►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
We've
got
a
great
schedule
lined
up
today
in
the
first
block,
we've
got
myself
giving
a
presentation
just
to
introduce
today
and
juan
bennett,
the
founder
and
ceo
of
protocol
labs
speaking
about
the
y
division
and
introducing
the
day
two
there's
been
a
slight
change
to
schedule
in
so
much
as
I'm
going
to
be
starting
off
and
juan's
going
to
be
going
next.
A
Following
a
break.
We'll
then
move
on
to
two
of
the
engineers
working
on
the
saturn
project,
to
introduce
it
in
more
depth
and
go
through
some
more
of
the
technical
details
in
terms
of
the
flavor
of
today.
It's
going
to
be
quite
you
know
a
technical
and
architectural
look
at
saturn,
rather
than
anything
sort
of
more
product
focused
and
then
yeah
for
anyone
who's
here
in
person,
we've
also
got
the
added
bonus
of
a
workshop
after
the
live
stream
is
finished.
So
thanks
for
showing
up
in
person,
and
with
that
I
will
begin
so.
A
I'm
going
to
talk
through
start
off
with
the
motivation
of
why
we're
talking
about
retrieval
here
at
all.
A
lot
of
the
a
lot
of
the
events
of
protocol
labs
are
focused
around
storage,
but
retrieval,
in
my
opinion,
is
equally
as
important,
so
we'll
look
into
why
that
is
we'll
then
talk
about
what
a
retrieval
market
is.
It's
these
words
are
used
in
the
title
of
today's
workshop,
but
it's
good
just
to
get
that
definition.
A
Okay,
so
some
motivation,
this
statistic
is
taken
from
the
cloud
industry
and
it
goes
that
by
2025
the
global
content
delivery
network
market
is
expected
to
be
twice
as
large
as
the
cloud
object,
storage
market
so
for
cdns,
think
akamai,
fastly,
cloudflare
cloudfront
for
cloud
object,
storage.
I
think
aws
s3.
A
So
we're
saying
that
actually
fetching
your
file
from
cloud
services
is
a
market
twice
as
large
as
storing
them
in
the
first
place.
When
I
first
saw
this,
this
kind
of
surprised
me,
I
kind
of
felt
that
whenever
I
stored
stuff
in
cloud
storage,
the
the
acceleration
of
that
content
in
a
cdn
was
like
an
added
extra
benefit.
But
actually,
when
you
think
about
it,
some
more
it
starts
to
make
a
bit
more
sense,
because
people
around
the
world
using
their
browsers
using
games
and
and
more
recently,
interacting
with
meta
verses.
A
If
you
think
about
how
quickly
you
get
results
from
the
internet,
it's
actually
mind-blowing
that
you
can
load
a
list
of
pages
which,
which
you,
when
you
do,
a
google
search,
for
example,
you
can
load
a
list
of
pages
quicker
than
it
takes
you
to
actually
search
your
own
computer
for
files
is
so
optimized,
and
so
people
expect
this
great
experience
and,
as
such
content
providers
are
happy
to
spend
a
lot
of
money
on
creating
such
a
good
experience
for
their
clients,
and
that's
why
this
retrieval
space,
these
retrieval
networks,
there's
actually
such
a
big
market,
and
I
think,
what's
super
interesting
here-
is
actually
in
the
web.
A
A
So
yeah
we're
talking
about
retrievals.
Today,
that's
essentially
what
I
was
getting
out
there
now
we're
going
to
talk
about
what
a
retrieval
market
is.
So
these
two
words
we
use
a
lot
well
in
in
my
work,
and
I
just
want
to
make
sure
that
we're
all
on
the
same
page
as
to
what
we're
talking
about.
A
So
what
is
a
retrieval
market
to
me?
It's
a
setting
or
a
place
where
individuals,
entities
nodes
can
contribute
their
resources
in
order
to
provide
data
retrievals
to
people
who
are
demanding
that
data
and
trying
to
optimize
the
experience
for
people
who
are
demanding
the
data.
So
that
means
we're
looking
at
good
performance,
good
reliability
and
economic
retrievals
of
data.
A
Now
I
think
the
cdn
providers
that
I
spoke
about
previously
akamai,
cloudflare,
cloudfront,
etc.
They
operate
in
a
retrieval
market
and
that's
the
sort
of
web
2
retrieval
market
as
we'll
refer
to
it,
but
in
particular
in
the
retrieval
market
working
group
and
at
protocol
labs.
We
focus
on
decentralized,
retrieval
markets,
and
that
adds
a
lot
of
extra
interesting
complexities.
A
So
there
are
far
chrome
retrieval
market,
as
I'm
saying
is
a
setting
where
we've
got.
We
can
see
an
emergence
of
these
content
delivery
networks
and
the
far
chrome
retrieval
market
is
the
emergence
of
a
decentralized
content
delivery
network
around
the
far
coin
storage
network.
A
So
yeah,
when
we
zoom
in
a
bit
and
think
about
it
a
bit
more
like
okay,
what
actually
are
the
building
blocks
of
a
decentralized
retrieval
market?
So
we
start
to
come
up
with
these
different
topics.
We
can
look
into
on
the
right
hand
side
here:
we've
got
this
research
to
development
scale.
This
pipeline
of
how
developed
this
particular
topic
is.
So
I've
tried
to
match
each
one
of
these
building
blocks
on
the
left-hand
side
onto
this
scale,
and
I'm
just
going
to
quickly
rattle
through
them
so
cryptoeconomics.
A
We
need
to
find
ways
in
which
nodes
are
going
to
join
a
retrieval
market
or
they'll,
be
incentivized
to
join
the
network
because
of
the
rewards
they're
going
to
get
for
providing
a
service,
and
we
also
have
to
think
about
making
sure
that
crypto
economics
are
aligned
with
the
goals
of
the
network
and
the
goals
of
the
network
here
are
to
provide
those
fast
retrievals
and
it
can
often
be
quite
challenging
to
completely
align
the
the
crypto
economics
with
the
actual
goals
of
a
network.
A
The
second
point
is
retrieval
mining,
so
this
is
sort
of
a
sub
branch
of
crypto
economics
and
more
of
a
research
focus,
but
are
there
ways
in
which
we
can
have
block
rewards
where
we
can?
We
can
reward
these
retrieval
providers
based
on
the
services
they
provide
in
sort
of
a
probabilistic
way
or
like
a
more
of
a
lottery
style
approach,
payment
channels.
So
again,
we've
got
to
keep
reminding
ourselves
that
this
is
a
decentralized
network.
A
If
I'm
going
to
be
serving
your
retrieval,
I
want
to
earn
for
it
and
either
that
can
happen
from
block
rewards
from
from
a
blockchain
or
it
could
happen
in
payment
channels
that
are
managing
off-chain
payments,
which
could
then
be
later
resolved
on
an
on-chain
network
on
chain,
blockchain
reputation
systems
we're
in
a
trustless
setting.
So
I
don't
know
who
I'm
talking
to
so
it'd,
be
great
for
me
to
understand
each
node:
are
they
actually
trustworthy?
Have
they
been
serving
give
retrievals
for
for
a
long
period
of
time?
A
Is
this
someone
I
want
to
do
business
with
indexing?
So
can
I
find
data
very
quickly?
Are
there
services
in
the
network
which
are
going
to
allow
me
to
find
data
quickly,
and
could
I
and
there's
content
indexing
where
I
could
use
search
terms
to
try
and
find
data
just
based
on
those
words,
so
essentially
google
search
for
web3?
A
We
look
at
data
transport
and
transfer
protocols.
So
how
can
we
improve
bit
swap,
which
is
the
the
main
protocol
for
ipfs
and
graph
swing
graph?
Sync,
the
main
protocol
for
filecoin?
And
how
can
we
also
add
other
data,
transport
and
transfer
protocols,
for
example,
http,
which
can
allow
us
more
access
to
browser
retrievals
moving
on
to
0.7
or
perhaps
other
more
interesting
protocols
like
webrtc,
there's
some
research
going
into
that
at
the
moment
which
we'll
come
to
later
and
then
finally
network
monitoring
network
monitoring?
A
A
Cool
so
now
I'm
going
to
move
on
to
the
file
coin,
retrieval
market
in
particular,
and
just
talk
through
the
different
parts
of
the
falcon
retrieval
market.
So
far,
so
in
this
diagram,
on
the
left
hand,
side,
we've
got
a
retrieval
client
who
is
going
to
be
trying
to
retrieve
a
file
from
a
file
coin
storage
provider,
which
is
in
green
on
the
right
hand,
side,
and
we
call
this.
The
primary
retrieval
market.
A
Primary
retrieval
market
means
that
this
is
a
a
deal
and
a
transfer
that's
happening
against
a
file
coin
storage
provider,
rather
than
any
sort
of
secondary
caches
or
retrieval
providers
and
as
of
the
time
of
far
coin
launch.
This
was
the
approach
that
the
retrieval
client
would
have
to
go
through
in
order
to
make
a
retrieval
if
it
was
a
retrieval
that
they
wanted
to
pay,
for,
they
would
have
to
start
off
by
setting
up
a
payment
channel,
which
is
an
on-chain
transaction.
A
They
would
then
transfer
vouchers
to
the
storage
provider
in
exchange
for
bytes
in
this
optimistic
fare
exchange
protocol.
So
the
client
would
send
one
voucher
in
exchange
for
one
byte,
then
it
would
say:
okay,
this
is
working.
I
can
build
a
little
bit
of
trust
in
the
storage
provider,
so
I'm
going
to
send
a
few
more
vouchers
for,
and
so
let's
say,
two
vouchers
for
two
bytes.
A
Now,
why
are
we
using
vouchers
rather
than
just
file
coin
transactions?
The
reason
is,
we
don't
want
to
have
to
wait
for
an
unchained
transaction
every
time
we
want
to
receive
some
bytes.
We
need
this
to
be
very,
very
quick,
and
so
we've
got
this
this
on
off
change,
setting
where
we
can
exchange
vouchers
for
bytes
in
all
of
this,
we've
got
quite
a
lot
of
requirements
for
placing
on
the
retrieval
client.
A
So
it's
not
really
a
browser
couldn't
really
sit
in
this
blue
box
because
it
wouldn't
be
able
to
make
that
on-chain
transaction
for
a
payment
channel
or
do
this
graph,
sync
style
protocol
to
exchange
vouchers
for
bytes,
and
so
there's
yeah,
there's
quite
a
lot
of
heavy
requirements
on
the
retrieval
client.
A
So
yeah,
as
I
say
that
was
the
last
flight
was
how
far
coin
launched
and
now
what
we're
starting
to
see
emerge
is
what
I've
tried
to
capture
in
this
diagram.
So
we
have
content
publishers
down
in
the
bottom
left
who
are
now
speaking
to
what
I've
called
deal
and
distribution
services,
for
example
estuary.
So
they
say
to
s3
we'd
like
to
store
this
data
and
we'd,
also
like
it
to
be
retrievable
and
the
deal
and
distribution
service.
A
For
example,
estuary
makes
storage
deals
with
one
or
more
storage
providers
and
then
also
allows
an
ipfs
retrieval
story
through
the
ipfs
gateway.
So
this
is
great.
We've
got
some
archival
storage
covered
with
far
coin
and
we've
also
got
an
ipfs
refusal
story.
However,
I
suppose
it
leaves
a
few
things
to
be
desired
in
terms
of
sort
of
the
decentralization,
the
web
three-ness
and
also
the
crypto
economics
of
the
system.
A
If
we
think
about
the
sort
of
web
2
value
flow,
it
would
be
from
the
content,
publishers
to
the
the
website
or
web
app
providers,
and
then
they
would
serve
to
those
clients,
and
perhaps
the
clients
would
provide
value
back
to
the
content,
publishers
by
buying
things
or
interacting
with
marketplaces.
A
That
sort
of
thing
and
we've
got
a
similar
sort
of
thing
that
can
happen
here
where
the
content
publishers
are
transferring
tokens
or
value
to
the
deal
and
distribution
service,
which
then
goes
to
storage
providers,
but
it
doesn't
really
link
up
to
the
ipfs
story.
Here.
We've
still
got
people
providing
ipfs
nodes
without
sort
of
the
incentive
to
do
so,
and
the
person
who's
providing
the
ipfs
gateway
is
picking
up
the
bill
here
to
provide
the
retrievals.
A
And
then
what
we're
starting
to
see
emerge
on
top
of
the
previous
diagram
are
these
content
delivery
networks
which
are
essentially
cache
missing
to
the
ipfs
gateway?
So
to
mention
a
few
we've
got:
nft
dot,
storage
gateway,
which
is
racing
ipfs
gateways
to
on
on
a
cache,
miss
and
then
is
populating
the
cache.
With
the
content,
we've
got
meson
network
who's,
building,
a
decentralized
cdn
which
cache
misses
the
ipfs
gateway
and
the
media
network,
which
is
also
cache
missing
to
the
ipfs
gateway,
but
the
actual
cdn
is
more
decentralized.
A
A
So
that
takes
us
now
on
to
the
secondary
retrieval
market,
which
I've
still
got
the
content,
publishers
down
in
the
bottom
left
and
the
deal
and
distribution
service
and
they're
still
talking
to
the
storage
providers.
But
I've
replaced
the
ipfs
retrieval
story
with
this
ring
of
retrieval
providers
and
also
index
providers.
A
The
idea
here
and
why
it's
a
secondary,
retrieval
market
is
that
the
retrieval
providers
yeah,
so
the
primary
retrieval
market,
if
you
remember,
was
talking
to
the
storage
providers
and
making
deals
and
transfers
with
the
green
nodes
in
this
network.
The
secondary
retrieval
market
is
where
retrieval
providers
could
talk
to
each
other
and
make
deals
and
retrievals
with
each
other
and
it
sort
of
distances
it
from
from
the
storage
providers.
A
One
of
the
roles
of
these
retrieval
providers,
similar
to
the
ipfs
gateways,
is
we've
now
got
a
translation
layer
between
the
clients
that
we
use
every
day
like
browsers,
game
consoles
and,
as
I
say,
meta
versus
coming
soon
to
the
the
graphing
transfer,
the
graphing
protocol
and
the
ways
that
we
can
talk
to
storage
providers.
So
we've
got
this
sort
of
translation
layer.
A
We've
also
got
this
indexer
provider
coming
into
the
architecture
here.
This
is
work
that
was
recently
carried
out
by
the
data
systems
team
on
the
indexer
node,
and
that
can
help
to
inform
the
retrieval
provider
how
to
make
a
quick
retrieval
from
storage
providers
in
the
primary
retrieval
market,
and
we
can
also
see
now
that
the
retrieval
providers
could
then
act
as
and
and
fulfill
the
same
roles
as
web
2
cdn,
so
they
can
start
to
handle
traffic,
spikes
and
other
sort
of
attacks
on
the
sort
of
front
line
of
the
internet.
A
Another
thing
to
mention
is
this:
can
now
allow
us
to
have
much
lower
hardware
requirements
for
someone
to
contribute
to
the
network
to
be
a
storage,
firecoin
storage
provider?
There's
you
have
to
have
a
pretty
beefy
node
and
there's
some
barriers
to
entry
in
terms
of
the
hardware
you
need
to
have,
whereas
retrieval
providers,
if
we
can
allow
anyone
to
join
with
old
devices
or
their
current
desktops
and
laptops,
we
can
start
to
enable
anyone
to
join
the
network
and
provide
value
to
the
network.
A
So,
as
I
mentioned
on
the
last
slide,
we've
got
these
cdns.
That
are
cached
listening
to
the
ipfs
gateway.
We've
now
got
these
we're
going
to
mention
a
few
other
networks
which
are
merging
more
and
more
towards
this
architecture
and
I've
written
down
here
in
order
of
when
they
sort
of
started
working
in
this
space,
we've
got
the
the
miele
network,
we've
got
saturn
and
we've
got
titan
and
we'll
be
hearing,
as
I
say,
from
the
saturn
team.
A
bit
later,
I
think
also
titan
is
a
moon
of
saturn,
which
is
interesting.
A
That
we've
got
two
saturn
based
team.
Well,
yeah
planetary
teams.
I
don't
think
miles
related.
I
think
it's
to
do
with
like
something
in
your
brain,
but
I'm
not
entirely
sure
where
that
name
comes
from.
A
Okay,
so
I've
tried
to
map
out
all
of
these
different
networks.
I've
mentioned
so
far
that
are
emerging
in
this
space,
quite
crudely
onto
this
web
2
to
web
3
axis
so
a
firstly
disclaimer
that
I
apologize
if
anyone
who's
watching
live-
or
here
in
person
disagrees
quite
with
the
placement
on
this
on
this
axis.
A
But
I
think
it
sort
of
captures
the
point
that
there
are
trade-offs
that
you
have
to
make
in
order
to
talk
to
the
clients
who
use
everyday
like
browsers
in
order
to
and
that
trade-offs
make
you
more
web
2
focused.
A
And
then
there
are
teams
like
miele
who
are
working
on
the
no
trade-off
approach
where
everything
is
pure
web
3.
But
it
makes
it.
I
guess
a
little
bit
more
difficult
to
integrate
with
with
everyday
devices
and
to
manage
the
network
and
apply
the
crypto
economic
mechanisms
which
are
starting
to
form.
A
Okay,
why
do
we
even
need
a
web
3
retrieval
market?
So
I
guess
the
first
answer.
That
question
is
just
because
it's
web
3.,
if
we
have
a
web
3
stack
and
it
then
has
a
web
2
part
of
it,
then
for
a
lot
of
people
that
just
kind
of
breaks
the
whole
chain
and
you're
only
as
good
as
your
weakest
link,
but
the
the
web
2
offerings
like
cloudflare,
etc.
They
offer
a
pretty
good
service
for
very
small
files.
It's
very
performant,
it's
reliable
and
it's
pretty
economic.
A
So
why
is
it
we
want
to
reach
for
a
web,
3,
cdn
and
delivery
network,
so
I've
got
at
the
top
these
feature
requirements.
These
are
things
I
think
that
any
web3
cdn
has
really
got
to
focus
on
and
I
believe
from
these
features
and
requirements.
We
get
the
value
proper
propositions
below.
A
So,
firstly,
it's
got
to
be
permissionless.
Anyone
can
join,
anyone
can
start
to
contribute
to
the
network
and
that
sort
of
also
links
to
this
lightweight
devices
can
join.
If
lightweight
devices
can
join,
then
anyone
can
join
and
you
don't.
You
allow
the
network
to
be
sort
of
market
oriented
and
just
anyone
can
be
part
of
it
and
I
think
that's
really
important.
A
It's
got
to
be
content
addressed,
though,
because
if
you're
retrieving
files
from
a
node
that
you
don't
trust
or
has
just
joined
the
network,
if
you
ask
for
a
very
large
file
and
they
just
return-
you
a
tiny
file
or
just
the
wrong
file,
you
need.
You
need
a
way
to
sort
of
check
that
you
that
what
you've
got
back
from
them
is
what
you
asked
for,
and
that's
where
we
lean
on
content,
addressing
as
we
do
so
often
in
ipfs
and
farcoin
networks.
A
It
just
really
wouldn't
work
with
location
addressing
because
you
just
wouldn't
know
what
you
get
back
is
the
right
thing,
and
that
also
links
back
to
here.
One
of
the
one
of
the
key
things
that
makes
one
of
these
networks
more
towards
the
web3
scale
is
the
verifiability
when
we're
fetching
things
in
the
ipfs
gateway
as
a
raw
file.
A
We're
not
verifying
that
it's
the
right
file,
we're
implicitly
trusting
those
gateways
to
give
us
back
the
right
file
and
as
we
move
more
towards
the
web
three
end
of
the
axis,
we're
not
so
much
trusting
those
nodes,
I'm
having
to
verify
those
files.
So
that
means
that,
let's
say
we're
in
a
browser.
That's
an
added
thing.
The
browser's
got
to
do
which
it
doesn't
normally
do
to
the
files
it
retrieves.
A
And
crypto
incentivized
there
have
been
peer-to-peer
networks
for
retrieval
in
the
past
that
have
had
great
success.
There
have
been
some
now
which
are
starting
to
launch
for
crypto
incentives.
I
think
that's
an
important
part
of
creating
an
economy
which
will
thrive
and
continue
for
a
long
time
and
the
value
propositions.
This
gives
us,
as
I've
already
sort
of
mentioned,
completing
it
a
bit
with
the
requirements
is
the
low
barrier
to
entry
anyone
can
join.
A
A
We
can
really
do
with
those
files
once
we
retrieve
them
apart
from
just
view
them,
but
in
this
web
3
context
you
could
retrieve
and
then
you
could
start
to
earn
by
sharing
what
you've
retrieved
with
others,
just
enabling
users
to
earn
by
being
more
symmetric
and
more
powerful
nodes
in
the
network,
we
can
get
multi-peer
retrievals,
something
which
the
bittorrent
networks
did
previously,
but
we
can
start
to
rather
than
just
always
fetch
from
that
one
point
to
presence.
A
We
can
start
to
get
parts
of
the
file
from
different
nodes
and
that
links
to
sort
of
network
resilience
points
further
down.
If
nodes
start
to
go
offline,
we
can
start
to
fetch
files
from
different
parts
of
the
network
deduplication
because
it's
content
addressed,
we
don't
have
to
store
the
same
file
again
and
again
just
for
different
content
publishers.
A
We
know
that
it's
just
a
particular
block
or
a
particular
file.
We
can
store
it
once,
even
if
many
different
clients
want
to
retrieve
it
or
many
different
content.
Publishers
want
to
store
that
same
file,
so
it's
just
kind
of
a
nice
optimization
and
then
it's
market
oriented.
If
we
start
to
see
spikes
in
particular
parts
of
the
world,
or
even
in
sparsely
populated
regions
that
are
far
away
from
current
points
of
presence
in
web
2
cloud
providers,
then
we
can
start
to
see
people
spin
up
their
nodes
there
and
provide
a
good
retrieval
service.
A
A
So
I
brought
back
in
the
the
different
building
blocks
of
topics
that
we
we
touched
on
earlier,
which
we're
looking
at
as
as
part
of
a
retrieval
market
and
going
through
some
of
the
teams
that
we've
got
working
in
space
and
I'll
just
very
quickly
introduce
what
they're
working
on
first
and
also
I'd
like
to
apologize
here.
If
I
haven't,
got
a
particular
team
down
just
apologize
in
advance,
I
think
I've
covered
most
of
them
magmo.
A
This
is
a
team
which
is
part
of
consensus
and
they're
working
on
multi-hot
payment
channels.
So
how
can
we,
rather
than
having
a
payment
channel
in
between
each
retrieval
provider
and
storage
provider,
can
we
have
sort
of
a
payment,
channel
manager
or
payment
channel
provider
that
can
start
to
match
retrieval
providers
with
storage
providers
and
allow
this
to
happen
in
a
performant
way?
A
Top
right?
We
have
ken
labs
who
are
working
on
reputation
systems
in
particular,
this
metadata
service
called
pando
and
they're
just
about
to
start.
The
second
phase
of
that
we
have
my
l,
who
are
working
as
I've
mentioned
already
on
a
retrieval
network.
They've
also
done
some
great
stuff
with
browser
retrievals
and
looking
into
that
and
also
looking
at
graph
sync
and
how
we
can
write
it
in
different
languages
and
make
improvements
so
that
folds
into
sort
of
data,
transport
and
transfer
protocols,
chainsafe
they've
been
working
on
content
indexing.
A
How
can
we
figure
out
how
fast
it
is
time
to
first
bite
or
time
to
last
by
across
all
of
these
different
retrieval
networks
and
find
ways
we
can
improve
the
networks
and
finally,
new
web
group
who
are
building
the
titan
network
where
they
want
to
leverage
all
of
the
unused
bandwidth
in
home
networks
in
order
to
create
a
thriving
retrieval
network,
and
we
have
currently
10
plus
grants
in
progress
lots
of
teams
working
in
this
space
and
if
there's
any
other
teams
listening
or
interested
just
get
in
touch,
and
we
can
start
working
on
grant
proposals.
A
So
yeah
get
involved
if
anyone's
interested
in
learning
more
head
to
retrieval.market
or
scan
this
qr
code
here
to
get
there
on
your
phone
or
just
reach
out
on
file
queen
slack
retrieval
market
channel,
and
we
can.
We
can
speak
more
about
it.
A
A
B
A
Okay,
so
actually
I
think
it's
better
to
describe
it
on
this
slide.
Can
you
see
my
mouse?
A
Yes,
so
the
primary
retrieval
market
is
where
we
can
fetch
stuff
retrieve
stuff
from
the
far
coin:
storage
providers-
that's
like
the
first
layer
of
retrieval,
but
the
secondary
retrieval
market
is
when
we
can
fetch
things
retrieve
things
from
these
sort
of
secondary
cash
layers
so
retreat.
That's
the
retrieval
provider
talking
to
a
retrieval
provider
and
that's
where
we
have
the
secondary
market.