►
From YouTube: IETF95-RTCWEB-20160405-1620
Description
RTCWEB meeting session at IETF95
2016/04/05 1620
C
D
Face
will
be
starting
in
just
a
moment.
So
if
you
can
find
your
seats
find
your
power,
we
heartily
recommend
you
stay
on
this
side
of
the
black
line
for
two
reasons:
one,
if
you
will
notice
all
of
the
AV
equipment
is
pointing
this
way,
so
you
will
actually
here
if
you
stand
on
that
side.
Second,
all
the
people
on
the
other
side
are
going
to
be
required
to
pay
for
the
black
wall
between
the
two
of
them.
D
E
F
All
right:
well,
this
is
the
RTC
web
working
group,
but
if
your
think,
you're
supposed
to
be
someplace
else
is
probably
good
time
to
leave
next
slide.
This
is
the
high
level
overview
of
what's
going
on.
It's
us
where
the
chairs
I'm
Sean,
that's
Ted,
it's
colin
behind
me,
Alyssa
Cooper
is
our
illustrious
ad
in
the
front
row
samee
deco
going
on.
We
might
have
some
remote
participants
and,
if
you're
presenting
make
sure
to
stand
in
the
pink
box,
we
have
a
red
button.
Just
for
you,
we
have
a
an
XMPP
back-channel.
F
F
That
Sean
thanks
the
bra
everyone,
who's
provided
comments
on
the
list,
github
issues
etc.
Thank
you
very
much.
I
guess.
The
next
thing
we
need
to
do
is
thank
Barry
who's.
Our
outgoing
area,
one
of
our
outgoing
arc
area
directors.
I,
do
not
see
him
here,
he's
not
gonna
come
well.
We're
gonna,
give
them
a
beer
the
next
time
we
see
him.
So
all
right.
A
F
We
drink
it.
This
is
the
note.
Well,
you
should
have
all
seen
this
by
now.
Note
it
well
read
it
there's
a
bunch
of
links
in
there
about
the
things.
Can
you
do
that
the
gist
is
that
things
are
going
to
be
recorded,
both
audio
and
visuals,
so
make
sure
if
you've
got
a
statement
that
you
understand
what's
going
on,
and
you
don't
want
to
follow
this,
you
need
to
go
next
and
here's
our
agenda
I
actually
uploaded
a
new
version.
F
That's
got
some
links
into
the
dress
that
actually
related
to
each
of
these,
but
basically,
we've
got
two
days
and
today
we're
going
to
kind
of
go
through
these
and
then
the
on
Fridays,
the
the
remaining
part
of
Jason.
So
we
got
going
to
go
through
some
work
group
status
through
some
feck
and
then
we
do
some
IP
I
just
handling,
and
if
we
have
time
permitting,
we
r
to
do
a
SUV
and
ngls.
D
D
H
I
J
So
so
the
latest
the
latest
was
that
I
missed
the
deadline,
and
you
said
yes,
those
changes.
Look
fine.
Please
have
been
the
next.
The
next
version
of
the
draft.
When
you
get
the
opportunity
I
haven't
had
the
opportunity.
I
will
do
that
right
now.
How
does
that
sound
great
and
then
we
can
just
move
along.
I
Also,
there's
a
couple
of
your
documents
are
in
a
publication
requested
state
for
more
than
a
year.
I
think
they
are
above
the
ones
that
you
are
looking
at
now
in
the
data
tracker,
so
you
might
want
to
go
over
those
as
well
they're
somewhere
in
the
days
checker,
so
don't
skip
ones
that
are
in
that
state.
Please.
H
Okay,
okay,
we
might
have
to
come
to
back
to
those
in
a
second
audio
codecs
shut
data
channel
one
of
the
many
ones.
That's
in
the
RFC
editor,
with
missing
references
or
four
hundred
and
fifty
three
days.
H
There's
I
have
a
draft:
that's
not
meant
to
be
an
hour
savings
just
to
draft
ginning
steps
or
something
where
you
can
see.
All
the
dependency
is
in
interdependencies
of
the
drafts.
Much
of
our
stuff
falls
back
to
being
blocked
on
all
of
it
being
done
at
once,
at
a
certain
degree,
so
I
forget,
which
ones
exactly
the
data
channels
blocked
on
it.
H
Maybe
the
security
which
is
blocked
on
the
other
ones,
but
we
there's
a
bunch
of
documents,
they're
just
blocked
on
references
and
will
probably
continue
that
way
until
some
of
the
long
until
the
ice
straps,
including
the
ice
trickle
drafts
complete
until
the
chase
updraft
completes,
which
the
you
see,
those
offers
hurry
up.
You.
B
On
the
data
channel
kind,
a
Christian
grows
with
regards
to
the
data
channel
draft
I
think
it
falls
down
to
the
a
ctp
end
data
as
being
one
of
the
critical
drafts
and
according
to
the
tsv
session.
This
prior
to
this,
that
should
be
going
to
working
great
working
group
last
call
in
a
couple
of
weeks,
so
hopefully
get
some
movement
on
that.
F
Subscription
so
so
the
security
drops
the
security
drafts.
You
remember
last
time
we
did
kind
of
like
a
foot
scrape
through
the
outstanding
comments,
and
basically
it's
trying
to
get
some
very
busy
people's
time
to
do
some
updates,
I
guess
when
I
talked
with
him
last,
they
basically
said
we're
like
ninety-eight
percent
of
the
way
there
we
dated
a
couple
of
tweets,
we're
not
going
to
rewrite
the
entire
draft
based
on
comments
from
I,
guess,
apps
area.
So
what.
K
I'm
a
bad
person
and
they
should
have
worked
on
this
harder
on
that
said,
I'm
not
going
to
so
at
least
not
probably
first
of
our
months.
So
on
someone
I
mean
it's
not
a
North
accion
Israel
to
be
modest,
but
it's
also,
I
just
don't
have
time
for
it
right
now.
So
if
if
someone
would
like
to
volunteer
to
like
help
with
putting
up
some
last
open
issues,
I
be
helpful
than
that
it'll
be
like
sometime
around
Berlin.
F
D
F
J
So,
while
you're
mucking
around
mum,
Thompson
fluffy
asked
me
to
have
a
look
at
some
of
the
identity,
sections
of
these
documents
and
I
did,
and
unfortunately
I'm
too
close
to
all
of
that
and
I'm
unable
to
see
any
of
the
problems.
That
I
would
expect
that
a
no
doubt
there
yet
so
I
need
I
need
help
on
that
as
well.
So
what.
J
I'm
sure
that
the
Sean
will
do
an
excellent
job,
I
think
that
if
we
can
find
someone
else
to
step
up,
that
would
be
great
too.
D
D
Basically,
you
need
to
go
find
Martin
asked
Martin
what
sections
of
a
particular
graph
to
talk
about
identity
need
review
by
somebody
who
hasn't
looked
at
it.
A
lot
I
think
you
qualify.
Would
you
be
interested
sure,
ok,
right,
Ian's
name
down
is
another
pressed
into
service,
possibly
not
actual
volunteer.
Thank
you
very
much.
G
D
D
Interestingly
enough,
the
state
tracker
doesn't
have
a
please
stop,
bothering
me
state,
but
if
it
did
surely
we
would
be
the
highest
user
of
it
among
current
art
working
groups,
but
this
this
went
back
into
a
last
call
after
a
small
number
of
editorial
and
not
so
editorial
changes.
So
it's
in
working
group,
/,
IETF
last
call
so
far
everything's
going
swimmingly
and
we
certainly
expect
it
to
continue
so,
at
which
point
it
will
return
to
its
current
state.
H
F
H
Mean
just
as
a
point
of
sort
of
I
mean
like
you
look
you
see
all
this
stuff
is
not
done,
but
just
in
a
point
of
the
stuff
that
it
has
gotten
that
when
you
go
and
like
look
at
the
I
mean
in
the
list
of
drafts,
we've
been
tracking
as
part
of
this
working
group.
H
I'm
were
up
to
20,
published
RF
seeds
and
not
directly
from
this
working
group,
but
the
pen
of
the
WebRTC
work
as
a
whole,
we're
in
a
bunch
that
are
a
past
last
call
like
probably
over
half
another
20,
maybe
and
another
you
know
10
or
so
that
are
not
yet
at
working
group
last
call
so
we're
closing
in
on
the
end.
Here,
okay,
we
can
see
the
light
at
the
end
of
the
tunnel.
H
Okay,
it
should
be
brief:
I
managed
to
misspell
or
miss
arrange
the
words
in
the
slide
title
of
the
actual
draft
name
yep
all
right.
So
don't
look
for
that
document
name
anyway.
Next,
so
Magnus
gave
this
a
very
thorough
review,
thanks
very
much
I
fixed,
basically
the
things
he
found
as
well
as
some
of
the
remaining
to
do
is
in
the
document.
Others,
one
of
the
things
that
was
missing
was
actually
what
the
overhead
is
of
using
fact.
H
The
internal
Oprah
spec
we
found
out
about
twenty
to
thirty
percent,
which
is
actually
much
better
than
the
other
possible
mechanisms.
So
it's
noted
in
the
document.
I
also
mentioned
some
the
limitations
of
the
Opus,
but
in
effect
mainly
that
you
can
only
protect
the
previous
audio
frame,
you
can't
do
multi
packet
fact
with
the
built-in
opus
faq.
So
if
you
want
to
do
that,
you
wrote
back
to
RFC
2198
style
fact.
H
Also
Magnus
asked
for
discussion
of
AMR
m-ary
band
because
those
are
things
in
their
recommended:
audio
codecs
for
interop
document,
which
is
a
working
group
document,
so
those
are
now
covered,
as
well
as
the
various
mechanisms
that
exist
for
for
AMR,
and
also
now
that
we
have
flex
pack
which
doesn't
require
the
SSRC
be
indicated
in
stp.
This
is
indicated
within
the
actual
payload.
We
no
longer
need
to
include
discussion
of
SSRC
group
and
a
equals.
Us
is
actually
attributes
in
this
document
and
it
describes
how
I
expect
we
use
for
the
demultiplexing.
H
Next,
so
open
issues
really
they're
very
few
there.
The
one
that
still
remains
in
the
document
is
what,
if
anything,
do
we
need
to
say
about
signaling
support?
For
fact,
in
a
mar
y
band,
morena
rmr
there
are
a
bunch
of
various
attributes
that
are
fine
in
a
more
document,
but
there
doesn't
seem
to
be
anything
that
is
like
on/off
switch,
that's
similar
to
opuses
use
in
band
feck.
So
I
want
to
sort
of
get
people
who
knew
more
about
any
more
to
comment
on
this.
So.
N
H
Okay,
well,
if
that's
the
case
that's
mandatory
to
implement,
then
they
know
we
don't
need
any
text
here,
which
is
what
I
was
hoping
there's.
Also
this
max
red
parameter
and
AMR
originality
to
indicate
like,
what's
the
maximum
depth
of
the
fact
that
we
should
support
that
seems
like
that
could
be
arbitrary.
We
don't
need
to
make
any
statement
here
this
document,
so
I
recommend
just
not
saying
anything.
Do
you
agree.
H
H
H
N
N
H
The
doctrine
says
for
mr
says
that
if
no
value
is
applied,
basically
it's
senators,
discretion
yeah
and
that
really
feels
like
that's
the
appropriate
thing.
I
mean
we
could
have
some
note
saying
that
if
you
receive
a
value
of
zero
for
max
red,
then
don't
send
fact,
but
that's
already
kind
of
redundant
in
some
ways.
So
if
you
want
to
investigate
and
report
back
I
I.
N
D
So
clearly,
it
would
be
nice
to
have
a
couple
of
focused
reviews
of
this
before
we
take
it
to
working
group
last
call
staring
out
into
the
number
of
people
who
are
staring
at
their
screens,
I'm
thinking
Tim.
Could
we
ask
you
to
do
it
sure,
okay,
so
Tim
mow
and
moe,
one
more
who
knows
another
than
a
bit
and
anybody
else
want
to
actually
raise
their
hand
and
not
force
me
to
call
her
oh,
come
on
now.
H
D
Oh
sure,
just
second
I
wanted
to
remind
those
of
you
who
are
on
the
left
side
of
the
room
from
our
point
of
view
that
you
have
put
yourself
into
the
dead
zone
from
the
point
of
view
of
where
all
of
the
speakers
are
pointing.
We
also
would
like
to
let
you
know
because
you're
on
the
opposite
side,
you
will
have
to
pay
for
the
wall
between
yourselves
and
the
people
on
the
right
side.
We
apologize.
H
Ok,
on
from
faq
on
to
handling
of
IP
addresses
this
time,
I
actually
got
the
document
name
right.
Next,
ok,
so
for
those
of
you
reading
email
and
we
discussed
in
Yokohama,
this
is
the
overall
goal
we
were
trying
to
cheap
at
this
document.
There
are
a
bunch
of
different
goals
that
we
were
trying
to
sort
out.
The
priority
of
the
main
thing
is:
prevent
I
dropped
web
pages
from
drive
by
gathering
addresses
using
it
for
identifying.
H
If
you
were
on
a
VPN,
we
also
had
a
very
strong
sort
of
goal
to
make
sure
that
at
least
England
this
is
on
by
default.
I
didn't
degrade
after
performance
I
in
actual
railroad
cases,
and
we
actually
roll
this
out
by
default
in
chrome
recently
and
found
that
it
did
not
in
fact
degrade
performance,
which
was
which
was
great
and
also
give
users
control
that
if
they
didn't
like
the
defaults
and
when
you
did
additional
privacy,
you
know
behavior.
There
is
well
defined
ways
of
what
they
could
actually
do
that
next.
H
Okay,
so
I'll,
let
you
read
this
for
a
little
bit.
The
main
thing
is
the
old
the
old
tax
base
and
made
a
bit
of
a
hand
wavy
thing
about
how
you
know:
access
to
the
actual
video
camera
information
is
more
sensitive
than
the
IP
address
information
and
therefore
I.
You
know
the
just
coupling
you
know,
therefore,
should
follow
directly
from
that
on.
H
Justification,
yes,
it
is
perhaps
the
one
slightly
salient
point
here
would
be
the
fact
that
perhaps
like
the
it
gives
my
advice
to
user
agents
that
they
should
indicate
they
are
going
to
in
the
prompt
for
accident
for
permissions,
mention
the
IP
gathering
stuff,
so
I
think
that's
spelled
out.
Yes
in
the
second
sentence,
that
we
say
the
permission
granted.
You
probably
also
make
mention
this,
but
if
it
is
worded
sort
of
carefully
not
that
it
gives
the
fair
amount
of
discretion
right
right
so
with
Cullen
Jenny,
so
as
individual,
I
did.
H
This
may
be
a
w3c
issue
or
even
a
non
standards
issue
a
UI
type
issue,
but
since
all
of
these
need
to
get
it
sort
of
accessible
privacy
thing,
I
just
want
to
sort
of
say
I
mean
I
can
hop
the
interest
to
see
how
this
is
all
phrased.
But
if
it's
like
it
is
right
now
where
somebody
asked
to
use
my
video
camera
and
then
that
grants
something
that
people
thinks
reveals
a
lot
about.
My
location,
I,
just
don't
think
the
argument
that
your
video
camera
is
more
sensitive
than
your
location.
H
Therefore,
if
you've
got
permission
for
the
video
camera,
you'd
get
permission
for
the
location.
I
don't
buy
that
argument.
If
the
grant
is
for
video
and
location
and
microphone
all
combined
in
one.
Yes,
no
objection
to
that.
That
type
of
idea
at
all,
but
I
think
that
we
need
to
somehow
make
sure
that
the
permission
being
granted
relates
to
the
attack.
That
is
changed
by
whether
that
grants
permitted
or
not
right.
So
I
think
that
the
new
text
should
address
exactly
that
concern.
B
H
J
I
Unless
it
could
wear
as
an
individual
I
think
it's
like
such
a
slippery
slope
to
start
talking
about
this
in
this
draft,
because
I
mean
you
the
way
you
can
read.
This
is
actually
not
that
anything
would
change
about
the
prompt
at
all.
It's
just
that.
You
know
what
this
is
saying
is
that,
implicitly,
when
you
accept
a
prompt
that
asks
you
about
the
camera
and
microphone,
most
people
are
willing
to
accept
the
addition
of
their
local
addresses
as
well
right.
A
E
I
H
I
mean
that
the
jacketing
does
sort
of
consider
the
whole
purpose
of
those
documents.
The
basic
consider
these
cases
so
III
think
the
doctrine
does
you
know
least
try
to
consider.
You
know
that
the
balance
of
trying
to
be
you
eat
out
here
of
you
know
how
we
can
sort
of
hassle
neither
understandable
and
also
provide
the
privacy
protection
that
we
desire
on.
H
You
know
it's
not
trying
to
be
super
prescriptive
on
exactly
what
you
I
is
put
in
front
of
user,
because
I
think
that's
really
outside
that.
You
know
the
scope
of
this
crew,
but
as
far
as
like
making
it
clear
that
it's
not
ideal,
I
certainly
think
the
new
text
is
better
than
the
old
text
and
you
know
perhaps
are
some
things,
the
new
text
we
would
want
to
consider
it.
You
know
wordsmiths
more,
but
I,
think
it
does
try
to
sort
of
explain
that.
H
Yes
is
not
a
simple
thing
like
yes,
if
you're
gonna
give
video
that
definitely
means
that
it's
okay,
to
give
you
know
your
network
information,
and
you
know
we
could
sort
of
mitigate
exactly
what
does
expanding
the
consent
to
closer
than
the
network.
Question
Korean
exactly
mean,
but
I
think
I,
don't
think
we
could
be
super
prescriptive
so
at
this
life,
so.
D
Would
you
be
okay
with
saying
explicitly
include
I
think
the
problem
that
people
are
having
is
that
it
is
an
implicit
inclusion
in
some
of
these
cases
and
that,
as
a
result
of
which
they
don't
believe,
people
understand
what's
being
consented
to?
If
it
said
explicit,
would
you
be
okay
and
would
they
be
okay,
but.
D
We
said
here
that
this
permission
grant
expanded
to
explicitly
include
consent
to
allow
WebRTC
to
so
that
the
consent,
change
on
camera
and
microphone
to
camera
microphone
network
location
was
an
explicit
part
of
the
consent.
However,
that's
presenting
you
I,
that's
outside
of
our
thing.
Would
that
change
your
opinion
of
where
we
are,
but.
I
D
H
H
Video
and
not
my
location
right
so
like
in
one
of
the
ways
that
we've
tried
to
approach
this
is
by
having
the
other
modes
that
people
who
knowingly
are
in
those
cases
could
use
to
have
that
specific
control
like
that,
no
information
that
would
perhaps
being
location,
sensitive,
would
ever
be
given
out,
but
whether
that
could
be
done
like
whether
whether
that
can
be
intimated
in
a
prompt,
I
think,
is
really.
I
got
unresolved
question
yeah.
H
H
The
rest
of
the
system
looks
like
and
we're
going
to
need
to
revolve
that
along
a
little
bit
and
see
it,
but
if
it's
a
system
where
no
saint
user
would
think
there
was
any
location
thing
here,
there
then
I
don't
think
we're
protecting.
Anything
in
this
solution
is
inappropriate
because
either
you
don't
care
about
protecting
anything,
in
which
case
it
should
just
be
a
different
mode
or
you
do
care
about
protecting
it,
in
which
case
this
fails
to
protect
it.
H
So
I'm
just
sort
of
flagging
that,
as
we
see
how
the
whole
solution
I
mean,
I,
think
the
draft
of
text
and
most
three
and
four
the
fact
that
browsers
are
providing
api's
for
things
that
definitely
know
they
want
to
be
in
a
certain
mode
like
tor
browser.
Tor
button
are
great,
but
whether
the
default
in
stream
privacy,
whether
this
works
or
not,
is
hard
to
say
without
understanding
what
the
prompts
going
to
look
like
and
and
the
rest
of
the
pieces
right.
I
guess:
I'd.
H
The
remit
of
this
working
group
to
be
designing
UI
for
browsers
totally,
but
evaluating
whether
this
works
or
not
to
meet
the
privacy
idea
is
within
the
context
of
the
system
is
going
to
be
used
in
which
includes
that.
That's
all
I'm,
saying
it's
hard
I'm,
not
saying
you
should
design
the
prompt
at
all
sure,
I'm
saying
we
need
to
understand
that
pump
a
little
bit
if
the
prompts.
What
that
prompt
look
like
would
greatly
influenced
this
discussion.
G
Other
sites
and
given
a
lot
of
thought
to
and
has
generally
come
down
on,
saying
no,
you
can't
dictate
to
what
the
you
I
should
be.
You
will
you
have
to
write
a
bison
on
what
you
want
the
UI
to
accomplish
and
I
think
that
and
point
one.
We
should
heed
that
and
say
what
we
want
to
accomplish,
rather
than
the
UI
should
look
like,
and
by
2
and
the
details
of
the
one
level
detail
down
should
be
in
the
eye
in
a
WC
documents,
not
in
the
IDF
segment.
K
I'm,
thank
you
a
little
surprise
for
having
this
discussion
so
sick.
Well,
okay,
six
months
ago,
we
were
like
all
totally
cool
with
just
excellent
rating,
all
this
data,
any
piece
of
JavaScript
on
the
web,
and
so
so
we're
now
discussing
the
conditions
under
which
we
will
not
expose
right
it
also
at
the
same
data,
which
we
were
totally
cool
with
six
months
ago,
like
exactly
what
level
of
inform
user
consent
we
want
easier
to
give
on.
K
Similarly,
when
we
discuss
this
last
time,
I
think
there
was
an
uneasy
agreement
that
they're
really
user
not
way
willing
to
spawned
a
more
than
one
product
and
that
no
sane
person
could
possibly
have
explained
to
them
like
this
is
going
to
reveal
the
IP
address
in
some
problem
to
make
any
sense
of
it.
I
mean
fine.
If
there's
some
?
they
click
to,
and
then
along
would
like.
He
doesn't
create
a
bailment
and
advising
liability.
K
It
also
says
there
goes
your
IP
address
but
like
it's
clearly
not
going
to
go
in
the
front
deck,
so
maybe
people
have
some
material
objection
to
warn
those
two
statements
and
think
that
the
system
should
behave
differently.
If
so,
then
great,
if
the
issue
is
that
the
rationale
it
just
happens
to
club
here,
for
why
this
quite
yucky,
but
nevertheless
I
think
a
reasonable
compromise
is
reached.
That's
like
really
quite
different,
and
if
that's
the
objection
then,
let's
just
was
a
good
move.
K
This
text
entirely
and
say
we
recommend
that
I
was
I
just
fine.
By
the
way
it
should
be
clear,
but
if
robbery
litigated
from
the
next
hour
as
every
simply
move
this
text
entirely
and
say
you
know,
browsers-
are
encouraged
to
to
get
some
obtain.
Some
method
of
user
can
send
their
discretion
before
before
moving
in
the
mode
three
or
four
whatever
it
is.
K
I
will
say
that,
basically,
from
my
perspective,
whether
that
consent
is
important
or
not,
any
dialogue
whatsoever
has
a
very
substantial
insurance
act
on
the
drive
by
on
case
that
we
are
most
concerned
with
here.
So
I
think
rattling
on
whether
the
user
like
it
knows
exactly
what's
going
on
when
there's
no
hope.
Actually,
this
venom
is
going
on.
It's
really
quite
silly.
B
J
Here
I
was
going
to
suggest
exactly
the
same
thing.
Perhaps
this
is
it
how's?
This
is
far
too
much
detail
and
far
too
specific.
Now
it
may
be
that
Colin
wants
to
have
us
have
this
level
of
specificity
so
that
we
can
perform
the
analysis
that
he's
looking
to
perform,
but
I
don't
see
I,
don't
need
to
see
that
I
would
just
like
to
see
this
a.
H
J
Making
the
point
is
that
the
the
browser
has
received
some
signal
from
the
user,
that
this
is
not
this
is.
This
is
an
exceptional
website,
in
the
sense
that
you
know,
in
this
case
you've,
given
them
access
the
camera
microphone.
That
is
an
extraordinarily
exceptional
thing,
sure
right
on
the
web,
and
so
that
is
that
is
a
signal
that
the
browser
can
confuse
to
interpret
as
saying.
Well,
maybe
these
guys
can
get
a
little
bit
more
than
the.
H
Weather
fault
right,
and
so
that's
the
ballot
about
driving
force
here
and
I
think
that
the
problem
we're
having
is
that,
since
that
covers
like
ninety-nine
percent
of
the
overall
sort
of
intent,
does
a
camera
the
case
where
I
did
one
percent?
Where
someone
says
oh
well,
I
don't
want.
I
wanna
use
this
app,
but
I
don't
want
to
give
away
up
my
location,
information
and,
like
you
know,
I'm
not
sure
that
will
ever
be.
I
probably
can
perfectly
resolved.
H
O
Mosin
at
ease
up
I'm,
not
really
seeing
how
difficult
this
is,
because
I
look
at
today,
I
pop
about
a
Firefox
page
in
and
start
web
RTC.
It
gives
me
you
know:
do
you
consent
to
share
your
microphone
and
fills
me
drop
down
of
what
microphone
so
a
little
camera
and
a
drop
that
book
I
couldn't
the
third?
O
Both
the
network
addresses
all
our
own
primary,
it
seems
so
simple,
and
even
if
you
don't
allow
selection
or
explicit
consent
to
be
combined
at
least
having
notification
that
when
you're
showing
these
things,
you
will
also
share
your
network
address.
That
seems
to
me
to
be
a
simple
thing
that
most
users
would
understand,
and
it's
not
rocket
science
for
use.
We
figure
out
what
they're,
what
they're
doing.
D
Ted
Hardy,
individual
and
no
doubt
heavily
cognitively
loaded,
sorry
about
that,
so
I
just
agree
pretty
strongly
with
Martin
Thompson's
characterization
of
this,
and
I
think
I
agree
far
more
strongly
with
eckers
and
I
think
the
problem
with
this
isn't
that
we're
you
know
desperate
to
go
and
get
people
you
I
it
clue,
because
we
don't
have
it
to
spare
truly
I,
don't
think!
That's
it.
D
What
we're
trying
to
say
here
is
that
in
both
this
one
and
that
one
you're
making
a
parallel
ISM
that
says
the
permissions
grant
for
camera
and
microphone
is
related
to
the
permissions
grant
for
the
IP
addresses
which
are
not
going
to
be
used
and
the
information
which
therefore
leaks
about
you,
and
that
is
actually
not
really
the
the
attempt
in,
in
the
left
hand,
side
to
say,
they're
much
more
sensitive.
That
I
think
you're
very
right
to
focus
on
other
things
than
to
try
and
expand
it.
D
D
We
we
use
only
a
specific
set
of
addresses
or
or
resolver
or
or
turn
servers,
and
those
are
therefore
revealed
and
then
three
and
four
we
provide
the
api's
for
other
things,
because
this
relationship
between
the
two
kinds
of
information-
just
isn't
there
and
I,
believe
that
what
alyssa
was
saying
earlier
when
I
finally
understood
it
was
if
you're
on
a
chat,
server
and
playing
Oh
medley
or
chat
roulette
or
sex
camming
or
whatever.
You
may
very
well
be
willing
to
show
somebody
your
face
or
your
genitals
and
not
be
willing
to.
D
Let
them
know
that
you're
in
the
dorm
room,
three
three
blocks
away
right,
it's
a
very
different
thing
and
if
you,
if
you
try
to
relate
them,
you're
going
to
always
be
in
this
muddy
moment
and
I
think
not
trying
to
relate
them
and
simply
exposing
in
in
the
modes.
What
will
happen
is
is
fundamentally
better
off.
So
I
would
personally
rip
this
all
out.
So
Ted
I
the
doctrine.
I
D
H
So
okay
I
mean
really
take
another
shot
of
this
say
we
said
that
being
like
mode
2,
1st
Armored
long,
because
it
provides
these
latency
advantages,
but
also
is
behind,
in
least
some
consent
rather
than
zero
consent,
which
is
really
the
drive-by
case
like
is
that
going
too
far,
I
mean
I
would
be
happy
if
I
mean
I,
basically
returning
the
text
that
was
more
similar
to
the
original
texts
of
striking
the
whole.
You
know
sort
of
rationale
about
well,
it's
more
sensitive.
It
would
basically
be
the
old
text.
I
Nobody
believes
that
there's
any
reasonable
way
to
obtain
informed
consent
about
this.
Then
you
shouldn't
say
anything
about
consent
and
I.
Don't
think
I
mean
other
than
mo,
who
just
got
up
here
and
said
that
I,
don't
think
anybody
actually
believes
that
you
can
feasibly
explain
this
to
someone
in
this
context
and
have
them
agree
to
it
and
have
it
be
meaningful.
I.
P
H
P
H
H
It's
it
seems
like
of
the
various
we
consider
various
options,
and
this
was
the
least
bad
option,
and
you
know
I
think
it's
also
very
workable
option.
You
know
we
actually
saw
you
know,
drive
by
usage
of
harvesting
go
down
substantially
in
chrome,
wants
they
actually
did
this
so
I
feel,
like
you
know,
if
we,
if
you
look
at
this
to
carefully,
we
may
actually
be
kind
of
overanalyzing
and
missing
kind
of.
Like
the
you
know,
the
fact
that
I
could
solve
in
the
core
problem
here
I
mean.
I
H
That's
quite
on
the
same
page
with
Alyssa,
okay,
so
modes.
Three
and
four
are
great.
You
get
into
them
with
your
sprint
request,
so
we're
only
talking
about
the
mode
one
or
mode
0
and
whether
a
certain
type,
what
type
of
permission
prompt
might
move
you
to
those
right
and
what's
the
default
behaviour,
the
core
question
one
or
two.
E
K
Okay,
sorry,
oh.
H
Sorry,
the
other
slides
are
drafts,
they're
numbered
differently.
I
understand
why
I'm
confused
now,
okay,
so
one
or
two
so
now
for
applications,
particularly
ones
that
are
like
a
chat,
application
or
peer-to-peer
application
that
don't
have
any
audio
or
video
involved
at
all.
I
mean
the
you
know.
There
is
real
costs
of
this.
Requiring
this
permission,
maybe
there's
a
sport
and
people
will
not
grant
it
because
it'll
just
be
creepy
that
your
IM
applications
act
is
asking
for
access
to
your
camera.
So
there
is
impact
on
doing
that.
H
So
this
is
one
of
the
things
that
I've
wondered
about
on.
This
is
we're
trying
to
solve
a
problem
which,
because
we
can't
explain
it
if
the
network,
if
we
could
explain,
network
location
in
the
prompt
and
the
words
network
location,
might
work
you're,
granting
camera
mic
and
network
location.
That's
three
different
things.
One
thing
or
location,
I
I
be
interesting.
Someone
really
dig
in
deep
and
see
whether
actually
users
could
comprehend
that
as
thing,
but
if
we're.
H
Not
protecting
it
all
from
the
original
problem
that
drove
us
to
this,
or
basically
just
punting
on
it
and
at
the
same
time
we
made
the
situation
worse
or
a
whole
bunch
of
applications
than
where
it
was
six
months
ago
and
so
I'm
just
not
sure.
That's
a
good
privacy
trade
off
on
that
and
I.
You
know,
I,
don't
know
where
we
can
go
on
them.
I
think
we
we
need
to
do
it
now.
H
J
J
H
J
I
shouldn't
really
respond
to
that,
but
I
will
know.
J
The
obvious,
the
obvious
counter-argument
there
is
something
I'm,
not
I
shouldn't
say
so,
that's
all
the
example
is
seen
that,
although
contrived
as
they
are
sort
of
run
caliber
to
that,
if
I,
if
I,
think
that
I
can
control
this
permission,
that
I've,
given
you
to
my
video
camera
that
points
at
the
wall
and
never
ever
points
at
anything
else
or
one
that
has
a
bit
of
tape
over
the
top
of
it.
That's
very
different
to
teething
access
to
surprisingly
something
I'm.
H
K
So
I'm
tired,
I'm
sure,
you're,
tired
and
so
I'd
like
to
see
if
we
can
close
this
on
so
I
think
we're
all
there
Suzy
widespread
agreement
that
the
technical
design
proposal
that
the
following
statements
are:
okay,
namely
that
I
browsers
should
it
the
president,
would
if
they,
if
they
do
short
browser,
should
be
allowed
to
go
in
the
mode
one
if
they
do
this
user
consent
for
the
camera
microphone,
namely,
this
is
user
interaction.
It
may
be
the
case.
K
The
browser
should
also
be
allowed
to
go
into
move
on
without
doing
that
or
not
so
I
guess
what?
What
on
the
primary
purpose,
as
far
as
I
can
tell
of
on
this
material
here
that
were
debating
is
two
things,
one
to
say
what
browsers
can
and
cannot
do
on
if
necessary.
The
second
is
to
rationalize,
why
we
wait,
say
they're
kind
of
kind
of
do
it
on
my
senses,
the
people
in
the
room
with
larger
vok,
saying
essentially
nothing
here
or
either
giving
browser
digression.
K
Do
they
wanted
or
on
or
alternatively
saying
the
browsers,
you
know,
might
construe
door
might
consider
a
user
her
action
prior
to
giving
access
to
mode
one
on.
Do
you
think
we
need
assisting
stronger
than
that,
or
do
you
just
think
I
mean
what
is
that?
What
are
the
the
normative
forces?
What
you're
trying
to
get
out
of
here
or
you
just
you
just
think
people
wrong,
and
so
you
want
to
argue
with
them
which
isn't
unreasonable
at
all
since
I.
Do
that
all
the
time.
H
That's
right,
good
man.
To
some
extent,
this
will
come
back
to
the
RFC
2119
discussion
that
we're
gonna
have
later
in
this
document
of
like
what
is
exactly
this
document.
You
know,
which
was
originally
intended
as
like
a
best
practices
document
and
now
is
being
so
recast
as
a
standards
track
document,
in
which
case
this
actually
does
have
normative
force
so
like
letting
us
assume
for
right
now
that
is
this
inter
strike
and
we
all
have
a
normative
force.
I've,
let
you
this
actually
say
you
know,
I,
guess
we
have
two
options.
H
One
we
say
like.
We
think
that
this
in
an
algorithm
sort
of
explained
here,
like
the
consent
that
would
be
derived
from
like
new
information
should
in
fact
we
use
as
a
hint
to
you
know,
engage
mode
1
702.
We
should
say
that
no,
we
make
no
explicit
recommendation.
You
might
want
to
consider
this,
but
basically
it's
not
the
browser's
full
discretion
to
decide,
and
you
know
with
with
whatever
amount
of
texts
we
wanted
to
say.
H
We
I
wrestled
with
this
issue
at
length
in
like
here's,
how
we
arrived
at
this
conclusion
or
not
they're,
like
those
two
points,
as
you
say,
I
think
our
separate,
like
I
started
doing
on
the
side
of
including
the
rationale
for
posterity,
but
I
I'm,
certainly
open
to
like
new
precedent
on
the
surface.
It
sure.
K
Yeah
face
ba
I
would
be
honestly
happy
either
we
leaving
browsers
that
discretion
or
not
the
embrasure
discretion
on
one
thing,
maybe
I'd
suggest
that
might
make
people
feel
better
about.
K
I
don't
know
worse
than
typical,
the
point
being
that
that
would
mean
any
dialogue
whatsoever
was
acceptable,
even
if
it
said
nothing,
it
was
a
quick,
he
wrote,
you
know,
you'd
make
him
out
the
coments
that
clicker
you
all
right.
Maybe
that
would
satisfy
people
to
know
the
drive
by
while
avoiding
this
order.
Well,
I
kind
of
consider
hair-splitting
about
what
the
sort
of
bigger
implications
are
of
various
hazards
that
right.
H
I
would
I
mean
that
bc
seems
to
make
a
recasting
of
this
text
here
or
the
text
I
was
on
the
slide.
Are
we
totally
fine
with
that
on,
but
I'm
not
sure
I
have
a
good
sense
from
the
room
of
whether
people
prefer
that
or
they
prefer.
Basically,
you
know
running
up,
hands
and
saying
here's
something
you
might
want
to
consider
and
you
know
basically
putting
out
the
stack
to
that
be
through
city.
I'm.
J
Going
to
take
what's
lot,
I
think
I'd
like
to
retain
some
ability
to
have
browser
discretion.
On
this
point,
we
may
discover
that
there
are
better
ways
to
acquire
signals
necessary
to
move
us
from
two
to
one,
and
it
may
be
that
we
decide
not
to
do
this
sort
of
thing.
I
want
to
avoid
the
prescription
in
the
document
as
it
stands,
and
I
want
to
avoid
the
implications
of
that
prescription,
which
I
think
probably
the
most
worrisome.
Ok,.
J
H
J
E
H
Well,
actually
expected
that
to
be
at
the
times
when
the
most
that
thing
spent
the
most
time
on
so
like
that
was
actually
fine
due
time
at
the
end,
to
kind
of
resolve
any
sort
of
terrifying
consensus
on
this
we
are
running
out
of
time
effect.
League
will
probably
have
to
do
that
on
friday,
okay,
alright,
so
we
do
new
proxies.
Then,
hopefully,
it's
been
less
contentious.
H
H
Probably
the
most
interesting
thing
is
that
mode
3
provides
most
the
benefits
that
you
would
want
on
when
you're
using
a
return
proxy,
because
the
proxy
will
still
be
used
for
your
trap.
You
don't
need
to
actually
all
the
good.
Oh
wait
go
to
mode
for
it
actually
force
use
of
the
proxy
and
I'll
show
this
in
some
diagrams
excellent
okay.
H
You
want
to
talk
to
you're,
going
to
end
up
actually
hair
printing,
that
traffic
eat
the
three
green
at
or
through
your
application
turn
server,
because
you
won't
have
that
local
address
that
you
can
use
the
talk
to
that
internal
client,
but
the
only
time
the
trap
actually
goes
to
the
proxy
is
in
mode
for
or
where
you
actually
say,
force
the
proxy
on.
Obviously,
everything
is
different
here.
If
the
application
firewall
are
sorry
that
the
enterprise
firewall
blocks
direct
UDP
out,
then
everything
will
off
the
funnel
through
adapter
and
server
and
the
proxy.
H
H
H
We
would
want
to
use
when
you
have
the
case
where
you
only
want
to
use
give
out
the
address
that
really
outside
people
are
already
seeing,
which
is
the
intent
of
mode
3.
Here
we
would
actually
want
to
use
the
return
proxy
and
no
internal
candidates
in
mode
3.
There's
no
local
candidates
being
exchanged.
Connections
between
local
candidates
would
in
fact
use
the
return
proxy
in
a.
H
C
case
the
traffic
will
actually
go
through
the
return
proxy
as
well.
So
you
know
if
you
compare
between
the
slide
in
the
previous
slides,
will
see
that
the
restricted
gathering
thing
is
a
little
bit
different,
in
which
case
the
return
proxy
will
actually
be
used
for
the
you
know
inside
the
firewall
case.
K
H
Slide
so
moving
along,
this
came
up
the
mailing
list.
What
do
the
words
must
and
should
mean
in
lower
case,
which
this
document
uses
largely
because
I
hadn't
really
taken
the
time
to
consider
exactly
how
I
best
contracts
as
document
should
be
written,
which
spawned
a
very
large
discussion
across
the
IDF
mailing
list,
which
came
to
I?
Think
the
conclusion
that
must
in
lower
case
must
actually
do
mean
different
things
which
in
some
ways
it
with
agonal
to
what
is
the
ultimate
fate
of
this
document
and
the
questions.
H
The
two
questions
to
be
considered
here
are:
is
this
a
standard
trac
document
or
not?
And
if
it's
a
standard
stretch
document
shouldn't
it
be
prescriptive
by
using
2119
language
about
the
recommendations
that
it
makes
on
harrell
proposals
that,
yes,
it
should
be
standards
track
similar
to
like
the
audio
and
video
requirements,
and
it
ought
to
actually
use
2119
language
and
some
examples
on
one
particular
example
I
was
given
was
the
the
topic
we've
discussed
that
nausea
I'm
around
Commission
granting
you
know,
ought
to
be
made
into
2119
language,
so.
D
Before
before,
we
take
the
line,
I'm
going
to
point
out
that
the
meeting
actually
is
now
running
over
time,
so
you're
missing
your
beverage
break,
but
I
have
a
question
for
our
illustrious
area
director,
so
illustrious
area
director,
given
that
this
could
have
been
part
of
j
sep
that
we
not
wanted
to
be
able
to
rev
it
independently.
Are
you
okay,
with
this
being
a
standards
track
document.
I
D
D
List
again,
given
the
previous
discussion
on
consent
and
I
thought,
let's
not
read
too
much
into
these
particular
examples.
I
think
that
the
right
thing
for
us
to
do
is
to
ask
the
author
to
take
into
account
the
obvious
consensus,
to
move
this
to
standards
track
and
more
prescriptive
language
as
a
result,
and
to
come
up
with
language
for
us
to
consider
in
a
pull
request
on
on
friday
or
on
the
mailing
list.
Is
everybody?
Okay
with
that
yeah.
H
Yeah-
and
I
just
want
to
point
a
point
out-
Justin
I
mean
even
even
if
we
never
were
prescriptive
about
which
mode
to
use
win
and
we
just
defined
what
the
modes
were
and
some
other
document
specified.
Those
it'd
still
be
perfectly
reasonable,
abstract
thought
so
sure.
That's
the
independent
decision.
Okay,.
D
I
think
that's
it.
Next,
there
is
a
nice,
oh
yeah.
So
sorry
definitely
they'll
be
publishing
a
new
version
at
some
point,
so
we
will
talk
to
Justin
offline
about
whether
we
might
hope
for
a
quick
pull
request
on
this
most
recent
slide
before
Friday.
Otherwise
will
happen
on
the
mailing
list.
Sorry
for
keeping
you
over
time.
Please
stick
around
for
Friday
more
fun,
then.