Cul-de-Sac vs. Inspiration Highway: Why It Pays to Be "Stupid" in Market Research

My last post received an interesting question that highlighted a technological aspect of the Reading Machine: the semantic-meaning space. This is also called the vector space, and we have found it to be extremely useful. The question points toward the inherent "stupidity" of vector technology. Indeed, it does not know that two very differently phrased comments can, in some abstracted way, be about the exact same thing.

While vector space approaches can be seen as a dead end, this view assumes a social theory Bakamo does not subscribe to. For the Reading Machine, it is precisely this vector limitation that provides the mathematical backbone of our honest qualitative proposition: it doesn't hallucinate meaning or force narratives; it just maps distance, forcing the human researcher to do the actual interpreting.

So instead of tech, let's talk about the social theory that leads us to shout about an epistemological revolution while others disappointedly pivot away.

Two ideas of what qual is for

At its core, this question is about what you construe qualitative research to be. This goes hand in hand with how you, your clients, and your methodologies are anchored, and how you select and process information. Broadly speaking, you have two options:

• Option A: Qual should scientifically respond to the questions the client wants answered.
• Option B: Qual should illuminate how people see the world around the client's interest areas.

Fundamentally, this is a replay of the age-old positivism vs. constructivism distinction. Is there one proper, scientifically sound reality, or are individuals allowed to live in their own bounded realities that occasionally overlap with others via culture?

I am not going to beat around the bush: Option A, if pushed to the extreme, can be blamed for most of the bad things in our current commercial world. Bad ads, disappointing products, hollow brand claims, disinformation, and zero trust in institutions. This is because Option A successfully sold an over-simplified version of reality to decision-makers, and decisions made on the back of over-simplified insights do not fall far from the tree.

That said, often (maybe even in the majority of cases) this is good enough, and exactly what clients need. With the right caveats, Option A has legitimate use cases. And while qual is reduced to a performance, what used to be the catering behind the two-way mirror of yesteryear is now the prowess of 100 AI agents concurrently video-interviewing people from around the world.

Why we pour the highway out of vectors

The Reading Machine, however, only supports Option B. This is also why vector technology is the concrete from which our Inspiration Highway is poured. It offers a mathematical way to topographize the entire discourse without any black box. Pure math, real science.

Our job is not to frame data into preexisting (and typically oversimplified or unsuitable) boxes. No, our qualitative job in social intelligence is to trace, observe, and interpret the structures that people themselves have built. We see ourselves as humble idiots who, once saturated with understanding, turn into jesters, delivering a readout of the unfiltered reality out there to the client.

To sum up: yes, vectors are a dead end if you follow a positivistic framing of qualitative research. Bakamo is shamelessly on the constructivist side of the debate. For us, this technology is a heaven-sent gift that fundamentally changes how we go about qual work.

It's an autobahn. You can go crazy fast, but you still need to know how to drive. The Reading Machine does not make a weak researcher into a superstar. It does not and cannot replace human analysis; it just gives us a way to do it at social media scale.

This piece first appeared on Daniel's Substack.