►
Description
Presented by Nadia Eghbal
Healthy, popular open source projects require significant time in the form of contributions. Where does that time come from? We’ll explore a brief history of successful contribution models, as well emerging philosophies for managing contributions in today's modern world.
About CodeConf
CodeConf improves the software community by providing a forum for thought-provoking talks and forging social connections. The fourth installment of the CodeConf series took place in Hollywood in 2016. This year's event focused on systems engineering projects, practices, and programs in the open source community.
For more information on this year's CodeConf, go to:
https://codeconf.com/
A
A
What
I
mean
by
the
market
is
people
that
are
using
open
source
software
and
people
that
are
contributing
to
open
source
software,
and
so,
as
the
amount
of
users
and
contributors
has
changed,
and
the
makeup
of
those
of
those
populations
has
changed.
We've
seen
sort
of
changes
in
the
models
themselves
of
how
we
manage
those
contributions,
so
starting
in
the
beginning
of
open
source
software
in
the
80s
and
late
1990s,
you
had
the
DD
FL,
which
was
sort
of
like
this
prevalent
model
for
managing
open
source
contributions.
A
Vdsl
is
benevolent
dictator
for
life
and
its
most
closely
associated
with
linux,
even
though
it
actually
came
from
python,
and
so
under
the
the
PDF
file
model.
You
have
people
that
can
anyone
is,
you
know,
able
to
make
contributions
anyone's
able
to
have
their
say,
but
in
the
end
you
have
this.
Like
one
person,
the
benevolent
dictator
who
has
final
say
on
everything,
and
this
made
a
lot
of
sense
early
on
at
open
source,
because
you
didn't
have
as
many
people
using
open
source
and
you
didn't
have
as
people
contributing
back
to
it.
A
So
in
like
the
80s
proprietary
software
was
still
the
default.
Open-Source
was
sort
of
like
the
counterculture
or
alternative
they
didn't
have
as
much
demand
on
the
software
and
then
on
the
contributor
side.
In
like
the
80s,
it's
a
lot
harder
to
learn
how
to
code
right
learning.
How
to
code
is
not
that
accessible
being
a
software
developer
is
sort
of
like
this
serious
profession
that
have
to
devote
yourself
to
so
you
don't
have
as
many
people
that
are
even
able
to
contribute
to
open
source
software.
A
Most
importantly,
there's
the
centralized
leadership
and
governance.
So
everybody
knows
who's
in
charge.
Everybody
knows
the
one
person
they're
supposed
to
go
to
if
they
want
to
see
anything
changed
and
then,
along
with
that,
you
also
have
a
more
centralized
roadmap
and
product
direction.
Because,
again,
you
don't
have
a
lot
of
people
that
are
like
trying
to
argue
about
what
the
direction
of
the
product
should
be.
You
have
one
person
who
has
final
say,
but
in
the
end,
because
it
is
a
benevolent
dictator
sort
of
by
definition,
it's
a
lot
less
democratic.
A
So
again,
open-source
is
not
like
the
default
in
the
80s.
Companies
were
nervous
about
contributing
to
open
source.
They
were
nervous
about
using
open
source
projects,
and
part
of
it
was
because
they
look
at
these
video
files
and
they're
like
okay.
So
there's
like
one
person
that
I
have
no
formal
relationship
with
and
I
have
no
formal
pathways
to
be
able
to
influence
can
make
arbitrary
decisions
that
affect
my
product,
that
I'm
trying
to
sell
to
other
people.
A
That
doesn't
really
make
sense
and
sort
of
on
this
high-level,
open
source
needed
company
buy-in
to
be
able
to
succeed.
So
those
companies
weren't
willing
to
use
open
source
software,
then
open
source
would
end
up
becoming
this
like
hobby
thing,
instead
of
becoming
to
default
that
it
is
today
and
so
that
ended
up
taking
us
into
the
meritocracy.
A
Meritocracy
was
largely
driven
by
the
Apache
Software
Foundation,
which
was
established
in
1999,
and
what
they
did
was
basically
slammed
the
btfl
model
to
let
multiple
actors
participate.
So
now,
instead
of
having
one
person
with
final
say,
you
have
like
multiple
people
that
are
weighing
in,
and
so
anyone
with
merit
can
contribute
to
this
project
or
make
substantial
contributions.
A
Anyone
with
merit
can
have
a
binding
vote
on
decisions
in
a
project,
so
this
was
really
great
for
a
company
buy-in,
because
now
companies
aren't
worried
about
well
what
happens
if
this
project
gets
really
popular
and
my
competitor
is
controlling
it.
They
have
a
path
to
be
able
to
participate,
to
have
a
say
in
the
project.
A
Apache
also
did
a
lot
around
legal
protections
for
people
that
were
contributing
to
the
project,
so
they
took
care
of
a
lot
of
these
things.
It
became
this
sort
of
like
mutual
safe
space,
and
that
is
because
we
had
the
meritocracy.
That's
the
big
reason
why
open-source
is
so
popular
today.
Part
of
it
is
because
companies
weren't
comfortable
using
it
before
and
when
they
had
this
like
DeeDee
FL
and
then
now
they
have
meritocracy,
they
feel
like
they
can
participate
and
now
they
feel
comfortable
using
it.
A
So
we
actually
have
the
meritocracy
to
thank
for
open
source
being
more
of
a
standard
today,
but
there's
still
the
problem
right
and
performative,
like
one
clause
that
I
used
earlier,
which
is
what's
merit
it's
like
well,
what
does
it
mean
to
have
merit?
How
does
one
get
merit?
I
pulled
this
from
Patchett
Software
Foundation.
Basically,
it's
saying
that
people
have
to
earn
the
right
to
have
merit.
A
Merit
is
a
privilege
that
needs
to
be
earned
and
when
they
talk
about
how
how
they
think
about
people
with
merit
that
they're
filtering
for
people
that
they
think
are
committed
enough
to
the
task.
So
they've
proven
that
they
actually
care
about
the
projects
and
that
match
the
human
attitudes
required
to
work
well
with
others.
So,
basically
saying
you're
not
going
to
rock
the
boat
so
they're,
you
know
they
have
this
sort
of
like
criteria
around
like
what
they
think
someone
would
merit
actually
means
so
basically
like.
A
If
you
want
to
be
part
of
a
meritocracy,
you
need
to
prove
yourself,
and
you
prove
yourself
by
spending
time
on
a
project
right.
So
the
more
time
you
spend
the
more
devoted
you
are,
the
more
committed
to
you
are
the
more
you're
able
to
prove
that
you
have
this
merit
and
then,
therefore,
you
have
more
of
a
say
in
the
project,
so
this
make
sense
in
theory
right
for
any
sort
of
like
membership
organization,
even
outside
of
open
source.
A
If
you
thought
about
starting
a
club
or
something
it's
like
well,
of
course,
it
makes
sense
that
only
members
or
people
who
have
earned
somehow
the
right
to
say
something
get
a
voice,
but
in
practice
this
ends
up
favoring
people
who
can
pay
to
play,
and
it's
not
just
paying
with
money
in
this
case.
But
it's
also
paying
with
time,
so
you
can
imagine
in
theory
a
company
that
can
devote
full
time,
employee
resources
to
a
project
or
somehow
create
full-time
contributors
open
source.
A
Those
people
are
going
to
be
listened
to
and
heard
a
lot
more
than
a
part-time
contributor,
even
if
that
part-time
contributor
is
really
passionate
about
the
project.
So
this
ends
up
becoming
a
problem
and
it
might
have
just
sort
of
stay
that
way
if
we
hadn't
seen
a
shift
in
how
people
were
contributing
to
open
source
today,
and
so
today
you
have
this
emerging
model
of
liberal
contribution.
As
far
as
I
can
tell,
that
term
was
coined
by
nodejs
and
they
really
helped
pioneer
this
model,
so
I'm
going
to
use
them
as
an
example.
A
A
I
saw
one
survey
that
said
that
the
number
of
companies
that
said
they
use
open
source
and
software
has
doubled
in
the
past
five
years
so
and
that
sort
of
matches
our
intervention
right
like
there
are
lots
and
lots
of
people
that
are
using
open
source
software
and
then,
on
the
other
side,
you
have
more
people
that
are
able
to
code
learning
how
to
code
want
to
contribute,
somehow
than
ever
before.
You
have
this
explosion
of
coding
classes
and
boot
camps.
All
this
has
happened
in
the
past
five
years.
A
So
now
you
have
lots
of
people
want
to
contribute
in
theory,
and
then
you
have
a
lot
of
people
who
are
using
the
software
there's
just
like
tons
of
demand
going
on
from
all
sides
and
if
you're
innocents
are
like
why
this
happened.
Basically
blaming
on
github
github
made
it
easier
for
anyone
to
jump
in
on
our
project
than
before
and
partially
enables
like
it
so
before.
If
you
want
to
contribute
to
an
open-source
projects,
each
one
has
its
own
website.
A
It
might
have
its
own
issue
trackers
or
tools
its
own
version,
control,
separate
mailing
lists,
everything
so
to
jump
into
each
one
of
these
projects
was
actually
a
lot
of
mindshare
and
today.
Obviously
there
are
plenty
of
projects
that
are
not
on
github,
but
for
the
ones
that
are
in
github
and
as
open
source
is
centralizing.
A
You
can
use
the
same
issue.
Tracker
same
version,
control
same
UI
to
contribute
to
different
projects
on
the
same
platform.
So
now
you
have
a
lot
of
people
that
are
dipping
in
and
out
of
projects
and
don't
have
as
much
commitment
to
each
one
and
sort
of
just
supporting
this.
Last
year
someone
did
a
survey
and
found
that
of
the
top
github
projects
across
languages
about
half
of
the
contributors
to
those
projects
only
made
one
contribution.
So
you
have
this
really
long
tail
of
like
very
casual
contributors,
so
yeah.
A
This
was
all
leading
to
more
noise.
And,
if
you,
if
you
think
about
up
to
this
point,
we
only
have
the
B
DSL
and
the
meritocracy
under
a
meritocracy.
All
those
noise
is
a
bad
thing,
because
the
whole
point
of
a
meritocracy
is
that
you
have
to
earn
your
right
to
have
a
say
in
this
project,
and
calves
are
contributors,
haven't,
earned
their
voice
and
so
they're
really
frustrated.
Noise.
Bsl
model
also
would
be
frustrated
by
this
noise,
and
their
reaction
is
basically
like
make
these
people
go
away,
filter
out
the
noise.
A
So
they
did
a
couple
of
things
in
2014,
no
drop
their
CLA
requirement.
Clas
contributor
license
agreement,
basically
some
legal
hoops
they
need
to
jump
through
before
you
contribute
to
their
projects
and
note
at
the
time
was
spirited
by
giant
and
so
Brian
Cantrell
saw
this
as
a
corporate
anti-pattern
who's,
basically
saying
that,
like
every
company
is
requiring
a
CLA
and
they
do
it
because
they
think
that's
what
they're
supposed
to
do.
But
what
happens
if
you
drop
it?
Would
we
see
a
difference
in
behavior
that
we
might
not
have
expected
to
see?
A
A
They
also
adopted
a
more
liberal
contribution
policy,
and
so
there
are
like
a
lot
of
different
components
to
this
and
they
happened
over
the
years,
so
I'm
not
going
to
really
go
and
tell
the
things
that
they
did.
But
I
think
this
quote
from
Michael
Rogers
sums
it
up
pretty
well,
which
was
getting
people
to
contribute,
is
hard.
Everybody
who
runs
the
project
knows
that
when
you
distribute
ownership,
people
contribute
more,
and
if
anyone
does
something
wrong,
it
is
good
at
removing
things
right
so
again.
A
There's
this
marriage
between
the
tools
that
enable
people
to
work
together
more
collaboratively
and
that
being
able
to
enable
this
more
liberal
type
of
governance
and
more
liberal
approach
to
contributions,
because,
if
anything
bad
happens,
we
have
the
tools
to
reverse
it,
so
they
did
all
that
and
then
some
fun
things
start
to
happen.
So
they
did
see
way
more
contributions
than
they
ever
had
before.
A
They
also
got
a
lot
of
free
marketing.
This
is
something
I,
don't
really
see,
node
talked
about,
but
it
was
just
something
I
noticed.
So
when
you
engage
all
these
casual
contributors-
and
you
tell
them
like
you-
are
part
of
this
project,
you
should
feel
ownership
over
those
projects.
They
get
really
excited
and
they
start
becoming
these
like
natural
word-of-mouth
ambassadors,
and
so
one
successful
project
came
out
of
that
and
they
think
really
embodies.
A
It
is
new
school
and
no
school
was
a
totally
community
run
project
bunch
of
tutorials
around
like
how
to
use
node
it
just
kind
of
spread
like
wildfire.
People
were
doing
those
goals
in
their
own
cities.
They
were
doing
different
translations,
while
he's
meetups.
It
was
really
really
successful
and
I.
A
So
all
this
was
really
great
and
it
started
kicking
off
a
lot
of
conversation
around
like
how
do
we
make
contributions
as
easy
and
commercial
as
possible,
and
so
it's
fill
a
new
model,
but
I
want
to
go
through
mention
another
example
of
a
project.
That's
adopting
the
liberal
contribution
model,
which
is
rust
and
rust,
is
interesting
because
he
actually
started
under
the
be
DFL
model,
so
it
was
started
by
one
employee
at
Mozilla
and
he
sponsor
Mozilla
worked
on
it
for
a
couple
years.
When
he
stepped
down.
A
He
means
another
VDSL
in
2013
and
that
person
decided
to
open
up
governance
to
a
core
team,
and
then
last
year
they
went
from
the
core
team
to
like
more
of
these
little
sub
teams.
So
they
basically
started
with
like
one
person
and
then
like
a
small
group
of
people,
and
now
they
have
like
an
official
governance
structure
that
is
even
wider,
and
so
rest
today
has
over
1,200
contributors,
which
is
a
lot
and
a
lot
of
them
make
regular
contributions.
A
A
And
just
sort
of
like
going
through
their
philosophy
a
little
bit.
This
quote:
comes
from
Alex
Creighton
who's,
one
of
them
and
teeners
when
he
was
describing
the
decision
to
go
from
a
core
team
to
like
all
these
extra
sub
teams,
he's
basically
saying
to
the
extent
practical.
We
should
include
a
diverse
range
of
perspectives
and
especially
include
people
like
our
users,
and
so
again
this
is
like
a
totally
different
shift
from
what
happened
under
a
meritocracy
where
it
was
like.
A
So
just
some
patterns
that
I've
seen
around
liberal
contribution
models
if
you're
interested
in
how
they
work
I.
Think
functionally,
like
a
lot
of
you
still
have
similar
governance
in
the
sense
of
there
is
some
sort
of
core
team.
There's
someone
in
charge.
It's
not
like
it's
total
anarchy,
but
the
difference
is
sort
of
their
approach
and
philosophy
to
how
they
think
about
contributions.
A
You
think
about
opening
everything
up
into
these
floodgates,
an
idea
being
that,
let's
say
in
a
bottleneck,
you
have
like
one
maintainer
who's,
trying
to
teach
like
ten
other
people
how
to
do
something
right:
the
ten
who
have
proven
themselves,
but
under
a
floodgate,
you
might
have
like
a
hundred
people
coming
in
who
are
all
really
excited
and
engaged,
and
hopefully
those
hundred
people
start
teaching
each
other.
And
then
the
original
maintainer
can
kind
of
like
take
a
step
back
and
not
be
as
involved.
A
So
it
ends
up
being
a
lot
less
work
for
the
maintainer.
Ultimately,
even
though
it's
sort
of
counterintuitive
another
feature
is
consensus.
Seeking,
rather
than
consensus,
so
under
a
meritocracy,
for
example,
you
have
people
that
are
taking
these
binding
votes
and
idea
is
whenever
there
is
a
decision
that
needs
to
be
made,
everybody
votes
under
a
liberal
contribution
model,
they
emphasize
the
discussion
and
the
process
rather
than
a
final
vote.
So
someone
might
say,
here's
an
issue
I'm
raising.
A
It's
saying
hey:
what
do
you
want
to
be
doing
and
how
can
I
make
it
as
easy
as
possible
for
you
to
get
that
done
and
finally,
I'm
just
defaulting
to
yes,
so,
starting
with
the
idea
that
all
contributions
should
be
accepted
by
default
and
if
they're
not
being
accepted,
then
it's
bonuses
on
the
maintainer
to
say
why
it's
not
good
for
the
project.
Even
if
functionally
this
doesn't
look
different.
A
The
idea
is
that,
instead
of
saying
again
like
you
need
to
prove
to
me
why
I
should
accept
your
contribution.
It's
saying,
yeah
I
want
to
accept
your
contribution.
Here
are
the
objections
I
might
have
so
putting
all
together.
I
just
wanted
to
go
through
a
couple
of
like
clarifying
questions
around
these
models.
So
first
question
is
fair.
Like
do
BSL
still
matter,
I
hear
some
people
saying:
oh
you
know,
the
PDL
model
is
totally
dying
out.
A
A
I've
looked
at
started
with
literally
like
one
person
that
worked
really
really
hard
on
the
project
up
front
and
then
maybe
the
model
changed
over
time,
but
like
they
almost
all
start
with
one
person,
so
a
B
D
FL
early
on
can
help
incubate
the
project
it
can
help
set
a
product
vision
in
this
way.
I
think
it's
actually
pretty
similar
to
company
structures
where
you
don't
start
a
company
with
like
200
employees.
A
You
start
a
company
with
a
couple
of
co-founders,
but
as
the
project
matures
and
gets
bigger,
I
see
a
liberal
contribution
models
helping
with
the
growth
side
of
things.
The
idea
is
at
some
point.
You
should
be
able
to
get
big
enough.
Your
community
gets
big
enough
that
you
can
decentralize
your
authority
just
an
idea
of
like
how
a
project
trajectory
might
look
like
today
or
how
I
envisioned
this
working
all
together.
A
You
have
sure,
like
this
incubation
phase,
early
on,
where,
like
one
person
has
an
idea,
works
really
hard
on
it
for
a
certain
period
of
time
and
then
as
they
grow
and
get
really
popular.
It
might
turn
into
sort
of
this
more
liberal
contribution
model,
but
again,
like
the
goal
is
to
grow
as
much
as
possible
that,
like
that
Authority
is
no
longer
centralized
on
one
person
and
there's
probably
something
in
the
middle
there,
where
it's
like.
A
You
know
going
from
one
people,
one
person
to
like
five
people
to
like
more
decentralized
and
just
as
examples
of
this
working
in
action.
Breaths
janggo
node
are
all
examples
of
big
projects
that
did
start
with
the
BD
FL
and
then
they
transitioned
out
once
the
community
got
big
enough,
so
Jacob
capital
Moss,
who
was
was
one
of
the
BDS
cells
of
Jango.
A
When
they
decide
to
step
down,
it
was
almost
like
an
afterthought.
He
stepped
down
because
he
realized
just
wasn't
really
needed
anymore.
Whenever
decisions
were
being
made
by
the
Django
community.
Without
him,
she
noticed
that
the
community
was
healthier.
He
noticed
that
it
revitalized
development
that
committers
were
more
excited,
and
so
he
is
sort
of
like
you
know,
I'm
still
part
of
Django.
But
why
do
I
need
to
be
the
person
in
charge?
A
The
community
is
so
strong
that
I'm
not
needed
anymore
and
so
that's
sort
of
like
the
ideal
success
model
right
and
so
on.
The
flip
side,
like
zubeidaa
fells
work
as
a
long-term
strategy.
Could
you
have
a
be
DFL
project
forever
answer?
Those
were
like
yes,
so
with
some
caveats.
So
closure
is
an
example
of
a
modern
project
that
uses
a
PDF
file
model
and,
as
far
as
I
can
tell,
has
no
plans
to
change
that.
So
there
DFL
is
rich,
Hickey
and
rich
does
makes
most
of
the
commit
to
the
project.
A
And
so
do
you
remember
what
the
rest
contribution
graphs
look
like
with
closure.
You
literally
have
one
person
who's,
making,
basically
everything
and
then
a
little
bit
sort
of
like
trickling
down
the
reason
why
this
works
and
as
far
as
I'm
going
to
tell
it
does,
work
is
because
closure
is
being
sustainably
supported
by
a
company
called
contacts,
so
Rich's
company
was
called
metadata
and
it
merged
with
a
company
called
relevance.
Relevance
had
a
very
deep
and
vested
interest
in
supporting
closure,
so
they
emerged
and
they
create
this
company
called
contacts
and
rich
works
there.
A
Everyone
is
like
perfectly
happy
with
situation
and
and
cognate
acts
for
the
foreseeable
future
is
able
to
continue
supporting
closure.
They
see
it
sort
of
as
closure
as
a
project
that
is
developed
by
cognitive
and
then
shared
out
with
a
community
which
is
really
different
from
a
community
built
project.
A
So
I
think
that
works.
It
seems
to
work
for
closure.
I
think
that
the
BDA
fellow
is
like
a
long-term
strategy
still
requires
getting
lucky
right
so
like
there
was
still
like
cognate
ex
is
kind
of
like
happen
to
help
support
the
development
of
closure,
and
so
it's
really
frustrating
to
be
able
tell
if
a
maintainer
says
sort
of
like
well.
What
is
my
long-term
strategy
here?
A
Telling
them
like
a
find
a
company
to
hire
you,
and
you
know,
enable
you
to
work
on
this
forever
is
sort
of
like
this
pipe
dream
or
like
crossing
your
fingers,
it's
like
if
someone
said
they
want
to
get
rich
and
telling
them
like.
You
know,
find
someone
rich
and
marry
them
versus
just
figure
out
how
you
get
rich
yourself.
A
Finally,
will
it
work
for
everyone
I?
Think
it's
still
too
early
to
tell
these
are
all
still
like,
create
the
unproven
theories.
The
real
contribution
models
still
really
new
and
not
every
project
will
fit
into
that
model.
So
if
you're
a
project
with
a
fairly
narrow
scope,
I
can
see
it
like
media
fella
could
kind
of
work
forever,
whereas
if
you're
more
of
an
ecosystem
like
node
or
rust,
then
it's
a
little
bit
different.
A
A
Yeah
I
think,
like
Russ,
is
actually
probably
example
of
a
project,
those
that
pretty
well
ended.
I,
guess,
like
the
caveats
of
all
these,
is
that
there
is
always
somebody
who
is
sort
of
in
charge.
There
is
some
sort
of
central
team
or
decision-making
body
to
help
facilitate,
like
all
like
anarchy,
mass
chaos
and
so
and
I.
A
Think
in
the
saga,
is
sort
of
conflating
governance
models,
with
contribution
models
and
like
they're,
not
necessarily
the
same
thing
but
I
think
they're,
so
closely
related,
so,
like
governance
being
like
who
has
final
decision-making
authority,
dancing
like
who
is
actually
like
making
sure
all
the
trains
run
on
time,
I
think
we're
all
is
going
to
need
somebody
to
do
that.
But
the
idea
of
like
having
this
longer
tail
of
contributions,
of
like
engaging
people
as
much
as
possible,
I
think
there's
sort
of
like
this
wide
base
that
can
be
created.
A
A
That's
a
good
question:
I
have
no
idea
how
it
affects
closure,
specifically
I,
think
people,
my
guess,
would
be
that
people
are
more
comfortable
with
it
now
than
they
were
then,
but
I
haven't
really
looked
into
like
how
that
affects.
Like
loaded
specifically
I'd,
be
really
curious
to
know.
Actually,
yeah
I
definitely
heard
like
conflict
from
companies
that
like
really
want
to
be
involved
or
make
contributions
to
projects,
and
they
feel
like
they're,
not
able
to
and
I'm
asking
that's
a
lot
harder
in
a
PDF
file
project.
A
Far
as
I
can
tell
Linux
is
still
VDSL,
but
they
also
have
the
Linux
Foundation,
which
basically
helps
support
its
development.
Lots
of
companies
paying
into
the
foundation
and
then-
himself
is
employed
as
a
fellow
of
Linux
Foundation.
He
doesn't
want
to
run
the
foundation,
so
there
are
very
I,
think
I.
Think
of
it
as
like.
Very
much
an
outlier
type
project,
just
the
size
and
scope
of
Life
Foundation
and
the
project
is
pretty
unusual.