►
Description
We had James Prestwich (https://twitter.com/_prestwich) from Summa (https://twitter.com/summa_one) join us for the Episode # 12 of our Whiteboard Series hosted by our Co-Founders Alex Skidanov (https://twitter.com/AlexSkidanov) & Illia Polosukhin (https://twitter.com/ilblackdragon)
More details about Summa: https://summa.one/
Download the video here: https://near.ai/ecfiw
------- About Whiteboard Series ---------
The Whiteboard Series with NEAR is a set of video interviews with Founders who are building scalable solutions for Blockchain. We deep dive into their story, what they're developing and how it is helping the ecosystem.
Follow the latest from NEAR Protocol on,
Discord: https://near.ai/discord
Medium: https://near.ai/medium
Twitter: https://near.ai/twitter
GitHub: https://near.ai/github
#WhiteboardSeries #FutureIsNEAR #Blockchain
A
B
My
name
is
James
Prestwich
I've
been
working
full
time
on
blockchain
stuff
for
the
last
five
years:
I
co-founded,
a
company
called
storage
back
in
2014,
which
is
still
going
for
a
year
in
a
bit
now,
I've
been
working
on
Summa
we're
a
design
and
implementation
firm
for
cross-chain
interoperability.
We
work
on
markets
that
link
Bitcoin
and
aetherium
and
other
chains
awesome.
B
There
is
there's
a
whole
like
history
of
interoperability,
solutions
here,
there's
like
so
many
different
approaches
and
nobody
has
written
it
down
and
nobody
seems
to
know
what
the
different
approaches
are,
and
so
we
actually
work
on
like
five
different
broad
tech
categories
for
interoperability,
but
explaining
them
is
a
little
difficult.
Sometimes
you
need
a
lot
of
background
information,
so.
B
B
The
right
direction
and
it's
a
set
of
transactions
and
a
block,
and
they
build
up
over
time
and
we
get
this
nice
state
out
of
them
in
Bitcoin
and
other
UT
x-bow
chains.
That
state
is
the
UT
Exocet,
which
is
the
unspent
transaction
outputs
in
aetherium.
We
have
this
nice
state
tree.
That's
everything
that
is
the
state
I,
don't
know
how
to
say
it
better
than
that.
That's.
B
Yeah,
it's
a
key
value,
store
encoded
in
a
Merkel
tree,
yeah
that
they
call
a
Patricia
tree
because
it's
some
like
a
special
expanded
Merkel
tree.
So
one
of
the
like
key
differences
between
Bitcoin
and
aetherium
is
that
in
Bitcoin
the
UT
Exocet
is
implicit.
It's
something
that
you
build
up
by
traversing
the
history
and
you
don't
commit
to
anywhere,
whereas
in
aetherium
each
block
commits
to
the
current
state
route
and
in
bitcoin
you're
building
the
UT
x
so
said
in
aetherium
you're
building
the
state
tree.
B
B
The
only
way
to
get
both
is
like
to
to
the
80
operations,
so
it's
really
expensive.
Okay.
So
what
we're
trying
to
do
is
take
these
completely
disparate
state
models
on
different
chains
that
have
completely
different
data
formats
and
layouts
really
about
the
only
thing
they
share.
Is
this
blockchain
structure
and
a
signature
scheme,
and
we
want
to
make
those
talk
to
each
other
in
meaningful
ways,
and
we
have
a
bunch
of
different
tools
for
doing
that
that
all
have
pros
and
cons
any
any
questions
so
far.
No,
it.
B
So
there
goes
my
marker
cap
I'm
not
going
to
worry
about
that
cross
chain
communication.
We
want
to
know
something
about
Bitcoin
from
aetherium.
For
example,
a
lot
of
these
can
be
generalized
to
any
pair
of
chain.
Sometimes
there's
really
specific
technical
considerations
here,
but
generally
we
just
want
to
know
something
about
Bitcoin
from
aetherium.
We
want
this
bit
this
etherion
block
to
reference,
this
Bitcoin
block
or
some
past
Bitcoin
block.
B
B
Spv
relay
and
merged
okay,
so
each
of
these
has
a
very
different
mechanism.
You
can
actually
like
mix
and
match
these,
so
you
can
combine
atomic
with
SPV
or
relay
each
of
these
has
very
different
chain
considerations,
so
you
can't
use
SPV
with
proof
of
stake
chains
at
all.
You
can
only
use
it
with
proof-of-work
chains.
B
B
B
Right
and
because
Bitcoin
uses
proof
of
work
and
proof
of
work
is
objective.
That
Oracle
could
be
anyone
if
we're
tracking
a
proof
of
stake
chain
using
relay.
We
have
to
follow
like
the
validator
set
changes
over
time,
which
is
doable.
It
just
can
get
a
little
weird
and
fuzzy
sometimes,
and
it
depends
on
which
consensus
system
we're
trying
to
track.
B
B
The
BTC
yeah
it's
it's
really
expensive
and
when
we
said
relays,
one-way
communication
like
eath
can
track
Bitcoin,
but
there's
no
way
to
make
Bitcoin
track.
Eath
anyways,
so
we
developed
this
is
our
contribution
is
SPV.
This
is
something
that
we
invented
back
at
eath
Berlin.
That
I
don't
think
anyone
else
has
ever
worked
on.
So,
instead
of
relaying
every
block,
we
relay
just
the
latest
few
blocks
to
prove
something
specific.
B
So
in
this
case
we
want
aetherium
to
inspect
some
specific
event
in
bitcoins
history,
and
we
have
the
etherium
contract
check
the
latest
few
blocks
and
that
Bitcoin
transaction.
So
you
take
the
transaction,
you
put
a
merkel
proof
of
inclusion
and
the
Bitcoin
header,
and
then
you
put
a
bunch
of
headers
on
top
of
that
to
show
that
there's
real
work
on
top
of
it
SPV
is
way
cheaper
than
relay
for
any
individual
transaction
for
any
individual
user.
Until
you
hit
like
several
thousand
regular
users.
B
A
B
B
But
it
does
know
about
this
T
X
and
these
three
blocks
and
it
can
determine
how
hard
those
blocks
were
to
make
how
expensive
they
were
by
checking
bitcoins
proof
of
work,
and
this
is
why
SPV
only
works
with
proof
of
work
chains
is
because
we're
actually
looking
at
these
blocks,
we're
parsing.
The
headers
were
examining
the
difficulty
and
we're
totaling
that
across
all
the
blocks-
and
that
gives
us
a
pretty
good
economic
idea
of
how
expensive
that
proof
was
to
make
how.
B
B
The
the
rule
of
thumb
right
now
is
that
six
Bitcoin
headers
at
the
current
difficulty
is
in
the
neighborhood
of
two
hundred
and
fifty
thousand
dollars,
and
so
as
long
as
this
transactions,
not
you
know
a
million
dollars.
Six
headers,
it's
going
to
be
more
than
enough,
and
it's
important
to
know
that
in
SPV
we're
not
doing
a
relay,
we're
not
tracking
the
whole
state
of
the
chain
over
time,
we're
just
getting
tiny
little
insights
into
the
chain.
And
so,
while
you
can
fake
proofs,
it's
very
expensive
to
do
so.
B
And
if
you
want
to
fake
a
proof,
you're
gonna
spend
a
quarter
million
dollars
faking
it.
But
if
I'm
an
honest
user,
the
main
chain
is
going
to
make
it
for
me
for
free
I'll
pay,
nothing
for
it
and
if
you
have
10%
of
the
hash
rate,
it's
going
to
take
you
ten
hours
to
make
six
block
headers,
but
the
main
chain
is
going
to
make
them
for
me,
the
honest
user
for
free
in
one
hour.
So
we
have
this
nice
economic
and
time
based
security
to
SPV.
B
A
B
A
B
You
would
lock
it
in
a
contract
and
specify
your
Bitcoin
payment
info
and
how
much
you
want
to
be
paid,
so
you
would
commit
to
a
transaction,
a
partial,
Bitcoin
transaction
that
pays
you
money
to
an
address.
You
control
and
anybody
can
come
over
here
on
Bitcoin
and
try
to
make
this
transaction
and
pay
you
that
money.
So
if
you're
selling
1/8,
maybe
you're
asking
for
a
point
O
to
Bitcoin
I'm,
not
actually
sure
what
ether
is
worth
today,
it
seems
to
change
a
lot
so
you've
committed
to
selling
your
ether.
C
B
It
to
the
contract
to
get
your
ether,
and
so
all
told
this
takes
about
an
hour
for
each
participant,
because
you
have
to
wait
for
some
blocks
to
confirm
it
can
be
actually
shorter
than
that,
and
one
of
the
supercool
things
is
that
anybody
can
try
to
buy
your
ether.
It
doesn't
have
to
be
me
like
with
an
atomic
swap.
B
B
B
B
B
You
could
steal
people's
Bitcoin,
so
you,
the
seller,
you've
locked
up
this
eath
and
you've
made
it
so
that
the
eath
is
completely
dependent
on
some
transaction
on
some
other
chain,
and
we
can
use
that
to
you
know
like
read
any
Bitcoin
state
that
we
want
as
long
as
it's
a
historical
state
and
we
don't
care
what
happens
after
that
point,
one
of
the
really
like
fun
tricks
that
we
use
in
production.
This
has
been
made
net
for
four
or
five
months
now.
B
A
B
What
if
there
are
two
people
trying
to
buy
at
the
same
time?
Basically
so
normally
this
would
be
a
race
condition,
but
we
have
tools
in
Bitcoin
to
make
transactions
mutually
exclusive,
and
so
we
can
use
a
Bitcoin
protocol
feature
that
as
far
as
I
know,
nobody
else
has
ever
used
called
sig
hash
single
anyone
can
pay
and
in
a
nutshell,
what
this
does
is
it
lets
you,
the
seller,
create
a
one-time
payment
invoice
where
anybody
can
pay
you,
those
Bitcoin,
but
only
one
person
can
do
so.
As
soon
as
one
person
is
paid.
B
Basically,
what
it
looks
like
is:
Bitcoin
transactions
are
a
list
of
inputs
and
a
list
of
outputs,
and
they
have
to
sum
up
to
the
same
amount,
and
so
you
as
the
seller
are
creating
this
transaction
that
sent
that
has
like
five
hundred
and
fifty
Satoshi
tiny,
tiny
little
amounts
of
money
less
than
a
penny.
And
it's
going
to
send
you
point
o2
out
right.
B
So
this
is
a
validly
formatted
transaction,
but
it
can't
go
on
chain
because
it
doesn't
balance,
and
so
anyone
can
come
in
and
make
it
balance
and
submit
it
to
the
chain
and
making
it
balance.
Pays
you
this
money
and
so
because
we're
including
this
tiny
little
dust
input
everyone
who
tries
to
make
that
TX
balance
and
tries
to
submit
it
to
the
chain.
They
are
all
mutually
exclusive.
Yeah.
B
B
And
right
on
goal,
so
the
if
we're
using
this,
we
need
to
specify
the
end
goal
if
we
want
to
make
it
market
based.
The
alternative
is
that
you
don't
specify
what
you're
going
to
be
paid,
but
the
smart
contract
refuses
to
dispense
your
ether
to
anyone
who
pays
the
wrong
amount,
and
in
that
case
the
buyer
has
some
risk
because
they
don't
know
whether
they're
paying
the
right
amount
and
there's
a
whole
like
bunch
of
other
problems
to
solve
there.
But
we
think
it's
doable.
B
B
Okay,
so
this
is
a
lot
of
what
we're
working
on
recently.
This
is
what's
new
that
we've
come
up
with
these
things
have
been
kind
of
known
for
a
while,
and
the
last
thing
we
haven't
covered
is
the
atomic
swap,
which
is
the
most
well
known
of
these
and
was
invented
in
2014
by
a
Bitcoin
talk
user
that
goes
by
tier
Nolan.
So
we
call
it.
C
B
B
C
B
C
B
Are
right
so
I'm
gonna?
Do
this
you're
gonna
see
me,
do
it
and
check
my
work
and
wait
for
it
to
confirm
so
it
doesn't
roll
back
and
then
once
you're
happy
we're
gonna
go
down
here
and
you're
gonna
make
the
same
commitment
with
your
ether,
except
instead
you're
gonna.
Do
it
in
Reverse
you're
gonna
pay
James
if
Kay
Yuliya
after
time.
So
each
of
these
is
you
know
kind
of
this
nice
symmetric
contract
right
this
one
pays
you
if
you
know
the
secret
and
this
one
pays
me
if
I
reveal
the
secret.
B
So
this
has
to
end
States
once
we've
both
escrowed
one,
is
that
I
reveal
the
secret
and
I
take
your
ether,
and
then
you
use
the
secret,
which
is
public
on
aetherium
to
get
my
Bitcoin
right.
The
other
is
that
I
don't
reveal
the
secret
and
I
just
wait
to
get
my
Bitcoin
back
and
you
wait
to
get
your
ether
back
there's
once
we've
both
escrowed
there's.
No
other
outcome.
Well,.
B
B
Apparently,
however,
on
a
blockchain
time
is
all
sorts
of
wibbly-wobbly
time
is
actually
set
by
miners
and
miners
can
influence
how
it
progresses,
and
so
we
have
to
set
these
timeouts
very
carefully
so
that
your
timeout
always
happens
before
mine
yeah.
So
mine
has
to
go
this
long,
and
yours
has
to
be
this
long
right,
so
that
ensures
that
I
can't
both
get
my
refund
and
get
your
money.
B
However,
the
lower
the
hash
rate
of
the
chain,
the
more
able
I,
am
to
influence
time
by
like
using
nice
hash
or
something
some
hash
rental
service,
and
if
I
can
speed
up
time
on
this
chain
enough,
then
the
security
assumptions
broken
and
while
it's
unlikely
time
is
probabilistic,
even
in
the
best
case.
So
it
might
be.
You
just
get
really
unlucky
that
day
and
I
get
to
scam
you
it's.
B
The
other
thing
to
know
is
we're
both
escrowed
in
here
and
so
for
this
whole
duration.
We
can't
use
our
money
for
anything
right
and
I
can
decide
because
I
know
Kay
and
you
don't
I
can
decide
when
in
this
period,
I
want
to
make
the
trade,
and
so
that
leads
to
what
we
call
the
free
option
problem.
The
trade
happens
at
my
choice,
so
I'll
just
delay
and
see
how
the
exchange
rate
goes
yeah.
B
So
a
lot
of
thought
gets
put
into
fixing
the
free
option
problem
in
atomic
constructions
and
I,
don't
think
anyone's
come
up
with
a
great
solution.
Yet
we've
designed
a
few
constructions
where
you
can
put
in
like
a
premium
payment
for
that
option,
and
you
can,
you
know,
intend
for
it
to
be
an
option.
Then
it's
no
longer
free.
Some
other
people
have
put
in
multi-phase
HT
LCS,
where,
if
I
don't
take
it
I
will
be
penalized,
but
that's
not
even
the
only
problem
with
this
basic
atomic
construction.
B
What,
if
you
just
never
escrow
what?
If
this
just
never
happens
down
here
now,
I've
escrowed,
my
funds
and
you
just
walk
away
and
I'm,
just
locked
up
and
I
get
nothing.
And
so
you
can
liquidity.
Do
s
me?
You
can
just
throw
me
and
convince
me
to
lock
up
money
for
no
reason:
yeah,
okay,
and
so
the
other
thing
is
that
we
we
have
to
agree
on
all
the
terms
in
advance.
I
can
just
never
do
this
and
then
you'll
never
do
anything,
and
then
we've
just
agreed
on
terms.
But
you
know.
B
C
B
A
B
So
for
atomic
stuff,
it's
like
super
cheap
to
verify
a
few
hundred
gasps.
It's
easy
for
SPV.
Our
current
implementation
does
the
verification
in
less
than
seven
hundred
thousand
guess,
and
so
what
this
is
doing
is
loading
in
a
Bitcoin
transaction,
a
merkel
proof
which
is
going
to
be
twelve
or
thirteen
hashes
so
somewhere
in
the
neighborhood
of
like
four
hundred
bytes
and
then
loads
in
usually
five
or
six
Bitcoin
headers,
which
is
eight
hundred
I'm.
B
Sorry,
four
hundred
or
480
bytes
give
or
take
it
does
30
40
shots
on
all
of
that
yeah,
most
of
it
most
of
the
expense.
Right
now
is
loading
in
the
data
and
manipulating
bytes.
Getting
the
right
byte
strings
together
totally
does
not
give
you
very
good
tools
for
slicing
and
concatenating
bytes
yeah,
but.
B
Did
a
blog
for
garage
it
won't.
This
is
something
that
Bitcoin
actually
does
very
well
is
because
the
block
time
is
long.
We
need
many
fewer
headers.
If
we
want
the
same
amount
of
security
on
aetherium,
we
need
something
like
200
headers
and
then
we
have
to
do
eath
hash
on
each
of
those
to
check
the
work
and
eath
hash
in
evm
costs
about
2.5
million
gasps,
and
so,
if
you
want
to
spend
500
million
gasps,
that's
how
many
blocks
again
each
block
is
like
8.
B
So
you
know
if
you
want
to
take
like
65
blocks
processing.
One
proof
like
that's,
that's
fine
go
ahead
and
and
do
that
it
doesn't
seem
very
practical.
So
remember,
I,
said
earlier:
there's
a
bunch
of
like
weird
technical
considerations.
This
is
one
of
them
is
actually
it's
going
to
be
cheaper
to
do
this
for
a
Bitcoin
than
for
basically
any
other
chain,
because
the
blocks
are
slow.
Yeah.
B
B
We
can
do
it
with
because
of
the
long
block
time
and
another
super
important
thing:
bitcoin
uses
sha-256
as
its
proof
of
work
and
if
every
other
chain
has
shot
yeah
and
so
litecoin
uses
s,
crypt
and
z,
cache
uses,
equi,
hash,
209
and
grin
uses,
cucit,
rootin
tootin
something
and
beam
uses,
equi,
hash,
150,
something
and
D
cred
uses
Blake
and,
like
we've,
got
all
of
these
proof-of-work
functions
and
solidity
can't
actually
check
any
of
those.
It
can
only
check
sha
too.
B
And
so
it
would
almost
be
better
to
have
a
OS,
or
it
would
almost
certainly
be
better
to
have
a
OS
checking,
Bitcoin
and
aetherium
checking
Bitcoin
yeah.
Unfortunately,
there's
some
limitations
in
what
you
can
do,
because
you
have
to
encode
data
into
Bitcoin,
somehow,
which
means
you
have
to
have.
You
know
inbound
communication
to
Bitcoin,
but
we
we
think
we
have
some
solutions
to
that.
It's
still
a
little.
B
We
haven't
like
written
it
up
or
formalized
it
yet
so
I
don't
want
to
talk
about
it
too
much.
This
is
all
of
this
is
ongoing
research.
Everyone
in
East
two
is
thinking
about
how
to
merge
consensus.
We've
had
multiple
relay
designs
that
have
failed
to
take
off.
Spv
is
brand
new.
We
came
up
with
it
six
months
ago
and
atomic
constructions.
I
guess
that
are
the
oldest
they've
been
around
for
five
years,
but
nobody's
actually
deployed
a
really
good
system
using
them.
B
A
C
B
I
think
you
could
probably
patch
that
in
after
the
fact,
if
you
willing
to
hard
fork-
or
maybe
even
soft
fork,
one
network
mark
freedom,
Bach
wrote
a
really
interesting
paper
about
how
you
can
soft
fork
anything
into
Bitcoin.
Literally
anything,
you
could
change
the
proof
of
work,
you
could
change
the
state
model.
You
could
just
do
anything
as
a
soft
fork,
but
nobody
has
really
talked
about
patching
this
into
chains
that
exists
very
much
and
it
seems
to
us.
You
know
that
nobody
no-nobody
like
has
expressed
a
desire
to
do
that.
B
Yet
a
lot
of
chains
are
really
tribal
in
their
community
and
if
you
talk
about
merging
aetherium
consensus
with
tezo's
consensus,
people
get
really
like
squirrelly
about
it.
For
for
us,
we
like
anything
that
promotes
cross
chain
interoperability,
but
we
have
to
deal
with
a
lot
of
partisans
and
a
lot
of
maximalists
and
sometimes
sometimes
it's
a
little
difficult
and
it
actually
does
I
think
have
a
chilling
effect
on
what
research
gets
done.
I.
A
B
A
B
It's
it's
really
interesting
because
everybody
is
going
about
this
in
different
ways.
So
far,
cosmos
is
all
building
around
IBC
and
making
it
easy
to
spin
up
a
new
zone.
Polkadot
has
their
pair
of
chains.
Eath
2
is
eventually
going
to
have
these
beacon
to
chard
cross
links,
but
we're
still
years
away
from
that
everybody's
taking
different
approaches.
Nobody
knows
what
the
right
model
for
crush,
shard
or
cross
train
communication
in
their
ecosystem
is
what
we
like
to
focus
on.
Is
you
know
when
you
don't
have
control
over
the
system
when.
C
B
Working
with
deployed
things
that
aren't
going
to
change
like
Bitcoin
and,
to
a
lesser
extent,
aetherium
is
not
going
to
change.
How
can
we
get
some
of
those
benefits
and
we're
also,
you
know,
spinning
up
research
about
how
to
improve
communication
in
merged
systems?
How
can
we
go
eat
to
cross-train
faster
than
for
consensus
cycles,
which
you
know
sounds
pretty
for
consensus
cycles
is
like
25
minutes
or
so,
and
eath,
supposedly,
which
sounds
pretty
good
cross.
A
B
B
Our
communication
is
going
to
be
extremely
important
for
like
system
composability
and
for
users
and
so
we're
starting
a
lot
more
research
into
that,
and
we
think
we
have
some
good
models
to
do
it
with
economic
security
much
faster.
But
again.
This
is
something
that
I
gotta
write
down
and
talk
to
protocol
designers
about
and
designing
for,
East
is
kind
of
a
kind
of
a
losing
game
right
now
anyway,
because
it
doesn't
exist
and
we
don't
know
what
it
will
look
like
when
it
does
exist.
We
have
a
specification
that
changes
regularly,
but.
B
B
D
B
B
B
B
B
B
That
makes
perfect
sense
we're
talking
about
like
hacking,
this
into
eath
to
after
the
fact
which
again
ties
back
to
my
earlier
point.
What
we
care
about
is
adding
interoperability,
where
it
wasn't
designed.
You
know
in
place,
and
this
always
means
anything
communicating
with
Bitcoin
Bitcoin
was
not
designed
for
interim,
but
at
the
same
time
Bitcoin
has
more
than
half
of
the
liquidity
in
the
space,
and
so
it
wasn't
designed
for,
interrupts
and
interrupts
with.
It
is
super
important
I
got
a
research,
your
BF
to
you
more
I
mean.
B
B
Have
thought
about
it
so,
with
our
current,
like
SPV,
swap
design,
there's
an
order
book
that
lives
on
chain
and
etherium
and
everything
is
filled
by
id
self
matched.
So
you
look
at
this
on
chain
contract.
You
see
what
orders
are
available
and
you
fill
the
ones
you
want,
they're,
all
like
a
fixed
price
and
a
fixed
size
which.
B
The
user
doesn't
need
a
coordinating
chain.
The
user
needs
access
to
liquidity,
that's
already
out
there.
They
just
need
to
figure
out
where
the
liquidity
is,
and
we
can
do
that.
You
know
just
as
effectively
by
looking
at
many
chains
using
like
clients,
so
our
app
today
has
a
Bitcoin,
Mike
client
packaged
in.
We
rely
on
API
for
aetherium
right
now,
working
on
an
aetherium
like
client.
B
Yeah
yep
one
of
the
nice
things
about
shorting
designs,
cool.
Oh
the
last
thing
that
we
kind
of
add
to
this
list
sometimes
is
Federation's
and
I'm
gonna
separate
that
from
the
rest
of
the
list,
because
what
we're
going
for
with
most
of
the
things
we
work
on
is
economic
assumptions.
All
of
these
rely
on
some
someone
loses
some
amount
of
money
if
they
go
wrong
right
and
we
have
Economic,
Security
Federation's
rely
on
honesty
instead,
which
is
not
my
favorite
assumption,
so
so.
A
B
We
we've
thought
really
a
lot
about
that.
The
main
question
for
us
and
something
that
we're
actively
thinking
about
and
researching
is:
can
we
make
it
so
that
those
validators
themselves?
You
know
these
custodians
holding
BTC
on
someone's
behalf?
Can
we
make
it
so
that
they
have
risk
of
loss?
Can
we
make
it
so
that
they
will
lose
money
if
that
BTC
gets
moved
in
appropriately
and
that
converts
it?
That's
what
you
know
convert
from
a
federation.