►
From YouTube: Development Metrics Working Group 5/15/2019
Description
Development Metrics Working Group Discussion
Agenda: https://docs.google.com/a/gitlab.com/document/d/1Y50uhpRW0zSGWI-TzPxHnwEHyOl7uWiyCzXtpRJd1_E/edit?usp=drive_web
A
So
I
just
started,
recording
so
yeah,
so
just
a
fYI
am
today's
meetings.
Gonna
be
more
abridged,
mainly
because
I
think
that
number
of
folks
are
getting
back
for
contribute
and
also
we
haven't
necessarily
been
working.
This
specific
set
of
topics
last
week
but
I
did
want
to
hit
on
a
few
high
points
just
from
certain
perspectives,
so
so
just
going
through
it
for
the
persistent
links,
I'm
gonna
go
ahead
and
show
you
the
screen
really
quick
and.
A
Let's
see
so
everybody
see
the
development
metrics
screen
right
now.
Okay,
I
got
one
out
of
the
head,
that's
good
to
see
ya.
So,
let's
start
with
here
so
average.
Mr
is
per
engineer.
Actually,
surprisingly
enough
for
the
middle
of
the
month,
we're
only
down
like
1mr
per
engineer,
which
is
a
little
a
little
bit
strong
I,
just
rapper
sugar,
usually
about
the
middle
of
the
month,
were
about
five.
So,
given
that
we
had
the
last
week
of
contribute,
that's
actually
a
pretty
good
result
from
that
perspective,
usually
it's
it's!
A
The
second
half
isn't
necessarily
double
that,
but
it's
it's
usually
kind
of
goes
up.
I
want
to
say
between
13
and
or
yeah
goes
up.
Basically,
if
it's
5,
it
goes
up
by
3
3
to
4
so
60
to
80%,
so
so,
based
on
that
we're
trending
fairly
well
from
that
perspective.
Similarly,
our
mrs
is
at
457
at
the
middle
of
the
month.
If
you
take
and
look
at
July
to
August,
there's
basically
about
a
77
percent
drop.
So
if
we
assume
no
product
to
the
improvements,
what
you
would
expect
to
see
from
6?
A
B
A
A
So
that's
on
that!
That's
it
on
the
persistent
items
and
I'm
gonna
stop
sharing
the
screen
again
and
we'll
go
back
to
my
hypotheses
and
donor
reason
why
I'm
talking
about
hypotheses
is
just
to
remind
ourselves
of
which
ones
we
still
are
working
on
and
kind
of,
not
expecting
any
update.
Here,
though,
if
anybody
has
any
updates
feel
free
to
jump
in,
but
I
know
on
the
first
one,
which
is
Rita,
maintainer
x'
has
dropped
over
time,
which
is
niggling
infected.
A
Haven't
had
a
chance
Alaska
to
basically
address
that
so
and
then
the
next
one
is
developers
with
less
tenure,
get
lab
in
with
different
methodology,
smaller
Mars,
compared
to
those
with
longer
tenure,
the
productive
implementing
you
had
an
action
on
this
and
yeah
I
haven't
done
anything
work,
but
we
cool
just
remind
it.
Just
a
reminder
right
so
I'm
doing
the
same
thing
for
myself
is
just
kind
of
forcing
myself.
Oh,
oh
yeah,
that's
it!
A
A
Look
into
this
one
I
kind
of
understand
understand
so
so
I'll
check
with
her
in
the
next
couple
days
to
see
if,
if
she
can
start
working
on
that
cool
and
then
let's
see
here,
current
meth
met
metrics
kind
of
have
that
information
in
there
around
the
579
versus
1366,
like
I,
said
it
feels
feels
roughly
about
right.
It'll,
be
interesting
to
see
kinda
how
we
kind
of
work
from
there.
B
A
B
With
the
date
I
mean,
the
Remy's
team
has
been
doing
all
this
since
the
beginning
already,
but
as
we
scale
I,
is
this
the
right
group
to
to
discuss
about
this,
and
if
it
is
and
I
mean
Remy
is
gonna
join
or
mark?
Is
gonna
join
this
this
this
group,
so
there's
an
interface
there,
but
this
is
the
set
of
their
current
responsibilities
to
note
that
velocity
and
productivity
here
I'm
talking
about
things
in
the
test,
automation
and
test
infrastructure
of
also
review,
apps
Tbk
all
that
stuff.
B
That's
count,
that's
where
I'm
coming
from,
but
as
we
know,
productivity
can
be
a
lower
term
I'm
using
history,
because
that's
the
industry
standard
for
how
people
do.
How
really
mature
organization
is,
do
quality
in
the
really
advanced
term?
And
that's
this
is
the
facet
of
work
so
that
that's
my
input.
If
there's
anything
else,
that's
other
teams
in
development
of
working
on
in
productivity.
Then
it's
a
share.
Responsibility
I
mean
I.
If
you
can
bring
it
up
a
discuss,
it.
A
So
you're
thinking
in
terms
of
how
do
I
increase
overall
engineering
productivity
as
opposed
to
development
productivity,
is
the
way
interpreted.
That
is,
that
a
fair
characterization
right
right.
It's
not
yeah,
okay,
all
right,
okay,
and
but
if
development
came
back
and
said,
here's
something
that
we
need
from
an
automation
perspective
and
it
could
potentially
be
included
in
this
charter.
Yes,
then,.
B
A
A
Do
a
survey
with
our
development
teams
to
kind
of
figure
that
out
you
know
like?
Is
there
anything
that
you
should
be
going
to
automate?
And
you
know
if
you
don't
have
time
state
that
as
well,
because
that
may
be
one
of
the
cases
where
I
may
come
back?
Those
team
say
you
know:
is
this
this
worthless
this
type
of
initiative,
false.
B
So
I
can't
I
can
provide
some
status
that
at
the
top
of
my
head,
so
there's
two
high
level
initiatives
that
we
working
on
his
the
the
flaky
specs
that
came
up
in
the
retrospective.
We
actually
paralyzing
it
in
a
more
deterministic
fashion.
Now
so,
hopefully,
that's
gonna
come
in
by
the
time
we
have
the
next
retro.
It
should
get
better,
so
the
flaky
red
master
and
then
relapse
and
making
it
a
gate.
So
that's
that
should
roll
out
sometime
next
week.
B
B
B
Tension
is
working
on
it
and
that's
also
ongoing.
We
are
cleaning
up
the
labels
right
now,
I
think,
if
you
infer
like
a
type
or,
for
example,
server
lists
or
a
code
or
IDE,
we
will
automatically
assign
a
team
and
stage
now
and
that
we
already
got
feedback
from
all
the
product
managers
and
that
change
is
also
going
soon
as
well.
So
it
is,
it
is
being
actively
work
on
a
small
chunks
in
all
areas,
and
now
we
have
two
more
headcount
and
I'm
looking
forward
to
have
it.
A
I'll
talk
to
Holly
about
that
because,
like
that's
one
case
of
you
know,
we
think
the
issues
in
merge
requests
we've
seen
them
go
up,
so
the
question
would
be
as
if,
like
the
triage
would
be
actually
a
help.
With
that
perspective,
we
don't
have
necessarily
have
to
work
causal
to
say
that
you
know
having
a
triaging
capability
will
speed
us
up,
but
it
feels
like
those
are
potential
areas
where
we
should
be
thinking
about
it
just
to
kind
of
help.
A
Our
overall
effectiveness
from
that
perspective
got
it
okay,
it
might
also
help
with
our
SSL
that
we've
supposedly
you
know
we
come
size
with
engineers,
which
is
you
know
two
code
reviews.
First
though,
and
you
know
that's
really
difficult,
if
we
don't
know
the
load
properly
associated
with
that
when
we
make
that
statement,
so,
okay,
okay,
cool
and
then
the
last
thing,
I
had
well
two
more
things
one,
and
this
is
really
brief.
So
a
better
story
next
week,
I'll
say
here.
Basically
it's
around
right.
A
B
B
That's
why
we
just
double
down
on
the
Collier
dashboard
and
that's
why
it
is
where
it
is
now,
but
from
your
point
of
view,
because
if
we're
like,
how
do
you
want
to
I
assume
this
is
to
solve
the
KPI
oakley
ours,
correct,
cuz?
We
need
to
bump
up
the
maturity.
Are
we
seeing
like
half
and
half
like
if
we
have
it
in
the
quality
dashboard?
Just
add
it
there
quick
and
dirty
and
get
it
done,
or
is
it
like
a
Oh
18
in
periscope
and
whatever
we
can
do
with
periscope?
A
My
take
is
is
actually
what
I
want
to
get
to.
Is
this
like
right
now,
if
I
would
have
said
it,
I
would
say
it's
probably
8020
80s
80s
and
in
the
quality,
dashboard
and
20%
is
in
the
is
actually
in
the
product
right,
because
you
you
have
delivered
on
some
product
feature
set
associated
with
that
right,
where
I
think
we
need
to
be,
as
is,
we
need
to
have
I'll,
say
60%
of
the
product
and
40%
periscope.
B
A
And
that
may
be
something
that
I
if
it
turns
out
that
getting
the
ETL
and
to
get
em
ours
over
the
periscope
which
gets
us
off
of
and
then
building
up
the
dashboards
gets
us
off,
that
that
might
be
the
quick
and
dirty
it
gets
us
off
of
yeah
off
right.
So
then
you
don't
have
to
maintain
that
infrastructure
and
it's
more
uniform
for
that
perspective.
At
least
that
would
be
my
argument
currently
for
that,
because
I
think
I
think
we're
know.
B
Can
definitely
see
the
challenge
of
them.
Data
that
needs
to
API
calls
to
get
everything
in
one
unit
like
the
first
activity.
Your
first
time
you
add
a
milestone
like
those
are
n,
plus
one
events-
cuz
Evie,
it's
not
there.
You
need
to
make
another
call,
so
you
having
an
ETL
just
pump
it
in,
but
it
might
might
be
faster
yeah
and
is
a
who
is
the
coolest
who's
investigating
the
ETL
on
your
side.
So.
A
Nobody's
nobody
is
investigated
to
get
I
have
a
book
for
it,
though
it
already
exists,
and
it's
pushes
data
in
just
what
I
remember
when
I
went
and
looked
at
it.
This
is
Christopher
looking
at
it.
It
didn't
include
my
Mars.
So
in
theory
already
all
most
of
our
product
information
is
getting
pushed
in.
We
just
need
to
figure
out
how
to
use
it.
A
Tim
Zalman
is
working
on
pushing
our
code
coverage
into
into
and
that
he's
kind
of
my
guinea
pig.
As
far
as
you
know,
like
figure
out
how
hard
or
easy
it
is
to
get
data
pushed
into
her
scope.
We're
on
that
and
I.
Just
you
know,
with
with
everything
else.
Last
week,
I
haven't
had
a
chance
to
catch
up
with
him
to
see.
If
he's
been
able
to
invest
any
time
in
that,
because
basically,
what
he's
gonna
do
is
is
right.
Now
we
have
a
job
that
runs
and
it
does
post
data.
A
A
There's
kind
of
two
different
things
go
another
one
is
how
hard
is
it
great
things
to
push
in?
That
part
is
how
do
we
update
an
existing
tool
package
to
represent
mrs
with
it,
which
is
you
know,
that's
the
part
that
when
we
figure
out
how
easy
it
is
that
we
can
do
the
other
part
of
it
from
objective
cool
last
thing
I
on
my
list
was
just
exit
criterion.
Writing
everybody
of
the
exit
criteria
in
this
case
I
think
the
first
two
of
the
exit
criteria
we've
got,
though
I
haven't
defined.
A
The
official
KPI
well
I've
defined
to
keep
you
guys
by
haven't
gotten
to
level
four
maturity,
because
I
don't
have
thresholds
associated
with
them.
Next,
one
on
the
list
is
training
around.
Those
KPIs
will
have
to
basically
develop
that
associated
with
that
and
then
I
think
the
new
ones
read
by
you,
MEC
yep,
I,
don't
know
if
you
have
any
update
there
or
if
you
just
want
to
say
classic,
haven't
gotten
to
him.
Yeah
yeah.
B
We
haven't
agreed
on
this,
but
essentially
we
do
want
to
lower
the
time
to
resolve
as
whenevs
tus
and
that's
a
metric
that
I
think
sitting
area
could
be
looking
at
charging
other
backlog
of
customer
facing
bugs.
This
may
be
a
one-time
event
where
we
just
find
out
to
everybody
to
tell
charge
and
get
just
get
it
in
right.
Even
have
a
severity
and
priority
you
don't
have
any
visibility
and
support
team
is
game
on
this.
B
I
think
that
the
next
one
is
probably
gonna
come
first,
because
if
you
I'm
probably
gonna
be
order
this,
because
we
need
the
the
data
first
and
we
can
ensure
that
the
number
is
going
down
and
then
the
second
to
last
is
already
in
the
works
now
away
for
month.
Let
you
to
update,
and
after
that
is
I,
think
a
mini
training
program.
We
probably
end
up
with
some
youtube
training
videos
and
make
sure
that
all
engineering
I
just
know
how
to
use
the
levers
effectively.
A
Yeah
around
that
that
feels
like
we're
getting
to
the
point
where
we
got
a
Star
Trek
and
the
stuff
and
some
of
the
stuff
in
the
shoes.
So
I
was
I,
had
an
action
from
last
week
to
or
two
weeks
ago
to
create
a
board
haven't
done
that
yet,
but
I'll
I'll
start
work
on
that.
So
then
we
can
put
issues
around
this
and
then
kind
of
Trek
them
in
there.