and Color Range; the fills differ entirely.
This is not a compromise. It is the only arrangement in which all six dimensions are present simultaneously without one of them being demoted to a caption. The shape is the reading. The color is the painting. Nothing is left over and nothing needs explaining away.
V. What the Pigments Taught Us
Building the thing produced a finding that no amount of thinking about it would have.
The first paintings we tried it on were blue — five blues, close together, a narrow and disciplined palette. The wedges were nearly invisible. The shape read as one muddy field with faint seams. The next was a painting of black and yellow and ochre, with a lower Color Range score than the blue one, and its wedges snapped into place instantly, every edge legible.
The lesson runs against the obvious assumption. It is not the number of hues that makes a fingerprint readable. It is the value distance between them — how far apart they sit from light to dark. Five blues at similar value produce mush. Black against yellow produces architecture, even if there are only four pigments and the Color Range score is a 2.
Sindoor Field proves the point from the other direction. Its Color Range is 2, near the floor of the scale, and its fingerprint is one of the most legible in the catalog — because vermilion to near black is an enormous span of value inside a very narrow span of hue. A low Color Range does not mean low contrast. Those are different measurements, and until the wedges existed there was no reason to notice that they came apart.
One further constraint emerged, and it is absolute: the wedges must be opaque. An earlier build let the painting show faintly through them, which looked handsome and reported something false — a cerulean composited over a dark ground is no longer cerulean. If the fill claims to be the palette, it has to be the palette. Ninety percent of the truth is a different color.
VI. How to Read One
The five spokes sit in a fixed order, always: Mood at the top, then clockwise to Abstraction, Texture, Form, and Color Range. The order never changes, which is what makes two fingerprints comparable at a glance. A shape that leans right is a painting committed to abstraction and surface. A shape that leans left is committed to color and structure.
Reach
How much of the outer pentagon the shape fills. A large fingerprint is a painting that is doing a great deal on every axis at once — highly textured, highly abstract, chromatically wide, formally clear. A small one is a painting of refusals. Neither is better. A great many of the strongest paintings in this catalog are small shapes.
Spikes
A single vertex pushed to the outer ring while its neighbours stay near the centre. This is the signature of a painting organized around one extreme commitment. It is the most distinctive fingerprint a work can have, and the easiest to recognize across a room.
Symmetry
A shape close to a regular pentagon describes a balanced painting — no dimension dominating. These are frequently the hardest works to describe in words, and the fingerprint is honest about that: there is no spike to point at.
The Fill
Read last, and read as a whole. Warm or cool. Wide value span or narrow. One wedge that disagrees with the rest — a single cold pigment in a warm field — is often the most interesting fact about a painting, and it is visible here before it is visible anywhere else.
A spike at Mood and Form, a floor at Color Range, and a fill that runs black to yellow. Recognizable from across a room.
VII. What Is Still Unresolved
One decision inside this instrument has not been settled, and it is worth stating plainly rather than hiding.
A painting that scores 1 on a dimension has that vertex almost at the centre of the chart. If two adjacent dimensions both score low, the polygon pinches into a narrow wedge and stops reading as a pentagon at all. Two different paintings with two different low scores can produce nearly the same silhouette.
The remedy is a floor: a minimum radius, so the lowest score still plots just inside the innermost ring rather than at the hub. But every increment of floor compresses the bottom of the scale — a 1 and a 2 begin to look alike — and it compresses the spread across the whole catalog, so strong and weak shapes converge.
Measured across three paintings at four floor settings, raising the floor from 0.14 to 0.38 costs about 28% of the separation at the low end but lifts a collapsed shape from 28% of the frame to 45% — the difference between a wedge and a form. It also narrows the range across the catalog from 2.2× to 1.6×. That is the trade, exactly, and it is a judgement about how these marks behave at thumbnail scale rather than a calculation with a right answer.
The current setting is 0.14. It will change.
VIII. The Wall
Everything above is argument. This is the part that is not.
Below is every painting in the catalog that carries a complete reading, rendered as its fingerprint, at the size these marks are actually seen. Sort them and the taxonomy stops being a framework described in prose and becomes a thing with visible structure — clusters, outliers, a shape that recurs, a shape that appears exactly once.
No one has to be persuaded that the framework works. They can look.
What the Wall Says
I built this to give abstraction a language. Scroll back up and look at what is there — words in a language that did not exist a few years ago, and every one of them a painting that had to be made before it could be said.
I have one word for it, and it is not a sentence. Awe.
Which is where every visitor has stood, in front of every one of these paintings, since I started making them. I built the instrument to answer that. The instrument has handed it back to me, one level up.
It will take time to give voice to the words. That is not a failure of the language. It is what a new language is.
Questions
What do you mean, a "language" for abstraction? Isn't that just a metaphor?
Partly, and it is worth being precise about where the metaphor holds and where it breaks. A language needs units that carry meaning, a way of combining them, and a form a stranger can read without having been present when it was written. The nine dimensions are the units. The fingerprint is the combination. The mark is the readable form. That much is literal.
Where it falls short: there is no syntax. You cannot say something new with a fingerprint, only describe something that already exists. That makes it closer to a notation than a language — nearer to musical notation, which is not music but lets a stranger reproduce a piece they have never heard. I use the word language because notation undersells the ambition, and I would rather be argued with than be safe.
What makes you an authority on this?
Nothing does, and I think it is the wrong question to ask of this particular thing.
Nothing in the taxonomy rests on my authority. Every score is a claim about a specific canvas, and every claim can be checked by standing in front of it. If I say a painting has a Texture of 5 and its surface is smooth, I am simply wrong, and anyone can establish that without needing to accept anything about me first. A framework that required trust in its author would be a worse framework.
What I bring is narrower than authority. I have made more than two hundred and sixty paintings, which means the dimensions came out of a practice rather than a theory — I built them because I needed them, not because I was proposing them. And I grew up in a gallery in New Delhi, in a house where paintings were argued about professionally by people who did it for a living. That explains why the absence of a language bothered me enough to spend years on it. It does not make me right.
How have you tested it?
Incompletely, and the gaps are the interesting part.
It has been tested one way that works: the Abstraction Engine reads a person's description of their room and what they are reaching for, translates it through the three dimensions of the encounter, and recommends a painting. When someone describes wanting yellow as a total atmospheric condition rather than an accent, the framework finds the one painting where that is true. That happens repeatedly. It is also anecdotal — there is no controlled comparison against a keyword search, and there should be.
It has been tested a second way that is early: every painting page asks visitors what happened when they saw the work, on a five-point scale from not for me to moves me. Those responses accumulate into a reading that no one assigned. The sample sizes are still small — often single digits per painting — and I publish the confidence number precisely so nobody mistakes them for settled.
And there is one test I have not run, which is the one that matters most. If two informed people scored the same painting independently, how far apart would they land? That is the question that would establish whether these are shared perceptual categories or one painter's private habits. I do not know the answer. Until someone runs it, that is an honest hole in the middle of this.
The scores are subjective. How is that a measurement?
It is not, in the sense a physicist would mean. There is no instrument. Texture 4 is a perceptual judgement made by a person looking at a surface.
What makes it more than an opinion is that the judgement is bounded and the criteria are published. The scale runs 1 to 5, each point is described in advance, and the description refers to observable properties — whether relief is visible at a distance, whether the surface reads differently in raking light. That is the same status a wine's tannin score has, or a film's classification. Not measurement. Structured, arguable description, which is a great deal more than abstraction currently has.
Isn't this just a radar chart?
The geometry is, and radar charts are old and frequently misused. Nothing about plotting five values on five axes is an invention.
Two things are not standard. The first is the sixth dimension: Palette does not plot, because a palette is not more or less of anything, so it fills the shape instead of occupying an axis. I have not seen that done, and it is the only reason all six dimensions can be present in one mark without one becoming a footnote. The second is that a radar chart is only ever as good as its axes — five arbitrary metrics produce a decorative shape and no information. The entire weight of this rests on whether these are the right five, which is an argument about the taxonomy, not about the chart.
You sell these paintings. Isn't this a marketing device?
It is used commercially. The Abstraction Engine recommends my work, and I am the one who benefits when it recommends well. That is a real conflict and it should be stated rather than glossed.
What I would point at is what a sales tool would not do. It would not include dimensions on which a painting can score low and be described as doing less. It would not record not for me as a valid and valuable response, or publish those responses. It would not tell a collector that a work is wrong for their room. This framework does all of that, because a system that recommends everything to everyone is worth nothing to the person using it — and I would rather have an instrument that is right than one that flatters the inventory.
Would it work on a Rothko?
In principle, yes. Nothing in the nine dimensions is specific to one artist or period. A Rothko has a texture, a form, a color range, a palette, a mood and a distance travelled from representation, and a person standing in front of it has a room, an intention and a feeling.
In practice, untested at any scale. The framework was calibrated on one studio, and applying it to a canonical work would produce scores that other people would immediately dispute. I consider that a feature. An arguable reading is better than an unsayable one, and the argument is the thing that has been missing.
Do I need to understand any of this to enjoy a painting?
No. Emphatically not, and if the fingerprint ever gets between a person and a canvas it has failed at its job.
It is for before and after. Before, to help someone find the work that is waiting for them among hundreds. After, to give them something to say about what happened. During — standing in front of it, in the moment something occurs in the body before the mind catches up — you should forget every number on this page.
What happens when a fingerprint is wrong?
It gets corrected, and the correction propagates everywhere at once.
Every score lives in a single catalog file. No page stores a copy. The fingerprint on a painting page, the one on a social card, the wall on this page and the engine's recommendation all read from the same record, so fixing a value in one place fixes it in every place it has ever been shown. This has already happened — a painting recorded at Texture 4 was reassessed as a 5, and there was nothing to update but the record.
Where It Appears
Every painting page carries its own fingerprint beside the work. The reading cards carry it. The social cards for @artthatlistens carry it, composited onto the painting itself. And the Abstraction Engine reads the same five values when it listens to what someone is reaching for.
One instrument, one set of numbers, rendered wherever the catalog is. When a value is corrected in the record, every fingerprint everywhere corrects with it — because none of them store anything. They all read from the same place.