# Eugene Wei And Kevin Kwok: Graphs, TikTok, And Emergent Creativity

Status: deepened imported milieu note

## Source

- Date imported: 2026-06-19
- Original intake date: 2026-02-06
- Source published date: 2021-05-21
- Source type: `video`
- Source title:
  `Tiktok, Emergent Creativity, The Limits of Social Graphs, and whatever else Eugene talked about`
- Source URL: <https://www.youtube.com/watch?v=xbnDay35L8I>
- Source show / channel / publication: Closing the Loop
- People: Eugene Wei, Kevin Kwok
- Imported source note:
  [2026-02-06-intake-018-eugene-wei-kevin-kwok-graphs.md](../imported/theme-theory-2026-02/2026-02-06-intake-018-eugene-wei-kevin-kwok-graphs.md)

## Neutral Summary

The old intake frames this as a discussion about TikTok, recommendation
systems, social graph limits, product architecture, creator incentives, and
emergent creativity. The important point is that platform outcomes cannot be
explained by "just machine learning" or "just the social graph."

## Theme Theory Relation

This is a strong source for the project's future theme-graph lane. Theme Theory
cares about what audience members are connected by, not only who they already
know or what the algorithm predicts they will watch.

The useful question:

```text
what graph is implied when an audience coheres around a shared desired state?
```

Recommendation quality may surface interest, but it does not replace the need
to understand the underlying object around which interest coheres.

## Deep Corpus Comparison

This source is one of the most important imported items for future relation
work.

The corpus has a recurring intuition that an audience may be organized by a
theme-like object, not only by demographics, social relationships, platform
categories, or explicit topics. Eugene Wei and Kevin Kwok's graph discussion
helps locate that intuition inside a broader platform question:

```text
what kind of graph explains why people gather, create, watch, imitate, remix,
and return?
```

Social graphs explain some behavior: who knows whom, who follows whom, whose
signals carry status. Interest graphs explain some behavior: what people click,
watch, search, or linger on. Theme Theory may need a different or more specific
layer:

```text
theme graph: people connected by shared relation to an audience-side desired
state or object of interest.
```

This does not mean building an actual graph database now. It means the concept
could help agents reason about relations:

- audience member -> desired state;
- creator value -> object;
- post -> theme satisfaction;
- product/tool -> support surface;
- external source -> concept edge;
- audience segment -> different relation to same object.

This source also connects to emergent creativity. Platforms shape what gets
made by changing feedback, discovery, and imitation loops. Theme Theory can
ask how those loops change when the creative is organized around an object
rather than platform-only incentives.

## Core Links

- [Object Of Interest](../../core/object-of-interest.md)
- [Linking And Relations](../../project/linking-and-relations.md)

## Candidate Concepts / Edges

- social graph limits -> audience may cohere by object, not relationship
- interest graph -> revealed attention pattern
- theme graph -> possible TT relation model
- emergent creativity -> platform architecture shapes what gets made

## Promotion Judgment

- Promote to core? `yes, later`
- Reason: likely important for future relation/graph work, but premature for
  the first core docs.