Edward Lue Chee LipEdward

Field notes / Question I keep returning to

Question I keep returning to

Systems are strange loops.

The same object · seven altitudes

One output,
seven layers of cause.

Trace an output into action, evidence, and the conditions of its next use. Each layer links to work or a question; the institutional extension remains a hypothesis.

Read the full argument ↓
What is encodedInput

Tokenization, retrieved context, and instruction order shape what enters the system. Missing relationships can keep evidence out before reasoning starts.

Recall Debt →

01

The same object from different altitudes

At one level I am studying tokens and prompts. At another I am studying memory and evaluation. Higher up, I am studying institutions, capital, and the social stories that coordinate them. Each layer constrains the others.

That is why the work keeps crossing domains. A local failure often has an upstream cause and a downstream consequence that only becomes visible after changing altitude.

02

Mimesis and model culture

René Girard’s mimetic theory is useful here: people do not desire or compete in isolation; they learn what is valuable by watching one another. Benchmarks can become mimetic objects. Labs chase the same scores, users learn the same expectations, and capital rewards the convergence it helped create.

03

Mutual poiesis

I build systems that change how I think, then use that changed thinking to rebuild the systems. The relationship is productive when neither side is treated as finished.

The next question is architectural: who can inspect and correct the loop when its builders are changing inside it? Inside the loop develops that hypothesis; this seven-layer map keeps the possible causes distinguishable.