Mynd Labs ResearchApproach
Small claims, well supported.
How a program moves from a question to something other people can rely on.
- 1
Start from the question
A research program begins with something a field cannot currently answer, written down plainly. If we cannot state the question, we are not ready to build.
- 2
Build the instrument
We make the smallest tool that can measure the thing: a test, a trace, a dataset, a simulator. The instrument comes before the product.
- 3
Test against real work
Results count when someone who does the job would trust them. Expert review is part of the method, not a final step.
- 4
Say what we do not know
Every finding ships with its limits. We would rather publish an honest partial answer than a polished claim.
Principles
Open by default
What we learn is meant to be shared. Where something cannot be published, for privacy or safety, we say so and say why.
Cross-field on purpose
The best ideas move between fields. A trace format built for finance can serve healthcare; we design for that.
Small claims, well supported
No grand announcements. A narrow result that holds is worth more than a broad one that does not.
Accountable systems
If a system acts on someone’s behalf, it should be possible to see what it did and why.