Data, AI, and Tokens: Ocean Protocol

Trent McConaghy
@trentmc0
Data, AI and Tokens:
Ocean Protocol

1000x more data
The Unreasonable Effectiveness of Data

Deep Learning: Neural Networks * Moore’s Law
≈1950s algorithms on 1000x+ more storage & compute
https://en.wikipedia.org/wiki/File:HTM_Hierarchy_example.png

Deep learning
models with >> capacity
Error 5% .. 0.01%Models with
limited capacity
Error 25% .. 5%
Another 1000x more data
Deep Learning Loves Data

Have lotsa data
(1000 enterprises)
Have lotsa AI expertise
(1000 AI startups)

Have lotsa data:
1000+ enterprises
Have lotsa AI:
1000+ AI startups
Have both:
Goog, FB, a handful
$$

Mo’ data
(and mo’ compute)
Mo’ accuracy
Mo’ $

Some tech has particularly powerful opinions
1. AI – goes to heart of “what is humanity”

Some tech has particularly powerful opinions
2. Crypto – defense is easier than offense (new)

Opinionated Tech
• “Designed for ethics”  Thoughtful approach for ethics
•“Ignored the ethics”  Not opinion-less. Simply bad design.

What Can You Do, As a Technologist?
• Determine your values, as best as you can. Set your moral
compass. “Know thyself.”
• Use this to guide your work. Infuse your work with your values.
• With each new collaborative project -- especially large-scale
ones -- agree on values up-front. Example:

Schelling point:
A common solution in absence of communication
Q: Tomorrow you have to meet a stranger
in NYC. Where and when do you meet?
A: noon, at Grand Central Station

Silo Mo’ Data
(and mo’ compute)
Mo’ accuracy
Mo’ $
This is the current Schelling point

Silo Pool Mo’ Data
(and mo’ compute)
Mo’ accuracy
Mo’ $
Change the Schelling point?

Have lotsa data
(1000 enterprises)
Have lotsa AI
(1000 AI startups)

Connect 1000+ enterprises
With 1000+ AI startups
All get lotsa data & AI

Connect 1000+ enterprises
With 1000+ AI startups
All get lotsa data & AI
How?
Data marketplace

Data marketplace
Have AI
(Want data)
Have data Have AI
(Want data)
Have AI
(Want data)
Have data
Have data Have AI
(Want data)
Have data Have data Have AI
(Want data)

https://www.flickr.com/photos/atelier_flir/1118912668
A single data storefront?

A single data storefront
= another centralized monolith
(oops!)
https://www.flickr.com/photos/walmartcorporate/5249869475

We want data marketplaces to flourish.
Cathedral Bazaar
How?

We want this. A substrate for a thousand marketplaces
to bloom. “SABRE for data”
Data
marketplace
DM DM DM
DM DM DM DM
DM DM DM DM
DM DM DM DM
DM
DM
DM
DM
Have AI
(Want data)
Have data Have AI
(Want data)
Have AI
(Want data)
Have data
Have data Have AI
(Want data)
Have data Have data Have AI
(Want data)

Complement data marketplaces with a data commons

The MVP
(Minimum Viable Protocol)

Structured Approach to Token Design
Steps:
1. Mission, Values, Governance.
2. Use Cases. Pricing.
3. Stakeholders.
4. Goals
5. Checklist (from goals)
6. Candidate designs.
7. Iterate to final

Step 1a. Mission
Ocean aims to unlock data, for more equitable outcomes for
owners and consumers of data, using a thoughtful
application of both technology and governance.

Step 1b. Values
• Unlock data— make data ever more accessible to a broader range of
organizations.
• Human right to personal data privacy & consent
• Spread of power— democratic ideals, transparent governance. Strive to
decouple capital from governance.
• Spread of value— avoid becoming new oligarchs, better to spread wealth
creation among the community. Rewards based on actions consistent with
values. Discourage speculation, encourage building & community.
As well as…
• Identity & reputation over anonymity.
• Sustainability— can last the ages
• Resilience— not contingent on any one party.
• Work with the law, not against the law

Step 1c. Governance
1. Governance of network. How protocol gets updated
2. Governance of Ocean Foundation. Growing the ocean
community

Step 2. Use Cases
Target verticals
1. Autonomous vehicles. Drives: Reconcile data commons &
proprietary data.
2. Medical data. Drives: solve privacy, data regulations
3. (And another 20 after that)

Step 3. Pricing
1. Free Data. For the data commons.
2. Non-Free Fungible Data. Easy: an exchange
3. Non-Free Non-Fungible Data. Reconcile via a mix:
• Fixed price
• Auctions
• Royalties
• Programmable

Step 4. Stakeholders
Data providers Data consumers
Keepers
Developers
“I can sell my data with
security, control, privacy
and flexible pricing.”
“I can discover and
buy data.”
”I believe in the sustainability of the network and will
be rewarded"
“I want to build on
Ocean Protocol”
Marketplaces
”I have a new channel to sell data around the world.”

Step 5. Goals
Token design = designing incentives for stakeholders to grow the value of the
network.
The main goal of Ocean network is a large supply of data.
• For non-free data: incentive for supplying more? Referring?
• For non-free data: good spam prevention?
• For free data: incentive for supplying more? Referring?
• For free data: good spam prevention?
• Does token give higher marginal value to users of the network, vs external
investors?
• Are people incentivized to run keepers?
• Is it simple? Is onboarding low-friction? Use incentives/crypto over legal
recourse?

Step 6. Write checklist
7. Propose candidate designs, compare on checklist
8. Iterate

Software stack
Ocean
Ecosystem
Ocean
network Ocean core software:
Registry of users, data, prices
Open-source
reference
data market
Data
marketplace
1
Data providersData consumers, e.g. AI folks
Data
marketplace
2
…
Data
marketplace
3
DM DM DM
…
DM
DM DM DM
…
DM
… … … …
Storage&
Compute

Ocean
Ecosystem
Ocean
network
Inside Ocean Core Software
Ocean core software
Data marketplace DM …
API. TX model is COALA IP for data. “ORC 20”
Data curation: tokenized curation market
Data access registry: tokenized read permissions
User registry: tokenized registry
Data pricing: registry of bids & asks, etc
DM
Storage&
Compute
Database, tokens: BigchainDB+IPDB
Blob storage: on-prem, FileCoin, ..
Biz logic
Processing: on-prem, ..

Data access: tokenized read permissions
Via role based access control (RBAC)
{
“asset”: {
“ns”: “ipdb.testapp.vehicle”,
“link”: “<hash>”
},
“id”: “<hash>”
}
Image Source: Pixabay. CC0 license

class TokenCuratedRegistry
def propose(data)
def challenge(proposal)
def vote(challenge)
User registry: tokenized registry
A whitelist of good actors, incentivized to grow the list
Propose
new user 300
tokens
(stake)
Challenge to
proposal
User status
Alice “Challenged”
Bob “Proposed”
Mallory “OK”
Trent “OK”
30
tokens
(stake)

Squeeze out the centralized personal data silos

>100x more data for health
care research

Tokenize the enterprise, for benefit of community.
Step 0 is to tokenize the data.

Self-driving cars: fewer accidents, more mobility

Self-driving, self-owning cars

Fleets of self-owning cars = car swarms

Nature 1.0: soil, trees, wind, flocks of birds
Nature 2.0: internet, metal, AI, swarms of cars
Planetary infrastructure for us all

MOOD VISUALS
Data is siloed with no incentive to share.
Let’s equalize the opportunity for data access,
while retaining privacy
How? Opinionated AI*crypto tech.
Ocean: a protocol for data marketplaces & data commons.
Result: Equal opportunity for data, $ based on actual value
Trent McConaghy
@trentmc0

Data, AI, and Tokens: Ocean Protocol

Data, AI, and Tokens: Ocean Protocol

More Related Content

What's hot

Similar to Data, AI, and Tokens: Ocean Protocol

Recently uploaded

Data, AI, and Tokens: Ocean Protocol