►
From YouTube: ‘AMMA’ aka ‘Ask Molly Mackinlay Anything’ - Open Q&A with Molly Mackinlay - Molly Mackinlay
Description
Open Q&A with the Head of Engineering, Product, & Research Development at Protocol Labs
A
I
did
not
know
that
we
were
going
to
name
this
literally
after
me,
but
I'm
down.
Let's
make
it
a
thing
going
forward,
I'm
very
happy
to
really
quick
context
for
people,
just
if
you
showed
up
and
you're
like
who's.
This
person
standing
on
stage
I
lead
the
pl
endres
team,
which
is
about
almost
200
Engineers
researchers,
infra
experts,
product
managers,
ux
designers,
you
name
it.
A
As
a
five
percent
open
source
contributor,
I
was
working
at
Google
full-time
in
education,
I
was
like
wow
I,
really
really
wish
that
all
of
my
teachers
and
students
when
we're
building
applications
like
Google
Classroom,
can
collaborate
offline,
local
first
and
that's
what
kind
of
spurred
me
to
start
participating
in
the
ipfs
community
from
there
eventually
quit
my
job
at
Google
joined
ipfs,
full
time
as
their
first
go:
ipfs,
product
manager
and
now,
four
years
later.
This
is
what
I
do
so.
A
That's
you
know
to
put
things
short,
I've
gotten
gotten
a
chance
to
work
on
a
lot
of
different
parts
of
the
stack
got
to
lead
the
ipfs
0.5
release,
which
included
a
huge
rewrite
of
the
ipms
DHT
lots
of
collaboration
between
ipfs
and
lip
P2P.
In
order
to
make
that
happen,
let
the
the
launch
of
the
filecoin
network
on
the
engineering
and
research
side,
so
everything
involved
in
making
sure
that
the
Lotus
implementation
worked
and
running
the
infrastructure
of
Space,
Race
and
anytime.
Anyone
in
those
last
couple
of
months
was
reading.
A
Writing
that
are
coming
up,
whether
it's
about
getting
started
in
web
3
and
kind
of
how
to
how
to
you
know,
accelerate
your
trajectory
from
there
whether
it's
growing
from
you
know
the
junior
pm
to
Leading
200
person,
engineering
teams,
I'm
really
happy
to
answer
any
questions.
This
is
an
Amma,
so
I'm
not
here
to
talk
to
you
you're
here
to
ask
me
questions
and
so
yeah
we'd
pass
it
off
to
to
people
who
who
has
a
question?
What
can
I
help
answer.
B
Now
the
challenge
is
like:
we
don't
have
enough
resources.
So
how
did
you
manage
for
PL
to
go
through
a
process
where
you
don't
have
enough
resources,
but
you
have
a
huge
ambition?
What
are
the
steps
you
took
till
where
you
are
right
now,
it's
a
much
relative
team,
where
you're
actually
giving
people
grants
right?
How
did
you
do
that.
A
Yeah
I
mean
ipfs
has
been
giving
people
grants
since
the
very
beginning.
So
even
when
you
didn't
have
enough
resources
in
terms
of
humans
or
in
terms
of
kind
of
like
the
like,
there
was
a
lot
of
support
for
open
source
contributors
and
a
lot
of
development
work.
That
was
happening
not
just
with
a
team
of
hired
Engineers
I.
Think
the
vast
majority
of
the
people
who
worked
on
ipfs
in
the
early
days
were
were
like
contractors
or
were
open
source
contributors
who
started
building
something
amazing
and
they
was
like
Hey
can.
A
Can
you
keep
doing
that?
That's
awesome
like
what,
if
you
devoted
a
little
bit
more
time
in
bandwidth
to
that
and
so
kind
of,
even
from
from
early
days
in
ipfs
Project
history
resources
like
the
allocation
of
hey,
we
have
X
amount
of
small
money
to
to
put
to
things.
A
good
fraction
of
that
was
going
to
people
who
are
contributing
in
the
open
source.
Community
were
maybe
working
some
other
full-time
job
somewhere,
but
we're
passionate
engaged
in
making
impact
and
many
of
those
people
now
we're
work
on
ipfs
full
time
which
is
fantastic.
A
Like
you
have
folks
like
discordian
who
I
don't
know
if
folks
got
a
chance
to
meet,
who
started
out
just
as
like
the
most
active
contributor
in
the
ipfs
Discord
like
that
was
just
he
was
just
fantastic
at
it.
He
was
answering
everyone's
questions
was
like.
Can
you
do
that,
like
maybe
more
like
what
else
isn't
taking
up
your
time?
Can
we
help
you
allocate
more
resources
to
this?
You
have
Lytle
who
wrote
the
entire
ipfs
desktop
and
companion
applications
as
open
source
projects.
It
was
like
I
just
want
to
work
on
these
things.
A
This
is
what
I
want
to
hack
on.
It's
like.
This
is
amazing.
Please
keep
doing
that,
and
so
there
was
always
kind
of
a
shared
resource
allocation
between
hey.
We
need
someone
to
do
X.
We
need
to
go
hire
someone
or
we
need
to
go,
find
someone,
and
someone
amazing,
has
already
done
this
thing
open
source.
We
want
to
reward
them
and
encourage
them
to
keep
doing
that
sort
of
thing,
but
prioritization
is
hard.
A
Obviously,
that
ends
up
with
maybe
slightly
more
chaotic
development
of
a
road
map
and
of
an
ecosystem
than
being
very
centrally
planned,
and
that's
kind
of
also
core
to
what
ipfs
is.
It
doesn't
always
grow
in
like
a
single
linear
kind
of
like
company
narrative,
it
grows
as
an
ecosystem,
and
so
other
things
that
you
know
we
prioritized
were
making
sure
that
you
know,
in
addition
to
you
know,
maybe
you're
trying
to
get
out
a
release
that
has
new
IPS
DHT
in
it.
A
That's
a
that's,
actually
a
really
important
part
as
you're
trying
to
build
kind
of
an
open
source
project,
an
open
source
Community.
You
have
to
devote
some
resources
to
that
Community.
Otherwise
it
can't
take
off.
Otherwise
people
will
be
blocked
and
won't
be
able
to
have
that
permissionless
Innovation.
A
You
can
do
things
actually
to
architect
around
it,
though
so,
for
example,
some
of
the
ipvm
stuff,
if
you
architect
your
your
code
base
or
your
implementation,
such
that
it
has
kind
of
a
package,
managery
feel
where
anyone
can
release
new
packages
and
can
deploy
them
to
run
within
a
VM
like
environment
that
can
avoid
there
being
a
gatekeeper
on
some
implementation
that
needs
to
be
released
in
order
to
fix
people's
bugs
so
that
that's
an
additional
area
of
focus
for
for
ipfs.
These
days
was.
B
A
I
mean
ipms
has
been
open
source
from
day
Zero
from
before
day
Zero
before
ipfs
was
ipfs.
Juan
was
building
this
in
the
open,
as
like.
A
data
pack,
data
data,
Version
Control
Data
package
manager
in
I
think
he
got
a
grant
from
like
an
open
science
project
that
anyways
like
hacking
on
this
and
then
eventually
was
like
wait.
A
There's
a
there's,
a
connection
here
between
content
address
data
and
the
ways
that
you
can
incentivize
distributed,
storage
networks
and
then
was
able
to
to
go
and
actually
start
protocol
labs
and
push
that
forward,
but
yeah
from
from
Day
Zero.
These
things
were
open.
I
think
it
was
just
Juan
as
like
the
only
member
of
like
protocol
labs
for
a
while,
and
it
was
Juan
and
open
source
communities,
hacking
on
things
and
and
being
very
Community
native
and
yeah.
B
C
I
have
a
bit
of
maybe
of
a
follow-up
question
from
that
to
use
the
chairs
prerogative,
while
people
sort
of
think
about
the
framing
of
questions.
But
it's
you
know.
We
talked
about
the
early
stage
work
there
about
how
you
then
sort
of
incentivize
the
community
to
start
to
contribute
to
define
something
once
it's
sort
of
semi-defined,
there's
a
bit
of
a
sort
of
Chasm
between
overly
defining
something
in
order
to
sort
of
enable
people
to
focus
on
the
priorities
and
and
suppressing
Innovation.
C
A
A
There's
going
to
be
this
community,
that's
going
to
go
focus
on
this
thing
and
other
people
are
welcome
to
come,
join
here's
the
ways
in
which
you
can
get
involved
and
collaborate,
but
make
it
so
that
you
don't
overly
set
goals
that
hamper
other
people
to
innovate
in
different
ways
across
the
project.
So
maybe
some
people
are
focused
on
content
routing
and
some
people
are
focused
on
content
deduplication.
It's
like
great
awesome.
A
We
need
all
of
those
things,
but
you
can
focus
so
you
don't
have
to
work
on
everything
all
at
once
important
piece,
a
other
important
piece
when
you're
setting
goals
don't
over
Define,
exactly
how
you're
going
to
achieve
that
goal.
So,
for
example,
the
content
routing
performance
Improvement
was
like
look,
we
just
need.
We
really
need
at
the
edges
of
the
network
to
be
able
to
find
content
very
quickly.
A
So,
like
that's
the
high
level
goal
how
we
achieve
that
goal,
it
actually
can
be
a
combination
of
different
things,
and
so
you
can
have
teams
researching
multiple
different
Pathways
experimenting
with
them
and
seeing
how
much
impact
they
had,
and
so
actually
it's
succeeding
at
having
you
know,
sub
three
second
95th
percentile
content
resolution
came
from
a
number
of
different
improvements.
There
was
both
algorithm
changes
to
the
DHT
itself,
but
actually
a
large
part
of
that
was
having
nodes
self-identify,
whether
or
not
they
should
participate
in
the
DHT,
and
so
it
was
shrinking.
A
The
number
of
active
DHT
participants
to
those
that
were
very
you,
know
high
quality
and
were
going
to
be
good
servers
of
content
and
routers
of
of
requests,
and
so
that
that
wasn't
something
we'd
initially
like
really
latched
on
to
as
like
how
we
were
going
to
go
and
solve
this
problem,
but
it
came
out
of
some
of
that
research
work
of
like
well.
Why
are
these
queries?
Taking
such
a
long
period
of
time,
I
was
like,
oh
well,
because
they
keep
backtracking
okay.
A
Well,
why
do
they
keep
backtracking,
oh
because
of
the
20
nodes
that
they
try
and
dial
for
the
next
step
of
their
DHT
query:
they're,
all
offline
or
they're
all
inaccessible?
And
so
you
have
to
go
back
and
you
have
to
dial
a
new
set
of
people
and
we're
just
spending
all
of
this
time
like
looping
through
that,
and
so
that's,
maybe
one
of
those
examples
of
like
don't
overly
Define
how
you're
going
to
get
to
a
problem.
But
you
know
synthesize
the
information
nation
of
really
what
is
core
to
the
problem.
A
So
you
get
just
enough
Focus
to
then
align
a
group
of
people
and
actually
be
able
to
succeed.
So
we
went
from
like
40
second
content
routing
time
to
sub
three
second
content
routing
time
within
four
months,
and
it
was
really
you
know.
It
was
three
months
of
work
and
then
a
month
of
launch
and
roll
out
and
then
by
the
time
that
it
rolled
out
pretty
effectively
across
the
network.
A
D
Thanks
so
much
centrism
Legacy,
so
it's
basically
a
follow-up
to
the
discussion.
You
just
had
I
think
what
we
maybe
it's
out
there,
but
we
haven't
found
it
yet
so
kind
of
a
community
road
map
or
just
more
transparency
who
is
working
on
what
right
just
to
make
different
concrete
example.
You
know
what
we
need
is
encryption.
We
need
access
control,
we
need
stuff,
our
own
identity.
You
know
a
bunch
of
other
things.
Promo,
storage
and
in
our
position
is
okay.
D
If
it's
not
there
we're
going
to
build
it
right,
but
our
team
is
about
that
big
to
do
and
just
in
general,
in
the
ecosystem
there
was
so
to
do
and
at
the
end
of
the
the
community
is
still
relatively
small
to
the
task
at
hand.
So
I
think
we'll
be
really
good
to
get
this
transparency
to
just
better
or
allocate
resources
to
the
most
boring
issues.
A
I
100
agree
and
if
anything
I
feel
remiss
in
previous
Road
mapping
so
for
context,
I
was
the
ipbs
Project
Lead
from
20.
What
was
that
20
2019
like
May,
through
June,
2020
or
so,
and
and
now
do
this
much
broader
role,
but
in
in
some
of
the
road
mapping
exercises
that
I
led.
It
was
a
lot
of
like
needs,
Discovery
or
problem
Discovery,
and
not
a
lot
of
teams
reporting
out
across
the
ipfs
community.
What
okay?
Where
are
you
going
to
focus?
What
are
you
going
to
go?
Work
on?
A
It
was
more
of
a
like
synthesis
of
these
are
things
that
we
think
are
really
high
priority
right
now,
someone
should
go
work
on
them,
but
not
a
whole
lot
of
okay
who's
who's
raising
their
hand.
What's
your
road
map,
let's,
let's
communicate
that
with
each
other
and
I
think
that
is
so
so
needed,
and
that's
exactly
where
the
Falcon
Community
is
at
right
now
as
well,
in
really
having
these
many
different
teams
that
are
taking
parts
of
the
problem
around
data
access
control.
A
Like
literally
there's
a
team,
that's
built
Building
Solutions
around
how
to
do
threshold
signatures
for
data
Access,
Control
right
now,
and
so
I
definitely
think
that
that's
exactly
what
we
need
as
a
project.
Right
now
is
have
each
of
these
amazing
groups
that
are
presenting
today
in
ipfs
Camp
to
share
kind
of
their
roadmap,
which
parts
of
the
problem
they're
working
on
also
maybe
a
little
bit
of
how
you
can
get
involved.
A
If
you
want
to
start
contributing
to
those
different
parts
of
the
problem,
just
to
plug
a
thing
that
is
not
doesn't
actually
exist
yet,
but
we
are
working
on.
It
is
kind
of
like
an
open
road
mapping
tool
where
everyone
can
kind
of
write
their
set
of
GitHub
issues
of
like
hey.
This
is
our
roadmap.
These
are
the
areas
we're
focusing
on,
and
this
is
the
impact
it'll
have,
and
then
you
can
visualize
it
across
many
different
teams
in
say
a
set
of
themes
like
identity,
Access,
Control
performance.
A
Is
that
and
the
other
thing,
and
so
that's
something
that
we're
starting
to
get
the
Kubo
roadmap
into
we'd
like
to
get
the
lip
P2P.
Let
go
JS
and
rust
lip
P2P
roadmaps
into
many
of
these
groups,
like
those
specific
implementations,
might
share
out
like
well.
Core
maintainers
of
X
are
working
on
on
this
area,
but
there's
so
many
other
teams
and
if
we
don't
find
a
way
to
visualize
all
of
that
work
more
broadly,
it's
difficult
to
know.
A
What's
coming
and
I
think
we
should
actually
really
invest
in
that
and
I
want
that
to
be
one
of
the
follow-ups
from
ipfs
Camp
to
like,
hopefully
have
a
0.1.0
of
this
road
mapping
tool,
get
more
different
groups
to
kind
of
share
out
their
roadmaps
and
then
hopefully
align
and
be
like
great.
If
you're
working
on
that,
let's
combine
forces
and
make
it
happen
with
a
little
bit
more
resources.
A
C
E
About
resources-
and
maybe
you
know
looking
at
it
from
a
slightly
different
perspective-
I
know
that
we
have
a
lot
of
Builders
here
in
the
room
and
I
know
that
you
are
doing
something
amazing,
called
Launchpad
and
would
love
to
kind
of
for
you
to
explain
the
intersection
of
how
builders
that
are
currently
in
the
community
that
are
hiring
and
onboarding.
Their
teams
can
like
Leverage
Launchpad
as
a
resource
to
kind
of
supercharge
I,
guess
their
success.
Yeah.
A
So
for
everyone,
who's
in
the
room
for
Katie's
presentation,
they've,
probably
got
to
know
a
little
bit
more
about
Launchpad,
but
Launchpad
is
something
we
created
a
year
ago,
almost
exactly
it
kicked
off.
Our
first
pilot
kicked
off
November
2nd
of
last
year
and
it
was
exciting
and
crazy
and
we
were
figuring
it
out.
As
we
went
entirely.
It
was
four
weeks
long.
We
had
no
idea
what
we
were
doing,
but
we
iterated
a
lot
since
then.
We've
now
run
six
I
guess
we're
on
our
seventh
cohort
yeah,
because
we
index
from
zero.
A
So
that
makes
everything
confusing
but
and
really
I
mean
there's
two
main
ways
that
Builders
can
benefit
from
Launchpad.
You
can
either
join
Launchpad
and
use
it
as
a
amazing
index
across
ipbs
Falcon,
lib
P2P
ipld.
How
these
things
work
under
the
hood,
how
you
can
contribute
to
them,
how
you
can
leverage
them
for
the
the
applications
you're
building
and
also,
how
do
you
utilize
resources
across
the
pl
Network?
So
it's
available
for
any
new
hire
to
any
team
that
is
kind
of
an
active
member
of
the
protocol.
A
Apps
Network-
and
you
know
again,
our
our
hope
is
that
it
helps
people
on
board
into
this
community
really
really
fast
and
more
deeply,
because
we
have
so
many
people
joining
these
days.
We
want
it
to
be
a
really
smooth
experience
on
the
flip
side,
if
you
are
say
already
a
part
of
this
community
and
running
a
team
or
organization,
that's
in
the
pl
Network.
A
You
can
use
it
to
maybe
offload
some
of
that
training
and
on
boarding
time
that
you'd
otherwise
spend
getting
your
software
Engineers
or
your
community
managers
or
your
technical
program
managers
or
whoever
up
to
speed
so
that
they
can
learn
all
that
knowledge
quickly,
but
also
build
really
strong
connections
to
multiple
kind
of
super
nodes
across
the
ipfs
space
you
get
mentorship.
You
also
get
to
work
directly
with
you
know:
core
Kubo,
maintainers
or
core.
A
Go
lip
P2P,
maintainers
or
different
different
areas
like
that
and
if
you're
trying
to
maybe
make
changes
to
some
of
those
code
bases
can
be
really
useful
to
know
exactly
how
decisions
are
being
made.
How
to
you
know,
connect
with
the
right
people
to
get
feedback
early
and
you
get
to
meet
a
lot
of
other
people
from
other
teams
that,
maybe
you
can
collaborate
with,
and
so
that's
that's
kind
of.
A
If
you
are
leading
an
organization
or
company
in
the
pl
Network,
you
can
send
your
new
hires
to
us
and
we
will
help
onboard
them
and
we
are.
We
also
have
a
recruiting
component
of
that
as
well.
Launchpad
recruiting
that
helps.
Pl
Network
companies
hire
amazing
new
people
to
their
teams
and
fill
their.
You
know
most
important
engineering
or
product
or
other
roles.
I
think
what
was
the
next
one?
A
We
were
highlighting
something
something
along
the
lines
of
like
startup
operator
as
well:
people
who
can
really
help
execute
a
growing
business,
because
many
teams
across
the
pl
Network
are
in
the
like
10
person.
You
know
two
to
11
person
stages
and
really
need
to
to
bring
on
their
next
Talent.
So
that's
that's.
My
launch
pad
plug,
definitely
feel
free
to
ask
me
more
questions
about
it
or
I
mean
how
many
people
in
this
room
have
gone
through
LaunchPad
or
sent
to
someone
on
their
team
through
LaunchPad
yeah.
A
F
Hello,
I
think
my
question
already
got
answered
so
I'm
going
to
ref
on
this
one
I
guess
I'm
curious,
so
Molly
you're
somebody
who's
been
in
charge
of
this
organization
that
has
almost
turned
into
like
two
very
distinct
projects
in
some
ways,
among
many
other
important
ones,
there's
kind
of
the
file
coin
ecosystem
and
the
project
that's
been
built
out
around
there.
There's
ipfs
and
they're
like
they're,
intimately
intertwined
as
well.
F
I
guess,
I'm
curious,
like
you
as
an
organizational
leader,
you've
had
to
build
out
the
organizations
and
had
a
strategy
and
thinking
about
how
each
of
these
evolve
in
their
own
way.
So
what
has
been
your
thinking
about
how
you
build
the
stream,
the
organization
around
filecoin
versus
ipfs
and
like
what
are
the
things
that
you
think
ipfs
is
just
like
really
uniquely
able
to
do
because
of
the
way
that
it's
structured
that
you're
most
excited
about.
A
Very
interesting
question:
it's
obviously
evolved
when
I
first
started
working
on
ipfs.
There
was
a
team
that
had
already
started
when
I
first
joined
full-time.
There
was
a
team
that
had
already
started
working
on
falcoin.
You
know
filecoin
team
and
they
were
like
off
in
a
corner
like
with
their
own
little
private
code
base
like
starting
starting
to
lay
the
initial
building
blocks
of
what
falcoin
would
become,
and
so
you
know
for
my
first
year,
I
like
really
didn't
think
about
Falcon
at
all.
I
was
like
yep,
that's
the
thing.
A
Other
people
are
working
on
it,
I
don't
have
to
I'm,
focusing
on
ipfs
and
my
ipfs
community
and
all
of
the
great
stuff
there
and
then,
as
as
Falcon,
really
started
accelerating
and
like
there
was
some
code
there
and
they
started
open
sourcing
their
repos.
Then
we
started
getting
a
lot
more
involved
of
like
hey
we've.
A
You
know
the
the
intersection
between
the
two.
You
can
get
ipfs
users
what
they
want,
which
is.
You
know,
content
persistence
that
you
don't
force
some
ipfs
node
to
like
stay
online
all
the
time,
but
that
you
can
delegate
that
to
other
parties
in
the
network
and
have
an
incentive
system
around
it,
while
also
harnessing
the
changes
you
wanted
to
make
to
support
many
other
groups.
A
So
yeah.
When
you
know
in
those
days
I
think
we
had.
You
know
ex
differing
layers
of
like
totally
separate,
not
paying
any
attention
to
each
other,
then
realizing
that
there
were
a
lot
a
lot
more
intersections
figuring
out
how
we
could
support
and
share
knowledge.
A
lot
more
between
different
groups
than
after
Falcon
was
actually
launched,
doing
a
lot
together
to
try
and
make
sure
that
kind
of
the
life
cycle
of
you
know.
A
Great
I
have
data
that's
stored
in
ipfs,
I'm
running
all
of
ipfs
Wikipedia
cool,
like
what's
my
user
flow,
to
have
that
data
stored
on
filecoin?
How
do
I
make
that
really
easy?
There's
been.
You
know,
Staffing
a
team
on
that
intersection
of
like
hey
this.
This
needs
to
interoperate
really
nicely,
and
we
need
to
think
about
that
kind
of.
Like
product
user
Journey,
otherwise
ipfs
users
aren't
getting
what
they
wanted
out
of
filecoin
and
vice
versa.
A
The
way
of
accessing
and
storing
data
and
filecoin
is
like
very
clunky
and
totally
separate
from
everything
we
love
about
ipfs
and
so
working
with,
with
with
many
teams
to
like
identify
and
like
product
improve.
That
interface
was
important,
but
I
think
it's
also
really
important
that
you
we
we
have
like
a
separate
core
maintainership
teams
and
actually
now
many
separate
core
maintainership
teams
for
different
ipfs
implementations
for
different
filecoin
implementations
and
then
Gathering
bodies
or
or
venues
where
we
can
get
together
and
talk
about
the
improvements
coming.
A
I
think
ipfs
actually
learned
a
lot
from
filecoin
about
that.
Like
the
ipip
process
being
like
cool,
we
can
we've,
we've
had
conversations
about
how
to
do
rfcs
in
ipfs
forever,
and
it
had
always
turned
into
a
bike
shed
of
you
know.
Everyone
wants
to
bring
their
favorite
open
source
RFC
process
and
there's
all
this
like
governance,
crofty
overhead.
A
You
can
innovate
and
and
iterate
on
ipfs
in
many
different
types
of
networks,
much
faster
than
you
can
on
ipfs
or
on
Falcon,
where
you
have
to
achieve
consensus
in
the
blockchain,
and
you
have
to
have
these
kind
of
like
major
Network
upgrades,
and
so
that's
that's
a
thing
to
harness.
You
know
when
I
ask
people:
why
are
you
using
ipfs
100
of
the
time?
A
First
thing,
content
addressing
I
need
content,
addressing
I'm
working
in
some
context,
where
I
need
the
verifiability
I
need
to
time
stamp,
something
into
a
blockchain
I
want
to
use
that
content
addressing
to
dedupe
data
between
multiple
versions,
like
that's
the
core,
that's
what
ipfs
does
really
really
well
is
like
taking
the
core
of
ipld
and
making
it
accessible
to
applications
that
want
to
then
transfer
data
over
it.
So
harnessing
that
and
bringing
that
to
more
places,
I
think,
is
a
really
good
thing.
G
G
A
My
goodness
I
get
asked
this
all
the
time
and
I
like
don't
I,
never
have
a
good
answer
for
it
like
one.
It
does
depend
on
the
decision
like
how
it
happens
where
it
comes
from
I
mean
we
benefit
a
lot
from
having
a
brilliant
Visionary,
like
Juan
Bennett,
like
at
our
Helm,
helping
guide
many
many
decisions
and
seeing
around
corners
like
I.
A
Don't
think
any
of
us
foresaw
how
impactful
something
like
ipfs
would
be
back
in
2014
like,
and
you
know,
that's
Juan
deeply
going
into
the
the
history
of
computing
and
harnessing
some
of
these
ideas
that
were
stuck
in
Academia.
These
were
not
new
ideas.
These
had
been
around
for
like
a
really
long
time.
People
had
been
working
on
Emilia
cheese
for
like
years,
but
actually
being
like
wait.
No,
we
can.
We
can
build
like
real
networks
on
top
of
this.
A
They're,
not
crazy
they're.
Just
you
know
the
things
we
do.
One
of
them
is
like
writing
down
kind
of
like
a
core
decision
tree.
This
is
actually
how
we
decided
the
strategy
for
launching
Falcon
mainnet,
which
was
surprisingly
complicated,
and
we
had
a
couple
of
different
ways
that
we
could
launch
the
network
and
we
built
a
spreadsheet
which
was
like.
Okay,
you
have
you
know,
strategy
a
strategy,
B
strategy
C.
So
like
strategy,
a
like
just
hard
cut,
the
Space
Race
network
turn
it
off
start.
A
A
You
know:
okay,
from
a
development
perspective
from
a
cost
perspective
from
a
security
perspective,
those
were
kind
of
your
facets
in
the
table
and
then
you
kind
of
wrote
at
the
intersection
of
each
of
those
things:
okay
for
option
one:
what
if
the
security
benefits
and
costs-
and
those
are
like,
like
you,
try
and
make
those
as
small
as
you
can,
like
six
words
Max
trying
to
sum
up
that
cell
in
that
decision
making
table,
and
then
you
color
code
it
red
for
bad.
This
is
a
bad
outcome,
green
for
oh!
A
But
trials
need
to
write
down
the
cells
that
you
actually
really
care
about,
and
that
makes
it
obvious,
and
then
you
can
share
that
out
with
your
entire
team
as
long
as
it's
not
you
know,
security,
vulnerability,
confidential
and
it's
obvious
to
everyone
else
how
you
made
this
decision
creates
really
strong
alignment.
This
is
actually
what
we
did
for
the
fvm
buildernet
the
the
way.
A
There
were
many
different
ways
that
we
could
run
an
incentivized
testnet
for
for
Falcon,
Virtual
Machine
did
this
and
we
ended
up
with
strong
alignment,
strong
outcome
on
a
decision
that
none
of
us
had
expected
per
se
at
the
beginning
of
that
conversation,
because
what
you
do
is
you
iterate
on
the
different
strategies
you
could
take?
You
come
up
with
new
options,
and
you
have
a
very
concrete
way
of
saying
so.