►
From YouTube: Crossing the valley with iteration | GitLab Design Talks
Description
Nick Post and Mike Long have a chat about iteration at GitLab and how we cross the valley from local optimum to the global optimum. Watch Nick's related video where he poses the question about iteration and crossing the valley: https://youtu.be/M56CeleqpDI
A
A
B
A
But
yeah
when
I
was
watching
your
video
about
crossing
the
valley.
A
A
Ventriloquism
reminded
me
of
when
I
think
I
first
joined
a
company
that
was
doing
something
really
significantly
complex,
like
65
teams
working
on
what
we
were
calling
like
just
one
one
product
right,
a
platform
to
run
applications
on
enterprise
infrastructure-
and
I
remember
thinking
you
know
like
when
we,
when,
when
the
company,
when
it
was
actually
an
open
source
project
as
well,
so
that
added
even
more
complexity
and
just
in
terms
of
like
number
of
stakeholders,
number
of
people
contributing
the
number
of
voices
that
were
involved
in
product
choices.
A
Right-
and
I
remember
when,
when
cloud
foundry
decided
to
adopt
kubernetes
as
a
container
runtime,
and
it
already
had
a
container
runtime
had
several
iterations
of
container
run
time
until
kubernetes
was
chosen
as
the
next
container
runtime.
A
And
it
happened
from
a
distance,
it
looked
like
it
was
happening
slowly.
I
think
it
happened
rather
fast
rather
quickly
and
in
terms
of
like
having
a
version
that
a
user
could
an
operator
right,
a
cloud
operator
type
person
who
runs
the
cloud
for
the
organization
we're
able
to
get
something
in
customers
hands
like
in
a
few
months.
A
I
think
minimum
right,
maybe
even
earlier
than
that
for
some
really
rough
ideas,
and
once
we
got
that
in
the
hands
of
customers,
things
accelerated
even
faster
to
the
point
where
no
one
was
even
looking
back
kubernetes
it
worked
was
being
adopted,
but
it
really
required
the
full
participation
of
the
organization
to
support
the
core
team
and
all
the
things
that
other
teams
would
have
to
do
to
unblock
them.
A
So
that's
what
I
put
in
the
slack
conversation.
Basically,
I
still
believe
that
the
I
the
premise
behind
iteration,
which
is
take
a
big
vision
or
a
big
problem
and
break
it
down
and
incrementally
approach
it
with
a
core
team.
That's
responsible
with
the
full
weight
of
the
organization
there
to
support
and
unblock.
We
can
do
anything
and
we
can
do
it
fast.
You
know,
I
think,
where
we
get
stuck
is
when
we,
the
vision
becomes
too
complicated.
A
We
get
distracted
by
the
nuance
and
lose
our
focus
from
the
feedback
and
responding
to
the
feedback
and
another
phrase
or
thing
that
I
like
to
think
about
a
lot
that
helps
me
kind
of
have
more
confidence
and
faith
that
organizations
can
pretty
much
do
anything.
Is
the
idea
that
if
we
want
to
move
sustainably,
that
means
like
we're
not
burning
ourselves
out
we're
breaking
problems
down
to
be
smaller.
We
have
inertia
and
that
energy
leads
to
tangible
results
that
customers
can
use.
A
A
I
understand
your
concern.
I
really
understand
your
concern,
though
I've
even
when
I
was
at
thoughtworks
too
I
I
yeah.
We
had
to
take
a
on-premise
like
install
and
run
yourself
an
actual
project
management
tool
and
put
it
in
the
cloud,
and
I
thought
there's
too
much.
Technical
debt
there's
a
lot
of
design
debt
and
we
had
it
running
on
the
cloud
in
like
a
month
or
two,
and
as
soon
as
that
happened,
we
started
getting
more
feedback
about
the
performance
about
the
user
experience
and
again
things
moved
really
fast.
B
It's
it's
not
yeah.
I
mean
it's
basically
a
a
theme
that
I've
seen
running
across
analytics
where
we
need
to
become
more
suitable
for
enterprise
type
customers
and
in
order
to
be
more
suitable
for
enterprise
tech
customers.
We
need
analytics
capabilities
that
span
the
entire
instance
rather
than
focusing
in
on
a
particular
group,
and
then
our
back-end
infrastructure.
B
Our
back-end
infrastructure
is
is
basically
not
configured
in
the
right
way
for
us
to
be
able
to
do
instance,
level
reporting
and
the
first
project
manager
the
product
manager.
I
worked
with.
She
dealt
with
this
by
like
little
like
avoiding
some
of
the
some
of
the
instance
level
things
and
then
offloading
some
of
the
complexity
into
the
user
interface.
So
the
user
had
to
deal
with
it
and
then
the
second
product
manager.
B
B
So
this
the
reason
I'm
asking
this
question
and
I'm
speaking
to
products
engineering
ux,
is
because
I
want
to
create,
like
a
clear
narrative
of.
I
want
to
set
this
next
product
manager
up
for
success,
and
I
want
to
tell
them
exactly
like
how
we
should
distribute
our
efforts
and
and
what
the
focus
is
on,
because
the
challenge
the
challenge
within
analytics
is
going
to
remain
the
same.
We
still
need
to
figure
out
how
to
do
instance,
level
reporting-
that's
not
going
to
go
away
because
our
customers
need
it,
but
we
need
to.
B
B
It's
it's.
It's
not
a
concern.
It's
it's
a
challenge
that
I
want
to
like
actually
address
this
time.
Yep.
A
There's
some
other
things
in
there
too,
like
feeling
feeling
like
it's:
okay
to
make
mistakes
like
knowing
we're
gonna
make
mistakes.
B
A
I
think
sometimes
when
people
they
think
about
a
vision
or
an
idea,
they
want
to
make
it
very
robust
out
of
gates
and-
and
that's
what
kind
of
leads
to
this
sort
of
hardened
thinking,
this
thinking
that
it's
difficult
for
other
people
to
participate,
because
we've
become
so
rigid
in
our
thinking
right.
A
So
I
think
again,
it
kind
of
goes
back
to
a
simple
concept
and
that
could
be
worded
in
any
way
it
could
be
worded
as
a
a
competitor,
something
you
know
we
want
to
best
a
competitor
or
we
want
to.
We
have
a
pithy
way
of
describing
an
experience
that
we
want
people
to
have
customers
or
users.
A
B
B
Yeah,
so
it
seems
like
there
is
a
a
transition,
that's
required,
but
over
the
course
of
the
transition,
a
series
of
experiments
need
to
be
run
in
order
to
understand
whether
the
the
grand
vision
pieces
of
it
were
worthwhile
pursuing
yeah.
I
think
there
was
the
I
forgot
what
the
amazon
quote
quote
was,
but
you
need
to
be.
B
You
need
to
be
hot,
like
concrete,
concrete
on
the
outcomes,
but
a
little
bit
more
flexible
with
the
with
the
with
the
details,
the
process
of
how
you
actually
yeah
yeah,
some
something
like
that
concrete
flexible.
I
need
to
find
that
and
maybe
I'll
post
that
in
there,
but
that's
a.
B
Yeah,
I
think
what
would
be
nice
is.
If
we
figure
out,
I
mean
it's
a
big
challenge
and
the
the
way
that
you
address
this
sort
of
challenge
will
be
unique
to
to
each
particular
use
case,
but
it
would
be
nice
to
articulate
this
challenge
in
our
handbook.
B
So
because
I
I
know
that
a
number
of
other
designers
I've
spoken
with
are
feeling
a
similar
challenge
like
dan
moore
has
talked
about
it
in
terms
of
like
compliance,
navigation
and
all
this
sort
of
stuff.
Holly
reynolds
came
to
me
yesterday
and
we
spoke
about
it
in
terms
of
like
extensible
issues,
so
I
think
a
lot
of
people
are
feeling
it
and
I
think
it's
nice
to
to
know
that
other
people
are
facing
facing
similar
challenges.
B
It'd
be
nice
if
we
had
a
bit
of
a
framework
and
the
way
that
how
we
as
git
lab
address
it,
because
it's
different
for
a
lot
of
other
companies
as
well
like
the
transformational
step
change
is
a
thing
the
1.0
or
the
yeah
yeah
yeah.
A
And
I
I
think
a
lot
of
things
happen
on
timelines
that
are
tough,
they're
difficult
for
us
to
wrap
our
heads
around.
You
know
we
want.
We
want
to
do
everything
in
one
release.
We
want
that
release
to
be
the
best
and
it
can
you
know
I
think
git
lab
has
released
over
100
times
now
in
the
past.
A
I
don't
know
what
it
would
be:
five
or
six
years,
that's
a
hundred
iterations
so
to
speak,
and
I
think
the
goal
would
be
to
bring
that
number
to
thousands
of
iterations
in
the
next
five
years
to
cross
that
big
hill
that
you
drew
right.
We
can't
keep
iterating
at
this
pace
if
we
want
to
take
the
next
till,
that's
my
opinion,
and
I
think
we
can
do
it,
but.
B
So
yeah,
that's,
that
was
what
was
on
my
mind
and
I'm
I'm
I'm
speaking
with
jeremy,
about
this
particular
problem
and
I'm
going
to
record
it
as
a
gitlab
design.
Talk
on
monday
he's.
He
seems
to
have
like
some
some
nice
analogies
for
it
as
well
in
terms
of
greedy
algorithms,
for
global
versus
local
optimization,
and
also
he
was
talking
about
how
you,
how
like
yeah
envisioning
like
steering
a
ship
and
there's
two
there's
two
strategies
you
could
effectively
take.
Is
you
can
you.