►
From YouTube: Peer-To-Peer Decentralized Networks Meetup Lisbon
Description
Fluence, Ceramic, Filmine, and Livepeer are sharing updates and having a discussion on the future of web3 infrastructure.
Building on the Web3 or peer-to-peer tech stack? Interested in ideating on how to solve the tech challenges for moving from Web2 to Web3?
Let's gather to get an in-depth look into the latest research & tech insights from decentralized and open-source projects. You'll learn about the developments in web3 computing and storage, collaborative peer-to-peer stack, data composability, decentralized infrastructure, and much more.
This evening is for developers and thought leaders interested in the Web3 tech stack and the challenges it faces today.
What tech stack do developers need to build awesome Web3 applications?
What does the tech stack look like today?
What's missing for Web2 to migrate over?
Bring your questions. Join the conversation. Listen to thought leaders.
At Village Underground Lisboa, Av. da Índia 52, 1300-299 Lisboa, Portugal
A
It
went
so
guys
come
closer.
You
can
sit
on
the
chairs
So
today
we're
discussing
the
future
of
appear
to
be
a
technology
and
also
about
Webster
infrastructure.
We
have
amazing
speakers.
So
book
will
start
our
discussion.
After
this
we
will
have
a
small
break
and,
after
the
break
we'll
have
panel
discussion
also
you
can
have
you
can
ask
questions,
be
active
and,
let's
yeah,
let's
support
our
first
speaker.
B
B
Hey
everyone,
my
name
is
Luke.
Should
we
start
yeah?
My
name
is
Luke
I'm
the
founder
of
aspina
from
protocol
Labs
called
the
field
mine.
So
basically,
what
we
are
doing
is
we
are
building
some
protocols
that
allow
different
actors
to
be
coordinated
and
provide
infrastructure
to
the
web
frame
before
joining
protocol.
Apps
I
did
a
bunch
of
things.
B
So
generally,
what
I
want
to
talk
about
is
how
to
think
about
where
we
are
going
with
the
decentralization
of
the
cloud.
If
you
think
about
what
the
cloud
is
today,
it's
basically
just
a
resource
pool
where
you
have
storage
and
compute,
and
if
you
look
at
all
the
details
on
how
like
the
cloud
evolved
in
the
past
20
years,
you
basically
have
storage,
which
is
backed
by
big
network
attached
storage.
B
You
can
buy
cores,
you
can
buy
memory,
you
can
buy
iops
on
the
some
local
ssds
and
then
you
can
interact
with
the
massive
storage
through
services
like
SJ,
every
other
Cloud
product
that
you
use
today,
like
kubernetes
or
any
load
balancer
or
any
machine
learning
framework
that
is
running
on
top
of
Cloud
providers
is
basically
running
on
top
of
those
abstractions.
So
there
is
not
nothing
more
than
ec2
and
S3.
B
If
you
think
about
the
fundamental
layer,
it's
it's
really
incredible
to
see
that
clouds
are
growing
at
30
year
over
year
after
17
years
after
they
were
kind
of
invented.
Even
though
we
have
so
many
issues
with
the
clouds
like
the
fact
that
they
force
you
to
be
locked
into
a
particular
infrastructure
provider.
What
I
mean
by
that
is
that
you
don't
have
an
option
to
be
running
on
a
particular
server
that
you
wish,
when
you
use
AWS
or
any
other
cloud
provider,
you're,
basically
forced
to
use
a
cloud
provider.
B
Oh
sorry,
infrastructure
that
is
backed
by
that
cloud
provider
and
usually
that
cloud
provider
is
based
in
a
particular
jurisdiction.
So
they
have
like
very
strict
rules
on
what
they
need
to
do
with
that
there.
For
example,
if
someone
needs
to
read
that
they
are
like,
they
are
happy
to
share
that
they
help
with
whoever
the
authorities
are
in
their
respective
jurisdictions.
B
The
issue
as
well
is
that
everyone
is
building
their
own
tools,
so
if
you
think
about
AWS
Google
Azure,
like
they
all,
have
the
same
products
with
a
different
brand,
and
they
are
basically
just
building
everything
from
scratch,
without
really
sharing
anything
between
them.
Now.
The
worst
thing
is
that
all
that
infrastructure
is
then
being
used
by
other
tools
built
on
top
of
those
tools,
and
then
what
happens
is
that
we
have
services
like
Firebase,
that
really
don't
even
give
you
an
option
to
choose.
B
Now,
if
you
compare
what's
happening
in
web
2
and
what's
happening
in
Grand
Prix,
and
we
look
at
how
fast
the
web
free
ecosystem
is
growing,
we
are
seeing
Falcon
growing
18
months
to
17
exhibits
of
capacity.
And
if
we
compare
that
to
what
happened
in
the
case
of
webto,
we
can
see
that
over
the
period
of
six
years
you
have
AWS
Google
cloud
and
Azure
collectively
have
15
exabytes
of
capacity.
B
For
example,
your
computer
at
home,
and,
ultimately,
what
we're
also
all
together
getting
to
is
a
internet
that
cannot
be
broken
because
if
you
look
at
what's
happening
today
in
various
jurisdictions,
we
are
very
likely
to
have
a
few
intranets
that
are
completely
separate
between
each
other.
And
the
reason
for
that
is
that
countries
are
kind
of
trying
to.
B
That's
fathom
executables
in
the
case
of
life.
Here,
that's
encoding,
video
in
the
case
of
a
Leo,
that's
doing
zika
snarks
and
everyone
is
focused
on
one
particular
use
case.
So
what's
actually
happening
is
that
in
the
Waptrick
context,
we
are
first
solving
the
different
products
and
we
yet
have
to
solve
the
infrastructure
that
is
aligned
to
those
different
networks
to
have
a
resource
pool
on
top,
on
top
of
which
to
execute
their
workloads.
B
If
we
focus
on
their
archival
Network,
what
we
have
is,
basically,
we
have
manners
providing
storage
in
this
case
to
the
Falco
Network,
by
having
big
racks
full
of
hard
drives
full
of
data,
and
also
having
that
particular
storage
packed
by
field
collateral,
and
in
this
case
it's
super
important
for
us
that
that
infrastructure
is
secure,
because
what's
backing
that
there
is
really
collateral.
That
is
worth
a
lot
so
in
in
the
case
of
Falcon.
That
can
be
for
one
petabyte.
B
You
might
be
required
to
have
like
50
000
USD
in
collar
backing
one
petabyte
update
one
terabyte
of
there.
Now,
if
you
need
to
do
retrieval,
what
you
want
is
more
a
small
machine
that
you
can
put
in
basically
a
hotel
room
where
you
have
a
lot
of
bandwidth.
You
have
a
machine
that
is
connected
to
a
residential
connection,
because
of
which
it's
a
very
affordable
connection,
but
you're
very
limited
in
terms
of
storage.
But
you
don't
really
care
about
massive
storage.
B
What
you
care
about
is
providing
a
particular
piece
of
content
to
the
final
user,
and
if
we
look
at
Computing
what
you
actually
care
about
is
executing
a
massively
scalable
jobs
at
a
very
low
cost,
but
at
the
same
time
those
compute
providers
usually
are
not
as
reliable.
So
you
really
want
to
pay
attention
on
what
kind
of
workloads
you
are
running
on
those
compute
providers
and
especially
because
they
have
a
very
limited
connection.
B
So
you
are
not
able
to
transfer
as
much
as
you
would
be
able
to
transfer
to
a
retrieval
provider
or
their
hello
provider.
Now
this
is
just
the
beginning
of
computing.
What
we're
going
to
see
is
a
very
specialized
compute
providers.
For
example,
some
compute
providers
might
have
high
iops
on
the
ssds
and
might
have
like
very
specialized
cover
doing
a
particular
workload.
B
Now,
if
we
zoom
out-
and
we
look
at
the
different
personas
that
are
actually
participating
in
this
ecosystem,
you
have
the
storage
providers
that
care
only
about
gank,
predictable,
rewards
not
being
exposed
to
the
price
of
the
cryptocurrency,
because
what
they
care
about
is
just
running
a
business
and
those
can
be
seen
as
hosting
companies
back
in
the
2000s
that
were
just
doing
hosting
they
just
care
about
hard
drives.
They
care,
of
course,
a
bit
about
the
networking.
Maybe
they
want
to
be
on
internet
exchanges,
but
they
don't
really
care
about
betting.
B
On
a
particular
scenario,
in
the
case
of
crypto
and
what
they
also
care
about
is
being
able
to
scale.
So
if
they
have
some
demand,
they
want
to
be
able
to
keep
adding
drugs
full
of
drives
and
be
able
to
scale
their
business.
On
the
other
side,
you
have
the
compute
providers,
which
are
basically
a
new
persona.
B
B
Context
is
the
token
holder,
because
web
tree
networks
have
to
think
about
how
to
generate
demand
for
their
token,
because
those
economies
need
to
be
sustainable
because
at
the
end
of
the
day,
most
of
the
Computing
and
storage
networks
are
rewarding
the
infrastructure
providers
to
through
inflation
and
if
there
is
no
demand,
for
that
particular
token,
it's
super
hard
for
for
that
economy.
So
those
sustain
now
I'm
going
to
make
a
few
examples
on
what
we
are
specifically
working
on
since
probably
right
now,
everything
is
a
bit
too
high
level.
B
One
of
the
products
that
we
launched
six
months
ago
is
the
liquidity
protocol
in
the
case
of
the
Falco
Network.
So
Falcon
does
not
have
native
sticking,
and
the
reason
for
that
is
that,
in
order
to
protect
the
data
that
is
being
stored
on
Falcon,
there
is
the
concept
of
collateral
locks
to
that
particular
there,
and
that
is
basically
allowing
Falcon
to
connect
the
liquidity
with
the
particular
utility
that
the
Falco
network
is
providing.
B
This
is
a
great
for
you
if
you
want
to
sort
the
data,
but
it's
also
super
problematic
for
the
economy,
because
you
need
to
very
precisely
coordinate
the
liquidity
and
the
utility
that
is
being
provided
to
the
network.
This
means
that
the
storage
provider
will
need
that
liquidity
in
order
to
provide
the
service,
in
this
case
the
utility,
and
they
need
to
be
able
to
access
the
liquidity
somehow.
B
This
is
super
important,
because
what
you
want
is
to
be
able
to
scale
horizontally
and,
if
you're
doing
both
Computing
and
search.
It's
super
hard
for
you
to
orchestrate
those
workloads
inside
the
of
a
data
center
and
if
you
zoom
into
the
compute
provider.
Basically,
the
compute
provider
is
just
getting
some
data
from
ipfs
or
filecoin.
B
It's
preparing
that
there
and
it's
doing
the
compute
workload
that
is
necessary
to
onboard
the
the
to
the
Falcon
Network,
which
is
called
ceiling,
which
is
basically
doing
profile
for
application
that
that
is
required
to
be
able
to
efficiently
prove
to
the
network
that
there
is
really
stored
there.
You
could,
as
well
think
of
like
multiple
workloads
running
on
those
computer
providers
like
lipier,
fluence
and
so
on,
and
to
wrap
up.
Basically,
what
is
the
the
web
free
Cloud?
B
Computer
yeah,
so
so
let's
get
back
here
so
in
the
case
of
lipier,
you
have
two
personas,
so
you
have
the
validator
that
is
Orchestra
in
the
case
of
Live
Peter.
That's
kind
of
how
it's
called
so
live
here
does
not
have
the
link
between
the
utility
provided
and
the
work
necessary
for
that
utility
to.
So,
basically,
you
can
sneak
into
our
Orchestra,
but
that
Orchestra
is
not
required
to
do
that
utility.
B
So
you
could
imagine
having
those
validators
ask
workers
which
are
compute
nodes
to
do
just
the
small
pieces
of
encoding
and
basically
with
that
help
the
orchestras
get
more
rewards
because
they
would
be
able
to
be
more
efficient
and
with
that
they
could
allow
a
higher
yield
to
the
people
taking
into
those
orchestras.
In
the
case
of
liberal
right
now,
it's
super
small
in
terms
of
like
demand
for
computing,
it's
more
about
building
the
reputation
on
the
Orchestra,
Orchestra
or
level
getting
enough
liquidity
into
those
orchestras
and
and
then
focusing
on
the
compute.
B
D
B
E
B
Yeah,
we
are
super
excited
about
the
work
that
fluency
is
doing
and
I
think
fluency
is
going
to
be
like
the
the
infrastructure,
for,
like
many
other
use
cases.
The
reason
why
ceiling
is
super
interesting
is
the
fact
that
it's
a
a
workload
that
has
like
unlimited
demand
because
of
the
way
that
Falcon
was
designed-
and
this
is
probably
one
of
the
very
rare,
the
workloads
that
have
like
unlimited
demand.
So
any
workload
that
has
a
lot
of
demand
and
can
be
easily
paralyzed
is
something
super
impactful.
A
H
Both
people
problems
all
right,
hi
everyone,
it's
so
great
to
be
here,
I
hope
everyone
had
a
great
week.
Last
week
there
were
so
many
great
events
in
Lisbon,
and
this
week
is
super
exciting
as
well
before
I
start
I
have
to
apologize
in
advance.
I
made
a
decision
to
go
a
dentist
today.
I
thought
it's
a
good
idea
to
do
that
on
the
same
day
as
I.
Do
a
talk.
H
Success
lead
at
Ceramic
before
joining
ceramic
I
worked
in
web
2
for
most
of
my
career
I
was
a
data
scientist
and
also
focused
on
machine
learning
and
conversational
AI
for
quite
a
few
years
before
joining
ceramic
team
and
now
I
bring
some
of
that
experience
to
web3
and
also
learning
so
much
I.
Think
web3
data
overall
is
such
an
interesting
and
fast
growing
field
right
now
and
I
believe
that
we
have
such
an
amazing
chance
to
do
something
really
exciting
and
hopefully
better
than
things
that
we
have
in
web
2..
H
H
H
The
team
of
ceramic
has
been
around
for
quite
some
time
now
since
2016
and
they
have
been
focused
on
building
out
the
infrastructure
and
the
apis
that
allow
developers
to
easily
store,
modify
access
and
use
data
on
decentralized
network
and
the
core
piece
of
ceramic
is
compostability.
We
really
try
to
make
sure
that
developers
can
easily
reuse
and
repurpose
the
data
across
the
network,
and
we
aim
to
build
out
a
network
of
applications
that
run
on
composable
data.
H
We
try
to
build
on
or
even
create
new
web
3
standards
so
that
we
contribute
to
the
space
and
making
sure
that
it's
better
defined,
more
standardized
and,
in
general,
better
to
build
on
so
yeah.
Let's
talk
about
composable
data
in
general,
I
think
most
of
the
people
in
this
room
are
already
familiar
with
compostability
from
software
perspective,
and
open
source
is
really
at
the
core
of
compost
ability
in
general,
because
we
have
so
many
applications
that
were
built
in
web
3
even
over
the
past
few
years.
H
That
have
been
built
on
some
kind
of
other
open
source
project
that
was
built
before
and
it's
great,
because
developers
can
build
faster,
can
build
better
tools
and
it's
helps
with
Innovation
and
I.
Think
that's.
That
was
one
of
the
catalysts
of
why
we
had
so
many
great
applications
built
on
web3
over
the
past
few
years.
H
Most
of
the
applications,
though
I,
will
dare
to
say
we're
quite
simple
in
terms
of
what
kind
of
features
they
have,
and
most
of
them
are
obviously
really
great
in
terms
of
financial
applications,
but
if
we
compare
them
with
regular
web
2
applications,
especially
things
like
I,
don't
know
web3
social
gaming
and
things
like
that.
They
are
a
little
bit
behind
and
the
reason
why
is
because
we
don't
really
have
good
tools
and
good
infra
to
handle
data.
H
That
is
not
financial
data
and
also
there
are
reasons
of
it's
expensive
to
store
data
on
chain.
It's
also
quite
slow.
So
at
Ceramic
we
try
to
solve
these
issues
by
providing
the
tools
that
are
built
on
on
a
blockchain,
and
it
goes
beyond
that.
If
we
look
at
how
things
even
are
built
in
web2,
there
are
some
issues
on
both
user
and
developer
side.
So,
for
example,
from
user
perspective,
I'll
give
just
one
example:
let's
say
I
want
to
use
a
social
app
again.
H
H
Colors
theme
and
things
like
that
and
later
on,
I
will
invest
my
time
in
building
out
my
social
graph,
so
people
that
I
follow
people
that
follow
me
and
things
like
that.
If
I
sign
up
on
another
social
application,
the
same
process
goes
once
again.
I
have
to
do
exact
same
things
again
and
again,
it's
not
only
tedious
to
do
that.
Every
single
time,
but
also
there
are
some
aspects
of
user
experience
that
go
beyond
just
having
to
create
new
account
every
single
time.
H
Let's
say
if
I
am
a
Creator
and
I
have
millions
of
followers
on
one
account
if
I
want
to
Branch
out
into
some
kind
of
new
application,
I
have
to
build
everything
from
zero
once
again,
I
think
that's
why
I
noticed
that
on
YouTube
we
have
so
many
YouTubers
that
have
one
massive
Channel
and
then
some
other
smaller
ones,
and
then
they
always
say.
Please
go
and
follow
me
on
this
other
channel
as
well
from
developer
perspective.
H
This
is
not
great
as
well,
because
you
have
to
do
the
same
development
over
and
over
again
for
every
new
application
you
are
building,
so
this
is
extremely
simplified,
but
this
is
how
things
look
like
most
of
the
time
in
web2
developers
have
to
build
out
the
systems
where
users
can
interact
with
some
kind
of
server
that
has
some
kind
of
apis.
They
pull
some
data
from
somewhere.
They
store
some
data
in
a
centralized
database
somewhere,
and
then
the
user
gets
some
kind
of
view
that
they
see
when
they
interact
again.
H
The
problem
is
that
developers
have
to
build
the
same
thing
over
and
over
again
for
most
of
the
applications
out
there
because,
for
example,
all
of
the
social
apps
they
need
features
like
user
profile.
They
need
features
like
like
follow
and
things
like
that.
So
it's
all
repeated
and
then
another
thing
that
I
personally
experience
while
working
as
a
data
scientist,
is
that
all
information
on
Webco
obviously
is
really
siled
each
company.
Each
application
has
its
own
centralized
database
where
they
store
the
data
from
machine
learning.
H
So
there
have
to
be
some
kind
of
better
solutions
that
would
make
it
would
make
life
easier
for
both
the
users
and
the
developers
and
at
Ceramic.
We
believe
that
one
of
the
solutions
for
that
is
thinking
about
data
in
web
3
through
the
lens
of
components
of,
for
example,
data
models
and
data
model
on
ceramic
network
is
really
the
core
piece
that
developers
can
use
to
build
out
their
data
layer
in
their
application.
H
So
the
core
idea
here
is
that
as
a
developer,
you
can
define
a
data
model
that
would
include
some
kind
of
description.
What
that
model
is
for,
for
example,
it
can
be
for
a
user
profile
and
then
in
the
core
of
that
data
model,
you
have
a
schema
which
defines
what
kind
of
data
will
be
stored
in
that
model
and
and
what
kind
of
maybe
relations
this
data
model
will
have
with
other
data
models,
and
since
we
built
on
decentralized
network
applications
that
are
built
on
it,
they
can
share
those
data
models.
H
For
example,
we
have
this
that
app
number
one
that
I
mentioned
before
it
will
use
that,
let's
say
user
profile
data
model
and
then,
if
we
have
another
app
that
is
built
on
the
same
network,
it
can
just
simply
take
that
data
model
developers
don't
have
to
build
from
scratch
and
the
data
that
is
on
the
decentralized
network
is
going
to
be
synced
across
both
models.
What
that
means
for
the
user
is
that
they
can
take
their
data
with
them.
H
H
That
is
available
very
important
part
about
all
of
this,
though,
I
really
like
the
idea
that,
with
a
lot
of
freedom,
comes
very
big
responsibility,
it's
great
that
we
can
make
all
of
this
data
available
on
and
on
the
network
that
it
can
be
synced
across
different
applications.
But
then,
at
the
same
time
we
have
to
make
sure
that
user
data,
which
quite
often
is
it,
can
be
quite
sensitive
data
that
is
secure
and
that
users
have
the
ownership
of
that
data.
H
So
another
really
important
concept
that
we
talk
about
quite
a
lot
at
Ceramic
is
how
we
can
make
sure
that
users
have
the
ownership
of
their
data
and
that
their
data
is
really
Sovereign.
Some
examples
of
that
Sovereign
data
is
profile
details
like
obviously
your
name,
surname
and
things
like
that.
But
then
it
goes
beyond
that.
That
includes
your,
you
can
include
your
social
graph,
verifiable
credentials,
nfts
and
and
the
really
there
are
a
lot
a
lot.
H
These
are
the
core
of
ceramic
as
as
a
tool
in
general
and
when
users
interact
with
the
ceramic
Network
and
the
apps
that
are
built
on
ceramic
Network,
all
of
their
data
that
is
generated
while
they
interact
with
those
applications,
are
connected
to
their
decentralized
identifiers,
meaning
that
only
the
user
and
the
user
only
can
write
to
data
streams
can
make
updates
to
those
data
streams
and
also
see
the
history
of
the
updates
and
so
on.
Obviously,
there
are
still
open
question.
H
H
They
are
encrypting
the
data
that
also
at
least
right
now
comes
at
the
cost
of
composibility,
because
you
can't
really
have
that
shared
infrastructure
of
applications
that
can
access
the
data
of
your
application,
but
that's
a
possible
use
case
to
store
data
in
an
even
more
secure
way
and
yeah
and
I
think
that's
really
the
the
key
that
will
Power
web
in
the
future-
and
there
are
quite
a
few
in
general
advantages
of
aiming
to
have
this
open
sourced
information
available
on
the
network.
So,
first
of
all,
we
deciled
the
data.
H
H
We
also
at
Ceramic
believe
that
all
of
this
should
be
built
in
in
the
open
and
Community
Driven.
So
we
want
to
make
sure
that
there
are
some
standards
around.
What
kind
of
models
should
be
available.
What
kind
of
consensus
is
around
developing
them
and
instead
of
having
then
hundreds
of
tools
out
there
to
provide
different
options
of
how
data
can
be
shared
on
the
network?
Maybe
let's
create
some
kind
of
standards
and
consensus
around
that
that
all
developers
could
share
and
use
and
yeah
and
also
user
agency
is
extremely
important.
H
Again,
as
I
said,
it's
cool
that
we
can
share
it
open
in
the
open,
but
we
have
to
make
sure
that
users
have
the
agency
of
their
data.
They
can
decide
what's
being
done
with
their
data
and
where
it
goes,
and
yeah
I
think
I
covered
most
of
the
slide
in
general.
As
I
spoke
about
that
it
will,
it
will
definitely
power
the
web3
as
we
go
forward
with
open
source
data
powered
through
organizations
and
identities.
H
We
can
have
much
better
version
control
on
the
centralized
Network,
because,
as
long
as
we
have
node
operators
running
the
nodes,
it
will
all
be
available
for
developers
to
use
and
just
a
few
notes
about
Ceramics,
specifically
how
we
try
to
solve
all
of
these
challenges
around
data
in
web3.
So
on
ceramic
you
can
store
Dynamic
content.
There
is
really
no
specific
limitation
on
what
kind
of
content
can
be
stored
on
ceramic.
We
have
companies
in
in
web3,
Social
doing
that
good
coin
passport
is
built
on
ceramic
as
well.
H
It
has
indexing
and
things
like
that
right
now,
it's
on
a
developer
preview
mode.
So
we
really
appreciate
developers
trying
it
out
and
sharing
the
feedback
on
what
can
be
done
better
and
the
very
top
in
this
list
is
data
composite.
So
you
can
build
out
different
models
for
your
application
and
then
basically
compose
them
and
Define
how
your
applications
architecture
looks
like,
and
yeah
I
mentioned
some
examples
of
companies
that
are
building
on
ceramic.
Today,
we,
as
I,
said
we
have
a
really
big
range
of
applications
for
ceramic.
H
Even
last
week
there
were
quite
a
few
interesting
conversations
that
we
had
from
different
companies,
for
example
insurance
companies
interested
in
using
ceramic
for
storing
data.
So
it's
really
exciting
to
see
what
kind
of
applications
we
have
dials
as
well
are
really
interested
in
something
like
ceramic
to
store
data,
and
especially
for
their
Fireball
credentials
and
reputation
and
yeah,
and
right
now
we
are
working
on
four
key
things,
so
we
are
improving
the
data
models,
how
they
are
defined
again.
H
I
mentioned
we
have
graphql
API,
which
makes
it
much
easier
to
Define
schemas
for
the
models
and
also
interact
with
the
data
on
the
network.
We
added
indexing,
which
will
make
data
retrieval
faster
and
now
we
are
focusing
as
well
on
scaling
the
network
so
that
it's
fast,
reliable
and
and
is
production
ready,
and
we
actually
added
the
sign
in
with
ethereum
and
did
pkh
options.
So
basically
that
means
that
anyone
that
has
a
wallet
now
can
have
a
did,
and
hopefully
the
ceramic
as
well
to
store
their
data
so
yeah.
H
E
Thank
you,
hi,
hey,
do
you
have
any
requests
for
from
underlying
layers
to
improve
the
data
model,
so
it's
really
cool
that
you're
using
graphql
and
so
on.
There's
a
lot
of
people
working
on
data
models
and
access
controls
and
deities.
If
you
were
to
like
wave
a
magic
wand
and
have
the
infrastructure
get
better
and
improve
for
ceramic,
what
would
you
ask
for.
H
That's:
okay!
That's
a
great
question!
I!
Think
in
general,
it's
like
the
feedback
that
we
got
from
developers
so
far
having
to
use
graphql
is
that
it's
just
much
easier
to
understand
what
is
necessary
and
how
to
define
data
models
so
I
think.
Another
concept
that
developers
have
to
understand
is,
for
example,
how
to
define
relations
between
different
data
models.
H
H
Oh,
that's,
that's
a
great
question.
I
would
definitely
love
to
see
some
gaming
web3
gaming
apps.
That's
my
personal
response,
because
I
I
worked
in
gaming
for
a
bit.
I
love
that
field
in
general
and
when
I
joined
with
three
I
was
so
excited
to
see
that
there
are
so
many
interesting
projects
popping
up.
But
after
trying
quite
a
few
of
them,
I
would
say
they
are
still
lacking
in
terms
of
what
kind
of
features
there
are
and
just
in
general
how
gameplay
feels
like
so
with
ceramic.
H
You
can
definitely
do
that
more
something
that
I
haven't
mentioned
is
that
we
signed
in
with
ethereum
and
tools
like
cacao
that
ceramic
also
built
out.
You
can
have
session
management
and
delegation
in
your
apps,
which
means
that
users
don't
have
to
approve
every
single
action
that
they
do
with
the
app
because
they
are
delegating
The
Entity,
the
app
to
run
the
action
on
their
behalf,
and
that,
in
my
opinion,
opens
up
so
many
possibilities
for
interesting
applications.
H
H
Yeah
I
hope
so
I
think
one
of
like
there
are
some
components
that
obviously
we
need
in
web3
that
are
present
in
web
2
as
well.
For
example,
how
do
you
make
sure
that
the
issuer
of
some
kind
of
credential
is
official
and,
and
things
like
that?
We
had
interesting
discussion
last
week
with
some
of
my
colleagues
about
just
in
general,
the
perception
of
reputation
in
web
2
that
it's
in
some
cases,
for
example,
your
credit
score
is
your
like.
H
B
K
All
right,
I'm
gonna,
get
started
thanks
everyone
for
having
me
tonight.
My
name
is
yonden
and
I'm.
One
of
the
co-founders
of
Life
peer
life
here
is
a
decentralized
video,
streaming
Network
and
I'm
going
to
get
into
a
little
bit
more
about
what
that
means
later
on
in
the
talk.
But
let's
get
started
by
going
over
the
agenda,
so
I'm
going
to
talk
about
three
different
concepts
today.
The
first
one
is
this
notion
of
video
compute
on
decentralized
storage.
K
Then
I'm
going
to
talk
a
little
bit
about
a
feature
that
we
built
recently.
That
falls
into
this
category,
which
we
call
ipfs
CID
video
streaming
playback,
and
we
think
that
this
is
a
feature
that
helps
illustrate
some
of
the
concepts
that
we've
been
working
on
in
this
realm
of
video
compute
on
decentralized
storage
and
then,
lastly,
I'm
going
to
talk
about
by
taking
a
step
back
this
decentralized
storage,
video
compute
pipeline.
So
the
CID
video
streaming
playback
feature
we
think
illustrates
flavor
of
this
pipeline.
K
But
this
is
a
more
forward-looking
conversation
about
what
this
pipeline
could
look
like
in
the
future.
So
let's
talk
about
video
compute
on
decentralized
storage.
So,
to
help
illustrate
this
a
little
bit
better
I
think
a
good
starting
point
is
nfts.
So
what's
interesting
about
nfts
are
that
they
are
on-chain
assets
that
have
on-chain
property
rights,
but
some
of
the
most
interesting
things
about
the
nfts
are
actually
off
chain
or
in
the
metadata
Associated
within
nft,
and
that
metadata
might
contain
things
such
as
traits
attributes
but,
more
importantly,
links
to
off-chain
content.
K
Whether
that
be
images,
videos,
text,
Etc
and
what's
been
really
cool
to
see
in
the
nft
space
is
that
we've
slowly
started
to
standardize
the
way
that
we
reference
a
lot
of
different
files
and
assets
and
content
in
this
nft
metadata
and
at
livepier.
The
type
of
content,
obviously
that
we
care
about
most
is
videos.
So,
what's
nice,
is
that
whether
it
be
through
ipfs,
cids
or
various
other
cryptographic
IDs
in
the
space,
we're
able
to
create
this
shared
verifiable
data
layer
by
verifiably
linking
a
piece
of
data?
K
That's
off
chain
with
an
nft
that's
on
chain?
And
this
establishes
this
shared
verifiable
data
layer
because
with
something
like
a
CID
with
something
like
a
cryptographic
ID,
you
can
verify
that
this
actually
points
to
the
correct
data
and
it's
a
verifiable
link,
as
opposed
to
say
providing
an
HTTP
URL
which
could
break
if
that
location
stops
serving
that
particular
content.
K
So
at
live
peer,
we
think
of
cids
as
base
ingredients
for
enhanced
Renditions.
Now.
What
does
this
mean
in
practice?
What's
interesting
about
video
on
the
internet?
Is
that
very
rarely
do
you
actually
watch
the
original
video
that
was
created
by
someone,
and
this
doesn't
mean
that
the
visual
content
is
different.
K
But
this
is
a
process,
that's
called
transcoding
and
that's
usually
the
starting
point
for
a
lot
of
this
compute,
where
you
want
to
basically
take
a
single
video
that
comes
in
and
translate
it
into
different
versions
so
that,
regardless
of
where
you
are
around
the
world,
what
device
you're
on
things
just
work.
So
that's
a
little
bit
of
problematic,
though
right,
because
if
we
have
a
standardized
reference
via
a
single
CID
to
a
piece
of
video
content.
K
What
do
you
do
about
all
these
other
Renditions
that
you
care
about
right,
you're,
only
pointing
at
a
single
given
file,
but
I
think
one
way
that
we
can
think
about
it
is
we
can
work
with
that?
We
can
use
the
CID
as
a
base
and
as
a
root
for
the
rest
of
a
compute
pipeline,
where
you
can
start
with
the
CID.
K
That
represents
some
original
video
file
that
you
care
about
and
that's
the
original
visual
content,
but
you
can
apply
compute
on
it
on
a
layer
on
top
in
order
to
create
something
more
interesting
and
create
an
enhanced
version
of
that
video.
So
the
example
that
I'll
go
through
here
that,
hopefully,
can
help
illustrate
why
you
might
want
to
do.
K
Video
is
because,
actually
last
year,
someone
thought
it
would
be
a
really
good
idea
to
create
a
better
version
of
the
Rick
Roll
video,
because
the
Rick
Astley
music
video
came
out
quite
a
while
ago
and
around
that
time,
the
encoding
Technologies
and
the
video
Technologies
weren't
able
to
capture
a
high,
a
truly
high-res
version
of
that
music
video.
So
if
you've
ever
been
Rick
Rolled,
it's
a
pretty
grainy
experience
and
someone
decided
that,
wouldn't
it
be
great,
if
you
could
Rick
Roll
people
in
4k,
how
would
you
do
that?
K
Well,
thanks
to
some
more
advanced
video
AI
techniques.
These
days,
you
can
actually
apply
a
technique,
called
super
resolution
and
upscale
and
increase
the
resolution
from
an
original
grainy
video
and
restore
it
back.
So
that's
pretty
cool.
You
end
up
with
this
enhanced
version
of
the
original
video,
and
how
do
you
deal
with
that
in
this
context,
where
we
have
a
single
CID
that
was
linked
with
the
nft
metadata?
Well,
we
can
visualize
and
see
this
compute
layer.
K
So
this
brings
me
to
something
that
we've
been
working
on
recently.
That,
hopefully,
can
help
illustrate
this
a
little
bit
more
concretely
and
less
theoretically,
which
is
this
feature
that
we
call
ipfscid
video
streaming
playback,
and
this
brings
me
back
to
this
topic
of
transcoding.
That
I
was
talking
about
earlier,
where
the
plumbing
of
video
streaming
online
today
really
oftentimes
starts
at
this
process
called
transcoding,
and
why
do
you
want
transcoding?
Well,
two
of
the
main
reasons
here
are
that
one
you
want
to
shorten
the
time
it
takes
to
play
back
a
video.
K
K
K
So
that
brings
me
to
a
little
bit
about
what
live
pure
does
so.
Life
peer
is
a
protocol
in
a
network
that
helps
establish
this
Global
open,
decentralized,
video
infrastructure,
Network
and
transcoding
is
one
of
the
most
important
tasks
that
nodes
on
the
network,
complete
so
they're
responsible
for
taking
video
in
doing
the
necessary,
compute
and
processing,
so
that
all
the
things
that
I
mentioned,
such
as
short
time
to
playback
adaptive,
bitrate
streaming
just
work,
and
it's
powered
by
this
decentralized
network
of
nodes.
K
We've
built
this
life
here,
Studio
product,
which
you
can
think
of
as
a
gateway
to
the
live,
pure
Network.
So
it's
a
hosted
API.
That
makes
it
easy
in
order
to
tap
into
the
functionality
that
this
network
of
nodes
offers,
and
today
we
offer
we
power
a
whole
slew
of
different
video
applications
in
both
the
web
3
and
web
2
space,
including
some
of
the
ones
mentioned
here.
K
So
this
brings
me
to
the
actual
feature.
So
what
we
want
to
be
able
to
enable
is
transcoded
videos
for
cids,
so
you
can
have
a
CID
reference
to
an
existing
video
file
on
ipfs
and,
as
is
the
only
thing
that
you
can
do,
is
play
it
directly
from
a
Gateway
or
maybe
that
you
can
fetch
it
from
one
of
the
retrieval
markets.
But
at
the
end
of
the
day,
all
you
can
do
is
play
back
that
exact
file.
K
K
We
can
actually
pull
that
video
in
into
live
peer,
apply
the
transcode
in
the
video
compute
Pipeline
and
out
on
the
other
side
comes
the
transcoded
Renditions.
So
in
comes
the
CID
for
mp4
file.
Out
comes
the
transcoded
Renditions
that
can
be
relayed
back
to
a
video
player
on
the
client
side
and
you
get
this
high
quality
playback
experience.
So
we
have
a
demo
page
called
play
a
video
from
ipfs
that
actually
illustrates
this,
and
hopefully,
if
this
works,
I'm
going
to
show
a
quick
demo,
video
of
the
actual
process.
K
All
right
so
I'll
narrate
a
little
bit
here
so
in
order
to
help
illustrate
how
transcoded
playback
works.
Some
of
the
Prelude
here
is
I'm,
going
to
actually
simulate
a
mobile
Connection
by
switching
down
to
a
3g
connection
in
my
browser.
So
we
want
to
be
able
to
watch
these
two
videos
side
by
side
on
a
crappy,
3g
connection,
and
that
should
help
illustrate
why
transcoded
playback
is
useful.
So
we've
switched
it
over
to
a
3g
connection.
K
K
So
this
is
actually
what
you
would
expect
to
happen
when
you're
on
something
like
a
3g
connection,
because
if
you
have
a
higher
res
video
like
this
you're,
just
really
not
going
to
be
able
to
download
data
fast
enough
on
your
bandwidth,
regardless
of
what's
happening
on
the
infrastructure
side.
If
all
you're
dealing
with
is
that
file,
so
what
we're
going
to
do
next
is
we
can
go
to
that
demo
page?
K
K
So
hopefully,
this
demo
can
help
illustrate
to
you
what
the
what
the
feature
looks
like
and
under
the
hood
what's
happening
is
that
we've
created
a
simple
JavaScript
player,
that's
compatible
with
react
applications
and
it
can
take
care
of
this
all
for
you.
So
all
you
need
you
need
to
do
is
if
you
have
existing
CID
references
in
your
app.
You
just
drop
it
in
with
a
couple
lines
of
JavaScript
code,
and
then
you
can
enable
an
experience
that
looks
like
that.
K
I'm
running
a
little
bit
low
on
time,
so
I
won't
show
the
next
video
but
I.
Just
this.
Video
here
also
illustrates
the
same
concept:
if
you're
using
another
storage
Network
like
our
weave
as
well,
so
instead
of
using
ipfs
cids
in
order
to
playback,
you
can
take
an
rweave
transaction
ID
for
an
existing
asset
on
our
weave,
pull
it
into
the
video
compute
pipeline,
transcode
it
and
make
it
available
for
playback,
and
it
works
the
same,
but
I'll
skip
over
the
demo
video
for
now
in
the
interest
of
time.
K
Cool,
so
what
I'm
going
to
wrap
up
with
is
taking
a
step
back.
I
just
talked
about
this
ipfs
CID
video
streaming
playback
feature
but
I.
Think
what's
interesting
about
that
feature,
is
it
represents
a
step
towards
what
we
like
to
think
about
as
the
decentralized
storage
video
compute
pipeline,
and
what
we
mean
by
that
is
that
transcoding
is
an
integral
part
of
internet
video.
However,
it's
not
the
only
part
of
video
on
the
internet,
there's
other
things
and
other
ways
that
you
can
enhance
the
video
experience.
K
Something
else
that
I
think
is
pretty
cool
is
if,
if
your
call
when
I
was
talking
about
the
Rick
Roll
example,
this
technique
called
super
resolution.
So
super
resolution
is
image
restoration
technique
or
a
video
restoration
technique
where
you
can
upscale
and
increase
the
resolution
of
a
video
after
the
fact
or
an
image,
one
of
the
more
popular
ways
to
do.
This
is
using
this
open
source
model
called
ESR
Gan
and
on
the
left.
K
K
An
additional
thing
that's
worth
mentioning
is:
if,
if
a
lot
of
people
are
on
Twitter
or
follow
the
latest
trends
in
AI,
then
you
should
be
familiar
with
things
like
stable
diffusion
and
generative
models,
and
these
models
are
super
cool
because
you
can
actually,
instead
of
having
video
coming
in
and
video
coming
out.
You
have
text
coming
in
and
video
coming
out
and
you
can
actually
have
machine
learning.
Models
in
neural
networks
actually
help
generate
the
content,
and
you
can
create
videos
from
that.
K
And
then
you
use
video
concatenation
technique
in
order
to
stitch
everything
together,
and
then
you
have
the
video
and
under
the
hood,
what's
happening,
is
there's
a
lot
of
standard
video
and
image
tools
that
are
being
used
to
create
this,
in
addition
to
the
actual
generative
process.
But
those
are
all
things
that
can
be
a
part
of
this
video
compute
pipeline
that
I
just
described.
K
So
this
is
a
visualization
of
what
this
compute
pipeline
could
look
like
and
can
look
like.
So
I
won't
go
through
this
in
detail.
But
what
this
illustrates
on
top
is
a
pipeline
where
you
can
have
video
input
come
in
and
video
output
come
out
and
you
can
also
have
a
pipeline
where
text
inputs
come
in
and
video
outputs
come
out,
and
just
some
of
the
examples
here
include
you
can
have
a
video
input
come
in.
You
can
decode
the
video
you
can
scale
it.
K
Well,
it's
because
kind
of
going
back
to
transcoding
transcoding
is
composed
of
two
parts.
You
have
a
video
decoding
part
and
you
have
a
video
encoding
part
and
you
are
going
to
do
video
encoding
every
single
time
that
you
need
to
produce
a
video
that
needs
to
be
transmitted
online.
So,
basically,
anytime,
you
have
a
pipeline
that
needs
to
create
video
you're,
going
to
do
half
of
the
transcoding
process,
no
matter
what.
K
Ideally
in
an
Ideal
World,
we
can
end
up
with
this
situation,
where
it's
verified
inputs
in
and
verified
outputs
out.
So
in
a
world
of
many
formats
and
Renditions
of
the
same
original
video,
given
a
verified
input,
whether
it
be
a
CID
or
some
signed
piece
of
data,
how
do
we
link
the
video
output
of
the
pipeline
back
to
the
input
and
that's
something
that
we
think
a
lot
about
at
live
peer,
especially
in
the
context
of
this
video
compute
pipeline?
K
I
won't
get
into
too
much
like
some
of
the
work
that
we've
been
doing
there.
But
some
of
the
things
that
we
think
are
possible
here
are
basically
trying
to
create
these
provenance
chains
and
audit
Trails
for
the
computation
in
one
of
these
video
compute
pipelines,
so
that
you
can
actually
get
verified
output
out
given
verified
input
in.
If
that's
something
that
you're
interested
in
talking
about
happy
to
chat
more
about
that
separately.
But
to
close
out
just
want
to
summarize
some
takeaways
from
this
presentation.
K
So
the
first
one
is
that
cids
and
cryptographic
IDs
are
really
interesting
because
they
can
serve
as
standard
references
for
videos
and
that
can
then
be
enhanced
with
video
compute
pipelines
on
decentralized
storage,
which
enables
developers
to
build
experiences
around
video
content.
In
a
permission,
permissionless
manner,
the
ipfs
CID
video
streaming
playback
feature
that
I
demonstrated
is
one
way
where
we
can
enhance
by
creating
transcoded
Renditions
CID
by
creating
new
types
of
data
from
decentralized
Storage
data.
K
K
E
Hey
that
was
awesome.
It
was
really
cool
to
see
the
the
transcoding
and
so
on
so
I
understand
correctly,
that's
in
a
JS
HTML5
player
already.
Have
you
also
considered
doing
like
a
Gateway?
So
you
could
have
a
Gateway,
because
there
are
a
lot
of
like
simple
HTML,
sorry,
simple,
HTTP,
clients
that
can't
can't
run
JavaScript
or
something
like
that
to
reroute
the
requests,
if
you
consider
doing
like
a
Gateway,
so
you
have
like
an
actual
service
that
just
does
all
the
transcoding
ahead
of
time.
K
Yeah,
so
the
what's
happening
under
the
hood,
for
that
is
all
the
all
the
transcoding
is
happening,
so
the
the
demo
that
I
demonstrated
you
have
to
first
pull
the
data
in
and
for
the
purposes
of
the
demo.
I
did
that
before
doing
the
demo,
so
you
would
always
have
to
do
that
first,
where,
wherever
it
lives,
you
have
to
pull
it
from
the
Gateway
and
then
transcode
it
and
then
make
it
available.
What
the
actual
JavaScript
code
does.
K
Is
it's
just
a
wrapper
around
some
of
the
open
source,
hls
libraries,
because
we're
doing
hls,
transcoded,
playback
and
those
are
pretty
standard
already,
and
we
just
have
some
like
syntactic
sugar
on
top,
so
that
I
can
understand
a
CID
and
then
translate
in
that
into
the
format
to
send
it
to
the
API.
To
then
pull
the
data,
make
it
transcoded
and
make
it
available
for
playback
I'm,
not
sure.
If
that
answered
your
question,
but
okay.
D
L
Super
cool,
Tech
I'm
curious,
if
you
consider
like
using
all
of
this
for
remote
gaming
right,
like
with
nvidia's
DLNA
and
amds
FX,
where
you
could
just
like,
enable
people
to
rent
out
their
gpus
so
that
other
people
in
the
network
could
like
play.
Video
games
and
use
live
peer
Network
to
stream
that
video,
but
then
like
upscale
it
when
it
gets
to
the
user's
machine.
K
So
that's
a
really
interesting
and
exciting
use
case
and
would
love
to
kind
of
explore
the
technical
approaches
required
to
do
that.
One
of
the
main
things
one
of
the
main
challenges
there
are
that
from
a
network
design
standpoint
like
we've
spent
a
lot
of
time,
making
sure
that
the
way
that
the
pipeline
is
structured
is
that
it
can
handle
failovers,
particularly
well.
K
So
if
you're
in
the
middle
of
transcoding
a
particular
chunk
of
video
and
there's
a
failure
in
any
point
in
time,
if
you
fail
over
to
another
node
on
the
network,
it
has
minimal
disruption
on
the
actual
end
streaming
experience
for
something
like
the
rendering
cases
that
you
mentioned,
that
type
of
like
failover
for
those
compute
workloads
might
work
as
well.
I.
K
It
needs
to
be
this
really
slick
experience
the
entire
time,
and
if
you
have
to
do
some
of
these,
like
rendering
jobs
on
the
Fly
and
some
of
these
tracing
jobs
on
the
Fly,
you
would
have
to
kind
of
adapt
those
workloads
to
work
in
this
type
of
environment
as
well,
so
I'm,
just
honestly,
not
sure
what
that
would
take.
But
it
would
certainly
be
interesting
to
look
at.
J
K
K
K
So
that's
definitely
something
that
I
think
would
be
interesting
to
explore
with
projects
doing
decentralized,
CDN
stuff
in
the
future,
because
I
could
definitely
see
a
future
where
we
can
support
that
faster
and
also
just
better
than
some
of
the
traditional
methods.
Right
now
and
transcoding
would
be
a
little
bit
more
tricky,
but
I
would
have
to
kind
of
figure
some
more
technical
things
there.
J
K
The
architecture
is
that
we
have
two
different
types
of
nodes
on
the
network.
One
type
is
the
orchestrator
which
actually
does
the
heavy
listing
of
transcoding.
It
actually
doesn't
need
to
know
the
end
format
that
the
client
requests
and
then,
on
the
other
side,
we
have
these
broadcaster
nodes,
which
you
can
think
of
as
like
running
a
full
node
in
the
context
of
like
a
blockchain
network
and
the
broadcasters
are
responsible
for
basically
routing
these
jobs
to
the
orchestrators,
repeating
it
together
and
then
based
on
what
the
client
needs.
K
They
can
generate
the
hls
Manifest
and
make
that
available
to
the
client,
or
they
can
also
make
other
decisions.
If
you
need
something
outside
of
hls,
but
hls
just
happens
to
be
the
most
common,
so
usually
it
just
does
the
master
playlist
generation
and
then
the
client
is
able
to
fetch
it
from
there.
J
K
K
K
Yeah
so
I
think
that
would
be
a
part
of
the
that
that
would
be
a
part
of
the
vision
here
where
right
right
now,
like
transcoding
on
the
network,
is
just
something
like
the
core
team
is
like
baking
in,
but
I
think
it
would
be
pretty
interesting
to
be
able
to
have
a
larger
Marketplace
for
being
able
to
run
different
types
of
filters
and
compute
tasks
in
the
middle
of
the
pipeline.
The
best
analogy
to
this
is:
if
anyone
has
ever
used.
K
Ffmpeg
ffmpeg
has
this
nice
filter
graph
system,
where
there
is
the
standard
set
of
filters
that
are
implemented
for
ffmpeg,
where
you
can
do
stuff
like
scaling
frame
rate
adjustment,
you
can
even
do
deep
neural
network
inference,
but
you
can
always
fork
ffmpeg
and
add
your
own
filters
as
well.
So
I
think
that's
a
good
analogy
here
and
I
expect
the
way
that
we
would
get
started
kind
of
building
out.
K
These
types
of
pipelines
is
working
on
the
first
few,
with
the
community
directly
and
then
from
there
identifying
a
good
framework
with
which
you
can
have
more
flexibility
in
inserting
your
own
things
into
the
pipeline
as
well.
I
think
the
most
important
thing
there
is
just
making
sure
it
can
like
slot
into
the
pipeline
as
well,
because
we
need
to
make
sure
that
it
can
consume
the
right
type
of
content
and
output,
the
right
type
of
content
and
continue
in
that
chain.
A
G
M
L
M
M
F
M
M
We
see
a
lot
of
you
know:
projects
semi-successful
around
identity
and
you
know,
and
I
just
wanted
to
talk
specifically
about
what
happens
around
Computing
and
how
we
see
like
the
future
of
decentralized
computing.
So
basically,
you
think
of
computing
like
a
blockchain,
is
a
big
part
of
this
industry,
because
that's
the
new
way
of
doing
computations,
it's
a
like,
very
expensive,
very
constrained
way
of
doing
computations,
but
they
were
created
for
a
specific
purpose
of
you
know:
Reinventing
money
in
like
financial
instruments.
M
And
you
know
the
point
that
that's
a
that's
a
long
concern
for
people
who
ever
built
anything
in
web
2
that
you
usually
wanna
build
the
the
you
want
to
create
some
new
applications,
some
new
business
as
fast
as
possible.
Then
that's
why
you
want
to
rely
on
some
tools
that
provided
by
other
companies
and
built
by
other
people,
especially
in
the
form
of
apis.
M
But
then,
when
you
use
these
apis,
you
basically
locked
in
into
proprietary
databases
like
proprietary
companies
that
own
these
databases
that
provide
API
and
then
like
they.
They
control
the
access
of
of
yours
to
their
API
and
then,
if
they
don't
like
you
anymore,
they
can
just
cut
the
access
at
any
moment
and
your
application
just
stops
working
and
you
have
no
like
really
no
control
over
it.
M
This
happened
to
Twitter,
it
happened
to
Facebook
and
you
know
a
lot
of
such
cases
in
in
web
space
and
also
like
on
the
infrastructure
level.
You
don't
want
to
be
locked
in
into
particular
cloud
provider
and
there's
a
lot
of
stories
that,
like
even
if
people
were
happy
with
some
cloud
provider,
were
happy
with
the
pricing
with
the
tooling.
M
But
when
their
business
starts
growing,
the
pricing
may
not
be
good
enough,
but
they
already
created
so
much
things
on
the
proprietary
stack,
so
it's
too
expensive
to
switch
so
people
paying
like
Amazon
a
lot
of
money,
especially
when
their
business
grows.
But
it's
just
at
the
same
time
it's
just
too
expensive
to
rebuild
everything
on
on
the
other
Cloud,
so
they
they
never
switch
like.
Sometimes
it's
just
very
painful,
but
in
web
tree
we
have
actually
like
in
web
tree.
M
M
But
in
fact,
a
lot
of
people
basically
like
because
the
in
in
particular
case
of
blockchains
and
blockchain
data,
the
data
is
organized
on
blockchain
nodes
in
a
very
specific
way
that
usually
it's
not
very
convenient
for
you
know
doing
some
queries
or
particular
data
extractions
from
it.
So
people
create
apis
on
top
of
blockchain
data,
and
these
apis
are
basically
like
some
code,
Sitting
In,
traditional
cloud
and
it's
centralized.
M
M
You
know
there
are
obvious
limitations
of
the
chain,
so
you
cannot
put
like
the
application
fully
Unchained.
There
was
always
some
smart
contracts
on
chain.
There
also
is
some
back
end
sitting
in
the
cloud
in
Amazon
doing
some
indexing,
API
user
data
and
stuff
like
that-
and
you
know
there
are
a
lot
of
interesting
stories
like
when
the
dial
makes
a
decision.
M
You
cannot
really
enforce
the
code
sitting
in
Amazon
to
follow
this
decision
of
the
dial
so
like
if
I.
If
the
dial
wants
to
update
the
the
off
chain
application,
that's
actually
not
really
possible
to
do
directly.
You
need
to
bring
someone
who
is
admin
of
this
Amazon
account
to
go
to
login
and,
to
you
know,
roll
out
the
update
after
the
dial
made
a
decision.
So,
like
you,
have
this
disconnection
which
may
lead
to
like
really
just
the
dial
like
managing
applications.
M
Biologist
doesn't
give
you
its
real
value
in
this
case,
yeah
and
and
again.
So
this
is
the
point
about
like
apis
that
built
on
top
of
blockchain
data.
There
are
like
many
of
them.
There
are
nfts
apis,
like
you
know,
apis
for,
for
the
wallets
defy
apis
and
stuff
like
that
and
like
if
you,
if
you
build
on
it,
you
have
again
the
same
problem,
your
subject
of
the
login
and
the
access
can
be
disabled.
Even
if
these
guys
are
cool
and
great
and
they're
in
battery
Community.
M
M
So
the
way
we
see
kind
of
how
to
solve
it
like
we
need
some
sort
of
peer-to-peer,
decentralized
Computing
that
would
be
somewhere
between
centralized
Computing
and
on-chain
computing.
That
would
also
will
have
no
middleman,
no
as
as
blockchain,
but
will
not
have
redundancy
and
like
cost
like
high
cost
like
like
launching
computer,
and
it's
basically
like
super
useful
for
many
use.
M
We
go
to
the
network
where
these
developer
development
platform
runs
and
the
network
is
the
Open
Marketplace
of
providers.
Anyone
can
join.
Anyone
can
run
a
note.
Anyone
can
be
provided
on
this
network
and
be
part
of
development
platform,
and
it's
all
kind
of
works
together
with
the
economic
model.
So
development
development
platform
is
basically
I
can
I
can
yeah
I
can
skip
it.
M
M
Basically,
we
built
our
own
tooling
for
the
distributed
workflows
that
can
orchestrate
this
execution
of
the
functions
and
we
kind
of
and
kind
of
we
try
to
enable
all
these
simple
pieces
that
allow
you
to
build
whatever
back-end
business
logic
you
want,
and
the
the
workflow
part
is
important
because
we
are
trying
to
like
when
you
build
the
more
or
less
complex
backend.
It's
always
contains
from
from
some
pieces,
and
if
you
wanted
to
scale
you
end
up
with
this
design,
it
contains
of
two
things:
you
can.
M
You
have
the
piece
of
code
to
execute
and
you
have
some
business
Logic
on
top
of
several
pieces
of
code
and
how
they
work
together
and
what
happens
if
something
fails
or
we
need
to
paralyze
something
and
stuff
like
that
and
usually
in
in
the
cloud
you
have
this
design.
So
so
you
have
a
centralized
API
Gateway,
where
all
the
HTTP
requests
from
the
clients
come,
and
then
this
Gateway
orchestrates
the
request
to
a
bunch
of
microservices,
and
it's
all
works
perfectly.
M
But
the
problem
is
someone
needs
to
own
and
run
this
API
Gateway
and
if
you
wanna
try
to
like,
if
you
want
to
avoid
this
gateway,
then
we
need
something
very
different
so
and
what
we
created
is
something
that
that
works
very
differently.
So
that's
the
basically
protocol
and
the
programming
language
that
allows
you
to
create
and
run
any
distributed,
workflows
or
network
algorithms
like
this
one,
for
example,
in
this
example.
M
Don't
have
any
coordination
or
like
a
Gateway
here
where
I
need
to
look,
you
know
which
keeps
the
list
of
nodes
or
which
keeps
the
list
of
like
the
order
of
the
execution.
So
this
script
is
basically
put
into
the
data
packet
that
initiated
by
the
first
user
and
every
node
has
the
runtime
that
can
parse
this
script
and
say:
okay,
I
need
to
execute
this
part
and
then
I
put
like
add
the
result
to
the
data
packet
and
then
I
send
it
further
so
yeah.
M
They
may
contain
plague
plugins
with
the
data
networks,
like
with
storage
networks,
with
like
external
they
can
access
external
world
like
we
are
Legacy
HTTP
calls
they
can
access,
blockchains
and
stuff
like
that.
So
basically,
on
every
on
any
node,
you
can
run
a
function
that
can
you
know,
take
data
somewhere,
do
some
work
and
put
data
somewhere
else
or
put
data
to
the
output
request
and
the
experience
of
it
like
it's,
it's
very
similar
to
the
to
the
serverless
stack.
M
M
But
it's
it's
something
that
just
just
we
need
to
build
as
a
part
of
the
platform
and
which
will
be
built,
as
a
part
of
you
know,
Aqua
language
as
a
like
libraries
of
the
language,
basically
yeah
and
again,
like
some
some
basic
features
available
already.
Some
like
data
adapters
available
already
and
a
lot
of
things
that
looks
more
familiar
on
the
level
that
you
used
to
see
in
the
cloud.
M
So
it's
a
very
it's
a
very
nice
and
developer
tooling.
That
is
like
targeted
for
creating
distributed
algorithms,
so
you
can
create
whatever
algorithms
you
run
on
the
back
end
and
very
important
topic
about
secure
model
of
all
of
this,
and
basically
like
in
in
blockchain
by
default.
You
have
secure
model
security
model
of
the
consensus.
So,
basically,
like
you
have
some
computation
you
deploy
to
a
network
and
then
every
node
has
to
repeat
it,
and
every
node
has
to
get
the
same
result
and
that's
why
it's
so
expensive.
M
That's
why
it's
so
slow
and
and
stuff
like
that,
so
fluents
by
default,
does
not
apply
consensus
on
the
computation
so
like
by
default.
It's
basically
trusted
computation
right
now,
so
you
deploy
a
function
to
the
node
and
then
you
trust
that
the
node
executed
it
correctly.
There
is
a
thing
called
truss
graph
that
we
have,
which
is
something
similar
to
the
concept
of
web
of
trust.
M
It
means
that
if
I
know
some
Authority
and
this
Authority
trust
this
node,
then
I
maybe
should
trust
this
note
because,
like
it's,
it's
like
some
other
people
say
and
it's
trusted,
but
obviously
that's
not
enough.
If
you
want
to
have
like
proper,
verifiable
decentralized
computations,
and
what
we're
allowed
to
do
is
to
like
its
optionality
so
like
if
the
consensus
based
computation
is
fine
for
you
in
terms
of
cost
and
performance,
you
can
apply
consensus
to
execution
and
some
function
and
you're
gonna.
M
It's
like
this
area.
Progress
is
pretty
fast,
now
Yeah.
So
basically,
like
this
development
stack
runs
on
the
network
and
the
network
is
heterogeneous.
Basically
like
every
node
is
different.
Every
node
is
a
hosting
provider
and
like
every
node,
runs
different
functions,
whatever
it
allows
to
run
under
Hardware
and
yeah.
So
we,
if
you
have
such
a
network,
if
you
have
such
development
stack,
we
can
really
build
applications
and
apis
that
you
know
can
be
not
really
can
can
have
similar
capabilities
like
applications
on
chain.
M
If,
if
some
part
of
network
disagrees
with
update
that
developers
pushes
the
network
works,
and
here
you
can
have
such
sort
of
application,
Forks
like
if
I
run
some
application
on
some
nodes
in
the
fluence
network
and
then
I
want
to
you
know,
you
know,
create
the
update
and
deploy
the
update
that
the
users,
don't
like
the
users
can
say.
Okay,
like
we
are
forking
the
application
we
are.
M
We
keep
running
the
older
version
on
the
notes
that
that
it
was
available
before
and
the
new
version
maybe
will
also
be
adopted
by
some
other
nodes.
But
that
creates
a
lot
of
flexibility.
So,
in
terms
of
like,
if
there
is
a
version
of
some
application
that
has
demand
the
user
demand
for
it,
it
will
exist
while
it
has
demand
so
like,
while
some
people
are
ready
to
pay
to
nodes
for
hosting
this
version
of
a
application
regardless.
What
the
original
author
of
this
application
thinks
or
does?
M
This
version
may
exist
and
may
be
available
for
users,
and
this
is
some
state
of
things.
So
the
we
have
the
network
up
and
running
right
now
and
and
the
like
you,
you
can
do
some
basic
stuff
with
that
the
be
launching
like
we
want
to
launch.
We
want
to
have
the
the
dial
to
make
the
network
to
be
governed
by
the
Dow.
This
is
should
come
this
year,
but
you
know
with
this
Market
we
we're
kind
of
not
sure
yet
and
we
will
work
with.
M
We
work
a
lot
on
the
all
of
this
incentives
and
the
economic
model
and
the
the
proofs
because,
like
it's
all
connected
like
the
the
proofs
of
computations,
is
tied
to
rewards
for
the
computation,
because
we
want
only
a
pay
to
those
nodes
who
executed
the
code
correctly
and
that's
why
we
need
the
the
proof
of
computation
for
enabling
the
the
payments
on
on
the
network.
We
don't
want
to
rely
on
just
reputation.
M
Yeah
so,
like
the
economics
is
very
interesting
topic
because
you
can
not
only
just
have
the
payments
for
computation,
you
can.
You
know
play
with
this
a
lot
of
unchain
Primitives
to
like
enable
even
some
sort
of
defy
on
top
of
computational
workloads
and
stuff
like
that-
and
there
are,
you
know
several
other
interests
in
things
here
that
that
can
be
enabled.
For
example,
the
whole
thing
runs
like
a
you
know:
decentralized,
open,
Cloud,
where
there
are
compute
providers.
There
are
some
application
developers
and
there
is
some.
M
M
The
clouds
gonna
adopt
it
and
they're
gonna
extract
all
the
money
they
can
and
they
are
not
really
incentivizes
to
incentivized.
To
pay
you
anything
back,
and
here
we
can
enable
like,
because
we
have
this
link
between
where
Revenue
goes
and
which
you
know
which
pieces
of
applications
generate
revenue
for
the
nodes.
M
We
can
extract
the
small
part
of
this
Revenue
stream
and
send
it
to
the
authors
of
this
useful
pieces
of
applications
such
as
database
or,
like
maybe
like
libraries
stuff,
like
that
they've
been
reused
across
different
applications
and
you
as
an
open
source
developer,
can
like
really
build
continue,
building
the
open
source
product
and
get
a
fraction
of
cloud
Revenue,
which
is
super
cool
yeah.
So
this
is
just
a
link
to
some
Network
dashboard.
Like
it's
still
kind
of
you
can
you
can
call
it
test
net?
M
I
M
So
right
now
the
network
is
like
pretty
small
like
because
it's
not
in
such
a
device
Network.
So
if
you,
if
you
know
how
ipfs
Network
looks
like
like
ipfs
is
known
and
centralized,
Network
for
storage,
basically
like
a
fluency,
is
a
similar
model
for
compute.
So
like
you
can
run
a
node,
but
because
you're
not
earning
money
for
any
node.
There's
no
point
to
write
a
note
like
there's
only
way.
M
The
only
point
to
run
a
note
if
you
build
the
application
yourself
and
you
run
it
for
yourself,
so
we
as
a
like
fluence
Labs,
we're
running
like
a
bunch
of
nodes
for
Community.
They
are
quite
reliable,
but
it's
not
like
not
a
huge
Network
at
the
moment.
So
like
we
for
us,
is
you
know
the
the
moment
of
scale
on
the
network?
Is
the
launch
of
the
economics
and
and.
I
Okay,
and
so
the
second
question
I
had
because
you
mentioned
how,
if
the
users
want
to
continue
to
support
an
application,
it'll
cause
a
fork.
So
one
thing
that
it
popped
in
my
head
was
like
as
far
as
like
support
of
that,
like
let's
say,
the
users
do
do
that,
and
they
want
that
application,
and
you
come
up
with.
I
M
In
those
version
of
the
application,
they
may
continue
development,
but
they
may
not
happen
and
it's
okay,
but
the
the
the
the
point
here
is
that
the
API
to
the
previous
version
application
is
always
available
until
there
is
demand
for
it.
And
this
way,
if
you
build
on
the
on
some
decentralized
Facebook
API,
and
then
they
decided
to
cut
access
to
someone
in
the
new
version
in
the
old
version,
you
will
still
have
the
access.
So
that's
about
like
like
where
you
have
the
access
to
and
no.
C
M
C
L
D
M
Basically,
if
you
there
is
a
way
to
if
you
don't
want
to
disclose
the
code
of
the
application
like
at
all,
there
is
a
way
to
do
it
if
you
run
your
own
node
and
like
well,
it's
it's
going
to
be
like
centralized
setup
because,
like
if
you,
if
your
code
sits
on
your
node
and
the
only
way
to
access
to
kind
of
run,
it
is
to
access
to
your
particular
node.
Then
it's
just
like
no
one
sees
the
code,
but
at
the
same
time
it's
centralized,
but.
M
C
N
M
N
Just
to
give
a
little
bit
of
perspective
is
what
I'm,
what
I
mean
when
I
say
distribute
scheduling
is
so
I
as
an
application,
developer
or
maintainer
or
whatever
provider.
I.
Think
that's!
That's
that's
the
word
so
I
want
to
provide
my
application
in
the
network
right
so
now
my
application
has
different
computations
like
each
of
the
Marine
things
that
you
that
you
talked
about
right
now.
These
can
run
either
on
a
single
node
in
the
network.
N
M
By
that,
so
if
the
question
is
like
how
do
I
know
where
to
put
my
computation,
how
to
you
know
schedule
the
the
execution,
so
we
kind
of
think
that
is,
you
know
there
are
two
ways
to
that.
There
are
two
approaches
to
do
it
if
you're
application
developer
like,
ideally,
you
want
to
just
pick
something
some.
You
know
practices
or
templates
or
something
that
give
that
says.
M
You're
okay,
for
this
function,
they're
the
most
like
the
fastest
note
will
be
chosen
for
this
particular
user
or,
like
you
know
something
that
is
pre-built
for
you
or
you
can
build
the
whole
custom
workflow
in
aqua,
based
on
whatever
you
think
the
needs
of
your
application,
I'm,
not
sure.
If,
if
it's
okay.
M
Sort
of
like
you
can,
you
can
express
so
you
can
in
your
application.
You
can
have
the
aqua
workflow
that
would
describe
how
the
computation
should
run
and
like
where
it
should
be
deployed.
You
have
full
control
on
where
it's
deployed
and
how
it's,
how
it
runs
in
terms
of
function,
calls
okay
and
you
can
schedule
this
stuff
as
well.
So
you
can,
you
can
deploy
the
piece
of
you
can
deploy
the
function.
M
You
can
schedule
the
function
on
one
node
and
can
deploy
the
function,
another
node
and
schedule
as
well,
and
then,
if
this
functions
and
then
you
can
have
the
workflow
that
connects
these
functions,
for
example
to
replicate
data
between
them.
So
this
way
we'll
have
the
peer-to-peer
replication
like
data
replication
protocol.
F
F
M
Or
maybe
it's
a
little
a
little
different,
so
there
are
two
parts
of
it
like
Aqua
proofs
right
now.
The
the
design
that
we're
doing
is
that
because
Aqua
like
there
are
a
lot
of
Aqua
is
basically
requests
right,
like
the
same
as
HTTP
requests,
but
going
on
over
curricular
Network
and
the
network
serves
potentially
like
billions
of
requests.
M
There
is
you
need
to
find
a
way
to
generate
either
very
light
proofs
or
for
every
of
these
requests,
or
you
have
some
probabilistic
algorithm
that
would
allow
to
only
check
you
know
once
per
10
requests
and
see
if,
if
it
was
correct
or
not
and
then
to
reward
the
the
notes
if
they
serve
correct
results,
so
for
aqua
we
do
probabilistic
proofs.
So
it
basically
means
that
in
order
to
withdraw
the
earnings
the
node
has
to
you
know,
there
is
a
like
algorithm
of
choosing
the
particle.
M
This
request
that
the
node
needs
to
submit
on
chain
where
it's
going
to
be
verified
the
you
and
you
know
if
the
node
submits
incorrect
particle
it
got
slashed.
If,
if
it's
correct,
it
can
get
get
money
and
there
is
a
the
the
Marine
proofs,
there
wasn't
proofs
we
still
kind
of
like
researching
this
stuff,
but
the
idea
is
ideally,
we
should
have
the
ZK
proof
on
every
execution,
because
this
code
can
do
like
some
State
transition
and
stuff
like
that.
M
So
it's
dangerous
not
to
have
proofs
on
every
execution,
but
you
need
to
optimize.
You
can
have
ZK
proofs
for
like
basically
running
every
instruction
of
webassembly
virtual
machine,
but
you
want
to
optimize
it.
So,
like
the
you
know
you
can
there
are.
There
are
ways
to
optimize
it
like
you
can
do
like
Suzuki
approval,
like
every
10
instructions
of
virtual
machines
and.
A
G
O
G
G
G
G
G
G
G
G
G
G
G
G
A
O
Okay,
let
us
get
going
well.
Thank
you
all
so
much
for
coming
this
evening.
We're
not
gonna,
have
a
quick
panel
on
all
the
fighting
speakers:
We
Have
Tonight
Justina,
vuc
evgeny
and
yonda
I'd
like
to
start
with
Justina
what
you
know
we
have
been
making
so
much
progress
in
web
3
and
really
advancing
the
state
of
the
art.
Yet
there's
still
obviously
a
gap
Beyond
just
the
stuff
that
all
of
you
are
working
on.
What
do
you
think
the
biggest
thing
holding
us
back
is.
H
I
think
there
are
a
few
things
here
really
I
will
try
to
make
my
answer
as
short
as
I
can,
but
I
think
one
of
them
is
that
in
general,
like
Tech
has
to
improve,
like
even
it
doesn't
really
apply
the
ceramic
specifically,
but
in
general
applications
that
are
being
built
in
web3
right
now.
I
think
I
mentioned
that
in
my
talk
as
well
that
most
of
the
things
that
I've
seen
so
far
are
really
good
in
terms
of
financial
standpoint.
H
H
I
was
at
Defcon
a
few
weeks
ago,
and
it
was
actually
a
talk
from
live
peer
that
I
heard,
and
they
mentioned
a
really
interesting
idea
about
that,
and
they
said
that
the
real
break
will
happen
when
someone
will
build
an
app.
That
is
not
only
better
than
what
we
have
in
map
2,
but
that
doesn't
even
exist
in
map
2..
So
I
think
we
really
have
to
keep
thinking
bigger
and
instead
of
just
trying
to
replicate
what
we
have
in
web
2,
which
is
obviously
great.
H
B
On
top
of
like
the
users,
they
were
using
other
services,
they
just
sold
the
VMS
for
much
more
more
higher
prices,
but,
on
the
other
hand,
like
I
think
what's
holding
us
back,
is
the
fact
that
we
are
really
not
collaborating
enough.
If
you
think
about
what
happened
in
ethereum,
most
of
yeah
sound
is
better.
Most
of
The
Innovation
came
from
the
fact
that
you
had
different
smart
contracts,
interacting
between
each
other,
and
if
you
look
at
computing,
what
we're
doing
today,
we
are
really
not
doing
that.
B
We
have
a
bunch
of
computing
networks
all
doing
one
particular
thing
which
cannot
by
itself
be
better
than
what
cloud
providers
provide.
But
if
we
combine
multiple
networks
and
we
have
workloads
interacting
between
each
other,
then
we
can
actually
do
things
that
were
not
possible
before
imagine
having
a
zoom
application
that
is
decentralized
and
then
having
the
ability
to
have
a
bunch
of
filters
that
are
built
by
open
network
where
you
have
a
marketplace
where
people
can
actually
decide
what
kind
of
filters
they
want.
B
Maybe
that
is
a
super
smart
way
of
like
removing
your
background.
Maybe
it's
a
way
for
our
argumenting.
Your
background
on
what
people
are
talking
about
in
that
particular
Zoom
call.
So
those
are
things
that
are
not
possible
today
and
opening
up
the
market
both
for
developers
that
can
build
those
backgrounds
and
filters,
but
also
for
compute
providers
that
can
run
those
and
earn
money
for
doing
those
and
then
reducing
further
the
cost
between
30X.
That's
something
that
definitely
is
not
possible.
Today.
That's.
O
Great,
so
on
in
not
to
put
you
on
the
spot,
but
you
know
Justina
mentioned
that
you
were
already
looking
Beyond.
This
book
mentioned
that
you
know
there
are
already
a
lot
of
monopolies
out
there.
Why
are
people
buying
into
it?
Why?
Why
are
people
going
and
buying
on
AWS
versus
taking
a
choice
on
something
that
is
decentralized?
Is
it
just
missing
pieces,
or
is
it
something
else.
K
So
great
question
I
think
a
big
reason
is
in
the
status
quo
and
in
the
status
quoic
development.
It's
just
really
convenient
and
it's
really
convenient
to
default,
to
known
quantities
and
AWS
and
cloud
service
providers
are
known
quantities
and
I.
Think
that
is
a
hurdle
that
can
eventually
be
surmounted
by
a
lot
of
the
infrastructure
that's
being
built
in
this
space.
K
At
the
same
time,
I
think
it's
important
for
infrastructure
Builders
to
work
hand
in
hand
with
app
developers,
even
when
that's
not
the
case
and
work
with
the
app
developers
that
are
especially
taking
alternate
approaches,
sort
of
to
what
you
were
alluding
to
before,
where
like
we
definitely
lead
better
versions
of
the
web.
2
applications
that
we
already
have
today,
however,
I
think
there's
a
huge
opportunity
to
take
a
different
route.
K
Initially,
as
well
and
identify
what
web3
and
crypto
actually
can
uniquely
enable
and
because
it's
Unique
and
fundamentally
cannot
be
achieved
with
web
2
applications,
there
is
room
for
there
to
be
a
gap
on
the
infrastructure
level
and
then
that
allows
the
applications
to
actually
drive
the
roadmap
of
the
infrastructure
because
the
applications
are
demonstrating.
Oh,
this
is
a
clear
use
case
and
this
is
clearly
something
that
infrastructure
and
web
2
cannot
do
so
it
makes
total
sense
for
you
to
improve
on
it.
O
M
Well,
I
think
it
was
a
good
question,
I
think
some
of
these
pieces.
Definitely
we
just
need
better.
You
know
better,
like
evolution
of
them
in
in
Webster
space.
Some
of
these
Pieces
Just
will
well
I.
Think
a
lot
of
this.
It
depends
on
the
scale
like
right.
Conceptually,
things
like
I
want
to
run
some
code
on
some
infrastructure
that
magically
runs
and
magically
scales.
M
I'd
say
like
I
mean
at
some
point:
we
will
get
to
the
place
where
we
will
have
quite
enough
infrastructure
to
build
proper
applications
that
competitive
with
webto
applications,
but
in
in
many
areas
we
are
not
there
yet,
and
you
know,
as
guys
rightfully
said,
that
we
need
to
try
to
really
work
really
tied
with
application
developers
to
build
those
infrastructure
and
enable
those
exact
use
cases
that
help
them
today.
So
they
can
today
onboard
more
users
and
drive
more
demand
for
the
infrastructure.
O
So,
let's
shift
a
little
bit
instead
of
looking
at
all
our
gaps.
What
are
the
things
that
we
do
uniquely?
Well,
just
you
know
what
do
you
when
you
see
people
coming
and
using
ceramic
at
like
scale
or
or
saying?
Oh
I'm
ditching
my
old
thing,
because
it
doesn't
offer
what
you
offer.
What
do
you
lean
into
there
or
what?
What
apps
have
you
seen
that
really
set
set
a
new
bar.
H
Why
I'm
such
a
big
fan
of
Open
Source,
because
not
only
you
share
what
you
are
working
on,
but
you
get
ideas
from
so
many
different
developers,
teams
that
are
working
on
the
same
problems
and
you
kind
of
you
you
have
that
like
togetherness
in
terms
of
working
and
building
on
top
of
things
and
and
sharing
that
so
I
definitely
see
that
and
I'm
such
a
big
fan
of
that,
and
also
communities
that
surround
every
single
project
in
web3
is
amazing.
I
love
that
part
now
specifically
about
let's
say
ceramic
and
things.
H
One
of
the
key
to
use
cases
that
we
see
right
now
is
web
3
social.
That
is
definitely
one
of
the
top
use
cases
that
we
see-
and
here
are
the
I
would
say
two
key
reasons
why
we
have
teams
like,
for
example,
orbis
if
anyone
in
in
the
audience
heard
of
them
they're
really
great
project
and
they
basically
allow
users
to
own
their
data.
First
of
all,
and
that's
the
biggest
advantage
and
difference
from
what
we
have
in
web
2,
because
social
media
is
so
powerful.
H
It's
it's
amazing
and
now,
with
things
like
a
single
sign-on,
you
open
up
even
more
opportunities
for
big
companies
like
Google,
for
example,
or
Facebook,
collect
your
data
in
so
many
different
areas.
So
by
building
on
web3,
you
allow
users
to
make
sure
that
their
data
is
not
stored
on
these,
like
fragmented,
centralized
places
and
then
at
the
same
time,
the
same
composability
and
and
the
building
in
the
in
the
open
with
other
developers
the
same
data
models.
H
They
are
being
improved
by
other
developers,
so
they
can
make
improvements
faster
and
and
develop
much
faster
and
then
another
one.
I
I
mentioned
that
in
my
talk
as
well
Dao
tools
as
well,
they
are
looking
into
how
their
Fireball
credentials
could
be
stored
in
a
more
secure
way
on
on
web3
and
how
they
can
build
a
necessary
tooling
for
for
Dao.
So
this
is
something
that
we
still
see
growing,
but
it's
definitely
a
lot
of
interest
from
that
end.
So
yeah.
O
Book
kind
of
taking
it
from
the
other
side.
Those
are
great.
You've,
obviously
built
some
enormous
Partnerships
and
worked
with
people
behind
the
scenes.
A
little
bit
like
what
evgeny
said
like
how
do
we
bring
Partners
to
Market
in
the
same
way
that
Justine
is
bringing
users
to
Market?
What
do
you
see
as
the
pull
through
there?
Why
are
Partners
signing
up
for
web
3?
What's
pulling
them
in
yeah.
B
I
think
it's
generally
because
for
the
first
time
we
are
able
to
coordinate
big
like
lots
of
people,
and
we
finally
have
the
ability
to
have
particular
actors
focus
on
only
one
particular
activity
and
have
everyone
focused
on
their
parts
and
together
as
a
society,
we
can
coordinate
in
solving
like
larger
problems.
This
is
really
not
possible
today
because
like
if
you
want
to
do
anything
in
terms
of
like
comparing
you,
you
always
need
to
create
Some
Cloud
account.
You
need
to
have
a
company.
B
The
company
needs
to
be
linking
Voice
by
that
particular
organization
selling
you
that
service
and
so
on,
but
you
don't
have
ways
of
like
maybe
pulling
together
resources
in
order
to
have
a
solution
for
a
particular
cause.
For
example,
you
could
imagine
having
compute
providers
focus
on
solving
like
a
particular
disease,
because
that's
something
important
for
a
particular
community
and
then
you
could
have
like
other
communities,
basically
sticking
funds
or
investing
in
that
particular
pool
and
having
a
incentive
for
particular
compute
providers
to
like
execute
the
compute
jobs
that
actually
help
discovering.
B
O
Yeah,
when
you
you
know
to
to
men,
to
extend
what
vuca
was
just
mentioning
about
collaboration
and
Justina
as
well.
Actually,
how
do
you
see
like
organization
happening,
particularly
around
common
problems,
that
we
need
to
solve
in
a
lot
of
ways,
you're
like
working
on
a
very
common
problem,
transcoding
and
you
know,
and
and
decoding
and
things
like
that,
what
components
should
we
be
collaborating
with,
and
what
components
should
we
be
differentiating
with?
O
K
So
for
collaboration,
I
think
just
using
our
domain
as
an
example,
we
spend
a
lot
of
time
thinking
obviously
about
video,
but
generally
about
how
media
content
is
handled
and
all
the
different
Computing
infrastructure.
That's
required
in
that
and
I
think
that
gives
us
a
fairly
unique
insight
into
like
some
very
domain
specific
problems
there.
But
if
you
zoom
out-
and
you
forget
about
the
fact
that
we're
dealing
with
video
and
media,
it
can
be
very
easy
to
just
say
that.
K
Oh
this
is
a
distributed
computing
project
and
because
it's
distributed
computing,
we
can
take
a
general
framework
and
have
that
apply
to
any
form
of
compute
that
you
try
to
do
and
I
think
that's
the
Holy
Grail
right
like,
of
course,
we
would
all
love
to
see
that
happen,
but
I
think
that
is
also
theoretical
and
a
little
bit
removed
from
the
reality
and
practical
reality
that
we
have
today
and
while
that
might
be
a
worthwhile
thing
to
work
towards
I,
think
that
is
a
gap
and
like
Chasm.
That
I
think
is
worthwhile.
K
Bridging
on
the
collaboration
front,
where
it's
like
taking
like
using
us
as
an
example.
The
time
that
we've
spent
kind
of
looking
deep
in
this
domain
about
like
what
is
specifically
quirky
and
weird
about
some
of
the
work
that
we're
doing
here,
that
makes
it
a
little
bit
different
from
other
types
of
distributed.
Computing
projects,
I
think
having
that
Insight
being
brought
to
the
table
and
making
that
clear
and
having
these
conversations
with
people
can
be
useful
because
then
it
it
helps.
K
K
Think
differentiation
should
be
around
like
once
again
going
back
to
that
specific
domain
like
it's
great
to
collaborate
on
these
like
shared
technical
problems,
but
at
the
end
of
the
day,
I
think
each
of
us
also
need
to
remain
laser
focused
on
like
our
users
and
that
specific
needs
that
they
have
so
to
the
extent
that
we
should
be
collaborating.
I
think
we
should
always
be
thinking
about.
K
O
Make
sense
if
Jenny
I'll
ask
one
more
question,
then
I'll
open
up
to
audience.
You
just
raised
a
round.
Congratulations
when
you
think
about
what
investors
are
looking
for
in
the
space
you've
heard
a
bunch
of
different
ideas
here,
differentiation
extension
and
things
like
that
when
you
went
through
your
process,
what
were
they
most
looking
for
you
to
talk
about?
What
did
they
want
to
hear
that
that
other
people
can
lean
into.
M
M
That,
like
like,
say,
for
example,
a
lot
of
these
examples
are
marketplaces
like
Life
beer
or
like
fluence
like.
Basically,
we
all
do
some
some
sort
of
marketplaces.
We
have
providers
and
we
have
customers
and
then
we
connect
them
and
we
create
the
value
and
the
token
basically
allows
to
subsidize
this.
The
bootstrapping
on
this
Market
or
like
put
additional
rewards
to
Providers
or
put
additional
rewards
to
the
customers
and
basically
the
the
enable
higher
efficiency
than
you
know.
M
It's
obvious
things,
but
you
know
instead
of
uber,
we
have
decentralized
Uber.
We
because
we
eliminate
the
company,
we
can
have
lower
margins
and
we
can
power
all
the
whole
economy
with
the
token.
So
investors
look
at
such
areas
where
this
model
just
makes
sense
and
if
it
makes
sense
in
in
your
area,
then
it's
a
good
opportunity
to
really
like
replace
the
the
big
Tech
with
their
current
monopolies.
We
are
thin
or
like
well,
you
can
call
it.
You
know.
O
L
L
O
The
local
market,
the
law
of
their
evolution
of
so
I'll,
repeat
the
question
and
then
I'll
toss
it
open
to
anyone
who
wants
to
answer
this
transformed
web
one
to
web
2.
Arguably,
what
is
the
device?
What
is
the
thing
that
will
unlock
the
evolution
of
web3?
Did
I
capture
your
question
right
yeah,
who
wants
to
answer.
B
Yeah
I
I
would
say,
like
probably
I
mean
how
you
should
think
about
web
free,
like
the
first
use.
Cases
are
going
to
be
like
this
super
slow,
super,
cheap,
Computing
workloads
and,
if
you
think
about
because
that's
where
you
can
actually
provide
20
30X
improvement
over
decentralized
providers.
So
if
you
think
what
are
those
workloads
most
likely?
That's
going
to
be
machine
learning,
maybe
it's
going
to
be
inference.
That's
a
use
case
where
you
can
imagine
a
bunch
of
traction.
B
Probably
it's
going
to
be
a
bunch
of
things
around
the
data
processing
and
so
on.
Those
are
not
as
sexy
as
having
a
iPhone
actually
through
an
application
calling
a
backend,
but
those
are
very
different
problems
and
those
are
the
kind
of
problems
that
we
have
today.
Maybe
what
you're
going
to
also
see
like
is
probably
protein
folding
and
things
like
that
that
were
basically
impossible
to
incentivize.
B
O
Else
have
a
question:
anyone,
okay,
so
actually
vuc
I
want
to
open
it
up
to
whoever
would
like
to
answer
this,
but
but
really
extending
that
where's.
The
pivot
point:
where
is
it
to
like,
when
you
think
about
going
Beyond
either
what
is
possible
in
web
2
or
not
convenient
in
web
2
from
a
scale
perspective?
What
do
you
think
that
is
yeah.
B
I
think
we
are
generally
in
a
stupid
situation
where
we
have
like
a
coordination
problem,
like
all
the
tools
are
there,
but
we
have
no
coordination
between
the
different
projects.
So,
for
example,
fluence
has
a
great
Tech
stack,
but
very
few
nodes
are
right
now
run
influence
because
they
focus
on
building
the
attack.
Rather
than
focusing
just
on
pure
token
economics,
then
you
have
like
filecoin
that
has
a
great
tax
tag
for
a
high
level.
B
Ipfs
is
still
not
incentivized,
so
like
it's
super
hard
to
actually
have
like
that
layer
that
is
required
for
fluent
jobs
to
actually
be
able
to
run
fluent
workloads,
and
then
you
also
need
to
have
like
the
separate
layer
of
Dev
tools
that
are
basically
using
fluence
technology
combined
with
ipfs,
combined
with
file
form
combined
with
a
compute
layer
that
is
actually
going
to
run
the
entire
thing
and
yeah.
We
have
so
many
layers
of
complexity
to
just
get
to
our
executed
job.
B
So,
generally,
what
my
opinion
is-
and
this
is
of
course,
the
reason
why
we
started
working
on
the
resource
layer
is
that
we
need
to
have
a
global
incentive
that
basically
allows
infrastructure
providers
to
run
all
of
those
networks
on
the
same
machines
and
in
that
way,
being
able
to
have
like
multiple
workloads
interacting
between
each
other,
because
we
might
not
be
able
to
incentivize
someone
running
ipfs
at
this
point.
But
we
might
incentivize
that
someone
running
ipfs,
if
that
is
required,
to
get
a
reward.
O
K
K
I
think
there's
an
opportunity
in
crypto
and
web
3
to
leverage
some
of
the
unique
value
propositions
of
crypto,
around
attribution,
Dynamic
and
automated
value
distribution
to
create
new
media
formats.
Centered
around
things
such
as
Community
code
creation,
I
think
one
of
the
most
interesting
things
that
infrastructure
can
enable
by
being
permissionless
and
open
is
allowing
communities
to
build
on
top
of
that
co-create
media
and
content
or
new
types
of
things,
and
then
having
that
value,
attribution
and
distribution
reward
them.
I.
K
Think
those
types
of
formats
are
the
things
that,
if
we
as
infrastructure
Builders
at
live
peer
can
enable
in
the
short
term,
will
be
the
big
differentiation
and
then
once
we
layer
on
the
cost.
Differentiation
as
as
well
will
demonstrate
the
true
value
of
web3
infrastructure
for
video
relative
to
the
status
quo.
G
M
M
We
just
have
to
keep
you
know,
trying
to
combine
pieces
together,
try
to
collaborate
in
and
see
whatever,
like,
whatever
really
gets,
traction
whatever
works
for
the
user,
whoever
user
is
like
whether
it
is
a
developer,
whether
it
is
like
a
a
smaller
medium
business,
Enterprise
consumer
mobile
users,
web
users,
like
it's
just
so
many
layers
here
and
huge
space
of
experimentation.
Just
like
quick
example
of
ethereum.
M
Had
this
vision
of
you
know
that
financial
application
is
going
to
be
built
and
the
token
is
going
to
be
built,
but
they're
also
like,
if
you
remember
like
2015
or
2016
and
17,
there
was
a
lot
of
conversations
about
use
cases
of
ethereum
much
outside
of
fintech.
Like
you
know,
all
these
smart
locks,
like
I,
can
send
transaction
to
open
the
door
like,
which
is
still
makes
sense.
M
O
O
H
I
think
probably
the
key
thing
that
right
now
makes
me
really
excited
is
overall
decentralized,
science
and
machine
learning
space
which
for
me,
when
I
started
working
in
map
3
about
seven
months
ago.
Actually,
not
not.
That
long
ago
was
one
of
my
biggest
questions
and
doubts
like
I'm
leaving,
web2
and
I.
Did
all
this
data
ml
kind
of
work
and
like
what's
happening
in
web3
and
I
thought
there
is
not
much
happening,
but
actually
there
is
a
lot.
H
It's
a
really
growing
space
and
the
people
are
asking
really
really
great
questions
from
how
funding
can
be
done
better
for
research
and
how
web3
can
enable
that,
so
that
research
is
not
locked
and
not
like
managed
by
some
specific
companies
that
can
fund
specific
projects
to.
How
can
we
do
machine
learning
and
all
of
this
computation
better
in
a
more
scalable
and
and
faster
way
as
well.
H
So,
like
all
of
these
projects,
that
I
see
around
the
three
data
that
are
popping
up
and
that's
one
of
the
reasons
why
I'm
I'm
part
of
ceramic
team
is
because
they
are
part
of
that
ecosystem
as
well.
I
really
hope
that
they
will
open
up
more
opportunities
for
data
to
be
more
open,
more
shared
and
also
the
same
about
all
of
these
amazing
projects
that
are
working
on
with
these
really
hard
computational
problems
and-
and
we
have
really
open
up
the
network
for
some
really
really
great
great
and
interesting
things.