►
Description
Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on Swarm, IPFS and more.
https://archive.devcon.org/archive/watch/6/mev-capturing-ammsmcamms/
Escape hatches are meant to be a way to exit assets or state from a rollup in the hopefully unlikely event that sequencers are offline. In this talk, we review which projects have these hatches and what plans are publicly available for future rollups or versions of these rollups. We list some ideas that these, and other rollups, may implement or adopt in the future, for more robust and secure rollups. The suggestions will be rooted in our concern for security and the issues presented by bridges.
Speaker(s): Jan Gorzny
Skill level: Intermediate
Track: Layer 2s
Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum
Learn more about devcon: https://www.devcon.org/
Learn more about ethereum: https://ethereum.org/
Devcon is the Ethereum conference for developers, researchers, thinkers, and makers.
Devcon 6 was held in Bogotá, Colombia on Oct 11 - 14, 2022.
Devcon is organized and presented by the Ethereum Foundation, with the support of our sponsors. To find out more, please visit https://ethereum.foundation/
A
A
I'm
really
interested
at
and
I
think
a
lot
of
people
in
the
security
field,
sort
of
should
be
as
well
and
that's
Escape
patches
for
Roll-Ups
you've
heard
a
lot
about
Roll-Ups.
So
let's
give
a
brief
introduction
about
what
they
are
and
then
we'll
talk
about
the
specific
feature,
I
care
about
and
then
we'll
talk
about.
Some
features
that
those
features
should
have
so
what
a
good
roll-up
might
want.
So
this
will
be
sort
of
similar
to
my
colleague's
talk
from
earlier,
except
not
Bridges,
but
okay.
So
what
is
a
roll-up?
A
Well,
if
you,
you
know,
attended
a
lot
of
these
talks
here
at
Defcon
or
elsewhere.
You
know
you
see
a
bunch
of
more
or
less.
You
know
mutually
consistent
definitions,
but
essentially
it's
some
sort
of
layer
to
scaling
solution,
or
maybe
layer,
three
or
layer,
four,
if
you're,
really
Forward
Thinking
and
it's
some
system
that
just
takes
transactions
and
executes
them
off
chain
and
then
batches
them
and
compresses
them
and
does
something
to
get
you
some
sort
of
savings
and
pushes
some
data
onto
layer
one
and
so
essentially
it's
a
system.
A
That's
replicating
your
layer,
one
blockchain
like
ethereum,
but
it's
doing
it
elsewhere
with
some
nice
advantage.
That
gives
you
you
know
scalability
in
in
some
sense
or
or
savings,
and
so
how
does
the
system
work?
Well,
usually
you
get
some
transactions
and
then
your
system
design
is
very
dependent
on
the
implementation,
but
typically
there's
something
called
a
sequencer
that
gives
an
order
to
the
transactions
that
are
taken,
and
then
this
order
of
ordered
set
of
transactions
is
executed.
In
some
State
transition
function,
the
state
is
updated
and
something
gets
pushed
to
your
layer.
A
One
chain
where
it's
you
know
eventually
considered
final,
but
this
is
great
when
everything
works.
But
what
happens
when
things
don't
work
in
particular
if
your
sequencer
goes
offline
or
some
other
component,
possibly,
but
typically,
these
Escape
patches
are
discussed
in
the
context
of
sequence
or
failures.
A
What
happens
because
if
you
had
a
bunch
of
money
or
assets
or
tokens
locked
up
in
one
of
these
Roll-Ups
and
suddenly
the
transaction
that
says,
give
me
my
money
back
is
no
longer
sequenced,
you're
kind
of
screwed,
so
the
functionality
of
an
escape
hatch
is
some
way
to
get
program,
state
or
digital
assets,
because
you
know
tokens
nucleus
program,
State
off
of
this
roll-up
system,
even
when
your
sequencer
or
some
other
component
has
failed.
Now,
obviously,
there's
some
sort
of
caveat.
Some
things
probably
can't
fail
and
always
work
with
an
escape
hatch
right.
A
They'll
suggest
that
that
various
roll-up
projects
right
now
are
thinking
about
this
and
trying
to
implement
the
solution.
Now
some
of
them
have
the
none
there
we're
not
going
to
talk
about
the
specifics
and
I'm.
Also,
unfortunately,
not
I.
Don't
have
enough
time
to
talk
about
even
one
of
these
approaches,
but
if
you
are
interested,
let
me
know
or
go
contact
the
projects
and
say
Hey.
How
do
you
do
this?
But
even
when
they're
implemented
they're
often
not
exactly
done
in
some
cases
the
design
is
not
well
understood.
A
And
then
you
know,
ideally
you
fix
it
and
your
rollout
gets
back
up
or
whatever.
So
it's
being
done,
but
maybe
not
done
all
the
time
or
entirely
correctly
and
so
I'm
going
to
discuss.
You
know
what
I
think
some
nice
features
would
be
in
a
roll-up
now
these
would
vary
and
just
like
in
the
cases
for
bridges
for
features,
some
functionality
is
probably
contradictory
and
you
can't
necessarily
do
all
of
these
at
the
same
time,
but
there's
some
things
that
probably
should
be
nice
and
obvious.
A
First,
you
know
they
should
be
well
engineered
components
of
the
system.
They
that
usually
means
you
know
very
modular
and
in
case
you
need
to
upgrade
your
roll
up.
You
can
upgrade
part
of
it
or
all
of
it
or
just
this
part.
Maybe
this
is
the
only
part
that's
concerning,
because
if
you
saw
on
that
list,
some
people
just
force
a
transaction
whatever.
That
means
others
suggest
proposing
new
blocks
whatever
whatever
that
means.
But
you
know
this
should
be
a
full-fledged
feature,
not
an
afterthought.
Second,
it
should
be
secure.
A
You
know
big,
we
always
think
about
security
at
Quantum,
and
certainly
you
know,
a
roll-up
is
essentially
a
bridge,
that's
even
more
complicated
and
we
don't
want
to
have
to
you
know,
deal
with
exploits
that
arise
for
the
roll-up
because
of
this
functionality.
If
you
do
this,
do
it
right
and
you
know
make
sure
you
test
it
and
out
of
it
and
all
that
and
finally,
it
should
be
correcting
you
shouldn't
really
have
to
use
consecutive
Escape
patches.
This
is
a
bit
confusing,
but
I
have
heard
some
people
say
well.
A
So
ideally
something
that
gets
you
sort
of
out
of
the
system,
maybe
crafting
is
not
the
best
name
is
something
that
I'd
like
to
see
personally,
but
beyond
that
we
can
think
a
little
bit
deeper
about
what
it
means
to
escape.
From
a
roll
head
rolling
in
particular,
if
you're
taking
assets
out,
that's
pretty
well
understood
and
a
lot
of
them
do
have
functionality
for
forcing
transactions
for
erc20
tokens,
for
example.
But
what
do
you
actually
really
want
to
escape?
Maybe
you
want
to
escape
more
than
just
a
year
or
20
tokens.
You
know.
A
The
obvious
case
is
what,
if
you
have
an
nft
minted
on
your
roll-up?
Obviously
you
probably
want
that
back
too
without
having
to
go
through
extra
Hoops.
This
also
very
much
depends
on
the
actual,
bridging
that's
being
done,
but
you
know
maybe
there's
some
other
things
you
want
that
are
valuable
to
you.
They
should
be
built
in.
A
And
finally,
it
should
be
transaction
efficient
on
the
layer
underneath
if
you
escape
from
a
layer
or
two
roll-up
and
everyone
else
is
trying
to
do
this,
there's
going
to
be
a
gas
war
and
things
are
going
to
get
expensive
and
things
are
going
to
get
messy.
If
you
can
do
this
in
a
nice
way
that
Aggregates,
you
know
many
user
state
or
does
something
clever,
I
think
that's
that's
really
the
way
to
go
now.
I'm,
not
saying
I
have
solutions
for
these,
but
that's
just
what
I
want
to
see.
A
Some
other
properties
would
be
really
nice
be
Global,
so
in
particular,
if
you
call
an
escape
patch
functionality,
you
don't
have
to
do
it
for
every
dap
that
you've
deployed
on
that
L2.
So
if
I'm,
using
uniswap
and
something
else
on
some
roll
up,
I
want
to
just
say,
Escape
everything,
not
Escape,
uniswell,
then,
and
also
Escape,
whatever
the
other
dap
is.
That
would
be
ideal.
I'd
make
my
life
a
lot
easier
and
it
should
be
automatic
and
live.
Ideally,
it's
always
available
because
who
knows
when
the
system
could
go
down?
A
Hopefully
never,
but
you
know,
if
it
does,
we
don't
know
when
it'll
be
and
it
should
be
triggered
hopefully
automatically
so
we're
not
depending
too
much
on
social
consensus.
I
mean
it
might
be
unavoidable,
but
it
should
be.
You
know
very
clear
when
you
can
trigger
this
functionality
and
when
you
want
to
do
that,
it
should
be
available
to
you,
or
at
least
the
community
should
be
able
to
make
the
decision,
but
automatic
would
be
you
know
ideal.
So
that's
what
I
think
they
should
do
have
I
missed
anything.