Emergent �Curation

w/ Tokens

Spencer Graham — @spengrah

Information is everywhere, but…

​

1. A lot of information (most?) remains untapped
2. Information often ends up in the hands of monopolies with outsized profits, control, and power

Spencer Graham — @spengrah

Why does this happen?

Data doesn’t come from nowhere. Somebody must intentionally decide to devote resources towards finding and organizing it.

1. Information is expensive to find and organize
2. The underlying structure of most data does not act as an incentive to find/contribute it

​

Ad sales <> a list of the most relevant websites, or

a list of the best bars

Civic duty <> a graph of detailed factual articles

​

Spencer Graham — @spengrah

Prices are different; they have a superpower

In a proper market...

• Prices are data structures
• Nobody owns prices
• Nobody controls or sets prices
• The data structure (prices) are the incentive to find/contribute information (money)
• Prices emerge from the behavior of market participants

Spencer Graham — @spengrah

But prices only hold one type of information

Intentional Curation 💩

Emergent Curation 😍

• Incentive to find/contribute data <> data structure
• Requires a central actor to fund and execute, who owns and controls the result

​

Most data lives here, e.g.:

​

Facebook

Twitter

LinkedIn

Google

Wikipedia

Yelp

Health & medical data

• Incentive to find/contribute data = data structure
• No central actor required
• Nobody owns or controls the result

​

Only a little data lives here, e.g.:

​

Value of stuff (prices)

​

Reputation (Reddit karma)

Spencer Graham — @spengrah

How could we give other types that power?

Narratives

​

Curvilinear relationships

​

Databases

​

Network graphs

Trees

​

Ranked lists (interval or ratio)

Ranked lists (ordinal)

Unranked lists (sets)

Prices

​

Karma

Intentional Curation 💩

Emergent Curation 😍

Spencer Graham — @spengrah

Put a token on it

tokenize the data itself → �curation market

Fully emergent

​

​

​

​

Continuous values

tokenize the data structure → token-curated registry

Partially emergent

​

​

​

​

Discrete set membership

tokenize future outcomes → �prediction market

Fully emergent

​

​

​

​

Probability

TCR image from Sebastian Gajek.

Bonding curve image from Slava Balasanov.

🔎

Spencer Graham — @spengrah

Token-Curated Registries

What is a token-curated registry?

• A list of items
• curated by a community
• coordinated by a dedicated token.

Spencer Graham — @spengrah

Example: good bars in Chicago

How do TCRs work?

100 bar tokens

100 bar tokens

3 types of actors:

​

• Consumers
• Community Curators
• Applicants

10 bar tokens per vote

free to read

There are three types of entities who will participate in a TCR

• Consumers - use the information in the list
• Applicants - want to be a member of the list; apply to the list
• Curators - evaluate and potentially challenge applications; vote on challenges (collect rewards)

Bar owners

apply()

challenge()

Bar critics/ enthusiasts

Adapted from Sebastian Gajek.

good bars in Chicago TCR

Bar goers

vote()

Spencer Graham — @spengrah

Registry design patterns

Unordered

Data structure: unranked list / set

Lots more

Spencer Graham — @spengrah

Registry design patterns

​

Ordered

Data structure: ordinal ranked list

Graded or Staked

Data structure: interval or ratio ranked list

From TCR Design Patterns by Matt Lockyer

From Graded Token-Curated Decisions by Sebastian Gajek

Spencer Graham — @spengrah

Registry design patterns

​

Combinatorial

Data structure: unranked list; narrative?

Layered

Data structure: ordinal ranked list

From The Layered TCR by Trent McConaghy

From TCR Design Patterns by Matt Lockyer

Spencer Graham — @spengrah

Registry design patterns

​

Nested

Data structure: network graphs; trees

From TCR Design Patterns by Matt Lockyer

Spencer Graham — @spengrah

Many potential challenges remain untested

• The token value must be increasing → the value of the list must be increasing
• Maximum value of the list isn’t high enough to sustain incentives
• Free-rider problem → there must be a short-term incentive for curators to be active
• The short-term incentive must align with the long term incentive
• Must be enough curator bandwidth to evaluate all applications
• Voting works as expected
• Herding problem → curators should be incentivized to vote the truth (not the expected consensus)
• Handling objective vs. subjective information
• Are token incentives sufficient? Need a form of reputation or “knowledge”?
• Curator UX
• How best to create a strong Schelling point?
• Bootstrapping problem → must ensure there’s minimal viable value in the list
• Vulnerability to attacks? → bribing, dark DAOs, collusion, griefing

Spencer Graham — @spengrah

If we get it right, the payoff could be big

Narratives

​

Curvilinear relationships

​

Databases

Prices

​

Network graphs: nested TCRs

​

Tree: nested TCRs

​

Ranked lists (interval or ratio): staked or graded TCRs

​

Ranked lists (ordinal): ordered TCRs, layered TCRs

​

Unranked lists (sets): unordered TCRs

​

Narratives: combinatorial TCRs?

Intentional Curation 💩

Emergent Curation 😍

Spencer Graham — @spengrah

Emergent Curation!