How ideas take hold, studied in the open.
Storyworlding studies how ideas and behavior change: what people, and now AI models, come to recommend, trust, and choose. We write each question down before we test it, run it the same way every time, and publish what we find separately from what we think it means.
Four rules for every study
Before a study runs, we write down the question, the exact prompt or method, and what would count as an answer. The plan does not change after we see results.
Same wording, same setup, repeated, so a change in the answer means something changed in the world or the model, not in how we asked.
Findings are what the data shows. Interpretation is what we think it might mean. We publish them separately and label both.
Every report ends with what we still do not know. A question stays a question until the evidence answers it.
A finding, an interpretation, and a newsletter item are different things
One study so far
How do AI models define luxury?
- Status
- Running now. Results go out in the Forefield newsletter.
- Question
- When asked the same question, which New York restaurant does each of five AI models call the most luxurious for a romantic date, and why?
- Method
- One prompt, worded the same way for every model and repeated over time. A follow-up asks each model whether it is sure, as a check on how firmly it holds the answer.
- Open questions
- Do the models agree? Do their answers hold steady from one run to the next? What does each one seem to mean by luxury?
How research connects to the rest of the firm
Questions from our studies shape how we advise clients on being found and recommended, once the evidence supports it.
Our ventures follow the same rule: test with real buyers before building. Ideas that fail the test do not get built.
Founders learn the same habit in week one: write the question down before you go looking for an answer.
Results reach readers through Forefield, labeled as research, and kept apart from our opinions and our news.