►
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 IPFS and more.
https://archive.devcon.org/archive/watch/6/a-new-defi-primitive-private-trading-with-aztec-connect/index
Trading on decentralised exchanges is missing a key primitive, **privacy**. In traditional markets, an OTC desk enables large orders to trade without moving the price on chain this is missing and everyone can see everything.
ZKP's deployed in production allow us to implement private trading on Ethereum. This talk looks into:
1. How to uses ZKP's to give privacy on Ethereum
2. How to build a private dollar cost averaging smart contract using Aztec Connect
3. A code walkthrough and a demo
Speaker(s): Joe Andrews, Jan Beneš
Skill level: Intermediate
Track: ZKPs: Privacy, Identity, Infrastructure, & More
Keywords: ZKPs
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
So
we
can
see
quite
commonly
that
people
use
their
ens
names
with
trading
because
they
apparently
have
no
idea
that
is
not
private
and
to
drive
this
point
home
like
basically,
there
is
this
ensname
docs.if
and
if
I
want
to
spend
some
time
on
it,
I
can
just
like
basically
figure
out
that
the
user
has
units
web
position.
I
have
a
position
and
how
amount
of
USD
available
for
collateral-
and
you
might
be
asking
why
this
is
bad,
because,
like
plenty
of
the
times
it's
considered,
the
transparency
is
like
a
nice
feature.
A
But
when
we
are
talking
about
financial
information,
it
can
be
bad
because,
like
it
allows
basically
anyone
to
assess
solvency
of
any
entity
on
ethereum
and
this
information
can
be
used
by
adversarial
parties
like
Mev
Bots,
and
then
there
is
also,
of
course,
a
risk
of
doxing
and
consequences
in
the
physical
world.
This
can
be
as
innocent
as
your
friends
shaming
you
for
your
bad
trades,
but
it
can
also
make
you
a
target
of
crime,
so
the
solution
at
Aztec.
We
think
that
the
solution
are
privacy,
preserving
roll-ups
and
our
privacy.
A
Preserving
group
is
called
Aztec
connect.
It
is
live
on
mainnet
and
we
like
to
call
it
VPN
for
ethereum
and
I
think
this
is
quite
a
nice
mental
model,
because
essentially,
what
VPN
is
doing
in
the
webto
world
is
that
it
replaces
your
IP
address
with
the
IP
address
of
the
VPN
provider,
and
essentially
what
we
do
is
that
we
replace
your
ethereum
address
with
if
ethereum
address
of
Aztec.
A
A
Then
we
will
create
your
account.
This
will
take
a
while
because
there
are
some
complex
computation
happening
behind
the
scene
and
then
you
deposit
the
amounts
of
funds
and
behind
the
scene.
That
actually
happened
quite
a
few
things
that
happened:
two
transactions
on
ethereum
one
deposits,
the
funds
into
a
royal
processor
contract,
another
approved
this
approves
this
hash
and
then
there
is
like
a
generation
of
cryptographic
note
on
L2
and
this
cryptographic
node
allows
you
to
spend
those
funds.
A
So
once
you
deposit,
this
is
what
it
looks
like
it's
like
a
normal
wallet
and
then
there
are
opportunities,
so
you
can
deposit
into
an
arbitrary
Bridge
we've
implemented.
So,
for
example,
if
I
deposit
to
Euler
I
have
this
screen-
and
there
are
quite
a
few
interesting
things
here
which
I
want
to
explain.
A
So,
basically,
when
there
are
like
10
users,
which
want
to
deposit
each
to
Lido
and
let's
say
for
Simplicity,
each
user
has
one
if
we
only
call
the
stake
function
on
Lido
contract
once
with
the
10
ether
and
then
later
on.
We
distribute
the
results
to
the
users
and
in
this
case,
like
the
results,
would
be
the
output
tokens
which
is
staked.
A
Yeah
I
would
like
to
briefly
pick
what
is
behind
the
scenes
and
we
use
utxo
model.
It
was
talked
about
by
quite
a
lot
in
the
Aztec
3
presentation
before,
but
basically,
what
is
a
utxo
model
is
that
you
can
imagine
when
you
have
banknotes
and
you
want
to
go
to
a
merchant
and
you
want
to
pay
with
the
banknote.
A
So
what
happens
is
that
you
can
like
split
this
banknote,
but
you
give
the
merchant
the
banknote
and
he
gives
you
the
money
back,
and
this
is
essentially
what
we
do
as
well
and
we
do
it
with
this
thing,
join
split
circuit
and
essentially
what
this
allows
is
that
you
have
two
possible
nodes
on
the
input
which
get
destroyed
in
the
process,
and
then
there
are
two
nodes
on
the
output
and,
for
example,
in
the
example
of
of
the
bridge.
Interact
of
the
D5
deposit
to
Lido
on
the
input
would
be,
for
example,
0.5.
A
If
and
in
case
the
original
value
held
in
your
account
was
one
note
with
one.
If
this
node
would
get
destroyed
and
on
the
output
would
be
this
claim.
Node
and
another
volume
node
that
value
node
would
hold
the
remaining
heat
balance,
so
it
would
be
0.5
if
and
that
claim
node
would
allow
you
to
later
on
claim
the
funds
resulting
returned
from
the
defy
interaction
yeah.
A
So
you
might
be
asking:
how
is
this
possible
that,
like
this
is
private
when
the
settlement,
like
the
interaction
actually
happens
on
ethereum?
And
this
is
a
good
point
and
here
in
the
left
bottom
corner,
we
can
see
what
it
looks
like
like
what
token
transfers
there
are-
and
this
looks
pretty
not
private,
but
the
thing
is
that,
thanks
to
zero
knowledge
proofs,
nobody
can
distinguish
like
who
is
actually
originated.
A
This
interaction,
and
also
there
is
this
aspect
that
there
can
be
multiple
users
per
interaction,
and
so
these
balances
get
even
more
obfuscated,
and
currently
there
are
more
than
50
000
unique
users.
Last
time
I
checked
it
was
like
51
000
something,
and
it
means
that,
like
the
amount
of
people
you
get
height
with
is
spiritual
Arch,
you
might
be
wondering,
does
it
scale?
A
Luckily,
this
cost
of
posting
a
roll
up
gets
split
between
many
users.
So
when
there
are
many
users
in
develop,
this
goes
down
and
the
cost
of
call
data
is
fortunately
going
to
down
a
lot
with
this
ethereum
Improvement
proposal
for
844,
so
hopefully
like
in
a
not
so
distant
future.
The
cost
of
interacting
privately
with
D5
will
actually
get
very
close
to
the
gas
cost
of
that
one
L1
call
divided
by
the
number
of
users.
A
Now
you
might
be
wondering
like
how
do
we
integrate?
Is
there
like
some
integration
needed?
And
unfortunately,
yes,
but
it's
not
so
complicated,
and
we
have
this
thing
called
Bridges,
but
Bridges
nowadays
have
quite
a
bad
connotation
and
in
our
case
the
bridge
is
just
a
simple
smart
contract.
So
there
is
no
multi-stick
risk
and
the
smart
contracts
tend
to
be
fairly
simple.
A
So
here
on
the
left,
we
see
the
ID
vibration
interface
and
there
we
basically
just
Define
input
and
output
tokens.
Then
we
have
total
input
value,
which
is
the
sum
of
those
two
users
input
balances
and
aux
data,
which
is
arbitrary
data
which
can
be
used
to
pass
to
the
bridge
contract
and
on
the
bottom
right
corner.
We
can
see
the
implementation
of
the
con
of
the
function
itself
and
in
this
case
it's
an
implementation
of
an
erc4626
bridge,
and
we
see
that
it
is
like
a
relative.
A
Now
you
might
be
asking
if
there
are
some
limitations-
and
there
are
quite
a
few
I
would
say
the
biggest
one
are
the
longer
settlement
time,
and
that
is
because,
when
you
you,
we
have
to
wait
for
the
batch
to
get
submitted
to
to
ethereum
like
luckily,
you
can
pay.
If
you
are
really
in
a
hurry,
so
you
can
pay
for
the
full
roll-up
block,
and
in
that
case
you
could
get
the
instant
settlement
time,
but
most
users
don't
want
to
do
it
because
it's
fairly
expensive
and
this
causes
like
some
problems
with
slippage.
A
So,
for
example,
you
can
imagine
when
you
trade
on
uniswap,
you
define
a
slippage
parameter
which
basically
protects
you
from
Mev
Bots,
exploiting
you
or
in
general.
It
can
protect
you
from
Market
movements
and
the
longer
settlement.
Time
can
also
cause
issues
with
liquidations,
because
you
can
imagine
if
we
are
close
to
being
click
related,
and
you
need
like
fast
to
exit
the
position
or
something
in
the
that
is
problematic.
But
always
you
can
pay
for
the
instant
settlement.
A
Then
there
is
the
limitation
by
the
amount
of
data
which
can
be
on
the
input
like
most
of
the
time.
This
works
quite
fine,
because
it's
64
bits
of
data
and
with
like
a
bit
of
smart
encoding,
we
can
actually
put
a
lot
of
information
there,
but
it
can
be
problematic
when
we,
for
example,
want
to
pass
in
ethereum
address,
because
that
is
160
bits.
A
Then
another
limitation
might
be
that
the
system
doesn't
work
that
well
with
esoteric
tokens,
because
those
esoteric
tokens
might
have
insufficient
anonymity
sets,
and
basically
what
that
means
is
that
there
is,
unlike
one
esoteric
token,
you
user
and
he
would
deposit
into
Aztec
and
he
would
be
the
only
one
like
working
with
the
D5
protocols.
It
would
be
pretty
obvious
that
is
that
one
user,
but
in
practice,
I
think
this
is
not
that
big
of
a
deal,
because
usually
users
want
to
use
the
high
liquid
tokens
and
I
don't
feel.
Personally.
A
A
So,
for
example,
when
when
those
protocols
just
work
with
tokens,
you
can
imagine
that
there's
like
token
on
the
input
and
token
on
the
output,
which
is,
for
example,
the
case
of
Lido
like
this
works
perfectly,
because
there
is
no
limitation.
We
just
have
tokens
on
the
input
and
then
we
distribute
the
results,
but
when
the
protocols
work
specifically
with
message
sender,
which
can,
for
example,
can
be
when
we
are
borrowing
LUSD
from
liquite.
A
Oh,
it's
pretty
complicated
to
do
the
implementation
and
how
we've
solved
it
is
that
we
deploy
these
multiple
Bridges
and
each
of
the
bridge
has
fixed
collateral
issue
and
like
if
the
bridge
was
like
close
to
being
liquidated,
user
can't
add
more
collateral,
but
the
user
can
always
repay
the
depth
and
exit.
So
even
though
there
are
limitations,
usually
there
is
like
some
way
how
to
work
around
them.
A
Now.
I
would
like
to
talk
about
advantages
and
I.
Think
those
advantages
are
pretty
huge,
because
Aztec
connect
doesn't
need
redeployment
of
smart
contracts
on
L2
because
we
work
directly
with
smart
contracts
on
L1,
and
this
means
that,
for
example,
liquidity
is
not
fractured
and
I
did
my
own
personal
experiment
when
it
was
blue
market
and
I
checked
like
how
much
it
makes
sense
to
do,
for
example,
swept
on
L2,
and
it
turned
out
that
for
not
so
large
amounts
like
over
five
thousand
dollars.
A
Even
with
the
high
gas
fees,
it
still
made
sense
to
use
ethereum
L1,
and
that
is
because
the
liquidity
was
just
so
much
higher.
So
the
slippage
would
be
just
so
much
slower
that,
like
the
piece
would
be
worth
it
and
another
Advantage
is
a
lower
risk
for
integrated
protocols,
and
that
is
since
we
don't
need
to
the
Target
protocols,
don't
need
to
like
redeploy
again
on
on
our
L2.
A
Then
I
would
like
to
mention
that
there
was
of
course,
95
cases,
but
the
fire
is
like
a
nice
password,
so
we
mentioned
it
here
and
I.
Think
very
cool
use
case
could
be
private
doll
voting,
and
that
is
because,
in
all
the
voting
systems,
privacy
is
very
important
and
another
interesting
use
case
might
be
nft
mending
yeah.
Thank
you
for
the
attention.
B
A
C
A
Okay,
so
our
contract
is
implemented
in
solidity,
so
we
can
deploy
it
to
any
evm
chain
and,
like
our
L2
infrastructure,
just
uses
generic
Json
RPC
so
like
there
is
no
fundamental
issue.
Why
that
would
not
be
possible,
but,
like
an
interesting
note,
is
that
it
might
not
make
sense,
because
usually
those
other
l2s
have
also
like
have
to
post
the
call
data
on
ethereum
L1.
So
it
means
that
if
you,
for
example,
were
running
over
optimism,
we
actually
would
not
save
much
gas.
A
F
I
got
the
mic
ready,
so
one
of
my
my
question
was
around
what
you're
doing
to
try
to
deal
with
this
settlement
like
waiting
times
and
stuff
and
I
was
wondering
if,
like
as
you,
integrate
with
other
protocols,
if
that's
like
a
breaks,
Community
or
brings
in
composability
challenges
and
kind
of
how
you're
thinking
about
solving
that.
In
the
long
run,.
A
Yeah,
so
there
are
no
composable,
it
is
challenges
because
it's
running
on
ethereum
L1,
like
the
calls
and
like
the
best
how
we
can
deal
with
the
settlement
times,
is
basically
having
enough
users
that
it
the
that
it
makes
that
it's
profitable
to
submit
the
roll-up
blocks
frequently.
But,
as
we
said,
if
there
is
like
some
user
with
your
origin
plan
needs
it
to
be
settled,
he
can
always
like
pay
for
those
instant
settlement.
Much.
G
A
Yeah
exactly
it
would
be
on
by
the
Aztec
address,
but
there
is
this
thing
called.
We
don't
currently
like
support
nfts
directly
in
our
UI,
but
we
have
this
thing
called
virtual
assets,
which
basically
allows
like
a
representation
of
any
ownership
so
like
on
ethereum.
Of
course,
it
would
like
stay
within
ether
within
the
bridge.
It
would
stay
within
the
bridge
contract
but
yeah.
The
user
could
then
interact
with
it
by
using
the
virtual
assets
virtual
assets.
Basically,
what
it
does
is
that
it
like
doesn't
exist
on
ethereum
L1,
but
we
create
this
another
value.
A
H
A
H
E
A
Yeah,
so
currently
how
we
approach
it
is
that
we
limit
the
deposit
side
size
and
basically,
like
our
goal,
is
to
avoid
the
use
of
hackers,
avoid
users,
hackers
using
our
platform
and,
like
the
big
issue
with
tornado
cash,
was
that,
like
70
percent
of
the
money
there
was
elicited.
So
basically,
we
are
just
trying
to
make
it
impractical
for
hackers
to
use
our
platform.