►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
I
was
thinking
about
how
we
go
from
the
network
that
kryptonet
lab
was
talking
about
at
the
beginning,
so
having
having
on
chain
escrow
or
staked
amount
of
money,
and
then,
if,
if
I,
can't
directly
get
my
retrieval
I
go
to
the
set
of
third
parties,
have
them
attempt
to
do
it
and
that
if
they
can't,
then
that
ends
up
slashing
out
of
that
and
and
causes
the
the
SP
to
lose
money?
A
And
so
then
there's
this
question
of
like?
Can
you
push
that
to
something
that
that
would
work
in
a
CDN
context?
So
one
thing
you
could
imagine
is.
A
In
in
this
design,
in
the
initial
design,
you
have
that
whole
process
of
establishing
who
these
Auditors
are
as
an
on-chain
thing,
that's
relatively
slow
and
then
also
you
have
a
pretty
heavy
weight
cryptographic
protocol
as
the
way
that
you're
verifying,
whether
that
happens
and
sort
of
have
a
majority
thing
and
the
whole
thing's
pretty
expensive
in
a
number
of
ways.
A
If
you're
thinking
about
what
is
the
economic
thing
that's
happening
here,
though,
a
client
is
trying
to
get
some
data
and
they
can't
right,
and
so
the
thing
that
the
SP
has
gained
is
it
has
saved
the
bandwidth
cost
of
that,
and
so,
if
your
fallback
is
some
other
trusted,
third
party
does
trigger
at
least
that
much
download.
A
It's
now
wasted
twice
the
bandwidth
versus
the
one
to
the
original
client,
and
so
it's
in
the
sp's,
like
economic,
better
interest
to
just
directly
serve
rather
than
do
that,
and
so
you
could
imagine
doing
things
like
saying
you
know
what
we'll
agree
on
is
like
here's,
this
pool
of
you
know
a
few
other
audit,
backups
I'm,
expecting
you
to
be
able
to
serve
at
at
least
10
gigs
or
10
gigabit
per
second,
and
so
like
I
should
be
able
to
see
from
these
Auditors.
A
At
least
you
know
four
Megs
per
second
or
something
like
some
reasonable
thing
on
this
on
my
individual
download
and
then,
whenever
a
client
doesn't
see
that
it
refers
back
over
to
the
Auditors
and
the
SP
has
to
do
it
at
twice
that
bandwidth
or
it
gets
in
trouble.
A
So
you've
got
a
couple,
then
issues
like
what,
if
a
bad
client
is
just
referring
everything
over
to
the
Auditors
like?
Are
there
ways
to
do
that?
And
so
you
can
imagine
at
the
beginning.
You
agree
on
a
couple
things
like
potentially,
you
could
agree
on
a
fraud
rate,
and
so
you
say
if
the
if
these
Auditors
are
actually
seeing
that
it's
all
working,
you
know
most
of
the
time-
and
it
seems
like
most
of
these
are
false
reports
in
that
the
SP
sure
seems
to
be
going
at.
A
You
know
quite
a
bit
of
like
we're,
seeing
a
large
fraction
of
these
tokens
that
have
been
redeemed
or
or
attempts
at
downloads
are
getting
put
over
to
fraud,
but
then
they're
all
working.
Maybe
that
causes
the
Auditors
to
either
give
some
free
passes
and
be
like
hey.
You
seem
to
be
doing
this.
A
This
seems
like
a
client
side
at
this
point
that
we've
identified,
and
so
we
can
agree
and
say
we're
going
to
do
this
fraud
test,
for
you
know
at
most
10
percent
right
so
like
we
can,
we
can
have
a
Max
percentage
of
the
overall
amount
of
downloads
that
we've
agreed
upon
in
this
sort
of
scope
of
contract
that
we
that
we
actually
will
have
the
Auditors
do
now.
A
So
you
need
that
to
still
that
that
percentage
needs
to
be
offset
by
an
additional
stake
for
the
SP
somehow,
so
the
SP
is
sort
of
like
getting
back
to
a
neutral
round,
but
it's
not
getting
the
upside
if
it
ends
up
in
that,
so
there's
potentially
a
way
where
the
SP
has
higher
upside
by
serving
to
the
direct
clients,
but
both
by
the
2x
audit
cost
and
then
by
potentially
only
getting
back
to
some
baseline
in
the
audit
case.
A
You
know
the
the
case
of
a
bad
client
that
you
made
the
deal
with
ends
up
being
sort
of
like
a
wash
to
you
as
an
SP
versus
a
win.
If
you
get
a
good
client
is
one
way
that
you
could
think
about
how
that
like
there
should
be
enough
flexibility
in
your
payment
options
there
so
I'm
interested
in.
Can
you
get
down
those
various
costs?
A
For
that
part,
where
you've
got
a
prepayment
and
agreed
upon
set
of
terms
a
pretty
lightweight,
not
fully
cryptographic
verified,
but
in
terms
of
the
amount
of
where
bandwidth
is
going
to
happen,
it
still
ends
up
incentivizing,
the
correct
behavior
from
everyone,
and
it
seems
like
there
might
be
something
to
play
with
there.
B
You
will
not
understand
anything
strong,
so
our
group
actually
spent
a
lot
of
time
getting
clear
on
exactly
what
this
network
we're
talking
about,
even
is,
and
what
its
structure
is,
because
it's
actually
quite
complicated,
but
first
I
think
I
actually
think
a
good
way
to
start
is
to
like
talk
about
like
what
is
what's
a
CDN
right.
B
So
we
have
like
CDN
has
three
parties
right:
we
got
a
Content
owner,
the
the
client
and
the
CDN
right,
and
this
this
is
the
the
person
sends
dollars
to
the
CDN
and
their
content,
and
then
the
and
then
the
CDN
sends
it
to
send
that
serves
that
to
the
where's.
The
trust
relationship
in
this
does
anyone
know
where's
the
really
big
trust
relationship.
B
Yes,
this
is
a
big
old
T
right
here,
right
yeah
and
if
you
look
at
the
Saturn
design,
so
this
is
important
to
understand
when
you
go
into,
because
when
we
go
to
CD
dcdn,
the
reason
it
all
gets
complicated
is
that
we
can't
just
put
a
big
T
here
right.
So
now,
let's
let's
go
ahead
and
go
to
Saturn
right.
So
now
the
trust
is
still
here,
but
the
trust
is
this:
guy
named
the
auditor,
and
what
that
enables
is
that
the
auditor
has
all
of
these
L
ones,
all
right.
B
Okay
and
now
this
person
is
sending
their
content
to
the
l1's
I,
don't
know
exactly
what,
when
or
how
that
happens,
or
even
if
that's
been
worked
out,
and
then
they
are
serving
it
to
this
and
our
l1's
here.
What
this
enables
is
for
a
bunch
of
l1s
to
sign
up
and
serve
content,
so
it's
no
longer
controlled
by
our
super
centralized
CDN.
But
there
is
a
centralized
auditor.
B
They
have
full
trust
and
these
guys
are
like
in
the
current
design,
they're
big
beefy
nodes
that
have
some
kind
of
kyc
process,
I
believe
right,
so
they're,
like
semi
I'm
gonna,
do
like
a
little
semi-trust
here,
but
there's
also
like
a
fraud
detection
right.
B
B
So
then,
now
we
get
into
like
how
this
network
is
theoretically
getting
to
SPS
right
as
we
get
so
now.
There's
this
whole
other
thing
of
like
L2
right,
and
these
are
random
people
running
a
program
on
their
home
computer,
and
the
idea
is
that,
since
these
are
they're,
only
I
don't
know
30
of
these,
maybe
maybe
like
they're,
going
to
not
have
a
lot
of
content,
so
they're
going
to
fall
back
and
get
them
quickly
from
home
computers,
that's
in
and
of
itself
a
little
bit
of
a
challenging
statement.
B
The
the
other
thing
is
these
are
now
we're
now
moving
into
the
no
trust
right.
This
is
like
there's,
no
trust
here
right
also
like
one
thing
that
we
identified
just
in
talking
about
this
is
why
wouldn't
they
fall
back
that
way
right
like
yes
and
the
l2s
are
Regional
hubs
with
the
ltus
that
they're
connected
to
would
be
like
Regional,
yeah,
no
I
know,
but
I
was
just
thinking
if
this
is
a
big
we
were
talking
about.
B
If
this
is
a
big
beefy,
node
right
like
it
might
actually
first,
they
might
actually
get
it
quicker.
The
other
way-
and
it
doesn't
really
matter
okay,
but
let's
assume
that
they
can
get
it
quickly
from
these
nodes
right,
but
they're
now
totally
not
trusted
on
the
upside.
We
are
no
longer
in
the
no
money
we
are
now
in
a.
A
B
Zone,
potentially
right
money
can
move
around
between
these
parties
right
because
there's
no
expectation,
it's
going
to
land
free
and
then
finally,
these
guys
theoretically
are
going
to
fall
back
to
SPS,
which
are
sort
of
like
the
L3
of
this
scenario.
Right,
though,
also
it's
possible
that
the
L1
could
just
fall
back
to
the
SP,
if,
like
the
l2s,
didn't
have
it
right
so
then
the
question
is
so
we
this
is.
This
has
all
been
worked
out
and
it
works
perfectly
for
this
layer.
B
Definitely
so
actually
no,
it's
really
awesome
like
that.
You
guys
are
at
the
point
of
launching
this
and
at
least
finding
out
if
it
works
at
scale,
that's
really
cool,
but
then
the
question
is
like:
how
do
we
deal
with
this
flow
right?
Because
there's
like
these
two
layers
right
and
there's
a
number
of
problems
in
it
like
like,
for
example,
if
this
person
starts
getting
paid
every
time
they
serve
a
retrieval
to
the
L1.
How
do
we
track
that
right?
B
So
the
L1
can
report
it,
but
the
L1
is
no
longer
fully.
We
could
do
like
a
reporting
system
like,
which
is
what
you
guys
are
doing
for
the
L1,
and
you
could
do
it
all
the
way
up
like
you
could.
You
could
basically
be
like
I
report.
This
person
serves
the
retrieval
and
also
sends
a
report.
That's
like
I
got
it
from
SP2
right
and
then
these
this
guy
could
go
like
you
could
report
up
and
he
could
be
like
I
got
I
served
a
retrieval
and
I
got
it
from
L2
number.
B
Three
and
I
got
that
was
they
got
it
from
SP
number
two.
So
you
have
this
like
chain
of
like
how
the
retrieval
got
there.
The
problem
is:
you've
got
all
these
layers
of
trust
and
opportunities
for
collusion
along
the
way.
There
are
a
couple
things
we
could
consider
right.
First
of
all,
like
we
could
put
egress
in
any
of
these
layers
right.
The
problem
with
that
is
like
if
the
L2
is
doing,
the
e
grass
is
paying
for
egress.
B
That
kind
of
sucks,
if,
like
my
little
home
node,
is
now
all
of
a
sudden
sending
that
it's
no
longer
only
deposits.
It's
getting
withdrawals
that
could
potentially
go
below
zero
onto
my
home
node,
so
that
doesn't
really
work.
On
the
other
hand,
if
the
L1
were
to
fall
back
directly
to
the
SP
like
egress,
why
wouldn't
that
work?
Why
couldn't
you
just
egress
to
if
the
L1
falls
back
directly?
B
Something
to
consider
the
other
thing
you
could
do
is
you
could
use
something
like
sort
of
like
the
non-chain
version
of
the
of
the
of
The
cryptonet
Proposal,
which
is
you
could
say?
As
this
report
comes
in,
we
got
from
L1
this
L1
this
L2,
this
L2
and
SSP.
You
could
say
for
10
of
the
requests
I'm
going
to
re-execute
them
and
I'm
going
to
re-execute
them
in
a
couple
ways.
B
One
is
I'm
going
to
try
a
different
L1,
but
the
same
L2
and
S
and
same
L2
and
same
SP
right,
and
they
and
then
see
if
that
L1
reports
that
they
they
actually
delivered
it
I
could
also
try
the
same
L1
but
I
would
say,
use
a.
B
Going
to
the
same
SP
and
find
out
if
the
SP
really
delivered
it
there
and
you
could
also
say
I'll
go
to
the
same
L1
but
use
a
different
else.
The
same
L2
and
different
SP
or
I,
don't
know
yeah
I
can't
remember
which
one's
better
but
like
you
could
essentially
create
layers
of
a
referee
Network
that
would
like,
with
like
say,
10
of
randomly
selected
additional
requests
that
you
could
use
to
find
out
if
everyone's
actually
reporting
correctly
so
yeah.
That's
that's
the
theory
yeah!
You
can
also
I
mean
like
and
I
guess.
B
If
you
misbehave,
you
could
get
kicked
out,
you
could
get
like,
depending
on
how
bad
the
misbehavior
is.
I,
the
the
the
the
biggest
one
of
the
biggest
problems
in
here
is
these
l2s,
like
they
can't
they're
home
computers.
You
can't
do
a
lot
of
like
they're,
very
untrusted
and,
like
I,
don't
think
you're
going
to
be
able
to
do
a
lot
of
weird
money
mechanisms
with
them
like.
B
Presumably,
if
I
install
my
you
know
software
my
station
software,
my
home
computer
I'm,
expecting
it
to
like,
have
a
little
counter
that
slowly
goes
up
of
money
and
never
goes
down
right,
actually,
I!
Guess
if
I'm
behaving
well,
that's
fine
right!
That's
never
going
to
happen!
I,
don't
know
so
yeah,
it's
not
a
well-developed
system,
but
it's
a
system.