Separate to see. Combine to make.

Key takeaways

• Thinking has two motions. Critical thinking separates: it measures, compares, and shows where things differ. Creative thinking combines: it takes parts that were never together and makes a new whole. The best minds do both in one movement.
• Mathematics can be seen. A line that climbs, a curve that bends, a cloud that splits. Stevens, Tversky and Moskowitz each drew a picture that changed what people could see. We work in their tradition.
• Contradictions are the material. When a headline scores +25 with one kind of reader and −2 with another, the average hides it. A picture of the separation shows it, and shows what to combine.
• Machines build. People judge. Ideas are cheap now. Knowing which one a person will choose is not. MindSurf puts real human judgment inside the creative loop, at the speed the loop now runs.

Why it matters

It is now cheap to make things. A team can draw fifty packs before lunch, write a hundred headlines, and render a logo in every style that has ever existed. The bottleneck has moved. It is no longer the making. It is the choosing.

Choosing is a human act, and it is hard, because the people who will choose are not in the room. A team guesses on their behalf. The guess feels like judgment. Usually it is taste, and taste is one person's position in a crowd that has many.

We built MindSurf to put the crowd back in the room, in a day, without slowing the makers down. The method is old and the speed is new. This document is about the method: what it lets a mind see, and why seeing is the start of making.

Two motions, one mind

Ask a hundred people what creativity is and you will hear about combination. A new idea is two old ideas that had never met. Ask them what critical thinking is and you will hear about separation. Breaking a claim into parts, testing each, keeping what holds.

They are taught as opposites. One is loose and the other is strict. One is for the studio and the other is for the lab.

We think that is wrong. You cannot combine what you cannot see as separate. A cook who cannot taste salt apart from acid cannot balance a dish. A designer who cannot tell what the photo does apart from what the headline does cannot fix a pack that fails. Separation is what makes combination possible. Combination is what separation is for.

Take it apart. See the pieces and the gaps between them. Put the pieces together in a way they have never been.

That sequence is the whole of what we do. We believe it is the highest form of human intelligence, because it is the one machines are worst at. A machine can produce a thousand combinations. It cannot tell which one a person standing in front of a shelf will reach for. Only a person can tell you that, and only if you ask the right way.

A line that climbs

In 1957 Stanley Smith Stevens, who ran the Psycho-Acoustic Laboratory at Harvard, published a short paper with a long reach. He asked people to put numbers on sensations. How bright is this light? How loud is this tone? Twice as loud as that one? He plotted the answers against the physical stimulus, on paper where both axes were logarithmic.

The points fell on straight lines.

A straight line on log paper means a power law: sensation grows as the stimulus raised to a fixed exponent. The exponent is the slope of the line, and the slope is the sense. For brightness it is about 0.33, so a light has to grow eightfold to feel twice as bright. For apparent length it is 1: a stick twice as long looks twice as long. For electric shock it is about 3.5, so a small increase in current feels like a large increase in pain.

Stevens' picture did two things. It showed that human judgment, which everyone agreed was soft, obeys a rule that can be drawn. And it showed that the rule is different for each sense, so one slope can be set beside another. Direction and separation, on one sheet of paper.

A decade earlier he had sorted every kind of number into four scales, nominal, ordinal, interval and ratio, and asked which arithmetic each one allows. That question sits under every score we report. A +10 is a distance, not a rank. It can be added.

A curve that bends

Amos Tversky drew a different picture. With Daniel Kahneman, in 1979, he plotted how people value gains and losses. The curve passes through a reference point, the place you are standing now. To the right, for gains, it rises and flattens. To the left, for losses, it drops and keeps dropping, roughly twice as steep.

That bend explains a great deal of behaviour with one shape. People refuse fair bets. They hold losing stocks. They rate the same beef higher when it is "75% lean" than when it is "25% fat". The same fact, framed as a loss, moves people about twice as hard as the same fact framed as a gain. Tversky and Kahneman showed this in 1981 with a question about an epidemic, and the finding has held up in replication after replication.

Tversky's lifelong subject was the gap between what people do and what a tidy model says they should do, and he was precise about the gap. Earlier, with Krantz, Luce and Suppes, he helped write down the conditions under which a judgment can be split into parts that add up. That is the mathematics under a MindSurf study. A person's response to a whole sign can be read as a sum of responses to its parts, and the parts can be measured one at a time, even though nobody ever saw them one at a time.

In 1977 he showed that similarity itself is lopsided. People agree that North Korea is like China more readily than they agree that China is like North Korea. Similarity is not a distance in space. It is a comparison of features, and it depends on which thing you hold still. Whenever we put two groups side by side, we stand on that work. What looks close from one side can look far from the other.

A cloud that splits

Howard Moskowitz took his doctorate under Stevens, then spent a career applying Stevens' methods to food. In the 1980s Campbell's asked him to find the best spaghetti sauce for Prego. He made dozens of variations, had people taste them, and did not look for the best one. He looked at the spread.

The scores did not pile up around one point. They split into clusters. One group of people wanted it plain. One wanted it spicy. One wanted it with big pieces of vegetable, and no sauce like that was on the shelf. The average sauce was nobody's favourite. There was no perfect sauce. There were perfect sauces.

That picture, a cloud of people splitting into a few clouds, is the one we see most often. In a study of a sign for an online farmers market, the average said: lead with an offer. The spread said something more useful. There were three kinds of reader. One hundred and twenty-one read the headline. One hundred and five read the small print beneath it. Eighty-six read the call to action at the bottom. Each group was nearly blind to the parts the others read. The same headline scored +25 with the first group and −2 with the second.

Moskowitz called these mind-sets. We do too. They are not demographics. A mind-set is a way of looking, and it turns up in any age and either gender.

How mathematics moves

Three pictures, three motions. A line that climbs gives direction: more of this gives more of that, at a rate you can read. A curve that bends gives asymmetry: the same distance is not the same distance both ways. A cloud that splits gives separation: the crowd is not one thing.

A MindSurf study makes all three at once, from one set of ratings.

Each person sees whole versions of the work, one after another, for a few seconds each. A sign with this headline, that photo, that line at the bottom. Then another. Nobody is asked what they think of a headline. They are asked whether they would stop, buy or click, and they answer with the whole thing in front of them, the way they will in the world.

The arithmetic runs the other way. Because the parts were combined by design, a regression can pull them back apart. Every element gets a number: what it adds to, or takes from, the chance of a yes. The baseline is what the work earns with nothing in particular on it. The score of any version is the baseline plus its parts. Fifty, plus ten for the discount headline, plus four for the line beneath it, plus six for the produce photo, plus two for the call to action. Seventy-two in a hundred would stop.

Then the same arithmetic runs again for every group: women, men, each age, each market, each mind-set. Now the numbers have direction, distance, and separation. The picture is a field of columns, some up, some down, laid out by element and by group. It is the map a team needs. It is also, in our experience, the first time most teams have seen their own work as parts.

Contradictions are the material

The map is most useful where it disagrees with itself.

A protein cookie's pack scored 65 in Greece and 90 in Saudi Arabia. The same pack. A headline for the farmers market, "Organic groceries, fair prices", added seven points with people between 35 and 54 and took ten away from people between 55 and 64. "Guess the price of our box to win" moved men thirteen points and women two. The call to action "Seasonal eating made easy" was worth +11 to one kind of reader and −10 to another.

A team that only sees the average sees none of this. The averages are +2, +5 and 0. Flat. The averages say the sign does not matter much. The separated numbers say the sign matters a great deal, in opposite directions, to people who will stand in front of it on the same afternoon.

Two true things that contradict each other, side by side, with the distance between them measured.

This is the juxtaposition we exist to show. A contradiction is not a problem with the data. It is the most informative thing the data can say, because a contradiction has a shape, and a shape can be designed around.

What to do with a contradiction

Here the second motion starts. Seeing that the crowd is split is critical thinking. Deciding what to make for a split crowd is creative thinking, and it has only a few honest answers.

Choose. Make the thing for the group that matters most and accept the cost with the others. The pack for Saudi Arabia. The sign for the headline readers.

Carry. Build one thing that gives each group the part it reads. The farmers market sign built for everyone scored 86 with the headline readers, 80 with the small-print readers and 62 with the call-to-action readers, because it carried a strong headline for the first, a strong line of small print for the second, and a call to action the third could accept. Three tailored signs would have added one point, six points and five points. Not worth three signs.

Combine. Take the part that wins in one place and the part that wins in another and put them on one object. The cookie pack that went to print was one concept with another concept's dark lettering. Nobody had drawn it. The scores said it would work, and it added ten points in every market.

None of these are found by averaging. All of them are found by looking at the separation and asking what could bridge it. That question is creative in the plainest sense: it produces something that was not there. And it can only be answered because the pieces were measured apart.

What we are building

MindSurf is a research company. We test creative work with real people, in the market where it will live, and we return the numbers in a day. A study is designed around the work, so the parts the team can change are the parts that get measured.

We are building it into something larger: a layer of human judgment that sits inside the creative loop and runs as fast as the loop. Every pack, headline, logo, price, name and claim that a machine can now produce in minutes can be put in front of the people it is for, taken apart, scored, and put back together before the next version is drawn. Not a focus group once a quarter. A measurement whenever there is a decision.

The bet is simple. The cost of making is going to zero. The value of knowing what to make is not. Whoever holds the fastest, truest picture of what people will choose holds the most valuable position in every industry that makes things for people, which is every industry.

We intend to hold it. The method is Stevens' numbers, Tversky's asymmetries and Moskowitz's clouds, run at the speed of the machines that now draw the work.

Human intelligence at the speed of AI.