►
From YouTube: IETF-QIRG-20220907-1200
Description
QIRG meeting session at IETF
2022/09/07 1200
https://datatracker.ietf.org/meeting//proceedings/
A
Tens
of
millions
of
dollars
in
that
level
of
yeah.
B
Okay,
I'll
start
with
a
note
well
now
and
then
I'll
introduce
you
stephen
and
then
we
can
start
we'll
start
with
the
screen
sharing,
but.
C
B
Hi
everybody
through
this
seminar,
it'll,
be
given
by
stephen
I'll,
introduce
him
after.
I
do
the
note
well
to
allow
people
to
join
in
online
and
fix
their
I.t
issues.
So
I'm
going
to
spend
the
next
two
or
so
minutes
just
going
through
the
noteworld
as
usual
for
irt
of
policies.
A
A
B
And
as
this
is
the
last
slide,
I
just
want
to
again
highlight:
what's
in
the
middle,
the
irtf
is
about
research
on
standards
development,
so
we
do
want
to
foster
research,
collaborations,
etc.
So
we
highly
encourage
asking
stephen
questions
and
he
can
then
clarify
his
introducing
himself
whether
he
wants
to
accept
questions
during
the
talk
or
to
wait
afterwards
and
I'm
sorry,
okay,
thanks
and
thanks
for
this.
I
will
stop
this
and
I
would
like
to
now
hand
over
to
stephen
who's
joined
us
and
stephen
is
stephen
diademo.
B
Is
a
research
scientist
at
cisco
and
the
quantum
research
group
of
aliyah
zachary?
He
did
his
bachelor's
of
computer
science
at
the
university
of
toronto
and
masters
of
mathematics
at
the
technical
university
of
munich
in
germany.
He
continued
with
his
phd
in
electrical
engineering
at
the
teu,
munich
in
the
group
of
janis
nozzle,
where
he
worked
on
developing
simulations
and
protocols
for
quantum
networks
and
developing
architectures
for
distributed
quantum
computers.
B
C
Question
yeah,
so
personally,
thank
you
for
the
introduction
and
for
hosting
me
here
for
the
questions
you
can
ask
me
during
the
presentation,
but
if
I
think
it's
better,
if
something's
coming
up,
maybe
I
postpone
answering
so
which
we
could
try
with
the
with
that
first
during
the
presentation,
all
right
so
yep.
So
thanks
again
for
the
introduction
and
having
me
so
this
talk
is
about
this
concept.
We've
been
thinking
about
a
little
bit
about
packet,
switching
in
quantum
networks
and
how
we
think
it
could
lead
to
a
quantum
internet.
C
C
Just
quickly
show
a
picture
of
my
face
in
in
the
sunlight,
so
I'm
stephen,
I
work
at
cisco.
I've
been
there
for
about
one
year
now,
a
little
under
a
year,
10
months
or
so,
I've
been
working
on
quantum
network
designs,
a
little
bit
information
theory
in
my
past,
distributed
quantum
computing
and
simulation
is
my
main
focuses.
C
So
this
presentation
I'm
going
to,
firstly
introduce
what
is
a
quantum
network.
I
know
probably
we
all
know
this,
but
I
just
want
to
make
sure
that
we're
all
on
the
same
page,
with
the
definitions
I
use,
I
want
to
explain
what
does
a
quantum
network
do
briefly,
then?
What
I'll
do
is
I
want
to
look
at
the
existing
architectures
of
a
quantum
network,
so
this
one,
I
said,
is
about
packet,
switching
quantum
networks,
but
what's
what
else
is
there?
C
C
So
to
start
we
talk
about
what's
a
quantum
network,
so
a
quantum
network
is
something
that
can
is
a
type
of
network
that
connects
quantum
capable
devices,
and
that
doesn't
always
mean
that
every
node
in
the
network
is
quantum
capable,
but
just
some
network
nodes
are
quantum
capable.
That
could
mean
you
know
a
classical
computer,
sending
classical
information
to
a
quantum
computer,
maybe
receiving
some
quantum
information
back,
could
also
mean
that
you're
connecting
quantum
computers
together
it
could
mean
that
you're
transmitting
quantum
states
using
some
kind
of
attenuated
laser
and
detecting
it
somewhere
else.
C
And
the
transport
medium
is
generally
photons,
so
you
encode
quantum
information
into
a
photon
and
that's
usually
the
the
type
of
quantum
system.
That's
used
for
transmitting
quantum
information,
although
somewhat
less
popular
models
are
based
on
ions
and
other
things
like
that.
But
usually,
when
I
talk
about
sending
quantum
information,
I
mean
encoding,
that
into
a
photon
and
and
the
stack
like
what.
How
does
a
quantum
network
work?
C
It's
generally
so
far
been
proposed
like
an
oz
layer
model,
so
you
have
application
layer,
transport,
layer,
network
layer,
link,
layer,
physical
layer.
These
kind
of
ideas
have
been
kind
of
integrated
into
quantum
networks
and
they
work.
C
So
that's
what
I
believe
is
a
quantum
network
and
the
kind
of
a
picture
of
what
a
quantum
network
looks
like
is
you
know
in
the
different
scales
of
of
the
network.
You
can
have
a
kind
of
local
area
where
you're
in
a
data
center
kind
of,
like
we
heard
just
a
little
bit
ago,
where
you're
connecting
quantum
computers
together
and
you're
not
going
out
of
the
the
room
you
just
all
the
traffic
is
isolated
to
a
small
area.
C
Alternatively,
you
could
also
use
satellites
to
go
a
little
bit.
Further,
I
mean.
Maybe
not
arbitrary
distance
depends
on
on
the
application
depends
on
the
configuration,
but
satellites
can
be
used
to
extend
the
range
of
quantum
networks
and
you
can
kind
of
mix
and
match
the
two
systems
together
to
scale
and
distance
and
scale
in
other
aspects.
C
C
So
what
does?
What
do
we
want
to
use
quantum
network
for
so
first,
I
think
the
most
popular
use
case
at
the
moment
is
for
quantum
key
distribution.
This
is
the
nearest
term
application
that
we're
seeing
seeing
a
lot
of
field
deployments
of
quantum
key
distribution
networks,
and
it's
mainly
because
it's
quite
it's
a
little
bit
easier
to
do
than
other
quantum
network
applications.
C
We
have
the
technology
to
implement
these
things
already.
So
this
is
the
main
thing
and
the
quantification
region
yeah,
as
we
probably
all
know,
is
it's
the
ability
to
transmit
secret
keys,
but
with
the
knowledge
that
no
one
else
knows
what
that
secret
key
is,
except
for
me,
the
person
sending
the
secret
key
and
the
receiver.
I
intend
to
have
that
secret
key.
C
Another
big
application
is
in
blind
quantum
computing,
so
this
means
you
are
performing
quantum
computing
over
network,
but
the
the
the
person
who
has
the
quantum
computer
doesn't
necessarily
know
what
algorithm
you're
running
on
their
quantum
computer.
So
you
might
be
sending
them
quantum
states,
they
perform
quantum
algorithm
on
those
states,
they
send
them
back
and
they
have
no
clue
what
what
algorithm
was
around
those
qubits.
C
C
So
what
we
are
looking
at
at
cisco
a
little
bit
right
now
is
less
on
the
application
layer,
we're
kind
of
more
focused
on
the
lower
parts
of
the
network,
we're
thinking
about
the
networking
aspects.
So
this
is
more
where
I'm
focused
on
the
protocols
that
we
use
to
transport
quantum
states
simulating
quantum
networks,
the
control
of
a
network
security
aspects
of
quantum
network,
and
then
we
also
think
about
the
hardware
aspects
of
a
quantum
network.
So
can
we
develop
better
optical,
routers
and
switches?
C
C
Technology
so,
as
you
mentioned,
we're
kind
of
a
young
group,
we
were
about
one
year
old.
I
mean
sorry,
but
I
started
one
year
ago,
but
the
group
started
maybe
two
years
ago,
but
these
things
are
kind
of
like
in
the
vision,
we're
not
selling
these
things
right
now,
but
you
know
this
is
a
long-term
vision.
What
do
we
want
to?
What
do
we
want
to
focus
on?
Are
these
type
these
topics
and,
of
course
we
work
towards
that.
C
So
now
I
will
look
at
the
techn,
the
designs
for
quantum
networks
that
are,
I
would
say,
the
most
popular
and
the
most
far
along
in
terms
of
you
know
how
much
work
has
been
done
around
those
designs
and
the
first
one
I
look
at
is
this
quantum
key
distribution
networks,
the
quantum
key
distribution
network
is
a
network
of
nodes
that
are
performing
quantum
key
distribution
with
each
other.
So
if
you
look
in
this
picture,
you
see
these
nodes
numbered
one
through
48
or
so,
or
zero
to
47
or
46.
C
I
don't
know
it's,
but
each
node.
What
their
goal
is
is
only
to
do
quantum
key
distribution
with
any
other
party
in
the
network,
so
it
could
be
one
in
seven.
It
could
be
one
in
20
could
be
2
and
20.,
so
any
pair
in
this
network
should
be
able
to
perform
quantum
key
distribution,
and
this
is
a
this
picture-
is
actually
kind
of
a
depiction
of
what
has
been
deployed
in
this
work
site
here,
and
this
quantum
network
is
a
quantum
network.
C
Moreover,
this
quantum
network
is
designed
to
work
kind
of
side
by
side
with
the
classical
network.
So
in
some
cases
these
fibers
are
deployed
simply
for
this
quantum
network,
they're,
not
using
the
shared
systems
that
exist
in
the
city
for
doing
classical
internet,
and
this
quantum
network,
as
I
said,
has
been
deployed
and
it
can
be
working.
It
works
with
available
optical
technology.
C
So
that's
what
that's?
How
cricket
networks
overcome
scaling
issues
in
general?
I
mean
not
in
general,
but
in
this
particular
setting
is
using
trusted
repeaters
trusted
relay
the
next
design
to
discuss.
Is
this
kind
of
entanglement
based
quantum
networks?
I
call
them
and
what
I
mean
by
that
is
it's
a
network
who
uses
entanglement
to
transport
quantum
information.
C
The
way
it
does,
that
is,
for
two
arbitrary
nodes
in
the
network.
They
firstly
need
to
establish
entanglement
between
the
two
pairs
and
once
that
entanglement
is
established,
then
they
use
the
protocol,
quantum
teleportation
to
transport
quantum
states
from
point
a
to
point
b
using
classical
information.
So
in
the
most
simple
case,
teleportation
use
two
classical
bits
and
you
you
consume.
Your
entanglement.
Entanglement
is
no
longer
available
once
it's
consumed,
convert
that
into
two
bits
of
classical
information.
C
C
So
the
way
that
the
routing
is
done
commonly
is
done.
Yeah.
Okay,
sorry,
I'm
not
talking
about
writing
yet,
but
I
want
to
talk
about
that
you're
using
a
classical
interface.
Oh
sorry,
vortec.
I
hope
I'm
not
misrepresenting
this
paper
by
the
way
so
feel
free
to
intro
me,
but
so
this
is
it's
working
side
by
side
with
the
classical
internet.
C
So
again,
this
kind
of
concept
of
putting
the
quantum
network
beside
the
classical
internet
is
something
that's
commonly
seen
where,
where
the
vision
we're
about
to
propose
kind
of
want
to
use
all
the
same
fibers
yeah.
So
this
last
point
I
want
to
make
is
that
this
type
of
quantum
network
is
general
purpose,
meaning
any
quantum
state
can
be
sent
in
this
fashion.
C
So
these
are
the
two
types
of
networks
I
think
are
most
commonly
seen,
and
what
I
believe
from
these
two
designs
are
that
they
seem
to
be
on
a
road
that
will
not
easily
lead
to
scalability.
You
know
they're
using
types
of
switching
mechanisms
that
are
a
bit.
You
know
things
we
saw
in
the
past
that
we
had
to
get
rid
of
as
a
circuit,
switching
model
or
static
switching
models
which
don't
really
allow,
for
you
know
many
users
in
some
cases,
but
maybe
maybe
you
can.
Maybe
you
can't.
C
We
don't
know,
study
that,
but
in
general
we,
you
know
we
swapped
away
from
circuit
switching,
but
now
we're
bringing
circuit
switching
back
and
we
don't
want
it.
We
might
not
want
to
do
that
and
maybe
they're
not
supporting
all
applications
of
quantum
networks.
You
know.
Okay,
that's
that's
a
thing.
That's
a
big
problem
to
solve
in
this
current
technology
era,
but
you
know
if
we
start
thinking
about
how
will
the
entanglement
based
network
support,
prepare
and
measure
qkd
and
vice
versa?
C
How
can
qkd
network
support
entanglement
based
network
and
kind
of
thinking
about
supporting
many
applications?
Now,
instead
of
you
know,
building
one
type
of
network
up
very
deeply
and
then
finding
out
wait,
we
can't
actually
run
another
application
that
we
care
about
in
the
future.
So
you
may
want
to
start
thinking
about
that,
so
the
design
vision
for
quantum
networks,
that
is
that
we're
seeing
is
that
future
quantum
networks
should
integrate
with
the
classical
internet
and
should
already
start
being
designed
accordingly
to
coexist.
C
So
what
do
I
mean?
So
we
want
these
kind
of
three
pillars
of
quantum
networks
is
that
they
should
have
universality,
meaning
they
should
serve
all
quantum
network
applications
in
a
more
or
less
fair
way.
They
should
be
have
transparency,
meaning
they
should
integrate
with
classical
technology
as
much
as
possible.
C
So
we
don't
want
to
put
more
fibers
beside
the
fibers
that
exist.
We
don't
want
to
have
independent
hardware
all
the
time.
Of
course.
That's
a
big
demand
to
ask
of
to
ask,
but
you
know
we
should
think
about
that.
Can
we
can
we
do
anything?
Should
we
should
we
start
thinking
about
that
now
and
the
scalability?
So
we
should
have
protocols
that
allow
for
scalability.
C
C
So
we
find
that
these
types
of
quantum
network
designs
are
application
specific,
so
qkd
networks
that
have
been
deployed
are
designed
to
do
qkd
only
of
course,
there's
exception
where
we've
seen
some
qkd
networks
that
coexist
or
classical
fiber
and
showing
some
examples.
But
this
is
kind
of
a
new
thing
that
we're
seeing
haven't
seen
like
full
scale
deployments.
C
Entanglement
based
networks
are
general
purpose,
but
maybe
not
well
suited
for
prepare
and
measure
based
protocols
on
bbd4
and
maybe
others,
because
we
have
to
rely
on
robust,
entanglement
distribution
and
storage
they're,
not
so
user
scalable.
So
that
means
we're
relying
on
switching
methods
that
we've
seen
in
the
past,
don't
scale
that
well
for
many
users
in
particular
network
settings,
and
we
we
don't
want
to
rely
on
always
entanglement
based
networks,
because
we
were
thinking
about
the
connection
oriented
all
the
time
we
want
to
move
away
from
that
a
little
bit
and
then
integrability.
C
So
this
means
how
well
do
these
types
of
designs
integrate
with
what
exists
in
the
classical
regime?
Do
they
integrate
the
classical
internet
do?
Can
we
do
anything
with
the
current
switches
that
work
classical
internet?
Can
we
do
anything
with
that
technology
and
has
it
been
proposed
in
those
designs?
We
see
that
maybe
it's
lacking
there
and
we
want
to
kind
of
integrate
the
network
stacks
to
be
a
bit
more
integrated.
C
So
we
want
to
think
about
the
same
routing
approaches,
maybe
same
same
types
of
protocols
and
some
of
the
layers
so
that
we
don't
have
to
reinvent
things
that
might
come
up.
We
can
start
thinking
about.
Can
we
do
some
some
task
using
methods
that
exist?
We
want
to
integrate
that
with
the
classical
with
existing.
You
know
development
developments
instead
of
reinventing.
C
C
What
would
it
take
to
do
packet,
switching
pack,
switching
for
quantum
networks,
and
we
want
to
investigate
the
main
challenges
for
this
type
of
network.
We
don't
think
that
this
is
something
that's
going
to
come
out
in
a
year
or
two
because
yeah
this
is
definitely
some
kind
of
long-term
vision
that
will
have
many
challenges
to
overcome
and
when,
in
the
near
term,
of
course,
we're
going
to
have
to
make
compromises
just
like
the
other
architectures
have
done.
C
Switching.
What
I
mean
is
is
when
you
want
to
move
information
from
point
a
to
point
b
in
a
network.
What
you
do
is
you
reserve
the
entire
route
so
such
that
no
other
users
can
use
that
route
while
you're
actively
transmitting
information.
While
your
protocol
is
active,
you
are
sending
those
those
data
frames
through
a
reserved
route,
and
the
benefit
of
that
is.
C
You
know
that
your
your
traffic
will
arrive
in
order
because
it's
I
mean
probably
there's
exceptions
for
everything,
but
you're,
mostly
guaranteed
that
your
frames
will
arrive
in
order
and
you
you
can
kind
of
yeah.
That's
one
perk
in
the
packet
switching
network.
Your
route
is
dynamic.
You
don't
reserve
any
channels
for
your
own
purpose.
Everyone
has
kind
of
a
fair
use
of
the
network,
maybe
there's
some
prioritization
of
packets,
but
we
ignore
that.
C
For
now
we
assume
that
everyone
has
fair
access
to
the
network
you're
sending
these
data
frames,
but
you
don't
know
what
path
ahead
of
time
your
frames
will
take,
or
your
data
will
take
and
you're
not
guaranteed
that
those
frames
will
arrive
in
order.
So
you
have
to
do
a
little
bit
more.
You
know
to
care
about.
If
you
need
those
frames
to
arrive
in
order.
C
How
do
you
prevent
you
know?
How
do
you
reshuffle
them
at
the
end
to
make
sure?
So
it's
a
bit
more
complicated,
sometimes
the
packet
switching,
but
you
get
that
dynamic,
routing
and
fair
use
and
then,
with
the
circuit,
switching
you
get
that
frames
are
having
an
order,
but
you
kind
of
lose
fair
use
depends
on
a
couple
of
things,
of
course,
the
traffic
models.
C
I
know
how
often
are
these
nodes
transmitting
there's
some
exceptions,
of
course,
but
in
general
you
know
these
are
the
terminologies
I've
been
using
yeah
and
especially
like
the
pros
and
cons
is
a
difficult
question:
it's
very
dependent
on
on
the
network.
C
What
we
proposed
is
you
have
this
kind
of
hybrid
data
frame
where
you
you
have
classical
information
in
front
of
a
quantum
payload
and
then
following
the
condo
payload.
Is
this
classical
trailer
part?
I
think
you
know
maybe
you've
all
thought
about
these
things,
but
we
tried
to
go
in
detail
with
this,
and
so
what
would
happen?
Is
you
would
put
that
the
payload
information?
You
know
not
the
packet
information
in
the
header,
so
you
know:
where
does
this
this
data
frame
have
to
go,
how
you
know
the
other
properties
we
can
think
of?
C
Are
should
this
packet?
How
long
should
this
packet
live
before
we
drop?
You
know
we
can
take
a
lot
of
analogies
out
of
the
classical
regime
and
use
those
in
those
those
things
that
we've
been
working
on
for
a
long
time
and
kind
of
integrate
them
into
the
quantum
setting
yeah.
So
what
we
call
a
hybrid
frame
and
now
what
you
get
is
you
can
route
this
quantum
information
using
dynamic
routing,
so
you
don't
necessarily
need
to
reserve
anything
you're
not
doing
circuit
switching.
You
know
you
can
potentially
do
packet
switching.
C
C
You
know
once
they're
constructed
they're
moving
all
together,
we're
sending
them
at
the
same
time,
and
the
way
to
do
that
is
you
have
to
use
a
particular
type
of
multiplexing.
So
how
do
you
generate
this?
Data
frame
depends
on
the
multiplexing
scheme.
You
want
to
use
or
maybe
use
a
combination
of
these
types
of
multiplexing,
so
we
have
the
header,
the
payload
and
the
trailer,
and
the
way
we
can
do
this
modulation
is
multiplexing
is
either
with
time
with.
B
C
Division
multiplexing,
you
can
use
other
other
degrees
of
freedom,
but
in
general,
we're
thinking
about
time,
division
multiplexing,
where
you
generate
this
classical
signal
ahead
of
the
quantum
payload,
so
you
send
off
the
classical
header,
then
you
start
generating
the
quantum
payload.
Then
you
send
off
the
trailer
and
that
moves
through
the
network
synchronous.
C
You
can
also
do
wave
division
multiplexing
where
they
arrive
at
the
same
time,
one
drawback
is:
maybe
you
spend
some
time
performing
some
classical
processing
and
when
your
quantum
payload
arrives
at
the
exact
same
time,
then
you
need
to
do
something
with
it
maybe
store
it.
Maybe
you
lose
some
information,
so
you
have
to
think
about.
You
know
when
and
where
to
use
these
types
of
multiplexing
schemes.
C
C
Then
we
trigger
the
quantum
source-
and
I
don't
draw
the
trailer
here,
but
you
would
generate
the
trailer
and
all
these
things
will
be
multiplexed
together
to
generate
a
frame
and
then,
when
this
frame
arrives
at
a
node,
we
have
some
kind
of
demultiplexing
scheme
where
we
firstly
might
have
multiple
input
signals.
So
we
separate
those
multiple
input
signals.
C
Then
we
need
to
separate
the
header
from
the
payload.
We
can
do
that
over
an
optical
switch
separate.
The
header
put
it
into
a
processor.
So
we
know
what
to
do
with
this.
This
data
as
a
data
frame
and
in
the
meantime
the
quantum
payload
can
either
go
into
a
quantum
memory
or
delay
line
something
to
slow
it
down
for
a
little
bit.
While
we
do
the
classical
header
processing,
then
the
header
goes
on
and
we
recombine
these
two
pieces
together
or
three
pieces.
C
C
So
this
is
something
we
could
we
could
build,
but
if
we
want
to
think
about
the
long
term,
like
I'm
saying,
we
might
want
to
think
about
like
a
quantum
reconfigurable
optical
drop,
multiplexers
or
rotom,
that's
doing
this
in
bulk
and
more
with
more
automation
and
with
more
configurability-
and
you
know
we
so
we
drew
out
this
diagram
of
a
quantum
rhodum
where
we're
doing
much
more
complicated
things.
And,
of
course
this
is
a
distant
future
thing,
because
now
we
have
quantum
memories.
C
We
have
the
ability
to
do
error
correction
with
the
quantum
processor
we
could
potentially
convert
the
frequency
of
the
payload
or
the
even
you
know,
whatever
we
need
to
do,
and
those
technologies
don't
currently
exist
in
a
robust
way
that
we
can
build
this
already.
But
you
know
these
are
things
to
consider
for
the
future,
especially
when
you're
doing
kind
of
a
third
generation.
C
Quantum
repeater,
where
you're
not
necessarily
using
entanglement
but
more
air
correction,
and
then
maybe
this
comes
into
play
when
when
we
have
that
kind
of
network
architecture,
so
of
course
these
things
are
not
easy
to
build.
So
I
want
to
emphasize
that
there's
many
challenges
to
overcome
so
because
we
don't
get
the
ability
to
do
amplification.
C
We
in
general,
I
mean
there's
some
schemes
that
do
exist
for
amplification,
but
those
are
not.
You
know,
deterministic
and
maybe
they're
not
going
to
work
all
the
time.
So
therefore
we
have
to
avoid
signal
amplification
and-
and
that's
this
is
mainly
targeted
towards
the
way
we
do
signal
amplification
using
you
know
dope.
C
C
Also
moving
quantum.
You
know,
moving
these
weak
signals
containing
the
quantum
payload
through
optical
hardware.
It
has
to
be
very
low
loss
because
these
signals
are
already
quite
weak
and
we
already
know
that
the
optical
signals
cause
a
lot
of
loss
and
we
can't
amplify
so.
Therefore,
our
you
know
our
problem
is:
we
need
very
low
loss
optical
hardware.
C
We
need
to
think
about
the
distance
limitations.
How
close
can
we
put
our
switches?
Imagine
we
can
do
error
correction.
Can
we
put
our
switches?
Every
10
kilometers
have
two
kilometers
50
kilometers,
there's
a
lot
of
consideration
there,
but
I
think
the
general
consensus
these
things
have
to
be
quite
close
to
each
other,
maybe
in
the
ten
in
the
single
digit
kilometers,
and
because
of
that
you
might
also
change
your
network
architecture.
Does
the
classical
signal
still
have
to
be
as
strong
as
it
is?
C
If
we
want
to
avoid
these
crosstalk
effects,
for
example
the
raman
noise
between
the
strong
classical
signal
and
the
weak
quantum
signals
that
are
sharing
the
fiber,
but
if
these
switches
are
only
two
kilometers
away,
do
we
still
need
the
same
strength
of
classical
signals,
or
we
think
about
that
kind
of
thing?
Maybe
the
raman
noise
is
reduced
enough,
that
we
can
still
do
quantum
applications.
C
Maybe
we
can
still
yeah,
so
we
think
about
things
like
that,
and
maybe
if
we
want
to
build
a
switch
that
does
both
classical
switching
and
quantum
switching
in
the
same
device
do
we
also
have
to
contain
this
device
and
like
something
in
some
kind
of
fridge?
Can
we
put
these
in
room
temperature
things?
You
know
what
kind
of
restrictions
do
we
have
for
building
a
switch
that
can
support
classical
and
quantum.
C
Yeah,
so
in
terms
of
networking
challenges,
of
course,
you
know,
storing
the
quantum
states
show
that
there's
a
quantum
memory
or
delay
line,
but
yeah.
So
this
is
something
something
to
consider
I
mean
how
good
can
we
store
quantum
states?
We
want
error
correction.
This
is
also
very
difficult
to
do
right
now,
we're
not
able
to
to
do
this
yet,
but
in
the
future
you
know
people
are
thinking
about
using
air
correction
for
quantum
repeater
technology.
C
Do
we
use
quantum
relays
or
do
we
use
quantum
repeaters?
So
relay
is
kind
of
a
dumb
device
that
doesn't
actually
do
much,
but
forward
things
on,
but
quantum
repeater
design
might
do
something
more
complicated,
like
error
correction
and
then
one
big
thing
that
we
kind
of
neglect
is
a
security
of
the
network.
What's
you
know
what
do
we
do
for
security?
So
that's
the
kind
of
a
something
that
I
think
is
less
commonly
focused
on
at
the
moment,
so
yeah.
C
So
this
packet-
switching,
I
was
talking
about-
takes
the
classical
header,
the
quantum
payload
and
the
trailer,
and
it
sends
that
information
to
the
network
all
at
the
same
time,
and
when
you
do
that,
you
know
when
you're
processing
that
header
information
you
like,
I
said
you
have
to
store
that
quantum
payload
somewhere
and
maybe
the
technology
is
not
good
enough
right
now
to
store
anything.
Maybe
too
much
loss
too
much
insertion
loss
from
the
obstacle.
Switching
too
much
you
know
loss,
that's
always
the
concern.
C
So
what
we
wanted
to
do
is
think
about
something
you
can
do
instead
of
storing.
Maybe
you
can
send
the
classical
header
ahead
of
time
to
prepare
the
switch
so
that
when
the
payload
arrives,
you
don't
need
to
store
it.
You
can
just
send
it
right
through,
and
this
is
not
a
new
concept.
This
is
called
burst.
C
Switching,
so
we
want
to
make
a
compromise
at
the
intermediate
stage
so
that
we
can
still
do
some
quantum
networking,
but
we
might
lose
some
of
that
scaling
and
kind
of
pillars
that
we
discussed
and
just
to
show
a
little
animation.
What
is
burst?
Switching
you're
sending
the
header
ahead
of
time
so
that
when
the
payload
arrives,
the
switching
decision
has
already
been
made.
C
So
that's
something
we're
thinking
about
for
intermediate
stage
before
we
go
full
packet
switch
and
burst.
Switching
I
should
mention
also
is
a
bit
of
a
compromise
between
packet,
switching
and
circuit.
Switching,
because
if
you
see
that
animation,
you
know
you're
sending
the
header
over
the
channel
and
you
still
have
to
resume
in
the
time
between
sending
the
payload
and
the
header.
The
channel
has
to
be
reserved
just
one
channel,
not
the
whole
route.
C
But
you
still
have
this
reservation
aspect
and
it's
it's
a
bit
less
of
a
it's
a
bit
more
demanding
than
pocket
switching,
so
you
lose
some.
You
lose
quite
a
bit
by
doing
burst.
Switching,
but
you
gain
in
that
you
don't
need
to
store
the
quantum
payload
if
done
correct.
Of
course,
there's
challenges
involved
with
that.
How
much
burst
time
do
you
need?
C
So
what
I
talk
about
burst,
switching
is
saying
you
need
a
precise,
classical
quantum
transmission
scheduling,
meaning
you
need
to
know
precisely
more
or
less
or
else
you're
using
yeah.
You
need
to
know
precisely
how
sorry
you
need
to
know
precisely
when
you
send
the
classical
header
ahead
of
time
before
you
send
the
quantum
payload.
The
precision
is
important
because
I
said
you
waste
resources.
When
you
don't
do
it
right,
you're
sending
this
classical
control
just
ahead.
You
want
to
do
it
just
in
time.
C
You
send
the
header
just
in
time
such
that
this
routing
decision
is
made
and
the
quantum
payload
doesn't
have
to
store
in
any
memory.
And
if
you
have
robust
quantum
memories,
then
what
you
gain
is
that
that
burst.
Switching
timing
doesn't
have
to
be
as
precise,
because
you,
if
you
mess
up
the
timing,
then
you
store
the
state
instead
and
if
you
can
do
that
robustly,
then
hopefully
you
don't
lose
anything
but
in
the
yeah.
So
the
precision
is
difficult
in
some
sense
because
you
have
to
accommodate
for
the
number
of
hops
in
the
round.
C
So,
as
you
see,
you
lose
the
burst
time
hop
by
hop.
So
this
is
not
a
trivial
problem
to
determine
the
burst
switch
time
and
it
might
take
you
know
some
more.
It
definitely
takes
more
research
to
determine
the
optimal
burst,
switch
aspects
for
different
types
of
networks,
different
types
of
applications
yeah.
So
that's
something
to
think
about.
C
Also
with
the
burst.
Switching
we
can
think
about.
Is
it
can
mitigate
the
crosstalk
effect?
So
if
you
send
the
classical
signal
ahead
of
time,
then
later
send
the
quantum
payload,
then
maybe
the
classical
signal
is
not
interfering
with
the
quantum
payload
as
heavily.
So
that's
also
an
aspect
to
think
about.
C
Maybe
we
want
to
do
that
in
any
case,
because
we
would
mitigate
a
crosstalk
effect-
and
I
already
mentioned
yet
determining
this
guard
time
this
this
birth
time
is
difficult
because
it
there's
many
parameters
to
consider
so
one
other
design
aspect,
I
wanted
to
mention
it's
a
bit
futuristic,
but
why
not
is
this
concept
of
a
hybrid
network
and
these
these
concepts
I've
been
talking
about
this
burst?
Switching?
Is
packet
switch
quantum
network
without
a
quantum
repeater?
You
still
are
not
able
to
overcome
the
distance
limitations
without
doing
yeah.
Without
that
quantum
repeater.
C
You
still
are
restricted
to
maybe
at
best
a
metropolitan
network.
So
what
can
we
do
to
kind
of
get
the
best
of
both
worlds?
Maybe
we
consider
these
hybrid
networks,
where
we
go
back
to
those
models
I
showed
before,
taking
that
entanglement
based
network,
and
now
we
could
do
internet
working
by
taking
this
packet
switch
network
with
another
packet
switch
network
and
then
join
them
together
using
this
this
near
term
quantum
repeater
technology,
first
generation,
quantum
repeater
technology,
and
that
we
call
it
a
teleportation
channel.
C
C
Yeah
we
could
how
this
would
work.
Is
you
would
do
this
pack
switching,
let's
say
this
node
here
I
hope
you
can
see
my
mouse.
No
I've
been
using
the
wrong
screen.
Let's
say
this:
node
here
wants
to
communicate
with
a
node
over
here,
and
what
would
happen
is
you
would
do
packet
switching
up
to
the
the
egress
node
here
then
we
would
teleport
those
qubits
through
the
teleportation
channel,
then
re-prepare
the
frame
and
then
do
again
packet
switching
until
the
destination
is
reached,
and
now
that
this
node
is
only
doing
teleportation
at
one
node.
C
Let's
say
this
eagers,
maybe
in
the
meantime
this
this
channel
is
just
its
only
duty
is
to
build
up,
entanglement
and
teleport.
It
doesn't
have
to
wait
for
users
to
want
to
use
it.
It's
just
always.
You
know.
Maybe
this
is
like
the
more
efficient.
That's
all
open
question:
when
is
that
practical?
When
is
it
not
practical?
Is
it
possible?
C
These
are
things,
but
this
hybrid
scheme
is
something
to
think
about,
and
I
think
you
know
we're
not
the
the
only
ones
to
have
proposed
this.
So
this
is
kind
of
something
to
think
about.
Like
how
do
you
scale
these
networks
up
so
that
we
can
yeah
so
that
we
get
everything
we
want?
And
this
is
something
we
thinking
about
a
little
bit.
C
Okay,
so
now
I
want
to
just
present
a
little
bit
of
some
simulation
results
we
looked
at
for,
like
can
anything
run
over
a
packet
switched
quantum
network?
Is
it
too
much
loss?
Is
it
too
difficult?
So
we
took
some
some
ideas
of
some
suitable
applications
which
we
think
are
the
ones
that
are
somewhat
lost:
loss
tolerant.
That
means
you
can
still
achieve
the
application.
You
can
still
achieve
the
protocol
goal,
even
though
some
of
your
information
is
lost
in
the
process.
C
Some
of
those
of
those,
the
key
ones
that
we
thought
about
were
qkd
you
can-
and
this
is
the
distribution
of
entanglement.
So
we
did
a
little
bit
simulation
on
these
things
and
what
we
found
yeah.
So
we
modeled
this
this
kind
of
system
where
we
do
bb-84
and
we
are
only
sending
these
bb-84
states
over
a
line
network,
meaning
you
have
a
network
with
two
nodes
at
the
end
and
in
between
is
just
a
line
of
switches.
C
I
don't
want
to
get
too
far
into
the
secret
key
rate
calculations,
but
you
know
we
use
a
kind
of
standard
formalism,
but
we've
added
an
additional
parameter,
which
is
this
k
parameter
and
the
k
parameter
represents
the
probability
that
the
node
is
available
to
continue
the
transmission
when
the
payload
arrives.
C
So
it's
very
simple
model.
We
don't
think
about
heavy
traffic
load.
We
just
want
to
show,
can
you
route
quantum
payloads
using
packet
switching
and
so
what
we
found
is
you
know
with
with
the
number
of
hops,
of
course,
you're
losing
rates
you
your
rate
is
diminishing
because
you're
losing
every
time
you
enter
a
node,
but
what
we
see
is
yeah.
Okay.
This
is
still
positive,
so
in
the
simplistic
model
it
could
be.
C
C
And
the
way
we
did
this,
we
use
net
squid,
so
net
squid
with
nesquid.
We
we
wanted
to
to
focus
on.
How
long
can
we
store
these
bb?
These
bell
pairs
such
that
at
the
end
of
the
protocol?
C
There's
still
some
entanglement
left
so
in
this
simulation,
we're
really
focusing
on
the
memory
aspect
and
and
also
you
know
how
much
burst
time
we
could
could
get
by
with
the
minimum
burst
time
we
can
get
by
with
if
we
can
store
quantum
payloads.
C
So
we
assume
that
we
have
noisy
quantum
memories
and
noisy
quantum
channels,
and
we
assume
that
we
can
store
these
these,
these
qubits
in
memory
and
what
we
measure
as
a
cost
function
is
just
the
the
output
fidelity.
So
when
the
protocol
succeeds,
what
was
the
fidelity
of
those
cubits
in
comparison
to
the
perfect
state?
C
So
we
tried
two
different
topologies
here,
where
we
have
the
entanglement
source
in
the
middle
and
we're
measuring
this
fidelity
here
and
you
see
with
short
distances.
Even
if
you
know
we,
we
put
some
memory
parameters
that
are
maybe
optimistic
but
achievable
in
the
future,
and
the
fidelity
could
be
potentially
usable
for
long
distance.
Of
course,
this
is
still
optimistic.
I
think,
but
still
you
know
promising
at
least
it's
not
zero
entirely.
C
We
tried
the
same
topology
as
last
time
where
you
used
to
generate
the
entanglement
at
one
party,
send
the
remainder
of
the
entanglement
through
the
network.
Again
we
measure
the
cost
ek
you
lose
lots
of
distance,
because
now
the
entanglement
is
generated
on
one
side,
but
we
see
that
you
know.
We
see
that
potentially
it's
possible.
C
We
did
an
analysis
of
the
burst
switching,
so
we
did
a
analysis
against
the
length
of
the
fiber
versus
the
memory
quality
and
that's
what
these
plots
are
showing.
So
your
fidelity
versus
length
of
the
fibers
per
hop.
C
Oh,
no,
sorry,
the
total
total
distance
that
the
entanglement
is
traveling
and
the
rate
and
the
t1
t2
time
of
the
quantum
memory.
C
So
it
seems
like
in
a
millisecond
regime,
is
where
things
start
to
become
a
bit
positive.
So
we
would
need
potentially
millisecond
storage
for
doing
this
protocol
and
then
we're
looking
at
the
burst
time.
The
the
duration
of
the
burst.
Sorry,
sorry,
this
is
the
duration
of
the
time
the
packet
stays
at
the
node
per
hop.
C
Okay,
how
much
time
is
it
okay?
So
I
just
quickly
go
over
these
open
problems
and
we
should
have
ten
minutes
for
questions
so
yeah.
The
things
that
we
focus
on
the
most
is,
of
course,
the
crosstalk.
How
do
we
design
the
network
and
the
protocols
to
coexist
with
data
transmission
to
minimize
these
crosstalk
effects.
C
C
How
do
we
generate
the
frames
efficiently?
How
do
we
choose
that
guard
time
for
the
burst?
Switching?
Can
we
reduce
the
header
processing
time
in
the
quantum
network?
You
know
we've
seen
some
results
that
it
takes
about
200
microseconds.
In
some
cases,
can
we
make
that
half
as
much?
Do
we
need
as
much
data?
Can
we
simplify
the
protocols?
C
C
Now,
because
you
know
we
might
take
a
direction
too
far
and
find
out
that
we
lost
something
along
the
way
and
this
this
idea
of
the
hybrid
approach,
where
we're
taking
different
types
of
network
architectures,
mixing
them
together,
so
that
we
could
get
a
bit
of
the
the
benefits
of
each
is
something
I
think
that
could
be
important,
and
I
wanted
to
point
out
one
other
aspect,
that
kind
of
does
it
that
does
a
great
job
of
of
presenting
future
outlooks
for
quantum
networks.
Is
this
paper?
C
This
is
called
designing,
tomorrow's
quantum
internet.
I
think
this
is
a
very
good
read.
It's
also
promoting
this
kind
of
idea
that
we
have
to
move
away
from
circuit.
Switching
because
you
know
history
has
taught
us
we
we
we
do
better
with
packets,
which
depends
on
some
aspects,
of
course,
there's
of
course,
parts
of
networks
that
are
still
circumstance
today,
because
it's
better.
So
it's
always
about
consideration,
and,
but
I
think,
keeping
things
in
mind
for
designing
the
quantum
mechanisms
is
important
to
think
about
now.
C
So
that's
the
end
of
the
talk
and
in
conclusion
I
wanted
to
just
say
these
last
four
points
that
packet,
switching
quantum
networks
could
be
viable
means
for
scaling
quantum
networks
up
burst.
Switching
can
be
this
kind
of
intermediate
stage
thing
where
we
do
in
order
to
be
able
to
implement
some
quantum
network
applications
packet.
Switching
and
quantum
numbers
could
also
lead
to
better
network
utilization
that
we
saw
in
the
past
and
in
the
intermediate
future.
We
could
think
about
these
hybrid
networks,
so
yeah
thanks
evan
for
the
attention.
C
A
All
right,
thanks
steven,
you
wanna
run
the
q
a
or
july.
A
Okay,
good,
let's
see
so
there's
been
several
questions
or
comments
here
in
the
in
the
chat.
I
think
most
of
them
were
actually
answered
pretty
well
by
wojtek.
A
bunch
of
them
were
from
scott
scott.
Did
you
want
to
do
you
want
to
ask
anything
in
that
that
you're
still
unclear
about,
or
you
want
to
bring
up
with
stephen
instead
of
me
and
and
why
dick.
A
Okay,
anyone
else
questions
you
want
to
bring
up
in
via
voice
or,
you
can
add,
add
new
comments
into
the
chat.
A
B
A
Okay,
so
steven
question
from
jessica
in
the
chat
are
the
classical
headers
separated
from
the
quantum
payload.
If,
yes,
how
does
the
coupling
between
the
header
and
the
payload
take
place?
How
do
you
manage
a
payload
or
or
or
a
classical
header
loss?
Now,
that's
a
good
question.
C
A
C
Scheme
we
consider
two
different
ways
to
separate.
Well,
I
mean
they're
separated
and
depends
on
the
multiplexing
scheme,
so
we
separated
using
time
and
we
separated
using
wavelength
so
they're,
not,
for
example,
using
the
same
quantum
signal
we're
not
using
qubits
to
represent
the
classical
header.
We
want
to
do
the
classical
header
in
the
traditional
way.
So
the
I
guess
the
answer
to
that
part
is
yes.
C
If,
yes,
how
does
a
coupling
between
header
and
payload
take
place
so
there
we
again
we
consider
some
multiplexing
scheme.
This
is
kind
of
a
general
thing
right
now.
We
just
want
to
investigate
this
further
because
it's
you
know
we
don't
we
didn't
build
it,
yet
we
don't
know.
What's
the
best
way
to
do
it,
but
the
coupling
is
with
some
kind
of
either
optical,
switching
or
multiplexing
scheme,
where
we
can
can
couple
these
things
under
the
same
fiber.
C
I
think
that's.
If
I
understood
correctly,
that's
what's
the
question
and
how
do
you
manage
payload
or
classical
header
loss?
So
this
is
a,
I
would
say,
application
independent.
I
think
I'm
just
scheming.
I
didn't
notice
the
chat
during
the
talk,
but
what
we
do
for
loss
depends
on
what
the
protocol
is.
C
If
the
header
is
lost,
then
not
much,
you
can
do
because
the
quantum
payload
itself
has
no
information
about
itself.
It
has
to
be
dropped,
but
if
you
drop
the
payload
and
you
have
the
header
and
then
you
can
somehow
observe
that
the
payload
was
lost,
you
might
ask
for
a
re-transmission
if
the
application
allows
for
it.
For
example,
when
you're
using
qkd
the
information
itself
is
represented
in
a
quantum
way,
but
in
a
sense
it's
just
classical
information.
You
could
regenerate
it
and
there's
no
problem
with
entanglements
the
same
you
lose
entanglement.
C
It's
not
like
you've
lost
an
important
aspect
of
information,
but
if
it's
something
like
quantum
computing,
when
you
take
time
to
prepare
the
quantum
state,
then
you
transmit
the
quantum
state
and
it's
lost.
You
can't
just
say
retransmit
that
quantity
you
have
to
run
the
algorithm
again.
Potentially
we
don't
know.
So
that's
something
also
asked
an
aspect
to
consider,
but
of
course
it
depends.
A
A
Certainly
would
certainly
mean
that
you're
encoding,
a
some
logical
qubit
on
a
set
of
photons,
and
presumably
all
those
photons
would
travel
together.
C
Exactly
so
that
quantum
payload
we
tried
to
keep
arbitrary
you
could
you
could
generate
some
kind
of
graph
state.
Do
your
you
know,
air
correction
scheme
transmit
the
whole
state
over
perform
their
correction
scheme
and
then
forward
on.
This
is
something
that's
yeah.
This
is,
I
think,
it's
independent
of
the
protocol,
because
if
we
want
to
do
just
direct
transmission
without
packet,
switching
with
just
third
generation
quantum
repeater
same,
I
guess
the
same
challenge
would
apply.
A
We
actually
wrote
a
paper
on
those
lost
probabilities
and
what
what
are
necessary
some
years
ago,
which
which
we
actually
haven't
gotten
officially
published
yet
in
peer-reviewed
journal
I'll,
put
the
link
in
the
chat
other
questions.
How
are
we
doing
on
time?
It's
52
52
after
so
we've
still
got
plenty
plenty
of
time.
A
C
A
C
No,
like
other
scope,
is
you
know,
what's
out
of
scope
for
quantum
networks,
I
don't
know
it's
hard
to
say
so,
but
I
think
any
any
type
of
scheme
that
is
taking
some
dynamic
routing
approach
or
some
kind
of
package
approach
is
something
feasible
that
we
want.
Basically,
we
want
that
scaling.
However,
it's
achieved
is,
is
the
way
we
go
right.
So
there's
something
to
consider
right.
I
don't
know
yeah,
I
would
say
no,
not
at
all.
A
A
A
Up
while
other
people,
let's
see
we
have
one
here,
okay,
just
thanks
from
patrick.
I
actually
have
one
well
two
comments.
I
guess.
A
Two
weeks
ago
we
were
at
the
workshop
for
quantum
repeaters
and
networks
in
chicago
and
farzam.
Was
there
guy
farzan?
A
What's
prism's
last
name,
I'm
liking
his
last
name,
the
he
works
for
siena,
which
makes
optical
components
and
he's
been
working
on
their
approach
to
to
quantum
stuff
and
he's
adamant
that
the
quantum
channel
is
going
to
have
to
be
run
at
13
10
nanometers
with
all
of
the
classical
traffic
up
around
1550,
otherwise,
you're
not
going
to
get
you're
you're
not
going
to
get
enough
extinction
of
the
the
classical
channel,
even
with
fairly
narrow
filters
on
it.
A
I
know
there
has
been
some
work
on
multiplexing
qkd
with
classical
traffic
into
the
same
fiber,
but
I
don't
remember
off
the
top
of
my
head.
Who
did
it
and
I
don't
remember
what
wavelengths
they
they
were
working
at.
C
Yeah
one
so
one
observation
we
made
like
you
know
this
vision
for
being
so
far
in
the
future.
It
doesn't
restrict
us
from
using
technology.
That's
kind
of
available
today,
but
maybe
not
far
develops
over,
like,
for
example,
I'll
be
specific
about
hollow
core
fiber,
and
probably
some
of
us
have
seen
these
experiments
where
they're
transmitting
classical
signals
at
full
strength
and
doing
qkd.
At
the
same
time,
maybe
you
know
what,
if
the
quantum
network,
what
if
the
entire
internet
was
built
up
hollow
core
fiber
you
know.
Can
we
then
consider
this?
C
A
I've
I've
been
pushing
for
well
not
like
in
a
serious
level,
but
I've
been
pushing
people
over
here
to
think
about
holocore
fiber.
For
years
you
know
they
they'll
go
to
tremendous
effort
to
cut
10
milliseconds
off
of
the
latency
between
tokyo
and
london,
when
they
could
cut
40
or
50
off
of
it,
but
by
going
to
hollow
core
fiber,
but
the
holocore
fiber
has
still
has
loss.
That's
you
know
measured
in
db
per
meter.
You
know
not
not
fractions
of
a
db
per
kilometer.
It's
still
measured
in
db
per
meter.
C
So
yeah
I
mean,
but
it
had
the
prop
the
nice
property
was.
You
can
mix
the
states
and
basically
natural
crosstalk,
but
like
that's,
I
guess
kind
of
a
younger
thing
right,
so
we
will
see
how
that
progresses.
Maybe
that
comes
into
play.
It
could
be
quite
important
for
for
these
kind
of
technology
for
future
quantum
networks.
A
I
thought
bruno's
british
has.
The
experimental
work
that
was
done
was
actually
done
using
cisco
demolition
multiplexers.
How
nice.
C
Well,
this
one:
oh
yeah,
this
one,
okay,
there's
two
there's
two
papers.
At
least
I
know
this
one
and
one
more
that
I
was
using.
I
don't
know
if
they
had
any
any
yeah,
I
have
to
see
how
they
did
theirs
again,
but.
A
Bruno
does
the
paper
say
or
do
you
know
easily
what
the
wavelengths
were
for
the
classical
and
and
the
quantum
channels
and
those
experiments
does
it
say
someplace
readily
accessible.
A
He's
looking,
he
says
all
right,
I
actually
have
you're
putting
just
on
my
personal
hat.
You
know
not
nothing
to
do
with
q
irg
chairing.
I
actually
have
a
request.
The
my
research
group
has
created
a
set
of
icons
for
node
types
for
for
quantum
networks
and
we
would
love
to
have
cisco
adopt
them.
I've
been
talking
to
people
from
other
research
groups,
both
in
the
us
and
europe,
and
we've
gotten
we've
gotten
some
who
have
said
that
they'll
use
them
and
I
think
it
would
be.
A
I
think
it
would
be
a
big
win
if
we
all
sort
of
standardized
on
one
set
of
icons
will
make
it
a
lot
easier
to
share
information.
C
Yeah,
if
you
send
me
the
link
I'll
start
using
them,.
A
A
I'm
sure
that's
over
standard,
fiber,
okay,
standard
single
core
fiber,
I
presume.
A
All
right,
it's
one
minute
to
the
top
of
the
hour
and
people
are
already
starting
to
drop
off.
I
know
I'm
sure
a
lot
of
people
have
their
next
meetings
or
whatever
stephen
we
should.
We
should
all.
Thank
you
for
the
for
this
presentation.
I
thought
it
was
really
fantastic.
I
learned
quite
a
bit
about
what
you
guys
are
doing
and
the
way
you're
thinking
about
it.
B
No
thanks
a
lot
stephen,
that's
about
it!
We'll
try
to
keep
up
with
these
seminars.
I
think
two
times
here
seems
about
the
right
frequency,
so
I
will
keep
you
all
posted
on
the
mailing
list
and
if
any
of
you
have
any
more
questions
or
discussions,
please
just
continue
on
the
mailing
list,
we're
all
on
there,
and
so
it's
stephen
and
so
are
all
of
you.
Hopefully,
so
you
can
always
just
continue
that.
A
Looking
forward
to
seeing
all
of
you
all
in
tokyo
in
march,
colin
any
irtf
business
or
anything
you
want,
you
want
to
say,
and
you
want
to
stick
anything
in
before.
We
hang
up.
A
He's
not
responding,
so
I
presume
that
means
no.
Yes,.