# What Is a Research Memex?

URL: https://research-memex.org/docs/introduction/what-is-a-research-memex
Description: Where the name comes from, and the way of working the book builds on it.



*The name comes from a machine that was never built.*

In 1945, [Vannevar Bush imagined the "memex"](https://www.theatlantic.com/magazine/archive/1945/07/as-we-may-think/303881/), a desk that would hold a person's books, records and letters and let them follow trails of association between them. Bush called it "an enlarged intimate supplement" to memory.

A research memex is that idea with AI models in it. Models now take on some of the work Bush gave the machine, such as finding, sorting and connecting, and some he left to the person, such as summarizing and drafting. This book asks, page by page, which work stays with you.

## Interpretive orchestration [#interpretive-orchestration]

The book's way of working is [interpretive orchestration](/docs/introduction/core-principles#1-interpretive-orchestration). You do not just prompt a model for answers. You arrange the work, delegate parts of it to models, and judge what comes back.

```text title="diagram:interpretive-orchestration"
+---------------------------+
| Human researcher          |
| design / direct / judge   |
+-------------+-------------+
              |
              v
+-------------+-------------+
| Specialized AI roles      |
| discover / analyze        |
| synthesize / critique     |
+-------------+-------------+
              |
              v
+-------------+-------------+
| Research environment      |
| sources / tools / notes   |
+-------------+-------------+
              |
              v
+-------------+-------------+
| Research output           |
+-------------+-------------+
              |
        reflection loop
              |
              +--------> Human researcher
```

<Aside>
  Copy the diagram into an AI chat, document, or text file. Its plain-text form works in any monospace environment.
</Aside>

As the orchestrator, you:

* design the workflow: break a research goal into steps that can be checked
* direct the models: give each step to the model or tool that suits it
* keep the judgment: evaluate what comes back, name how it fails, and hold the theoretical direction

The diagram ends in a loop back to you. What a model returns shows you something about the question you asked. [Core Principles](/docs/introduction/core-principles#2-the-mirror-effect) calls this the mirror effect, and it is how the book expects you to develop [research taste](/docs/introduction/core-principles#the-goal-developing-research-taste).