►
Description
Whiteboard Series with NEAR | Ep: 32 Tim Moreton from Celo
Celo Engineer Tim Moreton (https://twitter.com/@timmoreton) and NEAR Co-Founders, Illia Polosukhin (https://twitter.com/ilblackdragon) and Alex Skidanov (https://twitter.com/AlexSkidanov) do a deep dive on permissionless platforms, new “ecologies of money”, and how Celo’s technology works in the 32nd episode of the Whiteboard Series.
Celo is an open platform that makes financial tools accessible to anyone with a mobile phone.
------- 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,
Website: https://nearprotocol.com/
Discord: https://near.ai/discord
Medium: https://near.ai/medium
Twitter: https://near.ai/twitter
GitHub: https://near.ai/github
#WhiteboardSeries #Blockchain #FutureIsNEAR
B
Great
to
be
here
and
so
I'm
Tim
Morton
I'm
on
the
engineering
team
at
say
way
before
this
I
ran
a
bunch
of
teams
at
Apple
running
their
cloud
infrastructure
I
have
a
background
in
distributed
systems.
So
when
you
guys
start
grilling
me
on,
like
the
economics,
we'll
see
how
how
about
guys
yeah
I'm
gonna
talk
about
selling
sellers
a
permissionless
platform.
B
B
Then
bank
accounts
like
there
are
1.7
billion
people
without
access
to
regular
financial
services,
and
if
we
can
close
that
gap
a
little
bit
and
empower
some
people
to
be
able
to
like
send
money
between
friends
and
pay
at
merchants
and
like
have
access
to
all
the
things
that
we're
used
to,
then
that
would
be
like
a
great
achievement.
Yeah.
B
Them
yeah
opponents
and
oh
yeah
yeah,
so
one
things
is
interesting
about
sello
in
the
spaces,
we're
very
much
like
a
product
company.
We're
focused
on
like
trying
to
make
an
end
user
experience,
which
is
so
seamless
that
you
don't
even
really
know
that
there
is
a
cryptocurrency
or
a
decentralized
back-end.
B
B
Like
a
low-end
mobile
device,
so
we
care
about,
like
you
know,
we
connectivity
how
much
data
we're
transferring
how
much
processing
power
and
if
that's
like
you,
present
some
interesting
technical
challenges.
So
to
do
that.
We've
sort
of
taken
this
like
whole
stack
approach.
You
can
sort
of
think
of
cells
like
three
things,
so
the
bottom
layer
is
like
the
blockchain.
So
you
can
call
this
like
solo,
blockchain
I'll
come
back
and
like
describe
these
things
a
bit
more
detail
in
a
second,
then
we
have
like
a
smart
contract
layer.
B
Really
like
most
of
the
business
logic
of
the
stack,
and
then
we
have
a
seller
wallet
and
the
obviously
most
of
our
users,
like
sort
of
sit
here
and
using
the
header
using
their
sort
of
devices.
This
you
can
sort
of
think
of
this
as
like
a
platform
in
itself,
there's
like
a
lot
of
opportunities
to
build,
like
other
things,
both
directly
on
top
of
the
smart
contract
layer.
Anybody,
of
course,
could
deploy
smart
contracts,
the
platform,
but
also
we
have
a
so.
We
have
like
a
couple
of
things
to
make
that
easier.
B
We
have
I,
guess
you
can
sort
of
think
of
it
as
being
like
here.
This
is
where
the
diagram
gets
a
little
confusing.
This
is
like
I
guess,
contract
kit,
which
is
a
layer
to
make
it
really
easy
to
like
access
the
sort
of
first
party
contracts
that
existed
in
the
business
logic
layer
here
and
then
the
wallet
also
uses
that
and
then
the
wallet
as
well
exposes.
B
B
B
C
B
Push
America
exactly
exactly
and
so
like
then,
basically
talks
to
a
service
he'll
talks
to
like
Google
or
Apple
or
whatever,
and
sends
it
push
notifications.
While
it's
saying
hey
tin
has
just
paid
you
like
whatever
and
so
there's
a
couple
of
things
there
so
they're
really
around,
like
optimizing
user
experience
and
around
performance,
but
broadly
everything
is
like
these
centralized.
So
the
smart
contract
layer
is
like
the
sort
of
business
logic
layer,
I'll
come
back
to
like
dig
into
that
and
radicand
and.
A
B
Talk
about
yeah,
let's
get
to
the
heart
of
it,
so
the
blockchain
is
so.
The
blockchain
is
actually
a
fork
of
the
Go
etherium
client.
This
is
an
incredibly
like
ambitious
project.
There's
a
lot,
there's
a
big
big
surface
area.
One
of
the
things
we
wanted
to
do
was
because
we
care
about
user
experience
was
to
try
and
like
get
to
market
as
quickly
as
possible,
and
so
we
looked
around
rather
than
building
like
a
blockchain
from
scratch
ourselves.
We
look
to
sort
of
build
on
some
of
the
great
technology.
B
But
yet
broadly,
it's
around
like
changes
for
usability
and
changes
for
mobile
being
mobile
first
and
then
also
like
I.
Think,
since
the
theorem
was
arranged
like
the
thinking
in
general.
In
the
space
around
proof
of
work
versus
proof
of
stake
has
moved
on
a
lot,
and
so
one
of
the
big
changes
we
made
was
an.
A
B
B
Do
we,
how
do
we?
How
do
we
do
that?
So
we
started
off
some.
You
know
I,
remember,
there's
a
there's
a
quote
on
I
think
it
is
I'm
gonna
get
this
wrong.
I
think
it
is
the
curator
recipe
page
for
using
zookeeper.
So
you
know,
I
have
a
lot
of
experience
with
a
zookeeper.
A
lot
of
experience
like
I,
mean
my
sins
I
once
implemented
like
twice
actually
for
my
sins,
implemented
Paxos.
It's.
B
Be
applied
to
like
distributed
consensus,
it's
really
gnarlie.
There
are
tons
of
edge
cases.
There
are
very
few
people
in
the
world
who
really
like
deeply
understand
it.
Maybe
you
guys
do
I
think,
like
you
know,
obviously,
obviously
I
think
we
I'm
just
selling
ourselves
short
I,
think
we
we
have
gained
that
experience,
but
in
all,
in
order
to
like
get
further
fast,
that
we
sort
of
looked
around
to
see
where
there
was
a
good
starting
point.
B
Right,
that's
right,
not
the
not
the
Istanbul!
That
has
the
sort
of
hard
folk
that
the
fork
of
etherium
that's
happening
right
now,
but
actually
a
separate
implementation
of
a
Pratt
who
buys
an
time
from
tolerant,
like
consensus
mechanism
on
the
existing
sort
of
consensus
interface
inside
the
go,
a
theory
implied
and
so
lovely
the
go
theme
client
house,
this
sort
of
I'm.
Nobody
to
you
yeah,
because.
A
B
Proof
of
authority,
it's
got
Eve
hash
and,
like
yes,
there's
already
like
a
few,
a
few
implementations
in
there
and
Istanbul
plugged
into
that,
but
it
was
never
merged.
The
code
was
like
18
months
out
of
date.
It
was
also
incorrect
because
it's
really
hard
to
write
these
yes
kudos
to
the
original
authors
they
did.
They
did
a
great
job,
but
yeah
there
was
some
edge
cases
we
wanted
to
pick
up.
It
was
also
like
piggybacked
on
a
theory.
A
B
It
ended
up
just
basically
generating
way
more
messages
and
cesarean,
and
then
they
were
also
like
a
few
security
things
in
there
as
well.
So
not
only
were
they
like
direct
security
bugs
and
like
just
just
straight-up
bugs
that
would
affect
security.
There
are
also
like
a
few
opportunities
for
denial
of
service
which
were
sort
of
in
in
the
codebase.
B
A
B
E
B
Right,
so
we
actually,
we
did
look
at
whatever
is
cool
at
the
time
if
the
mint
I
think-
and
we
talked
I-
think
to
the
people
involved
in
that
project.
It
was
a
really
nice
proof
of
concept,
but
it
really
that
the
API
around
tender
men
was
actually
very
like,
while
in
like
separately,
it's
actually
quite
a
nice
way
of
plugging
in
block
chains
into
a
pbft
system.
It
unfortunately
was
very
tricky
to
integrate
that
into
the
code
base,
so
that
was
like
our
sort
of
looking
at
it
like
on
our
own.
E
E
B
E
D
C
C
A
B
The
the
trick
with
yeah
I
mean
so
like
you-
can
write
a
sort
of
a
pseudocode
implementation
of
pbft
in,
like
you
know,
a
few
hundred
lines,
but
the
devil
is
in
the
detail
in
the
implementation.
Absolutely
an
Istanbul
was
not
straight-up
pbft
either,
but
instant
also
have
the
advantage
that
there
were
two
other
teams
like
in
the
broader
ecosystem
who
were
using
it
scrutinizing.
It.
D
B
E
B
Not
hot
stuff,
yeah,
so
I
think
hot
stuff
is
like
a
really
interesting
paper,
so
yeah.
So
let's
talk
about
like
the
plan
right
now
and
then
the
plan
like
going
forward,
it's
a
pbft
in
general,
be
ft
house
lighting
limitations
in
terms
of
the
scalability
because
of
like
the
complexity,
pedestrian
well,
but
it's
very
proven
algorithm.
That's
been
implemented
a
number
of
times
in
this
space
and
outside
of
this
space
has
good
pedigree.
From
that
perspective,
hot
stuff
is
really
interesting.
B
It
we
like
obviously
Libra
being
one
of
the
first
to
actually
like
really
implemented
in
the
space.
Of
course,
that
project
is
not
fully
live
yet
either,
but
I
I
think
at
the
time
that
we
are
making
this
decision
like
it
felt
it
was
very
like
it
was
very
tentative
right
and
it
would
be
very
interesting.
You
could
imagine
in
future
like
swapping
out
pbft
for
a
hot
stuff
based
implementation
and
getting
a
bunch
of
benefits
for
that.
B
B
Don't
you
have
something
working
yeah
like
don't
keep
improving
performance,
don't
keep
trying
to
like
add
extra
features
like
get
to
market
talk
to
your
users,
understand
what
matters
to
them
right
and
like
unless
you
can
draw
a
really
really
clear
line
from
what
this
person
here
cares
about
down
to
here
then
you're
wasting
your
time
and
for
us
like
hot
stuff,
has
likes
a
bunch
of
really
interesting
potential
benefits.
So
we
have
this
two-track
approach.
B
But
at
the
same
time
there
is
you
know,
we've
been
doing
a
bunch
of
work
around
an
algorithm
called
bf
tree
which
aims
to
use
hot
stuff
to
be
able
to
scale
permissionless
a
permissionless
approach
to
consensus
for,
like
potentially
tens
or
hundreds
of
thousands
of
participants
where
they
can
just
like
join
form
a
sort
of
tree
structure,
and
in
so
doing
we
don't
have
to
have
a
lot
of
the
complexities
of
a
proof
of
state
mechanism.
On
top
of
a
BFT
algorithm
to
determine
who
the
participants
in
the.
B
B
To
be
the
other
benefit
of
like
nailing,
this
down
early
was
that
we've
been
able
to
do
a
ton
of
like
low
testing
failure,
testing
on
like
very
large
clusters
and
being
able
to
like
put
this
through
its
paces.
In
terms
of
many
like
practical
sort
of
settings
and
like
we
found
a
ton
of
issues,
none
of
them
no,
no,
but
like
almost
all
of
them,
were
outside
of
the
scope
of
the
core
algorithm.
It's
like
yeah.
E
B
B
Usual,
you
know
exactly
yeah
and
so
contract
here,
in
fact
like
embeds
web
three,
and
you
can
also
go
and
do
your
own
thing
with
that,
but
the
smart
contracts
like,
of
course,
people
will
add
smart
contracts.
There's
no
restriction
on
that,
but
like
the
the
ones
that
sort
of
are
like
out
of
the
box
and
so
I
fall
into
like
such
three
categories
says
first,
which
is
like
around
proof
of
stake
and
governance.
B
D
B
D
A
B
A
B
Right
so,
like
you've,
seen
the
situation
in
theory
and
where
miners
are
holding
like
the
rest
of
the
network,
I
won't
say
to
ransom,
but
you
know
what
I
mean
like
they
have:
control
over
that
particular
parameter.
So
this
this
sort
of
puts
these
things
out
in
the
open
and
means
that
you
can
change
a
bunch
of
these
parameters
without
having
to
like
hard
fork
or
even
in
many
cases
update
their
their
client,
which
is
which
is
quite
nice.
B
So
yeah
Koufos
taking
governance,
so
this
is
really
the
so
yeah.
So
leaving
aside
governance
for
now
proof
of
stake
here
is
like
the
mechanism
by
which
you
decide
who
gets
to
participate
in
this
protocol.
As
you,
as
you
guys
know,
we
have
implemented
a
quite
interesting
like
proof
of
state
I,
guess.
Protocol
we've
been
thinking
about
the
scheme
that
we
want
to
use
for.
B
D
B
D
B
The
network
and
also
like
give
them
the
opportunity
to
potentially
win
prizes
that
can
convert
into
like
gold
on
maenette,
potentially
sex,
yeah,
so
yeah
so
yeah.
So
let's
dig
into
the
proof
of
state
protocol,
so
validators
are
basically
elected
by
hey.
Let's
move
over
here.
Violators
are
basically
elected
by
his
validator.
B
B
A
B
That's
an
interesting
question
around
the
naming.
It
is
definitely
not
pegged
to
gold,
it's
definitely
not
paid
to
gold
yeah.
So
it's
the
native
utility
token
of
the
platform.
Sello
gold
holders
are
essentially
the
voters
in
the
system
that
can
select
the
validators,
but
they
do
so
via
a
election
mechanism
in
which
they
vote
for
validated
groups
too.
Obviously
we
have
a
situation
whereby
you
know
this
is
a
decentralized
platform.
Anyone
can
create
any
number
of
accounts.
We
can't
do
like
one
person,
one
one
user.
B
You
know
the
only
way
that
we
in
the
community
know
how
to
do
this
right
now
is
to
like
you,
have
have
voting
by
balances,
yeah
and,
of
course,
to
prevent
people
vote
with
the
balance
and
then
transfer
it
to
a
different
account
and
then
vote
with
that
same
balance.
Again,
essentially,
we
require
them
to
like
lock
up
yeah,
so
we
don't
cool
one
interesting
difference
with
so
low
compared
to
other
networks
that
you're
probably
expecting
is
when
seller
gold
holders
lock
up
a
balance.
B
They
are
not
putting
that
balance
at
risk,
mm-hm,
so
no
matter
who
they
vote
for
that
balance
can't
be
slashed.
So
we
we
think
that's
an
important
decision
because
seller
gold
holders-
don't
have
you
know
you
don't
have
if
you're
just
an
average
vote.
If
you're
this
person
up
here,
you
probably
don't
know
and
don't
like
you.
B
B
You
basically
like
lock
some
gold
in
gold
contract
and
then
that
allows
you
to
vote.
But
what
you're
voting
for
is
actually
a
validated
group
and
a
validated
group
is
this
sort
of
new
concept
which
I
don't
think
any
other
network
really
has,
which
is
I,
guess
a
wrapper
around
a
set
of
validators?
And
now
you
might
imagine
these
things
coming
about
in
two
different
ways.
B
Maybe
there's
a
like
a
group
of
like
a
valid,
a
valid,
an
entity
or
an
organization
that
runs
validators,
maybe
more
like
the
sort
of
professionalized
validators
that
you're
seeing
cosmos
where
they
may
want
to
run
more
than
one
validator,
and
so
they
may
want
to
run.
A
validated
group
have
valid
like
have
validators
in
in
their
validator
group
and
potentially
be
able
to
swap
these
validators
I
out
if
they
need
to
and.
B
B
B
So
the
idea
is
that
these
things
have
like
a
reputation
like
name
recognition
like
long
term
persistence
in
the
network.
These
are
folks
that
you're
voting
for
so
yeah,
maybe
like
it's
like
someone
loose
analogies.
They
may
be
like
a
political
party
whereby,
like
if
you
vote
democrat
like
maybe
there
are
some
people
who
vote
democrat
or
republican
or
whatever.
B
Marketing,
maybe
marketing
but
I
was
gonna,
say,
vetting
right
so,
like
you
know,
they
have
a
reputation
to
preserve.
They
are
not
gonna
put
somebody
on
this
late,
who
hope
you
know.
I'm
gonna
put
somebody
on
this
late.
That
is
actually
actually
not
right,
and
so
they
have
this
a
better
ability
as
well
to
make
sure
that
a
validator
is
actually
doing
what
you
would
expect
them
to,
because
the
validator
to
be
elected
they
need
a
validator
group
right.
B
So
there
is
a
sort
of
admission
control
system
here
whereby,
like
a
validator,
basically
needs
to
say,
I
want
to
be
polished
but
later
grouping,
the
validates
group
needs
to
add
them,
and
what
that
means
is
that
they
have
the
opportunity
to
do
things
like
okay.
Why
Co,
let
me
come
and
do
a
security,
or
are
these
things
that
you're
actually
doing?
Let's
look
at
that
yeah.
C
B
So
yeah
at
the
moment,
probably
we're
gonna
have
like
five
validators
as
a
limit
in
in
a
validated
group
and
yeah
validator
groups.
Other
things
that
need
to
put
a
mistake
and
violators
also
need
to
put
up
the
steak
and
the
validator
group
steak
is
per
validator
as
well
so
yeah.
If
they
have
like
you,
know
three
validators
here
say
to
be
three
then
like.
Maybe
they
need
to
put
up
like
three
units.
E
B
Each
oh
absolutely
yeah,
absolutely
so
the
steak
is.
The
steak
is
like
the
combination
of
this
amount
here
and
this
amount,
and
the
way
that
you
put
up
steak
is
that
you
do
you
go
through
this
same
mechanism
here,
so
you
look
gold
and
then
that
gold,
the
benefit
of
having
a
unified,
lock
goal
mechanism,
is
then
that
you
can
also
use
it
to
participate
in
governance.
You
can
use
it
to
participate
in
validation
elections.
You
can
also
use
it
as
a
stake
or
validate
to
group
or
validator
cool.
B
B
Originally
we
had
one
but
like
you
can
have
vote
for
3,
and
the
reason
for
this
is
that
we
recognize
that
the
security
of
the
system
really
depends
on
well,
let's
have
a
look
at
a
normal
election
right.
So
if
this
is
like
the
distribution
here
of
votes
that
are
received
for,
like
you
know,
parties
or
candidates
in
an
election
right,
like
I,
don't
know
this.
B
B
Right
so
say,
you're
a
bad
actor
and
you
want
to
get
a
validator
elected
yeah
so
that
you
can
start
doing
malicious
things.
Of
course,
a
lot
of
the
security
of
this
protocol
here
depends
on
making
sure
that
one
third
of
validators,
no
more
than
one-third
of
validators,
are
malicious
yeah
and
that's
critical,
like
the
that's.
The
the
behavior
of
individual
nodes
is
more
important
than
in
comparative
proof-of-work
scheme.
E
B
Third
is
extremely
bad-
hopefully,
hopefully
very
small
numbers
anyway.
So
to
get
your
validator
elected,
you
don't
need
to
beat
this
candidate
here.
You
need
to
beat
like
number
100
say
we're.
Gonna
have
100
validators,
you
need
to
like
have
more
votes
than
this
candidate
here
right
and
so
the
ideal
graph
distribution
from
a
security
point
of
view
would
actually
be
much
more
like
this
right.
So
votes
up
here
for
these
guys
are
sent.
B
Like
super
force,
they're
wasted,
they
don't
contribute
to
the
security
of
the
system
like
especially
because
we,
from
a
decentralization
point
of
view,
it's
important
to
us
that
we
have
a
bunch
of
different
organizations,
individuals
participating
and
contributing
operating,
validators
and
validated
groups.
And
so
we
have
this
like
limit.
And
so
basically,
we
actually
limit
number
of
votes
that
evaluator
group
you'll
receive
such
that
they
can
only
receive
the
number
of
votes
required
to
elect
one
more
than
the
number
of
validators
that
they're
like
fielding
in
the
election
right.
E
E
B
B
E
E
B
E
E
B
Three
separate
votes:
we
and
you
can
set
different
bound
at
any
one
time.
You
you
have
a
bunch
of
rock
gold,
some
of
it
can
be
non-voting,
and
then
you
can
take
some
of
your
lung
voting
gold
and
place
it
for
a
vote
for
a
group,
and
that
will
be
accepted
so
long
as
we're
below
that
cab.
And
then
you
can
take
some
away
and
move
it
to
a
different
group.
D
C
B
A
B
E
A
B
That
state,
so
this
is
one
of
the
one
of
the
things
we
wanted
to
do
with
this
system,
is
like
create
like
duty
incentives
for
validators
to
contribute
to
the
security
of
the
system.
We
wanted
to
make
sure
that
the
the
rewards
were
interesting
for
professional
validators.
You
like
go
to
the
effort
of
like
properly
operating
and
monitoring
like
hardware
having
hardware
and
Kolo
facilities
with,
like
you
know,
potentially
hardware
security
modules
or
like
good
security.
B
Setups
cameras
pointed
at
the
cage
of
the
datacenter,
and
so
on
that
doesn't
come
cheap,
but
at
the
same
time
we
also
recognize
that
costs.
The
info
structure
costs
here
are,
in
effect,
taking
rewards
out
of
the
pocket
of
the
people
who
were
building
this
service
for
as
well
right.
So
we're
trying
to
balance
these
concert'
that
the
the
concerns
of
different
constituencies-
they're,
like
you
know,
having
building
such
a
big
platform
like
we're
like
there
are
many
different
like
there
are
many
different
like.
D
B
A
B
Controls,
yeah
and,
and
what
are
we
gonna
be
start
with
we're,
not
sure
yet
we're
thinking
I
mean
it's
gonna,
be
comparable
to
the
sort
of
stakes
that
you're
seeing
and
existing
networks
and
the
rewards
will
be
comparable
and
interesting.
I
think
the
the
other
thing
to
notice
that
validates
and
validated
groups
as
well
as
obviously
what
you've
like
set.
This
is
a
state
you
earn
rewards.
We
haven't
talked
about
this
yet,
but
you
actually
earn
like
rewards
for
like
locking
up
the
gold
here,
and
we
do
that
so
yeah.
Let's
talk
about.
B
Let's
talk
about
that
bit
so
so
you
lock
up
gold,
you
vote
for
some
groups.
Like
you
there's
an
election,
we
run
a
an
election
which
is
exactly
the
same
as
like
a
proportional
representation,
an
election,
as
done
in
say,
like
the
European
Parliament.
It's
the
same.
We
use
ad
on
our
room,
which
is
the
same,
which
I
think
is
used
to
like
allocate
houses
seeds
in
the
House
of
Representatives,
from
like
census,
data
and.
E
B
It's
widely
used
for
this
sort
of
purpose.
It
basically
will
like
determine
based
on
their
votes,
the
set
of
validators
that
are
going
to
be
the
next
validator,
sir.
We
have
an
epoch
length,
which
is
one
day
at
the
end
of
like
the
last
block
of
every
epoch.
We
go
through
this
process.
You
can
continue
to
vote.
You
continue
your
lock
or
unlock
anytime.
A
B
Right,
there's
like
a
snapshot,
and
at
that
point
you
can
do
one
thing
to
say
is
that
there
is
like
a
three-day
unlocking
period
here,
so
you
unlock.
Then
you
wait
this
interesting.
So
we
went
back
and
forth
a
lot
on
this
specific
time.
Like
you
unlock,
then
you
have
to
wait
three
days.
The
reason
it's
three
days
is
because
an
epoch
is
one
day.
Yeah.
A
B
We
want
to
make
sure
that,
if
you've
participated
in
an
election
that
you
still
have
skin
in
the
game
right,
so
we
want
to.
We
want
to
ensure
that
you're
down
just
do
that's
right.
If
I
want
to
create
a
tach
on
the
network
by
by
likes,
you
know
affecting
the
results
of
the
validator
election
by
borrowing
borrowing
funds-
you
know
maybe
trying
to
short,
sell,
Oh
or
something
then
use
that
to
purchase
a
lot
of
seller
gold
and
then
try
and
swing
the
election
by
that
vote.
E
B
B
E
D
A
E
B
We
run
this
right,
so
we
run,
we
run
this
PR
election
as
proportion
upset
and
so
that
chooses
the
set.
Then
we
shuffle
the
set,
because
it's
important
that
we
don't
want
folks
who
we
don't
want,
folks
who
can
to
be
able
to
engineer
the
number
of
votes
they
get
to
be
able
to
determine
the
order
in
which
the
proposer
rotates.
B
So
then
we
shuffle
we
have
sort
of
on
chain
randomness
as
well,
and
that's
what
we
use
the
shuffle
and
then
basically,
the
diff
of
that
validator
set
is
like
encoded
in
the
extra
data
of
the
header
yeah,
and
so
the
light
concerts
are
one
of
the
things
we've
actually
done.
You
know
going
back
to
like
with
mobile.
First
he's.
One
of
the
things
we've
actually
done
is
like
this
gives
you
some
interesting
opportunities
to
make
like
the
light
client
sync
protocol
a
lot
more
efficient,
so
you
so.
E
A
B
B
B
Yeah
and
so
in
certain
other
places,
you
know
in
certain
other
situations
or
certain
other
sort
of
pass.
The
protocol
we
end
up
doing
things
like
waiting
a
certain
like
the
chance
that
you
can
influence
the
proposer
more
or
the
sequence
of
proposes
for
more
than
n
consecutive
blocks
ends
up
like
falling
off
extremely
rapidly
and.
E
B
Is
true
with
the
theorem
right
there's
I
mean
it's
very
hard
to
prove
a
negative
like
it's
very
hard
to
force,
somebody
to
include
a
transaction
there's,
no
like
mechanism
to
like
prove
that
they
actually
did
receive
no
action,
but
so
the
proposer
does
receive
transaction
fees.
We've
changed
the
gas
pricing
mechanism
for
compared
to
what
a
theorem
does.
We
use
like
a
sort
of
surge
pricing
mechanism
which
actually
Vitalik
proposed
in
an
EIP
a
number
of
years.
B
So,
both
of
them
stepping
back
over
here
onto
my
other
shard
of
the
whiteboard
shards,
zero
I,
will
yeah
so
like,
rather
than
have
a
block
like
this
and
have
transactions
in
the
block
filling
up
up
to
here
and
basically
picking
the
highest
price
transactions
first
and
just
filling
to
a
fixed
limit.
What
we
do
is
we
set.
We
upsetting
to.
B
C
B
E
B
You're
right
from
the
point
of
view
that
the
validators,
like
the
two-thirds
elected,
validate
like
basically,
you
can
do
anything
you
like
when
you're
when
you're
a
sort
of
Byzantine
quorum
of
like
elected
validators
during
the
period
in
which
you're
elected
so
yep
you
can.
The
M
is
there's
sort
of
no
way
around
that
right.
It's
a
PFT
assumption
that
two-thirds
of
your
validators
are
honest.
E
A
D
B
Actually
I
mean
I.
Think
if
you
look
at
this
yeah
I
mean
if
you,
if
you
look
at
this,
like
the
there's,
probably
going
to
be
some
distribution
on
the
boats
right
and
so
the
validators
at
the
top
of
the
distribution
they
have
no
incentive
to
like
exclude
votes
that
are
not
gonna
like
you
can
calculate.
Is
this
vote?
Gonna
push
me
out
with
a
validator
set
and
that's
gonna
be
false
for
everybody,
except
for
the
like.
D
E
B
So
so
the
question
is
and
remember
that
sorry
remember
that
if
even
if
a
transaction
does
not
appear
in
one
particular
block,
it's
exactly
everybody
well
2/3,
2/3,
neater
center
and
so
like,
because
there
is
no
transaction
fee,
sure
it's
relatively
modest,
but
of
course
you
have
to
balance
that
against
like
making
it
affordable
for
people
to
send
transactions
yeah.
So.
E
E
B
B
D
B
C
A
Yeah
no
I
just
wanted
to
understand
yeah.
If
what
what
happens
when
you
don't
like
pretty
much
I
mean
there's
two
questions.
One
is
like:
if
I
just
don't
show
up
right
or
doesn't
approve
something,
and
then
that's
a
question.
If
it's
60
days-
and
you
said
you
can
only
release
only
after
you
are
not
in
the
selection
and
you
weren't
selected,
how
do
you
actually
stop
being
selected?
That's.
B
What
we
have
to
do,
then,
is
like
look
at
so
we
starts
at
so
it's
Li
Haos
like
like
a
fixed
total
supply
cap
and
there's
a
certain
amount,
that's
allocated
in
the
Genesis
block
and
then
the
rest
of
this
is
allocated
over
time
through
basically
epoch
rewards.
So
we
don't
have
block
rewards
like
a
theorem
does
we
have
rewards
given
out
at
the
end
of
every
epoch,
yeah
and
yeah?
Okay,
like.
B
C
A
B
To
basically
divert
some
of
the
epoch
rewards
over
to
that
and
so
yeah
and
in
any
given.
So
that
means
that
at
any
given
a
POC
you,
you
can
end
up
having
different
like
needs
on
your
funds
right,
maybe
you
need
to
pay
more
to
the
reserve.
Yeah,
maybe
like
all
the
validators,
had
really
terrible,
uptime
and
you've
decided
not
to
pay
them
very
much
or
maybe
sell
our
gold
holders
as
well.
Are
gonna,
get
less
money
and
I'll.
Explain
why
and
say,
but
you
this
is
your
sort
of
target
schedule
right.
A
B
A
B
Payments,
the
validators,
which
is
say
this
number
of
sello
dollars,
and
then
we
either
we
adjust
it
up
or
down
based
on
like
overspend
or
under
spend,
and
so
we
say,
okay,
so
actually
you
know,
we've
been
spending
a
ton,
so
we
can
only
afford
this
much
and
then
what
we
do
is
we
look
at
a
couple
of
things.
So
the
first
thing
we
look
at
is
like
uptime
and
the
way
we
look
at
uptime
is
for
every
block
that
we
are.
B
Every
proposer
includes
in
that
block
the
all
of
the
signatures
that
they've
collected
from
the
previous
block
as
well.
So
all
the
signatures
by
the
way
we
use
BLS
signatures,
absolute
validates
of
two
keys,
Devon
ECDSA
key
and
a
BLS
key.
We
used
BLS
signatures
because
we
can
aggregate
them
over
all
their
smaller
improvements,
the
like
client
protocol.
C
B
Now
we
use
a
sliding
window.
We
currently
we're
thinking
like
10,
so
long
as
you
speed,
your
signature
has
been
included
in
one
of
the
last
ten.
We
consider
you
up
for
that
block.
This
is
like
a
sort
of
simple
sliding
window.
The
reason
for
that
is,
of
course,
like
you
know,
if
I'm
a
malicious
validator,
you.
E
B
Is
we
don't
we
don't
penalize
that
right
now,
I
mean
we're
not
assuming
like
this
is
sort
of.
We
extremum
that,
like
most,
the
reason
that
most
honest
nodes
would
end
up
being
down
is
because
of
this
configuration
network
problem.
Like
you
know,
power
supply
failure
that
sort
of
yeah
hardware,
the
usual
things
exactly.
E
B
E
B
Yeah
so
currently
we're
set
to
three
seconds
plus
the
regular
blocked.
Oh
so
that
the
system
has
a
parameter,
block,
delay,
say:
hey.
We
want
to
make
a
block
every
five
seconds,
so
if
everyone
else
hasn't
seen
a
block
after
three
seconds
plus
that
time,
then
that's
when
everyone
so
amend
change
messages.
So
from
that
point
forward,
we
then
take
that
say
five
seconds
off
and
then
that's
the
basis
from
which
we
do
exponential
back-off
up
until
a
certain.
E
B
B
D,
that's
sort
of
what
we've
been
trying
at
the
moment.
We
can
easily
handle
a
hundred
validators
with
that
you
got
a
choice
of.
Do
you
want
to
increase?
The
number
of
validators
include
christi
centralization,
or
do
you
want
to
increase
like
reduce
the
latency
for
transaction
finalization?
For
these
says,
cool.
B
A
B
Right
exactly
and
so
sellers
selling
firstly,
is
like
a
sort
of
family
of
stable
coins.
I
would
say
it's
like
relatively
easy,
through
governance
to
propose
and
create
like
new,
stable
coins.
Right
now,
we've
only
been
thinking
in
terms
of
like
one
yeah
and
the
obvious
place
to
start
is
a
sellout
dollar.
Like
a
staple
coin,
that's
picked
up
a
dollar
also
interested
in
sable
coins,
which
are
not
just
necessarily
pegged
to
regular
fiat
currency.
A
A
A
B
B
Sello,
gold,
the
native-like
utility
token
yeah
of
the
system
with
which
is
obviously
decentralized
with
like
a
reserve
of
other
crypto
assets
as
well,
and
so
you
know
that
might
be
Bitcoin
might
be
like
eath
might
be,
other
things
might
be
other
stable
coins,
especially
initially
we're
thinking.
That
could
be
very
interesting
to
include
other
stable
coins.
B
A
B
So
we
have
basically
like
holdings
of
these
things
on
different
chain.
Maybe
it
exchanges
we
have
a
bunch
of,
like
you,
know,
algorithms,
which
look
at
this
reserve
decide
whether
we
want
to
rebalance
it
in
terms
of,
like
you
know,
the
various
volatility
or
prices,
the
prices
of
these
things
to
maintain
like
target
ratios
between
them
and
so
like.
This
is
going
to
be
centralized
to
start
with,
like
heading
towards
being
decentralized
and
it's
sort
of,
even
while
being
centralized
from
an
implementation
point
of
view.
B
A
B
That
I
couldn't
tell
you
I
guess
it
depends
on
the
value
of
this.
But,
like
you
know
there
are,
there
are
a
few
things.
What
we
basically
have
is
like
a
series
of
Oracle's.
We
do
depend
on
Oracle's,
of
course,
to
like
provide
price
signals.
Yeah
the
essentially
the
exchange
rate
between
like
selling.
D
B
B
A
B
Allow
the
price
to
move
by
more
than
a
certain
amount
each
time
you
know
a
bunch
of
other
bunch
of
other
things,
but
you
can
imagine,
like
you
know
these
things,
putting
up
stakes
and
getting
rewards,
but
yeah.
That's
an
area
which
we
sort
of
are
gonna,
be
yeah,
yeah
yeah
exactly
so.
These
things
provide
an
exchange
rate
and,
of
course,
like
you
know,
there's
several
ways:
we're
gonna
want
to
like
be
able
to
bolster
the
reserve
under
conditions
where
we
see
high
volatility
or
we
feel
were
under
collateralized
several
different
ways
of
doing
there.
B
A
A
B
B
B
B
And
so,
therefore,
you
can
deplete
the
result,
and
so
these
are
mechanisms
here.
Where
might
we
can
like
both
the
reserves
and
that's
also
valuable
from
full-width
under
collateralized,
because
volatility
increases?
This
is
a
sort
of
uni
swap
like
mechanism
whereby
it
sort
of
creates
this
negative
feedback
loop
around
price
shifts.
So
if
the
balance
of
trading
from
one
to
the
other
like
shifts
away
from
parity,
then
what
we
end
up
doing
is
increasing
the
price
like
against
that,
so
providing
making
it
increasingly
expensive
to
move
to.
B
A
B
A
B
B
A
B
Yeah
so
yeah,
so
then,
then,
the
question
is
like:
is
this
gonna
work?
In
practice?
We
have
done
a
ton.
We
have
many
people
like
far
smarter
than
I
am
like
looking
at
who
have
PhDs
in
economics
like
thinking
hard
about
this
sort
of
situation.
We've
done
a
ton
of
like
stability
simulation
analysis,
which
we
have
a
white
paper
on
which
talks
a
lot
about.
B
Like
you
know,
the
particular
situation
we're
concerned
about
is
that
if
you
have
a
basket
of
crypto
assets
and
a
sort
of
you
have
a
rapid
crash
in
the
market
like
like
they've
seen
like
we've
seen
absolutely,
you
know
what
does
what
does
that
do
to
the
peg?
And
so
we
plotted
a
bunch
of
those
scenarios
and
looked
at
the
paths
there
looked
at
the
sort
of
mechanisms
that
we
had
available
to
us
in
terms
of
keeping
the
reserve
we
were
able
to
like
like
keep
a
peg
and.
D
B
B
E
B
Which
I
pull
the
tile?
You
want
currencies
which
feel
familiar
to
you
and
whose
values
familiar
to
you
and
so
like
that's
where
the
staple
coin
mechanism
comes
in.
You
want
a
platform
where
it
just
works
from
your
mobile
device,
which
ever
a
lot
of
the
infrastructure
that
we
built
comes
in.
We
also
allow
you
to
pay
for
transaction
fees
in
any
currency,
so
we've
modified
like
sort
of
transaction
fee
payment
mechanism,
so
that,
like
validators,
can
basically
collect
transactions.
B
B
A
B
This
and
I
want
to
like
send
money
to
receive
money
for
my
contact
list,
so
to
do
that,
like
I,
give
the
app
permission
I
come
across
a
constant.
This
is
a
bunch
of
phone
numbers
in
it.
So
what
we
have
is
a
mapping
from
basically
from
phone
numbers
to
addresses
yeah.
We
of
course
want
that
to
be
secured
so
that
nobody
else
can.
A
B
C
B
The
user
can
request
verification
and
they
pay
a
small
verification
fee
that
covers
the
cost
of
sending
like
n
text
messages.
In
fact
they
can
choose.
What
n
is
the
client
can
choose
what
and
it's,
but
of
course,
from
like
a
sim
for
a
simple
user
experience
like
that
numbers
essentially
fixed,
and
then
what
happens
is
flow.
B
B
B
Then
so
later,
we
may
we'll
probably
be
doing
like
threshold
validation,
so
like
there's
a
bunch
of
different
like
nodes,
b3
b2
b1,
which
we'll
have
a
shot
of
like
the
actual
keys,
and
you
need
like
two
out
of
three
or
whatever
to
combine.
Two
signatures
will
also
support
multiple
proxies,
but
like
the
basic
set
up
in
territory,
the
attestation
service.
Basically,
you
find
out
the
vowel
datas
you
find
out
the
identity
of
the
validators
who,
when
you
ask
them
to
send
a
text
and
Mason
that
fruit
you
signed
by
there
after
station
key.
C
B
B
B
B
So
we
actually
have
like
an
accounts
mechanism,
so
this
is
not
what
we
have
right
now,
but
it's
like
something
we're
working.
So
we
have
accounts
mechanisms
that
this
is
like
the
way
the
validators
can
record
metadata
about
themselves.
They
can
claim
hey,
I've
got
existed,
this
DNS
I've
got
access
to
this
key
base,
account
and
so
on.
It's
also
the
place
where,
like
individual
account,
holders
can
record
a
public
key
mm-hm,
and
so
that's.
B
We
also
support
like
encrypted
comments
in
transactions,
so
that
I
can
like
send
you
a
bunch
of
emojis
to
indicate
that
I'm
paying
you
for
coffee,
and
so
that
ends
up
going
into
like
using
that
public
heats.
Basically,
encrypt
that
comment
that
same
public
key
can,
rather
than
basically
so
there's
a
mechanism
whereby
we'll
be
able
to
say,
hey,
actually
don't
send,
don't
send
to
this
address,
send
all
the
funds
should
be
the
center,
like
an
address
which
you
can
communicate,
which
you
can
basically
use
this
public
key
to
go
and
communicate
with
me.