►
From YouTube: RETMKT Builders - Textile Miner Index
Description
Andrew updates us on the textile miner index at the Retrieval Market Builders Mini-Summit in April 2021.
A
Hi
cool
so
yeah.
I
was
asked
to
talk
about
the
our
work
on
the
minor
index.
Today,
though,
I'm
I'm
feeling
like.
I
really
want
to
talk
about
our
new
project,
because
there's
so
many
cool,
conver
sort
of
side,
conversations
about
pieces
that
we're
interested
in
tackling,
and
so
I'm
looking
forward
to
doing
this
again.
A
Hopefully,
if
we,
if
one
of
the
followings
on
this
is
another
meetup
but
yeah
so
the
minor
index,
let
me
so
the
motivation
of
the
minor
index,
if
you're
not
familiar,
we
we
build
powergate,
and
so
powergate
is
an
ex
is
sort
of
a
library
that
runs
on
top
of
lotus.
Allows
you
to
has
some
opinions
about
how
you
can
manage
data
moving
back
and
forth
between
file
coin
and
ipfs.
A
It
also
has
some
helpers
to
help
you
make
deals,
and
one
of
those
helpers
is.
You
can
create
arrays
of
of
trusted
miners
and
you
can
set
parameters
on
creating
deals
with
a
certain
amount
of
replication
and
then
powergate
will
go
to
work
to
store
that
number
of
replicas
across
all
your
trusted
miners,
and
so,
if
any
of
the
miners
turn
down
your
deal,
it
will
move
on
to
the
next
miners.
A
Those
sorts
of
simple
tools
are
built
into
it,
but
we
did
a
lot
of
work
on
actually,
oh,
so
then
we,
what
happens
is
we
have
a
hosted
hosted
system
called
the
textile
hub
and
on
the
textile
hub.
A
We
have
users
that
are
creating
folders
directories
of
data
and
we
wanted
to
help
them
move
those
data
that
data
to
file
coin
and
that
to
do
that,
it
moves
through
a
power
gate
that
runs
on
on
the
hub,
and
so
when
we
were
exposing
all
those
systems,
we
realized
that
we
really
wanted
to
improve
the
success
rate
of
the
deals
going
to
filecoin.
A
That
are
likely
going
to
accept
your
deals,
and
so
the
first
stop
in
that
journey
was
actually
to
do
a
lot
of
sort
of
testing
with
people
on
the
file,
point
team
and
and
kind
of
try
to
figure
out
what
our
workflow
should
look
like
and
then
the
next
stop
was:
how
do
we
collect
all
the
information
about
miners
on
the
network
and
put
that
together
in
a
way
that
our
users
could
tap
into
easily
and
other
people
could
tap
into
as
well?
A
And
that's
where
the
minor
index
came
from,
and
so
the
minor
index
is
actually
in
its
courts
built
into
powergate.
So
anybody
that
runs
powergate
can
be
collecting
a
lot
of
information
about
the
miners
on
the
network,
and
so
it
will
use
both
on-chain
data
and
attach
a
lot
of
additional
metadata
on
top
of
every
miner
from
the
view
of
the
powergate,
and
then
we
also
run
this
on
the
textile
hub
and
expose
it
on
as
a
public
api
and
so
telemetry.
Well,
we'll
get
into
all
of
that.
So
so
right.
A
So
we
exposed
this
as
a
as
an
api.
You
can.
I
can
share
the
url
over
the
zoom
call
in
a
minute
and
ignacio
is,
I
think,
on
the
call,
still,
although
it's
kind
of
late
for
him,
but
he's
the
mastermind
behind
a
lot
of
these
pieces
and
so
he's
here
to
answer
questions
on
chat
as
well.
I
hope,
but
right
so
the
the
minor
index.
A
Here's
open
api,
specs
swagger
specs,
so
you
can
go
to
it
and
you'll
find
that
it
has
basically
every
minor
on
the
network
and
you
can
look
at
the
their
sort
of
history
of
creating
deals.
But
this
is
really
from
the
textile
perspective.
There's
some
on-chain
information
and
it
will
collect
information
about
their
their
storage
prices
and
all
those
and
you
can
filter
through
that.
But
then
it
also
has
the
the
perspective
of
the
textile
power
gates
history
with
these
miners.
A
So
you
can
use
it
through
the
api,
this
open
api
or
you
can
actually
download
a
cli
and
use
it
in
on
your
com.
You
know
on
your
desktop
or
in
your
any
any
tooling.
You
can
actually
script
that
and
build
into
other
things,
and
so
you
can
see
here.
Here's
a
nice
sample
of
the
data
that
comes
back
when
I
get
the
information
out
of
the
index
for
a
single
miner,
and
you
can
see
that
it
has.
A
You
know
their
advertised
costs,
but
then
down
below
you
can
actually
see
this
a
sort
of
hint
of
the
deal-based
metadata
that
will
come
back,
and
so
the
original
intention
here
was
we
knew
there
was
a
constraint
around
our
perspective
of
the
network
versus
what
another
client's
perspective
of
the
network,
which
is
a
challenge
I'll
bring
up
later.
But
you
can
see
in
the
index
it's
actually
exposing
region-based
perspectives
and
our
thinking
was.
We
would
always
run
multiple
of
these
nodes
that
could
collect
the
telemetry
about
miners
from
multiple
different
regions.
A
So
you
could
find
the
region
closest
to
you
and
get
telemetry
information
that
was
most
relevant
for
for
selecting
your
miners
and
you
you
can
query
it
and
you
can.
You
can
sort
by
a
lot
of
different
information,
and
so
actually,
if
I
just
jump
over
in
some
collaborations
with
jimmy.
Actually
I
put
together
this
little
observable
that
just
shows
how
to
use
the
api.
Quite
quite
simple,
I'll
share
this
link
as
well,
but
pretty
nice,
like
obviously
open
api.
A
So
then
I
just
do
a
bunch
of
json
filtering.
I
find
just
the
miners
that
are
advertising
free
storage
for
verified
clients.
I
find
miners
that
are
accepting
less
than
a
certain
threshold
of
data
size
and
then
I
I
list
out
the
top
eight,
and
so
that
way
you
know
over
here
I
have
a
function
just
to
get
the
top
one
at
any
moment
in
time.
The
default
sorting
in
the
minor
index
is
actually
based
on
our
last
successful
deals
with
the
miner,
and
so
that's
helped.
A
That's
meant
to
help
end
users
find
currently
active
miners
and
and
really
kind
of
relevant
latest
index
data
right.
So
there's
that
and
yeah
and
so
filtering
all
that
sort
of
good
good
stuff.
A
So
this
is
an
old
screenshot,
but
you
can
see
you
can
see
again
some
different
samples-
and
this
is
actually
this
is
still
a
subset
of
the
data,
but
just
kind
of
fit
nicely
in
the
in
the
ui
here
also
interesting
in
the
index
you
could
get
per
minor
stats
and
so
in
the
tool,
for
example,
I
can
get
the
all
the
latest
stats
for
a
single
minor
here,
and
that
will
give
me
all
the
telemetry,
but
also
the
all
the
deal
history
that
we've
observed
with
that
minor,
and
so
the
telemetry
will
give
you
things
like
the
transfer
speeds
that
we
saw,
which
which
are
pretty
helpful
for
other.
A
You
know
other
users
of
our
system
to
find
good,
good
miners
for
their
deals
and
then
another
thing
in
there
that
we
thought
was
very
useful
that
we
had
intended
to
build
into
some
nice
user
interfaces,
as
well
as
just
simple
deal
calculators,
but
specifically
deal
calculators.
That
would
facilitate
deal
creation
so
right.
A
So
I
can
put
in
the
parameters
of
a
deal
that
I
want
to
make
with,
and
I
can
list
a
a
set
of
minors
that
I
found
on
the
index
that
that
I
want
to
make
deals
with,
and
then
I
can
get
all
the
calculated
costs,
but
also
I
can
get
these
outputs
for
some
of
the
tools
for
how
I
can
quickly
turn
that
calculator
output
into
a
real
request
for
storing
the
data
with
that
miner
right.
So
then,
finally,
I
can
actually
just
copy
paste
it.
A
In
this
example-
and
here
you
can
see,
I'm
using
the
the
the
tooling
on
the
textile
hub
to
move
directories
to
filecoin
and-
and
here
I'm
just
doing
an
on
the
fly
update
of
my
storage
parameters,
to
set
it
to
these
three.
These
three
miners
that
I
found
with
fast
fast
retrieval,
flagged
yeah.
So
here's
just
kind
of
a
zoom
in
on
on
the
way
that
works,
but
we
don't
actually
have
to
get
into
that.
A
This
is
out
there
and
and
very
ready
for
people
to
use
and
jimmy
showed
one
example
of
some
of
the
things
that
he's
done
exploring
this,
and
he
has
a
couple
of
good
ones,
but
I
thought
maybe
it
would
be
really
useful
for
this
group
to
actually
talk
about
some
of
the
challenges
that
we
saw
with
this
going
forward,
and
so
the
biggest
the
biggest
and
most
obvious
one
is
that
like
collecting
data,
doesn't
equal
reputation
and-
and
we
can
say
that-
and
miners
will
definitely
say
that,
but
one
thing
that
we
found,
for
example,
is
miners
filter
for
clients.
A
This
is
a
great.
This
is
a
great
case
where
this
is
true,
just
because
a
minor
makes
a
deal
with
us
and
we
measure
them
as
being
high
quality
or
high
speed
doesn't
mean
that
the
another
miner
or
another
another
client
is
even
necessarily
necessarily
going
to
get
a
deal
with
them
at
all.
And
then
the
telemetry
is
is
very
affected
by
temporal
load
on
clients,
and
this
is
something
our
latest
work
is
actually
trying
to
improve.
Quite
a
lot.
A
So
there's
a
problem
right
is
if,
if
a
lot
of
people
start
using
our
index
and
all
continuously
have
good
experiences
with
the
same
miner,
this
can
actually
be
bad
for
the
miner,
because
they're
going
to
not
be
able
to
have
much
control
over
their
load,
so
they're
gonna
have
to
either
turn
down
deals
or
fail
to
to
get
deals
on
chin,
quick
enough,
and
so
there's
a
there's
a
funny.
There's
a
funny
mismatch
here
between
motivations,
yeah
and
then
things
like
property
properties.
Like
the
minor
location.
A
We
actually
use
the
max
mine
database
to
do
ip
matching
of
all
the
miners
on
the
network
and
create
locations,
but
this
is
a
pretty.
This
is
a
pretty
brittle
field
that
seems
probably
gameable,
and
so
those
those
kind
of
things
would
be
very
nice
to
figure
out
better
ways
of
collecting
them
and
yeah
we
haven't
got.
We
haven't
really
focused
on
retrievals
and
getting
that
data
in
here.
So
that
would
be
a
next
step
if
we're
gonna
take
this.
A
But
I
think
a
lot
of
this
is
gonna,
be
leapfrogged
by
the
the
work
that
mike
is
gonna,
talk
about
and
work
that
I
think
digital
mob
is
working
on
now,
but
hopefully
a
lot
of
this
can
can
inform
that
and
then
catch
up
to
where
they
get
to
but
yeah
like
miners
miners
feedback
was
a
little
bit
scared
at
first
when
we
started
pushing
this,
and
some
of
it
was
valid
like
back
to
the
point
of.
A
If
we
increase
load
on
a
single
minor,
it
actually
can
just
cause
factors
that
are
hard
for
them
to
deal
with,
or
external
factors
can
lead
to
them
being
showing
up
as
bad
in
a
reputation
index.
The
way
that
online
deals
work
that
can
be
really
tricky
and
so
taking
that
into
consideration.
A
The
second
set
of
challenges
are
really
around
the
disconnect
between
reputation.
What
you
really
want
to
get
out
of
an
index
like
this
and
the
economics
of
filecoin,
that
there's
kind
of
a
mismatch
here.
That
is,
that
is
only
loosely
connected.
So
just
some
examples.
Details
within
there,
like
the
prices
that
are
advertised
by
miners,
that
they're
accepting
deals
that
aren't
necessarily
going
to
reflect
reflect
what
you're
going
to
get
what
you're
going
to
get
on
your
final
deal.
A
They
can
have
a
lot
of
reasons,
but
filecoin
itself
is
a
very
volatile
market,
and
so
that
can
be
really
tricky
and
miners
are
going
to
be
a
lot
slower
at
adjusting
what
their
costs
are
to
the
market
rates
of
file
coin,
and
so
you
have
to
account
for
that.
When
you're,
looking
at
an
index
like
this,
that's
somehow
supposed
to
be
a
source
of
truth.
A
So
that
can
be
pretty
tricky
so
like
knowing
the
true
and
then
also
that
leads
into
how
you
convert
this
back
to
your
understanding
of
like
us
dollars.
How
much
am
I
really
paying
for
the
price
and
there's
a
lot
of
other,
like
kind
of
more
meta
issues
there
to
solve
that?
But
but
that's
there
and
then
yeah,
and
so
this
actually
gets
back
to
the
point
of
if
this
index
we're
being
used
by
a
lot
of
clients
and
putting
a
lot
of
pressure
on
the
best
miners.
A
A
a
small
miner
might
show
up
on
the
index
and
actually
be
their
quality,
might
be
reduced
by
that,
and
so
there's
something
there's
something
tricky
there
needs
to
be
figured
out
because
of
that
mismatch
between
the
incentives
and
the
and
the
data
that's
shown
here,
yeah
that
leads
into
this
problem,
which
is
being
something
like
a
king
maker,
which
is
the
exact
same
thing
I
keep
saying
which
is
like
if
the
minor
is
used
and
successful
at
highlighting
a
single,
really
good
minor
that
people
keep
finding
in
the
index,
it's
just
going
to
create
them,
make
them
better
and
better
and
better
in
the
index.
A
Potentially.
Is
that
a
bad
thing
for
the
filecoin
network?
I'm
not,
I'm
not
sure.
I
guess
there
could
be
arguments
either
way,
but
something
to
think
about
with
these
kind
of
consolidated
views.
The
third
challenge
is
again
around
the
economic
discontinuity,
but
the
reputations
aren't
driven
by
the
economic
incentives.
Yet
so
like
there's
like
no,
no,
no
miner
really
mind
so
much
about
this.
A
They
really
are
focused
on
how
how
they're
performing
on
their
the
things
they
have
to
do
on
chain,
and
so
that's
something
to
consider
is
how
we
can
we
can
bridge
reputation
to
to
the
on-chain
activities
more
closely
and
then,
like.
I,
keep
wondering
this
about
many
different
competing
reputations
is.
How
do
we
rank
the
reputation
makers?
There's
no
reason.
Somebody
has
to
trust
textile
as
being
honest
here
and
that's
a
really
tricky
thing
to
figure
out,
and
then
we
didn't
put
a
reputation
on
on
our
system.
A
It's
really
an
index,
because
the
fair,
like
scoring
system
doesn't
seem
obvious
to
me,
and
at
least
yet-
and
I
I
think
that
there's
there's
obviously
some
good
proposals
out
there,
but
making
minors
feel
like
they're,
being
measured
in
a
fair
way
is,
is
pretty
important,
I
think
they're
all.
I
think
many
of
them
are
working
really
hard
to
make
the
network
better.
A
So
how
do
we
actually
like
treat
them
in
a
fair
way,
so
that
they're
they're
show
they're
showing
these
index
in
a
good
way
and
then
obviously
not
to
add
tricks
to
the
index?
That's
that's
a
challenge
like
we.
A
We
could
layer
in
tricks
to
try
to
make
the
reputation
the
reputation
system
here,
but
those
tricks
are
going
to
often
be
matched
by
the
tricks
of
the
miners
themselves,
because
if
they,
if
they
end
up,
then
having
the
incentive
to
be
showing
up
what
as
high
quality
in
the
index,
that
just
creates
a
game.
So
those
were
kind
of
the
sample
of
challenges
that
we
kind
of
debrief
on
this
project
and
and
started
thinking.
How
do
we
get
to
the
next
level
here
and
think
through?
A
How
do
we
really
organize
high
quality
minors
to
accept
our
deals
in
a
like
high
success
rate
sort
of
way?
And
that's
what
I
want
to
talk
about
next
time
so
who
will
get
there?
And
then?
The
last
point
is
just
that
this
that
there's
this
additional
problem
that
our
our
index,
for
example,
only
knows
the
miners.
A
You
know
that
it's
made
deals
with,
and
it
knows
all
the
miners
that
are
on
chain,
but
things
like
telemetry
telemetry
are
really
tricky
from
our
point
of
view,
because
we
have
to
then
go,
spend
filecoin
and
spend
a
lot
of
time
testing
every
miner
that
we
find-
and
we
didn't
set
it
up
to
do
that
on
this
first
version,
it's
something
that
could
be
done
but
again.
A
I
think
that
some
of
these
other
projects
are
going
to
leapfrog
where
we
got
to
so,
hopefully
we'll
all
learn
something
from
them,
but
yeah
that
was
our
minor
index
project.
It's
still
running.
Definitely
we're
still
collecting
data
lots
of
good
stuff
there
until
so
until
people
leapfrog
us.
I
think
that
there's
a
lot
of
cool
stuff
to
build
on
here,
and
ideally
this
will
migrate
into
the
next
group
of
standards
wherever
we
go
next,
so
hopefully
that's.