►
From YouTube: CasperLabs Community Call
Description
Rewards Distribution presentation & status update.
A
A
A
A
All
right
so
good
morning,
everyone
good
day
evening,
wherever
you're,
watching
I'm
reporting
to
you
back
from
San
Diego
huddled
here
in
my
home,
not
quite
in
lockdown,
but
we're
good
we're
good
here
in
Southern,
California
I
hope
everyone
else
is
doing
fine
with
the
outbreaks
that
are
happening
worldwide.
Our
thoughts
in
poverty
and
everybody
stays
safe
out
there,
I'm
gonna,
quick,
give
a
quick
engineering
status
and
then
Alexandre
or
lemon
table
is
gonna
share
an
update
on
the
economic
side.
A
So,
on
the
engineering
side
we
are
working
on
our
next
release
in
preparation
for
alpha
test
net
team
is
very,
very
busy
I'm
cutting
that
release
we're
going
to
cut
the
release
next
week.
Our
original
plan
was
to
cut
it
this
week.
We
are
pushing
the
release
out
by
seven
days,
because
we
need
to
do
a
couple
of
things
one.
We
need
to
get
some
additional
time
for
developer
testing
under
our
belts
to
make
sure
that
the
products
really
solid,
as
you're
may
or
may
not
be
aware,
highways
come
and
hot
off
the
presses.
A
That's
how
we've
gotten
the
performance
up
to
somewhere
in
the
orders
of
200
transactions
per
second
on
a
single
core,
depending
on
the
speed
and
size
of
your
core,
which
is
fantastic
right.
This
gets
the
performance
way
up
into
the
realm
where
we
wanted
it
to
be.
You
know,
if
you
think
of
next-generation
software
very
excited
about
that,
and
we
are
proposing.
A
The
release
is
going
to
go
out
the
19th
of
March,
and
that's
not
right,
we'll
take
that
out
that
here
and
moving
along,
we
talk
about
our
current
focus,
so
testing
and
debugging
castro
labs
highway,
getting
ready,
moving
the
standard
payment
to
the
host
side,
like
I
indicated
we're
working
really
hard
with
s
test.
Last
week
we
had
last
update.
A
We
had
mark
Greenslade
come
in
and
give
you
guys
a
demo
of
s
test
and
we
are
integrating
the
s
tests
into
our
what
we
call
our
LRT
test
bed
or
long-running
test
bed
long-running
test
test
bed.
But
basically
s
test
is
going
to
form
the
basis
for
our
production
engineering
effort
where
we're
going
to
be
spinning
up
networks
of
different
sizes
and
different
configurations
for
those
of
you
that
are
familiar
with
highway.
You
know
that
highway
is
a
tunable
protocol
right,
there's,
it's
it's
flexible
right.
A
It
round
lanes
can
change
arrow
durations
can
change
depending
on
conditions
in
the
network
for
the
purposes
of
what
we're
doing
in
alpha
testing
a
the
round.
Duration
will
be
fixed,
but
they'll
be
set
at
the
validator
level
right.
So
when
you
sit
boot
up
your
validator,
you
set
your
round
duration.
The
error
length
will
also
be
fixed
for
this
iteration
of
alpha
test
net.
A
However,
in
the
future
they
will
become
more
flexible
right
as
we
learn
about
how
the
network
behaves
and
as
we
learn
about
you
know,
what
are
the
key
indicators
that
an
arrived
duration
needs
to
increase
or
shorten
or
look
around
latest
increase
or
short,
not
as
we
learn
about
that
will
make
those
now
those
pieces
more
dynamic,
but
we
still
need
to
figure
out
how
to
tune
the
network,
so
it's
stable,
and
so
that's
the
work.
We're
really
using
s
tests.
A
So
that's
that's
the
work
that
Michael
birch
is
doing
right
now
on
the
node.
We
are
working
on
stability,
optimizations,
metrics,
right,
metrics,
inform
us
how
the
node
is
processing,
processing,
work,
and
so
that's
going
to
help
us
tune
it
and
you
know,
find
opportunities
for
more
performance
and
then,
of
course,
stress
testing.
The
network
using
s
test
and
deploy
gossipping
deploy.
Gossipping
is
an
important
feature.
It's
not
going
to
be
there
in
alpha
testing
it.
The
team
is
pushing
really
hard
to
get
it,
but
I,
don't
I,
don't
want
it
for
alpha
test
net.
A
We
are
working
on
we're
going
to
remove
bonding
and
on
bonding
for
the
first
round
of
alpha.
So,
even
though
we
could
support
validator
set
rotation
during
switch
blocks,
we
we
believe
we
want
to
have
a
fixed
validator
set.
So
we
can
work
closely
with
the
validators
if
you're
interested
in
joining
the
validator
set.
Please
do
contact
us
at
hello,
a
counselor
lives
on
Io
and
we
can
work
on
having
you
participate
in
the
in
the
Testament.
We
will
have
incentives
as
part
of
the
alpha
test.
A
Net
milestones
so
feel
free
to
to
join
us,
but
you
got
to
come
in
at
Genesis
and
because
we
won't
have
alligator
cent
rotation
in
each
round
and
we're
documenting
assembly
script
and
getting
some
tutorials
in
there.
And
then
we
also
shipped
our
DAP
developer
guide
last
week
and
so
we'll
be
augmenting
that,
with
some
of
the
recent
features
which
is
supporting
support
for
the
type
system
in
the
ind
deploys
semantics,
basically
in
the
command
line.
A
The
cows
phillips,
client
command
line
will
be
including
that
in
the
dock
developer
guide,
as
well
as
documentation
for
assembly
strip,
once
this
is
done
and
we're
making
updates
to
the
proof
of
steak
contract
and
bring
it
in
line
with
highway.
There's
some
different
things
that
a
highway
that
we
need
to
make
those
updates
and
proof
of
steak
and
bring
it
and
keep
it
host
side,
tests
and
sre
again
same
theme,
production
engineering
and
using
the
s
test
environment
and
getting
ready
for
alpha
for
clarity.
A
We
want
to
provide
a
way
for
deploys
to
be
signed
and
set
from
the
browser
with
the
recent
CL
type
work
we
had
to.
We
have
to
rejigger
the
pull
request.
Pull
request
is
open,
but
it
needs
to
be.
Excuse
me
updated
to
use
the
new
type
system
in
the
client.
So
that's
there
are
some
changes
in
there
that
we
need
to
integrate.
So
there's
some
rework
and
then
we're
gonna
start
rebranding.
Clarity
clarity's
got
a
very
different,
looking
feel
from
our
new
website,
so
we're
going
to
bring
it
in
line.
A
A
We've
been
working
a
lot
on
pricing
models,
we're
starting
to
think
about
how
we're
going
to
price
transactions
as
you're
well
aware,
we
want
to
have
fixed
transaction
costs
and
there's
a
variety
of
different
ways
to
solve
the
technical
problem
of
not
bossing
the
system
out
with
with
by
removing
gas,
but
then
also
providing
predictable
transaction
fees
for
businesses.
You
know
scaling
aside.
We
do
want
to
provide
a
way
that
you
know.
A
Now
we're
using
that
simulator
to
learn
about
what
happens,
to
rewards
distribution
under
a
variety
of
different
scenarios,
we're
planning
on
engineering
off-site
a
full
off-site
and
mid-april
after
chestnut
lodge
in
San
Diego,
and
we
will
have
our
weekly
workshops
every
Thursday
morning,
8:00
a.m.
Pacific
and
4:00
p.m.
Pacific
bear
in
mind
that
we
have
gone
through
daylight
savings.
The
u.s.
has
undergone
daylight
savings.
A
B
A
A
B
B
Eventually,
we
will
have
to
you
know,
develop
some
kind
of
a
dynamic
rule
so
that
validators
can
use
the
truth
please
by
default,
but
user
exponent
in
such
a
way
is
that
you
know
everybody
is
you
know
somewhat
synchronized
right,
I
mean
you
can't
really
have
people.
You
know
running
ahead
of
the
train.
You
dude,
you
know
people,
you
know
being
late
for
the
train.
B
So,
if
you
know
your
exponent
rule
gets
you
into
a
situation,
you
know
where
it
is
exponent
rule.
You
know
it
has
to
be
designed
both
to
maintain
that,
and
you
know
to
make
sure
that
these
implicit
penalties
as
they're
water
alert
it
but
anyway.
So
let
me
actually
just
go
through
a
very
simple
script
that
uses
a
simulator
that
we
developed
to
study
the
effect
of
variation
in
starting
validator
exponent
parameters
on
you
know.
The
deviation
from
you
know
deviation
from
projected,
or
rather
expected
senior
trees.
Alright.
C
B
Course,
I
think
the
beginning.
Part
of
this
is
fairly
obvious
here
we
set,
you
know
the
sample
size,
so
the
number
of
validators
in
each
shrub
to
be
10,
we're
going
to
be
drawing
a
hundred
of
these
samples
everyone's
going
to
have
the
same
number
stick
to
the
one
is
going
to
have
the
same
initial
supply,
which
is,
you
know,
is
the
base
for
senior
age,
and
you
know
if
I
assign
situation
so
right
now
what
you're
using
so
right
here
this
helper
function
creates
a
particular
kind
of
validate
or
for
all
these
rods
right.
B
This
is
a
given.
You
know
this
is
whatever
exponent
lee
happened
to
draw.
This
is
really
you
know,
mostly.
You
know
useful
as
an
example,
because
you
know
even
the
initial
rules
that
wind
up
in
put
my
differs,
is
x1.
This
is
like
a
little
look
very
different
right,
but
because
I
wanted
to
see
what
happens
here
well,
you
know
we
already
had
this
implemented.
So
you
know
that
this
is
a
kind
of
validator.
B
Is
a
dynamic
exponent
rule,
so
I
just
use
that
for
now
so
well,
the
rest
of
it
is
just
more
or
less
set
up
for
you
know.
So
you
know,
for
example,
here
you're
making
our
drawers.
Let's
see
all
right,
we're
defining
our
validators.
You
know
the
actual
data
that
you're
going
to
be
collecting
and
then
we
for
every
single
of
our
hundred
samples,
we
instantiate
the
simulator
run
and
then
you
know
collect
everything
that
we
need.
So
let
me
run
through
this
and
see
what
happens.
B
B
Worked
and
well
before
I've
been
party
with
these
graphs.
So
before
these
graphs
right
and
you
know
what
you'll
see
for
the
first,
you
know
I'm
using
you
know
two
measures
of
central
tendency
and
the
you
know
two
measures
of
dispersion,
so
I'm
using
I
mean
and
the
variance
in
one
I'm
using
the
median
and
what's
called
this
index
of
dispersion
in
another
one
and
I'm
doing
you
know
two
types
of
grass
faziz
one
where
I
you
know
the
the
actual
variation
is
going
to
be
on
the
x-axis
and
the
other
one.
B
B
B
Never
mind
I
just
underestimated
how
longer
the
simulator
currently
takes
to
actually
run
these
simulations.
So
we
will
need
to
approach.
Take
about
a
couple
minutes,
so
we'll
need
to
do
some
optimizations
at
some
point,
because,
right
now
it
probably
takes
for
you
know
one
of
these
things
right.
So
one
run
with
something
like
ten
validators,
maybe
like
a
second
or
two.
So
if
you
want
to,
you
know,
run
a
lot
with
them.
You
know,
and
you
know,
if
you
want
to
do
a
lot
of
these
runs,
it
starts
adding
up.
B
Unfortunately,
so-
and
you
know
of
course
currently
you
know
everything
isn't
Python
and
is
not
the
greatest
thing
for
paralyzing
these
rods
and,
of
course,
naturally
they're
completely
parallelizable.
It
won't,
you
know,
rely
on
each
other,
so
we'll
try
to
figure
out
some
way
to
get
around
the
sole
limitation
Dobbs
line.
B
While
it
is
calculated
so
right
now,
what
you're
gonna
be
seeing
is
the
data
as
it
exists
at
the
error
level
right.
So
every
observation
is
a
you
know,
just
the
coolant
one
of
these
runs
and
because
these
rods
is
essentially
an
error,
so
one
improvement
that
I'll
be
looking
at
is
changing
as
a
simulator
so
that
it's
easier
to
collect
the
data
at
you
know
at
the
lower
levels
of
aggregation
right.
B
So,
for
example,
you
know
maybe
collecting
data,
it's
a
you
know
in
the
most
extreme
case
of
the
you
know,
validator
tic
level,
or
you
know
maybe
message
stick
level
because
collected
data
at
you
know
at
that
level
of
aggregation
would
enable
us
to
do
things.
Like
studies,
you
know
dynamic
evolution
of
Z's.
B
You
know
of
these
exponents
right,
which
is-
and
you
know
this
simulator
is
a
moksha
lesson
view
these,
and
you
know
trying
to
you
know,
there's
different
exponent
rules,
for
example,
or
you
know
real
hardware
or
in
some
you
know
in
our
more
complicated
simulators
at
the
whiticus
development.
So
you
know
like
once
we
could
collect
the
data.
We
can
rapidly
iterate
on
different,
different
exponent
rules
and
you
know
simultaneously
see
what
is
evolution.
B
B
B
I'm,
just
finding
all
the
things
right
now,
okay,
so
on
the
left,
you
know
you
can
see
the
so.
The
left
is
an
annual
senior
trade
after
burnin
versus
you
know
a
measure
of
you
know
of
variation
right,
so
the
top
one
based
was
the
median
the
bottom
one
based
was
the
mean
I
mean
I.
Just
did
both
for
just
in
case
they,
you
know,
looked
very
different,
but
usually
one
would
not
expect
it,
but
you
know
just
out
own
abundance
of
caution
right
and
there's.
A
few
of
these
points
is
you
know
the
darker?
B
It
is
higher.
You
know
the
median
or
Zameen
in
that
particular
sample
validators.
Every
single
point
is
one
round
simulation.
Is
that
then,
randomly
a
you
know
this
ten
validators
who
have
randomly
drawn
exponents
from
a
triangular
distribution,
centered
at
15
and
running
from
what
tempted
funny.
So
what
you
can
see
here
is
a
piece
of
like
a
diagnostic
force.
This
you
know
exponent
adjustment.
Rule
here
right,
I,
mean
remember,
is
that
one
of
our
goals
is
seniority,
stability
and
clearly,
here
you
know
what
potential
weakness
so
far.
B
You
know
this
particular
exponent
adjustment
rule
which
remember
very
only
use
it.
As
an
example
is
that
you
know
if
there
is
a
high
variation
between
validators,
then
you
know
you
rapidly
get
into
a
situation
where,
as
there
is
a
shortfall
relative
to
you
know
relative
to
you
know
what
we
want
the
seniors
to
be,
which
in
this
case
is
you
know
two
percent.
So
writing
in
says
the
moment
they
spread
around.
You
know
you
have
a
problem,
so,
of
course
you
know
B.
This
is
sort
of
in
a
way
intuitive
is.
B
B
So
you
know
the
median
and
the
mean
on
the
x-axis
that
doesn't
seem
to
actually
matter
very
much
right,
so
very
exactly
Eidos
that
they
either
whether
it's
a
higher
low,
at
least
no,
this
limitation
simulator.
It
doesn't
really
matter
right,
but
clearly
a
variation
matters,
a
lot,
but
so
I
guess.
This
is
just
a
very
simple
example
for
how
we
could
use
simulator,
even
without
any
further
modifications,
and
there
definitely
other.
B
You
know
variables
that
you
can
extract
from
this,
but
in
the
future,
what
you
might
see
in
one
of
these
schools
is,
for
example,
a
plot
of
you
know
something
like
these
exponents.
You
know
evolving
over
time
right
once
we,
you
know
enable
generation
of
data
in
the
lower
levels
of
aggregation
and
at
some
point
possibly
you
know
relatively
soon,
they'll
probably
make
this
a
simulator
repo
public.
You
know
provide
some.
You
know
sample
script
so
that
people
can
play
around
with
it.
A
Meet
yeah,
it's
really
cool,
so
the
this
will
help
inform
us
in
terms
of
our
I.
You
know
our
strategic
approach
towards
rewards
distribution
and
maintaining
you
know
getting
to
a
nice
Nash
equilibrium
in
the
system.
I
maintained
that
simulation
is
the
only
way
we're
gonna
actually
learn
how
this
thing
is
going
to
behave
in
the
long
run.
So
thank
you
very
much
for
that.
That's
terrific
I.
B
A
C
Like
being
put
on
the
spot,
hi
everyone,
as
mentioned
I
work
on
the
marketing
team
here
at
Casper
Lux
at
the
time
currently
I
do
want
to
announce
for
our
community.
We
are
hosting
a
series
of
upcoming
AMAs,
specifically
on
WeChat
and
on
QQ.
If
anyone
would
like
to
join
I'll,
be
posting
out
some
more
information
shortly
on
our
socials.
So
please
definitely
follow
us
on
Twitter
join
our
telegram
and
discord
for
more
opportunities.
Meta
will
be
personally
attending
these
AMAs
and
we'll
be
answering
all
things
test.