►
From YouTube: CHAOSS Metrics Models Working Group April 11 2023
Description
Minutes from this meeting can be found here: https://docs.google.com/document/d/1qE2duE73NbEOHzRRAbqgPSUSAfdH-AmvLruOu7uCIYg/edit
A
Hi
well
Welcome
to
our
metrics
model,
meeting
April
11th.
So
it's
great
to
have
you
here.
I
have
a
few
things
on
the
agenda
for
today
feel
free
to
include
anything
that
you'd
like
we'll
go
ahead
and
get
started.
A
So
the
first
thing
today
is
I
did
Issue
a
pull
request,
which
is
a
pretty
straightforward
one
for
the
metric.
A
C
E
A
The
next
thing
that
I
wanted
to
talk
about
was
the
influence
metrics
model,
so
this
had
come
up.
Actually
it
could
kind
of
be
down
here
with
the
ospo
meeting.
So
one
of
the
in
the
osbo
meeting,
one
of
the
things
that
I
did
was
I
kind
of
went
back
to
our
initial
sets
of
goals
that
we
had
established.
A
So
if
you
look
at
those
minutes
and
tried
to
identify
an
agenda
around
those
goals
like
what
are
things
that
we
can
do
to
try
to
address
those
goals
and
one
of
the
goals
in
the
aspo
meeting,
amongst
other
things,
is
to
develop
metrics
models.
So
I,
that's
what
this
goal
is
here.
A
So
this
is
straight
from
the
osbo
working
group
all
right,
so
we
had
talked
about
a
couple
different
metrics
models,
and
this
is
to
kind
of
address
the
you
know,
demonstrating
value
of
an
ospo
within
an
organization
that
that
conversation
continued
on
in
the
last
ospo
meeting
and
one
of
the
ones
that
had
come
up
was
or
that
I
had
brought
up
was
the
influence
metric
model
and
I
should
probably
put
it
on
here,
but
the
influence
metrics
model
is
one
that
was
brought
forward
by
shoya
and
Frank.
D
A
D
D
A
So
I
had
kind
of
kind
of
tightened.
This
metric
model
up
just
a
little
bit
so.
E
C
A
If
you'd
like
to
take
a
look
just
straight
up,
so
this
is
really
about
showing
an
organization
or
an
organization's
ability
to
have
an
influence
on
a
particular
project.
Sophia
had
brought
it
up
kind
of
as
a
as
a
counter
as
well
at
Google.
Sometimes
an
organization
wants
to
let
go
of
influence
if
they're
open
sourcing,
a
project
that
they
want
to
demonstrate
that
they're
actually
turning
it
into.
C
A
You're
interested
in
what
Don
has
to
say,
but
my
reflection
is,
is
that
these
are
the
kinds
of
metrics
that
you
do
indicate.
F
That
organization
has
influence
in
the
project
is,
if
you're
active
and
getting
your
change
requests
accepted.
Especially
I
think
that
that
says
that
there
is
a
signal
influence.
D
F
Is
the
influence
of
like
the
influence
that
a
project
has
relative
to
other
projects?
Is
that
this.
A
Would
be
so
like
in
this
case,
if
I
come
down,
I
think
to
this
one,
it's
identifying
the
people
who
work
within
your
organization.
E
A
A
E
A
Kind
of
like
through
your
employees,
you
know
affiliation
with
these
types
of
things.
Like
being
you
know,
participants
and
change,
requests
and
change,
request,
reviews
and
those
accepted
as
demonstrations
of
influence.
A
F
C
F
Is
it's
measuring
two
completely
different
things:
the
influence
of
projects
and
the
influence
of
people
and
I
think
that
that
carries
down
to
looking
at
the
metrics
and
the
metrics
model,
because,
like
Upstream
code
dependencies,
that's
a
repository
metric
and.
F
A
F
F
Different
things,
both
of
which
are
incredibly
important
but
I,
think
yeah,
so
I
don't
think
it's
possible
to
have
an
influence
metric
model,
because
because
the
influence
of
projects
is
just
completely
a
completely
different
thing
than
the
influence
of
people,
I
think
they're,
both
important
I
would
I
would
make
them
too
too
separate.
Okay,.
F
D
B
Well,
if
I
understand
correctly
the
influences
general
purpose
algorithm,
which
can
calculate
the
the
influence
of
a
node
in
a
graph.
Well,
the
graph
could
be
a
graph
of
contributors
or
a
graph
of
repositories,
so
I.
B
A
I
will
say:
I
mean
from
a
metrics
perspective.
We
do
have
metrics
that
are
oftentimes
very
similar
to
each
other,
like
I,
think
in
in
the
Dei
working
group.
We
have
like
an
event
correct
me
if
I'm
wrong
Elizabeth,
but
like
a
couple
metrics
that
are
really
close
to
one
another,
but
we've
taken
the
time
to
separate
them
just
from
a
Clarity
perspective.
E
Just
for
example,
documentation
we
have
like
documentation
discoverability,
we
have
documentation
accessibility,
so
it's
you
know
basically
all
around
documentation,
but
it's
yeah
very
specific
things
about
that.
A
And
that
yet
specific
to
that
conversation,
I
think
we
had
originally
one
metric
called,
like
all
things,
documentation
and
yeah
a
little
like
this,
like
all
things
influence
and
it
over
time,
we're
like
we
need
to
separate
this
out.
We
are
just
like,
as
we
walk
through
the
narrative
and
it's
confusing.
F
And
I
think
I
think
the
other
thing
that
I
particular
model
is,
if
you
look
at
the
you
know,
there's
a
whole
list
of
metrics
in
the
metrics
model
and
it's
not
clear
how
any
of
those
feed
into
that
influence
algorithm.
That
Liang
was
just
talking
about
so
I
think
the
influence
algorithm
is
absolutely
fascinating
and
I
love
the
you
know,
like
the
network.
Look
at
it,
but
I
don't
understand
how
how
we
got
from
this
list
of
metrics
to
that
particular.
C
F
Matt,
my
other
question
is
where,
where
is
this
in
the
process?
Has
it
already
been
pr'd
into
the
repository
okay,
yeah
I
I'm,
not
sure
what
to
do
with
this
I
find
I
find
this
metric
model.
F
Deeply
fascinating
and
insightful
and
I
also
find
it
incredibly
problematic
in
the
way
that
it's
written
and
so
I'm
not
quite
sure
what
to
do
with
it.
I
both
love
it
and
hate
it
at
the
same
time
is.
A
F
I
think
I,
so
I
think,
there's
I,
think
there's
two
things
like
when
I
look
at
the
list
of
metrics
and
that
we
had
at
the
at
the
top.
Some
of
those
were
people
metrics.
F
But
it's
not
clear
how
any
of
our
metrics
actually
fed
into
that
influence
model
because,
as
Liang
said,
it's
like
it's
like
a
network
craft,
that's
the
data
that
that
was
used
to
create
it,
which
is
different
than
than
what
we've
specified
so
so,
on
the
one
hand,
I,
on
the
one
hand,
I
wonder
if
this
is
even
I
wonder
if
this
is
really
a
metrics
model.
I
guess
I'm,
just
not
even
sure
what
to
do
with
this,
because
it
feels
like
too
many
different
things.
So.
D
Then
I
mean
I
think
so.
I've
I've
done
a
lot
of
social
network
analysis
work
with
this
kind
of
data.
In
my
day,
I
I
think
the
the
graphs
are
intended
to
represent
influence
from
a
sociogram
visualization
perspective
in
each
case
that
that
these
of
these
sociograms,
which
is
what
these
things
are
called,
are
probably
weighted
based
on
the
values
returned
by
each
of
the
individual
metrics
that
affect
those.
B
C
After
they
collect
this
video
from
people
to
people
they
will
they
will,
they
will
be
able
to
find
out.
Who
is
the
3D
Mentor
or
who
is
the
key
people?
Who
is
the?
C
Who
is
the
most
important
people
in
this
project
like
this
and
and
for
project
for
repository?
They
will
find
so
which
repository
is
the
most
familiar,
which
problem
maybe
and
will
be
more
popular.
D
C
Yeah
I
think
there
and
page
right
they
use
page
rank
algorithm,
so
I
think
maybe
we
can
separate
the
algorithm
in
some
place
and
we
can
use
it
inform
many
form.
Many
metric
model.
D
C
D
Like
maybe
we
should
so
I
think,
can
we
decide
about
whether,
because
I
think
the
implementation
could
occur
to
provide
both
of
these
perspectives
on
influence?
If
there
were
two
different
metric
models
like
I,
don't
think
that
affects
an
implementation,
I
wonder
if
we
can
decide
to
try
to
separate
the
implementations
or
the
metric
model
definitions
but
I
feel
like.
Maybe
we
need
to
wait
for
shoya
or
or
the
other
designer
of
this
model,
to
figure
Frank
to
figure
it
out
to
know
what
some
of
this
actually
even
means.
F
Yeah
and
I
I
think
I
think
it
would
be
better
just
so,
rather
than
talk
about
breaking
this
out
into
two
separate
metrics
models
right
now,
I
would
maybe
focus
it
on
on
the
people
side,
because
that
looks
like
what
they've
done.
You
know.
If
you
look
at
the
graphs,
you
know
people
people
have
influence
in
projects
right.
So
so
you
need
the
project
element
because
that's
where
the
people
have
the
influence.
F
So
so
that
bits
that's
important,
but
some
of
the
the
other
stuff
that
looked
like
it
was
added
by
the
the
osbo
group.
The
was
it
Upstream
code
dependencies
that
that's
more
of
a
project
against
project
influence,
which
feels
very
different
I
would
just
I
would
I
would
simplify
this
and
make
sure
that
what
we're
talking
about
is
the
influenza,
ultimately
the
influence
of
people
and
then
figure
out
whether
we
need
a
separate
metrics
model.
That's
the
influence
of
projects
against
each
other,
because
that
that
feels
like
it
would
be
completely
different.
E
Versus
like
internal
influence,
also
because
I
know,
we
have
a
metric,
that's
in
progress
called
contributor
development
influence,
which
is
like
how
influential
the
person
is
externally
like
how
many
followers
do
they
have
and
how?
How
much
does
their
weight
count?
You
know
in
the
in
the
project
ecosystem
in
general,
but
it's
more
externally
focused
so
I
wonder
if
there's
also
a
component
of
that
where
it's
like
the
project
is
externally
influential
because
of
all
these
other
projects
depend
on
that,
and
so
therefore
it's
an
important
project
and
it
needs
to
be
maintained.
F
A
Well,
let
me
I'd
be
happy
to
take
this
on
kind
of
as
an
action
item,
to
also
kind
of
go
through
the
text
where
necessary
to
continue
that
focus
on
people.
So
it's
not
confusing
reading
through.
C
C
A
So
would
you
I
did
also
in
compass,
I'll
put
a
hold
on
this.
This
was
just
to
validate.
You
know
what
I
mean
so
I'll.
Let
me
I'm
gonna,
just
close
though
I'll
just
say
should
I
close
this,
because
this
is
this
is
actually
kind
of
addresses.
One
of
the
questions
that
you
have
done,
which
is
how
how
do
you
implement
this
thing?
How.
C
D
D
D
Diverge
and
maybe
and
I
guess
really.
The
question
then,
is
listen
to
what
was
just
said.
Do
you
want
this
to
include
both
the
individual
and
the
organizational
or
the
project
influence
or
or
is
the
intention
of
the
original
metric
model?
Simply
the
individual,
which
sounds
like
is
actually
the
case.
F
But
but
I
think
it's
important
not
to
say
that
we're
removing
the
project
piece
because
right
because
we're
talking
about
people's
influence
in
projects
and
influence
and
repositories,
the
bit
that
I
recommended
for
moving
was
the
bit
about
Upstream
code
dependencies
where
you're
talking
about
projects
influencing
other
projects.
When
you
take
the
people
element
completely
out.
F
The
the
project
project
to
project
influence
is
the
bit
that
feels
out
of
place
to
me,
but
I
do
think.
You
still
need
to
understand
the
how
the
people
influence
various
projects,
because
that's
an
important
part
of
what
I
think
this
network
is
is
capturing.
A
Does
I
got
it
I
think
I?
Have
it
like
like
remove
the
project?
Department
I
think
was
the
point
yeah
okay,
I'll!
Do
that
I'll?
Keep
this
here,
I
will
see
what
they
say,
because
it
might
be
interesting
to
see
how
they
work
towards
implementing
it.
Just
what
the
initial
thoughts
are.
Okay,
great.
Thank
you
for
that
real
nice
conversation.
C
C
A
Great
another
one
that
came
up
so
we
have
another
one
and
not
quite
sure
what
to
do
about
this.
Was
we
have
project
awareness,
I'm,
sorry
I'm,
not
linking
to
the
actual
metric
on
the
chaos
page,
so
this
one
is
released.
C
A
Project
adoption
is
one
that
is
kind
of
sitting
there,
but
it's
not
been
released
so
as
pointed
out
here,
and
so
these
two
kind
of
were
brought
forward
again
with
respect
to
that
value
conversation
but
gosh.
When
you
start
looking
at
both
of
them
side
by
side
some,
they
feel
like
they,
they
land
on
top
of
each
other,
sometimes
and.
F
Okay,
so
I
didn't
want
to
talk
over
somebody
because
I
know
there's
a
time
delay
in
some
cheers.
I
see
I
see
adoption
as
a
subset
of
overall
awareness.
So
when
you,
when
you
look
at
a
Project's
overall
awareness,
you
you
see
things
like
you
know.
You
know
like
it's.
It's
presence
in
social
media
or
you
know
the
number
of
people
contributing
to
it
or
you
know
lots
of
stuff
around
it
and
then
the
actual
adoption
I
see
is
then.
F
This
is
probably
part
of
my
sansia
bias
is
that
we
tend
to
look
at
adoption
as
as
end
user
adoption.
The
people
who
are
actually
using
this
as
an
end
user
and
so
I
see
I,
see.
Adoption
is
a
piece
of
awareness
but
kind
of
a
cohesive
piece
of
awareness,
gotcha.
A
This
is
currently
released.
This
is
the
released
one.
A
F
F
A
C
F
Because
project
popularity,
so
awareness
can
be
an
indicator
of
adoption
and
also
adoption,
is
part
of
what
makes
a
you
know
provides
awareness
for
a
project,
so
you've
got
this
kind
of
circular
reference
that
yeah
I,
don't
know
what
to
do
with
that.
A
A
F
A
F
Maybe
include
some
language
that
that
talks
about
the
differences
between
between
the
two
and
then
I
would
maybe
just
kind
of
cross-reference
them
that
you
may
consider
adoption
as
as
part
of
looking
at
awareness
or
from
adoption.
You
might
want
to
also
look
at
awareness
as
another
another
indicator
for
for
adoption.
Okay,.
F
B
Yes,
thank
you.
Don
I
have
a
question:
is
it
possible
that
people
are
adopting
software
or
projects
without
knowing
that
he
or
she
is
adopting
it?
For
example,
it's
shot
as
a
part
of
a
software.
Is
that
possible.
D
Yeah,
that's
actually
in
the
risk
working
group,
one
of
the
things
that
we've
defined
are
dependencies
because
that
things
that
are
important
and
do
a
project
that
gets
shipped
as
part
of
it
where
developer
just
includes
a
dependency
that
is
not
on
the
list.
Unless
you
use
the
chaos
metric
and
I
know,
augers
does
three
different
scans
for
dependencies
within
a
delivered
package
and
sure
for
more
lab
and
other
software
does
something
similar.
F
D
Yeah,
like
I,
mean
there's,
there
have
been
ospos
who
have
focused
intensely
on
understanding
the
you
know
of
the
11
000
projects
they
have
which
dependencies
within
them
are
the
most
common
across
that
portfolio,
and
it's
a
important
and
difficult
question
to
answer
right
now.
I
mean
we
answer
it,
but
it's
not.
It's
not
like
auger
gets
distributed
with
Microsoft
Windows.
F
F
F
So
what
you're
really
talking
about
there?
So
so
this
GitHub
makes
the
distinction
between.
D
F
Yeah
dependence,
it's
projects
that
you
import
so.
D
F
Adoption
of
of
something
what
you're
not
you're,
not
looking
at
the
dependencies
that
that
something
has
you're
looking
at
what
is
Downstream,
what
other
projects
depend
on
this
particular
software
yeah.
D
F
And
it's
it's
not
in
the
API,
because
it's
incredibly
expensive
to
calculate
is
what
the
the
problem
is.
I
feel
there
are
ways
to
get
at
this,
so
criticality
score
counts
the
number
of
dependents,
but
I
poked
around
the
the
python
code
that
generates
that
and
they're
actually
using
last
I,
looked
anyways,
maybe
they've
improved
it
since
then,
but
they
were
they're
using
the
GitHub
search
API
to
find
it,
which
is
a
little
bit
problematic,
because
it's
not
particularly
robust.
F
But
that
sorry
that's
a
technical
challenge.
We
don't
need
to
talk
about
that
right
now.
F
F
C
A
A
D
A
D
F
But
that
doesn't
tell
you
how
many
people
use
it
from
that
package
manager.
It
just
says
that
they
put
it
in
a
package
manager
which
I
would
argue
everything
is
in
a
probably
a
soft
package
manager,
whether
it's
a
distro
one
or
or
something
else,
but
that
to
me
doesn't
indicate
adoption
just
that
it's
distributed
that
way.
F
Yeah
and
I
don't
know
the
answer
to
this.
Can
you
get
download
statistics
from
package
managers?
F
Yeah,
so
that's
the
problem,
that's
that's
why
you
can't
measure
adoption
is
because,
because
you
people
distribute
the
software
in
a
million
different
ways
and
you
have
no
way
to
get
download
statistics
from
those
million
different
ways.
That's
that's
the
whole
core
problem
of
measuring
adoption
right
there.
You
summarized
in
like
one
sentence.
A
Right
so
done,
because
you're
on
this
call,
I'll
I'll
work
on
the
influence,
one
and
kind
of
this.
What
we
had
talked
about
with
adoption
and
awareness
I
think
it
would
be
great
to
continue
this
conversation
with
the
hospital
working
group
next
week,
just
to
kind
of
show
how
these
are.
You
know
coming
together
in
ways
that
are
helpful
for
I,
suppose.
A
F
D
F
The
challenge:
the
challenge
is
that
if
we
cancel
it
on
the
20th,
oh
no,
never
mind.
I
was
thinking
that
the
next
one
would
conflict
with
ossna,
but
it
doesn't.
A
A
Great
okay,
Sean
did
you
have
a
comment?
I'm
sorry,
no
I'm,
I'm,
good,
okay,
good!
A
So
then
there
was
one
thing
that
did
come
up:
I
just
want
to
bring
this
up
here
in.
We
have
like
six
minutes
and
that's
these
last
two
things
won't
take
very
long.
The
this
is
again
from
the
hospital
working
group
that
the
metrics
and
like
the
stuff
that
comes
from
the
metrics
models,
the
numbers
aren't
super
helpful.
In
the
conversation
like,
we
need
to
consider
how
to
translate
those
findings
into
a
narrative
form
that
helps
in
like
conversations
within
the
organization.
So
I
think
we
do
that
a
little
bit.
A
F
Yeah,
this
is
Sean
and
I
talked
a
lot
about
this
actually
the
other
day
and
that
I
I
think
that,
in
order
for
the
metrics
models
to
be
really
useful,
we
need
to
figure
out
how
to
get
them
implemented
in
our
software.
So,
like
you
know,
where
do
you
find
this
metric
model
in
auger?
Where
do
you
find
it
in
grimmar
lab
how.
E
F
You
extract
this
data
from
our
from
our
own
tooling,
because
it
feels
like
our
our
tooling
has
gotten
a
bit
disconnected
from
the
other
conversations
that
we
that
we
have
within
within
the
chaos
project
around
metrics
and
metrics
models
and
I
think
it'll
be
an
important
part
of
thinking
about
about
the
next
evolution
of
of
our
software.
So
what
what
areas
do
we
focus
on
and
I
think
I
think
some
of
it
should
be
around?
You
know
new
metrics
and
metrics
models
so
that
people
can
actually
get
get
value
out
of
them.
F
D
And
I
do
think
that
something
simple
like
the
fact
that
we
have
stable
web
links
for
all
of
the
metrics
and
metrics
models
now
is
going
to
help.
It
is
early
on
in
auger,
tried
really
hard
to
create
that
link
and
the
links
can
be
changing
changing.
So
we
gave
up
and
that's
I,
think
both
grew
more
lab,
auger
Compass.
You
know
we
can
use
these
stable
URLs
to
know
that
if
we're
building
the
metric
model,
we
can
also
point
to
the
chaos
definition
within
the
software.
E
It
might
also,
oh
sorry,
Sean
yeah
projects,
though
this
this
idea
of
a.
D
Narrative
description,
so
there
are
so
many
mathematical
values
that
things
can
have
and
I
think
providing
narrative
descriptions
in
a
tool
set
or
data
scientific
analysis.
Where
you
know,
maybe
part
of
what
we
do
is
look
at
what
are
some
actual.
What's
the
actual
range
of
values
we
see
around
different
metrics
and
then
what
are
the
English?
What
are
some
common
English
language
descriptions?
D
F
So
I
love
their
narrative,
I
I,
don't
love
the
idea
about
assigning
specific
values,
because
every
project
is
is
different
and
I
think
that
that
becomes
a
sort
of
a
slippery
slippery
slope.
F
What
what
I
do
like
the
idea
of
doing-
and
it's
something
that
I've
done
with
some
of
the
VMware
ones
is-
is
talk
about
what
the
trends
mean
in
those
numbers,
so
this
number
going
up,
May
mean
this
or
this
depending
on
your
context.
This
number
going
down
might
mean
this
or
this
or
these.
You
know
this
number
going
up.
F
Well,
this
number
going
down
might
indicate
you
know
and
and
I
think
you
can
I
think
you
can
talk
a
lot
about
what
some
of
these
things
indicate
or
might
indicate,
depending
on,
depending
on
the
situation.
F
So
things
like,
like
you,
know,
big
piles
of
neglected,
pull
requests
that
could
indicate
that
you
don't
have
enough
contributors
working
on
a
project
that
could
also
indicate
that
everybody's
working
on
you
temporary
temporary
problem
that
will
resolve
itself
when
people
get
back
to
working
on
the
project
because
they're
not
all
working
on
an
event,
a
vacation.
You
know
other
other
things,
but
but
I
I
love
the
idea
of
putting
some
narratives
behind
behind
what
this
could
mean
that
the
interpretation
I
think
is
really
important.
F
A
Is
great,
though,
I
think
that
at
least
coming
from
a
lot
of
the
ospo
conversations,
it's
always
kind
of
like
there
seems
to
be
this
Gap?
That
needs
to
be
shortened
whether
it's
through
software,
whether
it's
through
narratives,
that
that
people
are
asking
for
help
in
shortening
that
and
the
more
we
can
think
about
ways
to
do
that.
I!
Think
the
better
for
everybody,
yeah.
A
So
we
are
at
time,
but
I
just
wanted
to
add
one
thing:
I
I
did
start
a
blog
post,
I
thought.
Maybe
we
should
do
a
starter,
Health,
metrics,
blog
post,
it's
they're,
very
simple.
Somebody.
A
So
if
you
could
change
the
Emoji
or
if
you
see
text
that
you
would
like
to
add,
that
would
be
great
I.
Just
think
it
would
be
nice
to
to
kind
of
get
something
out
there
to
help
people
just
think
about
exactly
what
this
this
model
is
trying
to
accomplish.