►
From YouTube: RATS Architecture Design Team, 2021-01-26
Description
RATS Architecture Design Team, 2021-01-26
A
A
A
B
No,
but
sometimes
people
want
to
adopt
want
to.
Sometimes
people
want
to.
You
know,
see
the
the
result
before
they'll
agree
to
adoption,
and
I
don't
know
how
I
feel
about
that.
So
I
I
feel
it
depends
a
lot
on
the
author.
I
think
that
that
I
think
that
the
zero
zero
should
be
exactly
identical
to
the
the
last
one
before
it
was
adopted.
B
A
A
C
B
A
C
B
B
D
Sentence,
and
so
if
you
wanted
a
low
hanging,
fruit
well
knock
on
wood,
one
at
least
full
hanging
fruit
in
terms
of
the
number
of
text
we'll
see
if
number
of
comments,
but
that
one
is,
there
was
a
comments
in
the
appendix
stuff
where
we
said
gosh,
the
parenthesis
stuff
is
not
even
explained,
and
I
responded
to
hank
saying
yes,
I
agree,
we
should
add
an
explanation,
and
so
that's
what
this
one
sentence
is.
This
is
back
in
the
first
example.
B
D
D
D
B
C
D
F
B
D
B
B
F
F
F
B
F
D
D
Fixed
an
error
in
my
wording,
and
so
the
notion
of
an
epic
counter
that
I
had
in
mind
is
where
your
handle
is
just
assigned
sequence
number,
and
so
what
happens?
As
I
mentioned
the
comment
here,
that's
just
the
placeholder,
not
in
the
green
text,
but
in
my
comment
to
where
yours
used
to
be
the
use
of
a
signed
sequence.
Counter
means,
if
you
ever
receive
a
handle
from
you
know.
Let's
say
a
verifier
sees
a
handle
in
evidence.
Okay
and.
A
D
B
F
A
D
C
D
So
thomas
is
adding
another
example
into
the
ore,
which
I
think
is
great,
which
is
the
buffering
incoming
messages.
You
wait
and
see
if
you're
gonna
get
a
handle
in
the
in
the
near
future.
So
and
you
have
some
maximum
buffering
time
after
which
you
got
to
give
up,
or
maybe
it's
until
you
get
another
one.
So
you
just
have
a
cache
of
one
that
you
don't
know
about
something
like
that.
So
then
you
don't
need
any
clock.
D
There's
nothing
in
a
handle
distributor
in
any
of
the
text.
Here
I
can
generate
three
different
handle
distribution.
Things
that
are
completely
independent
of
prop
delay.
Properly
could
be
a
month.
Everybody
gets
it
about
the
same
time.
The
thing
is
sitting
on
mars
or
sitting
on
jupiter
or
something
so
works
fine
yeah,
but
I
don't
care
what
it
was
you
only
the
difference
between
if
the
epic
changes
every
day,
but
it
takes
three
months
to
get
here
from
jupiter,
then
that's
fine
as
long
as
it
changes
every
day.
A
A
D
D
D
D
I
don't
know
that
it
has
to
be
the
same
length
every
time
as
long
as
it's
larger
than
the
prop
delay
between
the
tester
and
the
verifier,
the
prop
delay
between
the
verifier
and
the
relying
party.
I
don't
think
it
matters
whether
the
first
one
is
15
minutes.
The
next
one
is
30
minutes
the
next
one
is
two
hours
and
the
next
one
is
20
minutes.
D
B
D
B
D
A
D
Say
you're
a
verifier
and
you've
got
a
bunch
of
testers
scattered
around
okay.
The
question
is,
as
you
start
getting
in
evidence,
if
you
get
a
bunch,
if
you
get
like
three
pieces
of
evidence
back
to
back
all
from
different
testers
that
are
scattered
around
the
question,
is
how
many
different
epics
can
you
be
having
how
many
different
epic
candles
that
are
all
correct,
according
to
the
center
of
that
evidence,
and
you
receive
at
the
same
time.
D
Correct
and
so
the
higher
the
number
is,
the
more
ethics
you
have
to
remember
right.
So
do
I
need
to
remember
the
last
one,
the
last
two,
the
last
five,
the
last
ten,
and
so
that's
it's
not
a
clock.
It's
time
when
you
throw
them
out
it's.
I
need
to
remember
the
last
n
that
I've
gotten
from
the
handle
distributor
and
what's
that
and
how
many
of
those
do
I
have
to
save.
A
F
A
D
B
D
His
example
was,
you
could
buffer
it.
My
example
was,
if
it
is
signed,
you
can
just
accept
it
as
if
you'd
gotten
it
from
that
one
another
and
the
other
one
in
that
list
was.
You
could
also
drop
it
and
say
sorry,
this
is
rejected,
and
then
you
got
to
try
again
right-
that's
probably
the
least
good
one,
but
it's
possible
saying
I
haven't
seen
this
one
yet
go
away
and
then
the
intestine
just
has
to
retry.
Well
how
about
now
you
got
it
now.
B
D
I
think
it
does
avoid
the
problem
thomas
and
that's
because
either
you
have
to
block
all
p
all
evidence
coming
in
into
the
verifier
except
attestation
just
always
fails,
and
then
you
don't
know
it
doesn't
matter
whether
it's
handles
or
nonces
or
clocks,
or
anything.
If
you're
going
to
be
able
to
block
all
the
evidence
coming
in
and
you
can
attestation
right
now.
If
you
say
I'm
going
to
let
the
evidence
in
and
you're
going
to
prevent
the
handle
distribution.
D
D
That's
where
I'm
saying
using
a
signed
counter
works
if
you're
only
dosing
it
at
the
verifier.
In
other
words,
if
the
tester
gets
the
right
handle
and
you're
allowing
the
evidence
through,
then
you
have
the
channel
that
goes
to
the
verifier.
It
just
happens
to
go
via
the
tester.
Now
that
what
you
can
do
is
your
point
is:
if
you
could
dos
the
handle
distributor
such
that
it
can't
get
handles
out
to
anybody,
whether
it's
a
testers.
A
D
C
D
Delay
it
for
everybody
collectively
yeah,
because
that
is
not
going
to
be
detectable
if
you
can
deliver
it
for
everybody.
So
if
you
can
block,
if
you
can
insert
a
huge
delay
at
the
handle
verifier
or
at
the
handle
verifiers
router
or
something
like
that,
then
then
it's
undetectable
right
by
people
who
don't
have
clocks.
B
D
Yeah-
and
this
is
the
part
that
the
other
one
needs
to
be
based
on,
which
is
just
that's
the
red
text
at
1423
on
down
that
I
use
as
the
source
for
creating
the
text
in
line
up
above
so
I
deleted
the
the
text
that
I
said
should
move
up
and
replaced
it
with
the
summary
sentence.
Intro
sentence
that
matches
the
intro
sentences
of
examples,
one
and
two.
F
D
A
D
F
D
D
D
B
D
D
D
D
A
D
A
D
A
D
A
A
E
D
A
A
D
D
A
D
A
D
D
D
About
each
layer,
that's
from,
but
the
about
word
if
the
root
of
truth
thing,
if
they
collect
claims
about
a
root
of
trust,
then
right
now
it's
only
talking
about
a
tester
is
collecting
claims,
and
so
the
only
thing
the
tester
that
can
collect
claims
about
the
rooster
frost
is
due
to
trust.
So
that's
that's.
E
E
B
I
guess
it's
also
might
be
worth
re
recognizing
that
this
is
an
architecture
and
it
is
an
in
for
more
information,
informational
document
that
does
not
have
bcp14
language.
So
the
only
case
case
I
see
if
someone
gets
this
wrong
is
you
could
come
along
and
try
to
bash
them
over
the
head
and
said
you
got
it
wrong
and
here's
the
sentence
and
they
may
argue
backwards
the
other
way,
but
ultimately
the
guy
that
want
thinks
he
can
do
that.
Didn't
understand
the
whole
problem
anyway.
B
A
E
A
B
B
D
Look
at
it
because
the
the
other
technical
issue
that
I
had
with
it
was
still
that
discussion.
We
were
having
about
the
prop
delay
whether
the
prop
delay
is
relevant
and
whether
the
handle
transmission
time
is
relevant.
Where
I
claim
it's
not
that's
the
technical
issue,
and
hopefully
from
our
your
discussion,
you
at
least
understand
where
I'm
coming
from
so.
B
D
D
Any
objections
punctuation
wise
it's
odd-
to
have
a
colon
without
either
bullets
or
part
of
the
same
sentence,
so
maybe
include
the
following,
or
something
like
that.
So
that's
a
complete
sentence,
because
usually
anytime,
that
you
have
a
colon
there's
always
has
to
be
something
that
would
be
a
period.
So
if
it
was
a
bulleted
list
for
example,
then
you
know
the
last
bullet
would
be
the
thing
that
has
a
period.
But
here
there's
not
bullets
or
anything.
So.
D
B
D
D
D
D
B
B
B
B
C
D
D
You
don't
have
another
or
maybe
right,
and
so
the
or
even
actually
goes
with
the
verifier
may
and
then
the
even
carefully
secured
list
of
is
after
maybe
provided
with
either.
Then
there's
the
open
parenthesis
goes
between
with
an
a
database
and
if
I
were
to
sentence
diagram
it
right,
maybe
provided.
C
D
Like
that
may
be
provided
with
a
database
of
public
keys
or
the
verifier
may
be
even
before
the
word
provided,
but
yeah,
so
I
don't
know
if
you
need
to
rewrite
it
or
whatever,
but
that
that
that's
why
the
parentheses
need
to
go
away
all
right.
So
now
you
can
delete
those
extra
parentheses
that
were
for
diagramming
the
sentence,
but.
C
E
D
B
B
Yeah,
that's
not
the
one
I
wanted
to
have,
and
then
this
issue
here
almost
the
same
thing
as
the
same
thing,
description
of
mutual
authentication
and
ned
asked.
Should
we
have
this
term
and
I
took
a
a
half-assed
attempt
to
put
it
into
the
into
the
terminology
realized
that
with
the
reorganization
with
the
terminology,
I
had
no
idea,
which
section
to
put
it
in
and-
and
I
felt
it
was
such
a
big
thing
to
add
that
I
just
backed
off-
and
I
don't
know
whether
or
not
we
think
we
need
to.
B
E
D
D
A
Typical
mutual
attestation
are
totally
different
processes.
I
mean
yes,
cell
phones
or
as
a
private
evidence
to
cell
phone
and
cell
phone
is
providing
evidence
to
siphon
tower,
but
your
infrastructure
at
the
back
end
is
totally
different
when
the
the
the
the
trust
relationships
are
different.
The
trust
anchors
probably
even
are,
and-
and
so
so
yes
why
this
might
be
a
requirement-
it's
just
two
times
times
two.
B
So
first,
your
cell
tower
example
is
a
terrible
example,
because
that's
an
example
where
it
mostly
doesn't
happen,
but
anyway
never
mind
that
point.
Okay,
a
sim
card
aka,
was
a
is
a
stillborn
until
we
get
to
5g,
but
the
issue
with
mutual
attestation,
also
with
mutual.
I
think
more
so
than
mutual
authentication
comes
down
to
who
goes
first
right,
and
I
won't
I
I
won't
prove
to
who
I
am.
D
You
can
do
it
well,
one
at
a
time
and
that's
really
the
same
concept
which
you're
talking
about
with
attestation
right
with
privacy
sensitive
stuff.
I
don't
want
to
attest
until
the
other
guy
test
me
first,
so
I
know
that
I'm
only
revealing
my
sense
of
information
to
somebody
that
I
know
who
it
is,
and
so
I
think
it
actually
does
come
up
the
same
issue
about
you
know
the
same
time
or
one
at
a
time.
It
comes
up
in
both
authentication
and
attestation.
They
use
alternatives
so.
B
D
D
E
D
D
B
D
B
Well,
I
I'm
I'm,
I
don't
okay.
So
aside
from
the
fact
that
we've
said
that
there
are
cases
where
you
must
be
more
careful
and
therefore
your
protocol
better
deal
with
it,
we're
not
defining
the
protocol,
that's
going
to
deal
with
it,
so
I
don't
know
where
we
need
to
be
gets
that
far
into
it.
There
there
are
architectural.
B
B
I
think
you're,
alluding
to
you,
know
the
phone
first
as
a
class
attestation
that
it
is
a
particular
kind
of
phone
and
then
the
verifier
says
oh
and
therefore
you'll
trust
me,
because
I
have
blah
blah
blah
and
then
the
phone
will
reveal
some
pii.
After
that
point,
that
sounds
like
a
three-letter
three-way
handshake.
That's.
B
D
D
E
B
D
D
D
B
A
D
For
people
that
haven't
seen
this
pdf
before,
if
you
remember
one
of
our
use
cases
in
the
use
cases
section
is
concrete
competing.
This
is
a
deep
dive
into
the
confidence
competing
use
case
and,
of
course,
one
of
the
sections
is
about
attestation
for
that
and
that's
where
it
references
the
rats
architecture
in
that
section
six,
so
just
rats
architecture
is
more
general
right,
but
this
is
the
special
case
of
attestation.
That
section
six
is
a
special
case
of
at
the
station
four.
D
D
Different
white
papers
from
the
comprehensive
community
consortium,
this
one
is
the
technical,
deep
dive,
there's
another
one,
that's
much
more
higher
level
that
doesn't
go
into
all
the
you
know:
attestation,
details
and
stuff.
It's
like
what's
the
actual
use
case
for
it
as
if
you're
a
you
know,
c-level
person
trying
to
understand
what
it
is
or
whatever
or
you're
a
marketing
person
or
whatever.
Then
it's
very
high
level,
and
so
there's
another
one.
D
That's
actually
slightly
more
readable,
but
this
one
is
for
us
engineers
and
so
usually
engineers,
the
ones
that
read
rfc
is
not
marketing
people,
so
this
would
probably
be
if
we
were
going
to
reference
one.
This
is
probably
the
more
appropriate
one
given
the
audience
of
rfcs
and
it's
the
out
of
the
two.
This
is
the
only
one
that
references
the
rat's
architecture,
because
the
other
one
doesn't
go
into
that
level
of
technical
detail.