►
Description
Discussion with community stakeholders interested in the outcome of LIQ-2.0.
Liquidations 2.0 Technical Summary: https://forum.makerdao.com/t/liquidations-2-0-technical-summary/4632
Links:
Forum: https://forum.makerdao.com/
Chat: https://chat.makerdao.com/home
Governance and Community Events Calendar:
https://calendar.google.com/calendar/u/0/r?cid=bWFrZXJkYW8uY29tXzNlZmhtMmdoaXBrc2VnbDAwOWt0bmlvbWRrQGdyb3VwLmNhbGVuZGFyLmdvb2dsZS5jb20
A
Also,
just
for
the
record
just
so
you
have
it
hey
guys.
The
lpl
team
has
not
been
able
to
talk
to
the
smart
contract
team
about
liquidations
2.0,
and
this
is
the
first
time
we
were
able
to
talk
and
we're
just
doing
publicly
just
because
of
the
legal
restrictions
that
legal
put
on
us.
So
if
you
don't,
please
don't
if
you
have
any
criticisms
of
you
or
whatever.
Please
don't
take
it
personally,
because
we
we
we
had
to
do
this
in
the
public
forum,
according
to
legal
yeah,
that's,
okay!
We
have
great.
B
Well,
we'll
keep
a
thick
skin
on
yeah
and
I
will.
I
will
mention
that
any
sort
of
feedback
that
we
get
now
is
sort
of
critically
important,
we're
shooting
for
code
complete
within
the
coming
weeks.
So
we
really
would
value
any
sort
of
feedback
and
if
there's
going
to
be
last-minute
tweaks
they're
going
to
happen
during
this
time
period.
So
now
is
the
time
to
to
speak
is
kirkback.
C
Yeah,
I'm
I'm
back
seems,
like
I
didn't,
say
anything
important.
While
I
was
off
not
that
I
was
necessarily
saying
much
of
much
of
importance
ever,
but
yeah
really
excited
to
have
so
many
people
on
this
call
to
share
their
opinions
and
we
can
get
as
much
into
the
weeds
as
you
want.
Nothing
is
off
limits
here,
so
maybe
with
that
said
yeah
is
there
anything
somebody
wants
to
dive
into
right
from
the
start.
B
C
Yeah
I
mean
I
could
show
like
just
like
three
or
four
slides
really
quickly
to
just
demonstrate.
How
does
a
dutch
auction
work
or
something
like
that
if
that
would
be
useful,
but
I
don't
want
to
spend
a
lot
of
time
on
just
like
trying
to
communicate
the
information.
The
idea
was
that
people
would
come
prepared.
D
Okay,
let's
hypothesize
an
attack
where
a
single
actor
decides
to
for
a
long
period
of
time
bid
at
the
very
beginning
of
the
dutch
auction.
That's
when
the
calculation
function
has
produced
a
price.
That's
probably
unfavorable
right.
If
we
assume
that
the
calculation
function
starts
at
the
osm
price
and
decreases
what
happens
if
an
actor
comes
in
at
the
beginning
of
every
auction
and
actually
calls
the
take
method
for
the
entire
lot
size
of
the
auction
at
an
unprofitable
price
with
the
angle
that
they
want
to
discourage
other
participants.
B
So
the
the
presumed
attack
would
be
they
would
do
this
over
some
period
of
months.
The
ecosystem
would
basically
dry
up
until
they
had
sole
control
over
the
ecosystem
and
then
they'd
start.
You
know
allowing
the
sort
of
price
decrease
function
to
really
drop
and
really
drop
and
really
drop
until
until
prices
were
so
low,
they
got
a
steal.
So
I
would
the
the
sort
of
initial
gut
reaction
to
that
and
kurt.
B
Maybe
you
want
to
chime
in
too
at
some
point,
but
the
hope
is
that
we'll
have
integrations
from
other
like
dex,
aggregators
and
web
3
users.
That
will
be
able
to
sort
of
see
these
auctions
so
that
you
won't
need
this
sort
of
like
heavy
technical
infrastructure
in
order
to
spin
up
a
keeper,
which
is
how
you'd
be
expecting
to
get
away
with
the
attack
like
this.
B
So
our
hope
is
that,
even
if
somebody
does
this,
you
know-
and
you
know
even
trains-
users
to
stay
away
from
those
uis-
that
with
a
little
bit
of
messaging,
we
end
up
with
a
you
know,
an
easy
way
to
attract
a
bunch
of
web
free
users,
because
you
know
there
are
deals
and
then
we
also
end
up
with.
Hopefully,
if
dex
aggregators
are
just
integrated
with
them,
we'll
just
sort
of
naturally
get
sort
of
mixed
in
with
their
their
user
base.
C
You
know
whenever
somebody's
trying
to
buy
or
sell
particular
collateral-
or
you
know,
asset
types.
You
know,
as
you
know,
if
they're
just
trying
to
pick
out
the
best
price.
If
suddenly,
this
over
bidder
has
stopped
bidding,
that's
going
to
now
show
up
as
a
good
price,
and
hopefully
those
orders
will
get
routed.
D
Okay-
and
my
next
question
brings
up
something
that
long
for
wisdom,
discussed
kind
of
before
the
call
really
started
involving
front
running.
So
I
hypothesized
that
for
any
bidder
who
is
bidding
on
more
than
half
of
the
collateral
up
for
auction
at
a
time
if
they
get
front
run,
that
means
that
their
bid
would
essentially
be
blocked
because
that
liquidity
wouldn't
be
there
anymore.
C
Yeah
well
one
thing
I'd
say
very
quickly:
if
they're,
even
if
they're
bidding
on
more
than
half
the
liquidity,
the
way
the
new
auctions
are
implemented
are
kind
of
like
a
limit
order.
So,
like
you
could
say,
hey
I
want
to
buy
all
the
collateral
and
if
somebody
buys
out
75
of
it
before
your
order
gets
there
you'd
still
get
25,
of
course
that
doesn't
totally
mitigate
the
attack,
because
maybe
the
front
runner
just
buys
it
all
right.
But
you
know
it's.
C
D
C
D
C
B
Yeah
and
then
for
mitigating
front
running,
you
know
there's
for
keepers
that
are
gonna,
implement
bots
that
do
this
there's
any
number
of
strategies
that
they
can
try
and
employ
like
on
their
side
to
do
it,
but
we
have
no
specific
front-running
protections
in
in
the
new
liquidations
2.0
code
and
I
think
that's
primarily
because
there's
you
know
we
all
converged
on
a
similar
idea,
but
that
idea
came
with
trade-offs
that
that
seemed
maybe
worse
than
the
front-running
problem.
Well,.
C
To
be
very
specific,
the
trade-offs
seem
to
be
that
any
viable,
front-running
defense
seems
to
mess
up
single
block
composability.
To
some
extent.
You
know,
like
you
have
to
either
in
a
private
or
public
way,
for
example,
pre-commit
to
save
buy-in
at
a
certain
price,
and
then
you
know
since
you've
pre-established
those
rights
you
can't
be
front-runned,
but
that
presents.
C
The
the
idea
of
front
running
in
general
on
ethereum
is
you
see
a
transaction,
that's
going
to
be
profitable,
and
so
then,
you
very
quickly
before
that
transaction
gets
executed,
resubmit
the
same
or
a
similar
transaction,
and
this
can
actually
get
quite
detailed
where
front
runners
will
pick
apart,
a
transaction
and
reconstruct
a
new
one,
that's
very
different,
but
then
it
takes
advantage
of
the
same
arbitrage
opportunity.
You
submit
it
with
a
higher
gas
price.
C
Yeah,
you
know
first
priority
recapitalize
the
system,
so
there's
not
unbacked
die.
Second,
priority
is
like
return
as
much
collateral
to
the
you
know,
liquidated
vault
as
as
as
the
market
allows
for,
and
so
you
know
so
long
as
the
auctions
are
fulfilling
those
purposes.
You
know
we're
the
you
know
from
an
mkr
perspective,
that's
good
enough.
G
Okay,
so
it
sounds
like
this
is
going
to
be
an
ongoing
thread
right
and
I
think
that
there's
a
couple
different
ways:
game,
theory-wise-
that
this
can
that
this
issue
can
present
itself,
which
is
just
as
we
had
answered
in
the
first
question.
If
there's
an
actor
that
is
over
bidding
or
locking
the
majority
of
participants
out
over
time,
that
could
disincentivize
participants
until
there's
no
participants
or
very
little
participating,
and
I
think
that
this
will
just
come
up
again
again
again.
But
to
your
guys's
point,
I
don't
think
it's
a
bug.
C
I
mean
in
dealing
with
it
if
it
proves
to
be
a
problem
in
practice.
Dealing
with
it
could
involve
implementing
a
different
auction
system
that
you
know
hampers
front
running
in
some
way,
but,
like
we
said
that
will
come
off
come
along
with
trade-offs
like
we
currently
don't
know
how
to
do
that
without
spoiling
single
block
composability,
which
throws
out
flash
loans
and
dexag
integrations
and
stuff
like
that
stuff.
F
Explain
it
fairly
fast,
the
basic
theory
behind
this
is
again.
This
would
be
most
likely
when
you
know
there's
some
kind
of
market
event,
and
you
know
the
market
is
already
kind
of
weak,
but
some
extremely
large
player
could
essentially,
you
know,
take
the
major
major
exchanges
that
indices
are
based
out
of
so
typically,
most
indices
are
based
off
of
coinbase,
kraken
and
bitstamp,
for
example.
F
F
C
So,
just
to
make
sure
I
understand
you
know,
the
attack
is
push
down
prices
on
these
centralized
exchanges
that
feed
into
the
oracle
data
for
maker.
You
know
at
the
same
time
maybe
have
a
big
short
position
open.
Then
liquidations
happen,
you
know
they're
there.
The
design
does
include
a
configurable
buffer
above
the
current
osm
price,
for
whatever
the
starting
price
of
the
auction
is.
C
So
you
know
that
basically,
is
increasing
the
size
of
the
attack
in
the
sense
that
to
have
that
you
know
whatever
target
price
this
attacker
needs
at
the
start.
They
have
to
push
it
down
that
by
that
much
more,
but
it's
a
difference
of
you
know:
quantity,
not
quality.
I
suppose
you
would
say
yeah
and
then
yeah
buy
cheaply
and
and
cover
the
short
position
that
do.
I
have
understand
that.
F
Yeah,
I
mean
the
the
the
reason
he's
targeting
maker
and
the
maker
oracles.
Is
it
triggers
a
bunch
of
very
large
auctions?
So
the
you
know?
Normally
you
could
say
well.
The
attacker
could
do
this
normally,
but
the
advantage
of
attacking
maker
is
he
can
buy
a
very
large
amount
of
eath
at
these
depressed
prices.
F
C
C
Well,
it's
you
know
it's
set
by
you
know,
or
you
know,
parameters
are
recommended
by
the
risk
team
and
voted
on
by
governance.
So
it's
currently
15
million.
You
know
so
that
would
be
one
thing
that
would
prevent
them:
the
necessarily
the
profitability
of
this.
If
that's
set
low
enough,
where
the
attacker
can't
get
the
you
know
collateral
that
he
needs
to
eat
or
whatever
to
cover
this
position.
So
that's
you
know,
sort
of
balance
the
size
of
the
attacker.
F
C
C
Of
there's
also
an
element
of
where
anybody
can
bid
in
these
auctions.
So
even
if
he's
you
know
trying
to
buy
at
that
depressed
price,
everybody
else
will
want
to
buy
at
that
depressed
price
as
well,
so
there's
risk
and
that
you
know
he
might
not
necessarily
win
the
the
priority
gas
auction.
For
example,
although
presumably
has
a
larger
financial
stake
in
that
than
anyone
else,
so
he
would
bid
very
high.
F
Right
so
so
to
be
clear,
I'm
not
completely
clear
on
the
whole
auction
mechanism,
but
in
this
case
he's
willing
to
pay
a
higher
price
than
fair
market
value
at
that
time.
So
that's
why
he
has
a
higher
probability
of
grabbing
the
entire
15
million.
So
so
to
that
point,
let's
say
he
buys
the
entire
15
million
right.
C
F
C
F
B
B
So
this
so
my
next
comment
will
also
partly
address
this.
It's
not
necessarily
it
steps
back
from
the
auctions
and
just
says
that
the
the
whale
in
this
case
doesn't
really
know
how
many
of
those
vaults
are
available
for
liquidation
at
that
price.
They
can
only
conjecture,
because
a
number
of
those
vaults
are
going
to
be
protected
by
automation
like
defy
saver.
B
B
Yeah
you
could,
you
could
see
that
and
then
you'd
have
to
you'd
have
to
wonder.
I
mean,
like
I've,
been
impressed
with
some
of
the
pullbacks
we've
had
lately
by
how
little
or
almost
no
liquidations
we
got
during
some
of
those
pullbacks,
because
so
many
people
are
tooled
up,
at
least
on
the
very
edge
near
the
price
they're
tooled
up
with
like
defy
saver
or
some
other
automation
that
that
re
balances
their
their
collateralization
ratio.
B
D
D
That
delay
would
be
about
two
hours
right,
because
you'd
have
to
depress
the
price
enough
for
an
hour
to
actually
get
the
low
osm
price,
and
then
you'd
have
to
wait
until
the
next
one
rolls
off
for
the
liquidations
to
happen
on
it.
So
yeah
I
mean
the
the
actor
would
need
a
large
amount
of
collateral
or
a
larger.
D
C
Mean
I
think
there
are
a
few
angles
to
think
about.
You
know
reducing
the
risk
from
an
attack
like
this
one
is,
you
know,
making
sure
the
attack
is
costly,
which
includes
things
like
the
osm
delay
itself.
It
includes
the
buffer.
On
top
of
the
osm
price,
that's
applied
to
the
initial
price
of
the
auction.
C
F
F
D
E
F
F
B
E
B
E
Yeah
but
frankly,
isn't
that
just
a
general
thing
like
what
we
already,
we
already
do
that
right,
because
if
you
have
that
amount
of,
if
you
have
that
liquidity,
you
simply
you
put
it
in
a
vault
and
you
take
out
the
die
loan
and
we're
already
like
transferring
from
from
an
illiquid
acid
to
to
liquid
death
right.
So
yeah
yeah.
I
don't
know
if
this
is.
If
this
is
something
specific
to
the
auctions,
even.
D
C
You
know
right
now,
it's
just
the
cap
on
how
much
can
be
at
auction
at
a
time,
but
you
could
all
you
could
also
say,
like
you
know:
you're
not
allowed
to
liquidate
more
than
15
million,
or
you
know
five
percent
of
total
die
supply
or
whatever
per
hour,
or
something
like
that.
So
just
I'll
just
say
that
that
would
probably
need
more
offline
discussion,
but
just
to
make
sure
we
don't
miss
that.
E
I
I
would
even
argue
that
if
you
have
an
in
an
illiquid
asset,
then
the
debt
ceiling
should
be
low
enough,
so
that
it's
not
possible
to
to
ever
liquidate
the
the
amounts
that
we're
talking
about,
because
otherwise
you're
creating
the
exact
same
risk
on
the
side
of
simply
taking
out
the
die
loan
right
like
it's
a
suppose
that
I
suppose
that
I
own,
like
90,
of
all
of
all
tokens
of
any
liquid
asset
and
like
the
very
shallow
market
prices,
is
some
almost
random
number.
Well.
E
If
the
debt
ceiling
is
high
enough,
I
just
can
lock
it
in
and
sure
maybe
I'll
have
to
pay
like
a
50
premium,
because
the
liquidation
ratio
is
150,
but
then
still
that
would
be
much
more
than
what
I
would
get
from
the
market
because
they're
simply
my
liquidity
would
drive
the
market
to
zero
right.
So
I
think
this
one
really
comes
down
to
setting
the
right
risk
parameters.
F
D
I
wanted
to
go
into
a
little
bit
more
detail
on
whole
because
in
the
forum
post
it
discussed
that
there
was
both
an
overall
limit
on
the
amount
of
dye
that
would
needed
to
be
recovered.
That's
out
for
auction
simultaneously,
but
it.
It
also
said
that
there
was
going
to
be
a
collateral
type
specific
one
and
in
the
source
code
I
only
found
the
overall
hole.
I
didn't
find
a
like
a
an
auction,
a
collateral,
specific
hole.
C
I
I
checked
the
latest
version
of
the
code
because
that
has
been
added.
Okay,
make
sure
you're,
just
it's
the
liq-2.0
branch
and
dss,
and
the
latest
version
of
that
should
should
have
that
per
collateral
hole,
type
which
you
know,
governance
doesn't
have
to
use
if
it
doesn't
want
to.
But
you
know
we
sort
of
like
to
err
on
the
side
of
adding
lovers
if
we're
not
sure
whether
we
need
them
or
not,
because
it's
very
hard
to
add
them.
After
the
fact.
C
B
Okay,
I'm
almost
convinced
we'll
need
this
lever
because
of
something
like
mana
so
we'll
want
to.
We
may
need
to
cap
on.
You
know
again,
water's,
probably
right.
The
dc
should
be
the
first
line
of
defense,
but
the
second
defense
would
be
how
much
we
allowed
to
be
liquidated
at
once.
We
certainly
wouldn't
want
15
million
the
entire
amount
of
money
going
up
for
liquidation.
D
So,
on
that
note
considering
having
different
holes
for
different
collateral
types,
what
about
the
calc
function
when
we
choose
the
calc
function?
That's
part
of
the
clip
contract
itself
and
that's
going
to
be
different
for
each
collateral
type.
So
does
that
imply
that
we'd
likely
have
the
risk
team
determine
which
calc
function
is
most
appropriate
for
the
collateral
type
and
we
could
have.
C
Yes,
absolutely,
that's!
That's
the
idea,
and
I
mean
maybe
when
we
actually
deploy
these
for,
like
gas
cost
reasons,
we
won't
do
it
as
an
external
contract
and
we'll
build
it
right
into
the
clips,
but
the
I
you
know
right
now:
it's
intentionally
modular
that
yeah.
You
know
we
want
to
be
agnostic
about.
You
know
what
is
the
best
function
per
collateral
type
and
you
might
want
to
be
able
to
swap
it
in
and
out
even
for
a
given
collateral
type
based
on
market
conditions.
C
A
B
We
have
a
ticket
to
basically
aggregate
all
of
the
tunables
that
are
in
liquidations
2.0,
and
we
were
going
to
hand
that
off
to
risk
so
that
they
knew
what
they
should
sort
of
think
about.
Maybe
we
can
add
into
that
process
that
they
bubble,
that
set
of
parameters
up
to
governance
and
maybe
votes
on
a
so
governance
can
vote
on
a
package
or
something
like
that
as
well.
C
I
mean
you
know,
I
would
say
this,
this
sort
of
like
how
does
it?
How
does
you
know
once
the
technical
work
is
done?
How
does
that
flow
through
the
governance
and
risk
apparatus
and
turn
into
executive
votes
on
chain?
That's
sort
of
a
broader
discussion
than
what
this
call
is
intended
for,
and
so
you
know
it's
obviously
an
important
discussion
that
needs
to
happen.
E
C
What's
the
average
value
return
to
vaults,
you
know
like
what
are
there
any
like
catastrophic
failure?
Modes
discovered
and
you
know
obviously
no
simulation
and
modeling
perfectly
accounts
for
the
real
world.
But
the
hope
is
that
doing
it
is
a
lot
better
than
just
sort
of
being
blind
about
how
things
might
behave
and
can
inform
better
risk
parameter
settings.
I
think
there's
also
intention.
C
You
know.
Once
scotland
has
sort
of
developed
their
simulation
tool,
you
know
we'd
like
to
make
that
they
they
can't
open
source
it
because
it's
proprietary
technology,
but
it
could
be
exposed
as
like
a
sas
app
that
the
community
could
then
use
going
forward.
If
they
wanted
to
model
under
you
know
different
price
versus
time,
curves
or
different
risk
parameters.
C
F
My
other
question
is
like
how
is
this
going
to
be
released
like?
Is
this
like,
like
a
bang
release
like
all
the
collaterals,
will
kind
of
go
into
auction
v2,
or
can
it
be
staged
where
you
can
do
a
couple
of
collaterals
and
test
out
this
new
system
in
production
for
a
couple
months,
or
something
like
that?.
C
You
know
that
that's
still
under
discussion,
I
think
the
sort
of
current
thinking
is
actually
more
around
the
gradual
transition.
Just
from
the
perspective
of
you
know,
you
don't
want
to
introduce
too
drastic
a
change
at
once.
You
know
we
also
want
to
retain
for
sure
the
ability
to
switch
back
to
the
old
auctions,
at
least
for
a
period
of
time,.
C
H
Can
you
guys
hear
me,
I
think
yeah?
Let
me
know
if
you
can
okay
cool,
it's
it's
more
of
a
bit
of
feedback
in
another
direction
than
on
the
previous
point,
just
something
I've
noticed
with
the
price
versus
time
curve.
I
think
that's
going
to
be
the
most
important
factor
this
to
get
tuned,
just
right
for
the
worst
case
scenario.
So
one
worst
case
scenario
I
I
could
see
happening
would
be.
H
Let's
say
we
have
like
50
million
70
million
debt
ceiling
available
and
all
of
a
sudden,
the
die
price.
Just
crashes
are
not
to
that
price,
eath
price
and
all
of
our
collateral.
Other
crashes,
like
you,
know
black
swan
events,
an
attacker
could
have
or
a
team
of
attackers,
because
you
could,
you
have
multiple
people
do
this
attack
would
be
like
have
the
whale
attacker
part
of
it
draw,
draw
out
a
lot
of
dead
ceiling
and
fill
up
all
the
debt
ceiling
and
then
have
the
traitor.
H
Part
of
the
attackers
have
all
the
die
for
the
auctions
because
everyone
else
would
be
like
hey.
I
was
expecting
that
die
to
be
there
and
it's
not
because
that's
because
it's
not
there
also
coupled
with
this,
I
would
I
could
see
like
let's
say
I
have
50
million
in
negative
die
exposure,
because
I've
just
been
slowly
collecting
die,
puts
that
gives
me
a
hedged
position
to
go
very
long,
very
long
on
the
die
market.
So
now
like
since
I
have
so
much
50
million
negative
exposure,
I
could
collect.
H
Maybe
30
million
die
in
playing
this
attack
ahead
of
time
so
like
whenever
each
crashes,
I
don't
have
to
have
that
capital
to
make
youth
crash.
I
just
know
it's
going
to
happen
eventually
right
and
I'm
just
acquiring
puts
requiring,
puts
this
whole
time
and
also
acquiring
die
the
whole
time
with
the
expectation
of
filling
up
the
debt
ceiling.
C
H
That
that
could
be
part
of
it.
Yes,
that
would
be
the
even
worst
case
scenario,
but
it
doesn't
have
to
be
because
it
it
doesn't
have
to
be
a
cabal,
because
you
can
acquire
a
lot,
a
huge
amounts
of
negative
exposure,
because
the
price
right
now
has
been
0.01,
so
you've
just
been
able
to
stack
negative,
puts
or
puts
on
puts
on,
puts
that
have
been
sold
to
us
by
maker
protocol.
C
F
F
C
If
it's
just
around
hoarding
die,
I
mean
you
can
hoard
die
in
the
present
system
and
it's
much
worse,
it's
a
much
worse
attack
than
it
is
in
the
new
liquidation
system,
because
in
the
new
liquidation
system
you
can
take
advantage
of
you
know
dye
across
the
ecosystem.
You
can
do
flash
lending
much
so
you
know
the
liquidity,
should
it
should
strictly
be
better
or
a
harder
and
more
costly
attack
with
right
liquidations
2.0
in
place.
C
You
know-
and
you
really
have
to,
especially
as
the
dye
supply
continues
to
grow.
You
have
to
really
ask
like.
Is
that?
How
feasible
is
that
attack
for
someone
to
pull
off?
And
then,
like
akash
said
you
know,
you'll
have
collaterals
like
usdcb
that
can
be
used
for
very
short-term
liquidity
in
a
crunch
where
there
isn't
cheaper
liquidity
available,
because
you're
only
going
to
have
that
position
open
for
a
short
time,
but
it
wouldn't
necessarily
be
yeah.
C
Yeah,
that's
that's
a
matter
of
like
how
quickly
does
the
price
go
from
some
high
level
to
some
low
level,
and
it's
also
like
you
know,
is
there
like
a
phase
transition
within
the
price
curve
where,
if
it's
dropped
a
certain
amount,
does
it
start
dropping
significantly
more
slowly?
You
know
the
price
scores
are
set
up
to
have
endpoints
of
like
if
the
price
drops
by
a
certain
amount
or
a
certain
amount
of
time
is
passed.
A
C
C
C
I
can
give
my
personal
opinion,
which
is
that
we
probably
shouldn't,
because
you
know,
there's
no
real
reason
to
liquidate
those
positions
right
now,
they're
sort
of
like
you
know
they
represent
potential
profit
for
the
protocol
if
die
ever
goes.
You
know
significantly,
you
know
as
as
low
below
dollars.
It
needs
to
go
for
those
to
unwind
and
so
there's
no
like
real
pressing
needs.
So
long
as
we
aren't
spending
the
surplus
buffer.
C
We
don't
really
have
an
unbacked
debt
situation
as
long
as
we
keep
all
the
fees
sort
of
in
that
limbo.
State
of
we
haven't
spent
them
yet
so
we
can
cover
bad
debt
with
them
if
we
need
to
so
from
my
perspective,
there's
not
a
pressing
need,
but
you
know
that
doesn't
really
count
for
much
in
the
greater
scheme
of
governance,
things
so
open
question.
I
Yeah,
sorry
guys
I
was
muted.
I
totally
agree.
I
actually
wrote
about
this
today.
So
why
would
you
turn
the
fees
off
and
turn
on
liquidations
now
and
you
can
potentially
earn
much
more
so
there's
this
idea
that
we
could,
just
you
know,
leave
them
running
and
then
once
we
touch
hundred
percent
characterization
ratio,
we
could
disable
them
and
then
limit
depth
c
link
to
zero
and
then
open
another
similar
world
type.
So
for
all
the
new
worlds
being
co-point
we
we
can
still
charge
something
but
yeah.
I
There
are
pros
and
cons.
I've
actually
written
about
it
today
on
the
forum
post,
but
I
really
wouldn't
go
into
it
too
much,
but
they
have
another
question.
So
I've
read
the
forum
post
and
I
was
missing
like
a
really
simple
list
of
governance
parameters
needed
for
liquidations
2.0,
because
I
imagine
there's
a
bit
less
of
them
and
I
know
there's
this
price
curve,
probably
the
most
important
one,
but
I
guess
there's
there
are
others
as
well,
and
I
it
wasn't
completely
clear
to
me,
which
ones
are
they.
B
C
Mean
we
can
also
start
a
running
list
somewhere
with
the
understanding
that
it's
not
necessarily
finalized,
and
that
might
even
be
a
good
thing,
because
there
might
be
some
feedback
of
well
we'd,
like
you
know,
to
see
different
parameters
for
this
reason
or
that
reason
so
yeah
if
you're,
okay,
with
potentially
not
final
parameters,
which
I
think
might
even
be
better
because
then
you
still
have
the
opportunity
to
get
feedback
on
them.
You
know,
I
think
we
could
make
that
happen.
A
G
G
I
think
it's
worth
hosting
an
open
call,
just
like
this
one,
like
in
office
hours
to
discuss
the
risk
parameters.
Maybe
we
can
do
that
in
another
public
call.
I
think
that's
a
good
idea
as
well,
since
that
just
has
such
a
massive
effect
on
the
attack,
vectors
and
and
protecting
us
against
open
attacks.
A
D
I
do
have
one
more
question
if
I
may,
involving
the
bark
so
in
the
existing
auction
system,
and
it
won't
really
change
with
liquidations
2.0.
There
are
a
lot
of
resources
required
to
detect
opportunities
to
kick
off
a
new
auction,
and
this
thus
far
there
hasn't
really
been
much
incentive
in
calling
the
kick
right.
You
have
to
maintain
all
of
your
infrastructure
and
constantly
monitor
liquidation
ratios
of
vaults.
C
That
was
one
of
the
things
I
called
out
in
my
preamble.
Our
current
thinking
is
that
you
know
we're
not
a
hundred
percent
sure,
so
we're
just
going
to
add
something
to
it,
and
if
governance
doesn't
want
to
use
it,
they
don't
have
to.
I
think,
the
sort
of
like
tentative
first
way
we
try
to
do.
It
is
maybe
like
a
small
dive
reward.
C
You
know.
Basically,
you
would
call
suck
with
well
I'm
getting
a
little
technical.
If
you
don't
know
the
val
apis,
but
basically
that
would
be
a
thing
where
you
would
give
some
dye
to
someone
sort
of
up
front
and
create
a
little
debt
on
the
back
end
and
since
there's
like
a
large
liquidation
penalty
or
the
auction,
then
theoretically,
that
with
high
probability
gets
covered.
C
So
it's
just
it's
just
basically
mkr
holders
subsidizing
liquidations
with
some
sort
of
die
reward,
but
there
are
other
ways
you
could
imagine
doing
that
you
could
imagine
using
a
gas
refund.
You
could
imagine
just
shaving
off
a
chunk
of
the
collateral
and
giving
it
to
the
the
caller
of
mark,
etc.
So
very
interested
here,
if
people
have
like
specific
thoughts
on
how
that
should
be
done
or
how
it
would
work
well,
but
our
current
idea
is
to
add
something
so
that
it's
there
if
we
need
it.
F
So
to
this
point,
even
with
the
current
auction
system
or
the
future
auctions,
the
basic
fundamental
problem
I
see
is
you
need
one
honest
actor
right.
If
you
have
one
honest
auction
participant,
then
everyone
else
has
to
participate
honestly
and
they're
they're
putting
up
legitimate
bids,
but
I
could
easily
see
like
either
through
one
player
disincentivizing
all
the
other
players
that
it
somehow
ends
up
in
a
situation
where
one
player
controls
everything
and
then
can
eventually
become
dishonest.
So
I
guess
how
do
you
address
that
question?.
B
Yeah,
I
can,
if
I'll
take
this
one
so
so
outside
of
the
incentive
around
whether
or
not
we
call
you
know,
we
incentivized
somebody
to
call
bark,
let's
assume,
like
the
auctions,
actually
kicked
off
the
the
cost
to
manage
that
auction,
just
even
as
like
a
backstop
is
so
insanely
low.
Now,
because
you're
going
to
be
able
to
to
flash
lend
the
collateral,
that's
out
for
auction,
find
a
circuit
and
return
the
die
for
it.
B
That
even
you
know
it
would
take
almost
nothing
even
for
someone
to
just
run
the
infrastructure
to
do
that
particular
bit
and
they
would
probably
get
a
profit
if
they
set
the
parameters
right.
So
I
think
I
think
we're
going
to
have
a
much
easier
time
with
you
know,
honest
auction
participant
that
problem,
because
I
agree
that
problem
is
like
a
real
problem.
It's
it's
especially
a
problem
under
our
current
auction
system.
B
E
C
D
E
E
C
And
so
you
know,
if
you're
winning,
often
enough,
it's
worth
it
for
you
to
take
the
chance
and
trigger
it.
And
then
there
are
also
things
like
dex
aggregators
that
want
to
offer
the
best
possible
prices
to
their
users,
which
means
having
the
best
possible.
You
know
range
of
liquidity
to
choose
from,
and
so
maybe
there's
an
incentive
there
to
to
trigger
liquidations,
because
it's
good
to
have
you
know
more
options
for
your
aggregator,
but.
A
E
One
more
detail
to
add
there,
so
a
large
part
of
it
is
probably
of
the
cost.
I
mean
of
running
that
infrastructure
is
probably
monitoring
the
which
volts
are
underwater,
but
that
I
mean
there
will
be
data
apis
that
will
readily
provide
that
information,
and
I
I
think
that's
one
assumption
that
we
can
make
that
at
least
in
the
long
run
these
will.
These
will
be
available
right
so
that
cost
might
actually
already
be
minimized
so
yeah
I.
E
No,
no!
No,
because
once
the
initial
integration
is
done,
there
would
be
no
cost
to
to
participating.
That's
the
reason
why
you
can
drive
out
players
in
the
ecosystem
today
is
because
it
it's
costing
you
money
to
keep
your
keeper
online
right,
but
that
cost
would
be
reduced
to
to
virtually
zero,
so
yeah,
I
think
the
best
defense
you
can
probably
get
is
with
the
system
like
the
one
that
is
being
proposed.
E
B
To
be
clear
that
the
difference
here
is
that
the
bark
is
actually
expensive
infrastructure,
so
we
could
add
an
incentive
to
call
the
bark
but
bidding
on
an
open
auction
once
it's
out
there
once
it's
been
kicked
is
going
to
be
extremely
cheap.
So
that's
the
cost.
That's
like
near
zero
and,
I
think,
addresses
the
the
you
know
the
issue
of
like
bad
actors
and
auctions.
B
You
know
trying
to
like
control
those
auctions,
because
it's
it's
really
gonna
cost
almost
nothing
to
to
run
a
keeper
that
just
manages
the
options
it
can
find.
The
bark
still
there's
an
expense
there
and-
and
it's
slightly
worse
than
the
auctions
today.
So
if
we
bite
today,
usually
the
person
that's
going
to
bite
gets
the
you
know
they
can
call
bite
which
calls
kick
and
then
they
can
get
their
first
like
tend
in.
So
they
get
to
set
the
price
of
the
auction.
So
there's
a
pretty
good
incentive.
F
Right,
I'm
just
going
to
bring
this
point
up
again,
so
I
generally
agree
with
you
that
it
is
going
to
be
free.
But
let's
be
honest
here,
it
is
not
completely
free
right.
So
me,
as
an
independent
person
who
wants
to
bid
on
these
things
has
to
have
you
know.
A
computer
running
somewhere
has
to
have
access
to
a
node
that
actually
costs
money.
B
F
E
There
is
a
gui
included
in
the
initial
scope,
so
the
oasis
team
will
will
build
one.
The
one
thing
it
won't
do
yeah,
I
won't
be
complete,
I
believe,
in
the
in
the
way
that
you
say
it,
but
you
you
will
be
able
to
bid,
but
it
will
be
right.
Only
you'll
have
to
to
use
like
another
interface
to
see
which
auctions
are
going
on.
So
the
monitoring
aspect
of
it
is
not
included
but
yeah.
I
guess
that
can
be
done
too.
F
E
In
principle,
I
definitely
agree
too.
The
the
thing
here
is:
I
think
that
we
were
really
scared
of
that
scenario
now,
because
we
got
confronted
with
it
and
the
the
likelihood
of
that
happening
is
reduced
to
such
an
extent
that
I
think
that
there
are
probably
other
problems
that
will
be
bigger
than
this
one
and
we're
kind
of
we're
blinded
by
by
something
that
we
had
a
very
bad
experience
with,
but
in
principle
I.
I
definitely
agree
that
you
know
all
the
the
necessary
precautions
should
be
taken.
F
F
So
let
me
throw
out
another
use
case,
and-
and
I
can
see
you
saying
like
this-
is
useless,
because
the
bots
are
going
to
win
these
auctions
anyway.
What's
the
point
of
building
all
this
manual
gui
and
I'll
counter
it
with
this
gives
an
opportunity
for
manual
humans
to
buy
large
liquidity.
So
what
I'm?
What
I'm
claiming
is
that
manual
humans
might
be
willing
to
pay
a
higher
price
to
access
these
large
blocks,
whereas
a
bot
just
doing
pure
arbitrage.
B
C
Well,
yeah,
generally,
you
know
the
most
gas
intensive
thing
is
storage
and
any
configuration
variable
is
going
to
entail.
At
least
you
know,
one
storage
read
to
check
what
the
value
is.
On
the
other
hand,
like
it's
probably,
you
know,
there
are
already
a
lot
of
storage,
reads
and
writes
in
this
logic,
and
if
it's
a
thing,
that's
like
an
incentive
mechanism.
Obviously
the
incentive
can
cover
the
additional
gas
cost,
but
yeah.
C
G
Yeah-
and
I
mean
I
would
always,
if
I'm
as
a
maker
holder
right
or
on
behalf
of
maker
holders,
I
would
always
rather
have
the
system
pay
a
little
bit
more
up
front,
which,
if
it
can,
I
think
the
arbitrage
already
kind
of
does
this.
But
if
it
can
again
add
a
little
bit
more
incentive
to
encourage
that
participant
show
I'd.
Much
rather
have
the
system
lose
a
little
bit
of
capital
over
time.
G
That
way,
you
know
and
have
that
be
withdrawn
from
the
fees
we
collect
just
to
ensure
that
that's
there
and
you
know
we
don't
have
to
worry
about
that
and
don't
have
to
worry
about
a
black
thursday
event
or
people
trying
to
with
auction
participate
participation
altogether
by
performing
a
bad
action
or
by
forcing
other
people
to
not
participate,
because
I'm
over
bidding
things
of
that
nature,
getting
getting
it
right
at
fair
market
value.
So
I
don't
know
that's.