►
From YouTube: Filecoin Station - Miroslav Bajtoš
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
I'd
like
to
start
with
looking
a
little
bit
at
the
Computing
landscape,
where
we
are
now,
if
you
think
about
all
those
servers
running
in
data
centers,
let's
say
AWS
based
on
Public
Information.
There
are
about
3
million
machines
running
in
AWS,
which
might
seem
like
a
lot,
but
to
bring
this
into
perspective.
If
you
think
about
Raspberry
Pi's
that
have
been
sold
in
the
last
year
alone,
it's
10
times
more
units,
and
this
is
just
Raspberry
Pi.
A
If
you
look
at
all
personal
computer
species
sold
in
the
last
week
that
it's
even
10
times
more
again,
so
there
are
100
or
maybe
thousands
times
more
personal
computers.
Some
are
out
there
running
in
the
world,
not
being
participate,
participating
not
being
connected
to
the
decentralized
economy,
and
the
question
is:
how
can
we
fix
that?
A
The
second
part
is
how
actually
we
are
using
all
of
those
Computing
resources
we
have
out
there.
Most
people
are
not
using
their
CPU
for
like
are
not
using
all
of
their
CPU.
They
have
available.
Even
if
you
are
a
developer-
or
maybe
you
are
editing
videos,
you
are
still
using
your
machine
at
maybe
50
on
average.
A
Similarly,
if
you
think
about
network
connections,
if
you
have
a
household
of
four
people
who
are
browsing
or
watching
Netflix,
they
are
playing
games,
then
typically
they
need
maybe
40
megabytes
per
second
of
internet
speed,
but
an
average
American
household
has
a
connection
speed
of
127
megabytes
based
on
some
data
type.
A
It
was
able
to
found
so
which
it
means
like
we
are
using
our
internet
connections
at
30
percent
only
and
the
remaining
70
percent
is
there
waiting
to
be
used
for
something
better
and
now
the
question
is:
how
can
we
make
this
work
better?
How
can
we
put
it
into
work
to
make
the
file
coin
economy
better,
or
maybe
just
the
entire
world
better
place?
A
And
that's
what
we
are
trying
to
build
and
the
project
is
called
Falcon
Station
it's
a
or
our
vision
is
to
build
a
desktop
application
that
anybody
can
install
and
run
it
can
be
especially
it
should
be
a
non-technic
non-technical
people
like
your
parents,
your
friends,
this
application
will
use
your
idler
resources,
do
some
meaningful
work
and
it
will
reward
you
with
falcoin
in
return
so
perfect,
but
it's
very
high
level
and
if
you
are
thinking
of
how
does
it
work
under
the
hood
or
like?
What's
the
technology?
A
The
idea
is,
we
want
to
run
untrusted
code,
meaning
anybody
can
provide
codes
to
be
executed
on
stations
stations
are
operated
by
untrusted
people.
There
is
no
contract
saying
they
have
to
do
what
we
are
telling
them
and
we
want
to
pay
for
jobs
processed
by
station.
So
this
is
quite
tricky
and
we
are
dreaming
big.
There
are
so
many
use
cases
you
can
apply
to
file
coin
station,
for
example,
we
can
enable
scientists
to
run
distributed
computation
using
data
stored
on
filecoin.
A
A
And,
finally,
we
can
use
a
fleet
of
Falcon
stations
to
measure
performance
of
your
decentralized
network
and,
what's
very
interesting
here,
you
will
be
doing
the
measurements
from
Real
World
home
networks
all
around
the
world,
which
is
much
better
than
running
some
simulated
benchmarks
somewhere
in
the
cloud
using
simulated,
Network
conditions.
So
there
are
plenty
of
of
opportunities
here
now.
There
are
also
many
technical
challenges
which
are
in
front
of
us
that
we
are
actively
working
on
right
now.
A
The
first
one
is:
if
we
want
to
run
an
interested
code
on
on
users,
computers,
we
need
to
ensure
the
code
is
isolated,
that
the
code
cannot
exploit
any
resources
or
maybe
access
secret
data
on
user's
computer,
and
so
that's
the
first
part
to
solve.
The
second
part
is
making
sure
that,
even
if
there
is
a
security
vulnerability
discovered
in
our
isolation
mechanism,
we
need
to
fix
it
quickly
so
that
it's
not
exploited
because
we
were
waiting
for
too
long.
A
We
also
need
to
limit
resource
usage,
because
if
the
station
starts
using
all
your
computer
resources,
spinning
your
fan
eating
all
the
CPU
and
then
saturating
your
network
bandwidth,
then
people
will
just
shut
it
down
and
never
start
it
again,
because
they
want
to
use
their
computers
for
something
else,
not
only
for
the
station.
So
we
need
to
find
a
way
how
to
limit
everything
that's
running
inside
station,
to
keep
it
kind
of
not
noticeable
by
a
regular
user,
and
we
also
need
to
figure
out
how
much
is
still
acceptable.
A
A
So
if
I'm
on
a
slow
network
connection,
maybe
I
should
prefer
CPU
intensive
jobs
and
similarly
for
the
other
conditions
and
then
finally,
or
almost
finally
legal
issues,
for
example
with
Saturn,
if
I'm
caching
content
to
accelerate
retrieval
what,
if
my
computer
ends
up
caching,
illegal
content
and
then
police
comes
knocking
on
my
door,
hey
Miroslav!
Why
are
you
sharing
this
video?
That's
not
legal
in
our
country
who
is
responsible.
Is
it
me
because
I'm
running
the
station?
A
Is
it
protocol
apps,
because
their
Bluetooth
station
is
it
maybe
the
Saturn
team
because
they
are
operating
Saturn
or
maybe
it's
all
fine,
because
I'm
kind
of
just
being
like
an
ISP
just
giving
people
connectivity?
So
many
questions
to
explore
and
answer
and
then
once
we
address
all
of
this,
we
can
build
a
decentralized
Marketplace
which
will
allow
anybody
to
submit
any
job
any
computation
or
maybe
build
a
cash
note
and
pay
for
for
that
to
happen,
and
people
will
be
able
to
earn
more
money
and
contribute
in
more
different
ways.
A
That's
like
the
final
game
where
we
want
to
get
what
is
the
status
we
are
currently
running
an
alpha
version.
You
can
download
it
from
our
website.
It's
it's
an
alpha
version
right,
so
we
have
a
Saturn
module,
the
L2
cache
which
is
connected
to
test
net.
It's
already
handling
real
requests.
We
are.
We
have
synthetic
traffic
to
test
all
technical
parts.
So,
if
you're
on
the
station
now
you
will
see
you
are
handling
some
jobs
and
you
are
helping
us
to
calibrate
Saturn
Network.
A
There
are
no
file
going
earnings,
yet
that's
waiting
for
the
official
launch
of
L2
and
our
main
focus
right
now
is
to
iterate
on
the
ux
to
make
sure
the
station
is
easy
to
use
for
for
an
average
user,
and
we
are
also
researching
researching
more
ideas.
What
kind
of
modules
would
make
sense
to
build
and
run
as
part
of
the
station
so
that
we
can
make
station
a
better
deployment
Target?
A
Next,
we
are
aiming
to
do
a
better
release
which
will
be
almost
fully
functional,
but
maybe
with
small
basic
pieces
to
to
polish
before
a
GI.
It
will
have
built-in
file
coin
wallet.
So
if
you
give
this
to
your
parents,
they
don't
have
a
file
coin
address.
They
don't
even
know
how
to
create
it.
Then
it
will
be
all
handled
by
the
station.
For
them
we
will
have
Saturn
module
which
will
be
connected
to
the
mainnet,
so
it
will
be
black
real
in
production
and
in
production
traffic,
and
there
will
be
filecoin
earnings.
A
Our
current
plan
is
to
ship
this
in
a
March
next
year
cool,
so
that
was
kind
of
a
slideware
and
now
how
does
it
work
under
the
hood
for
real?
What
is
the
technical
architecture?
We
ended
up
with
a
kind
of
decoupled
or
motorized
solution
where
we
have
a
station
process,
which
is
an
electron
application,
providing
the
station
interface
user
experience,
and
then
we
are
thinking
about
supporting
multiple
modules.
Each
module
is
a
long
living
process,
managed
by
station
and
yeah.
A
So
again,
station
is
a
lightweight
electron
based
application.
It
has
basically
three
jobs.
One
is
to
integrate
into
the
OS
tray,
so
it's
just
running
on
your
computer.
Without
being
too
intrusive,
it
orchestrates
the
orchestrates
the
module
right
now
it
means
we
will
start
and
stop
the
module
process
in
the
future.
We
might
do
auto
upgrades
of
modules,
maybe
discovery
kind
of
app
store,
we'll
see,
and
it
renders
a
web-based
UI
in
the
electronic
window
so
that
you
can
have
a
nice
dashboard
and
then
this
is
an
example.
A
How
it
looks
like
it's
probably
better
to
download
it
from
the
website
and
see
it
for
yourself.
It's
too
small.
Sorry
and
what
is
a
module?
This
is
a
more
interesting
part
for,
or
it
might
be
more
interesting
to
you
at
the
moment.
A
module
is
basically
a
long
running
service
which
can
be
written
in
any
language
for
Saturn.
We
will
just
grab
the
go
binary
of
the
Saturn
L2
note
and
run
it
as
part
of
the
station.
A
We
think
we
can
keep
using
this
approach
for
the
next
two
to
three
modules:
to
learn
more
about
the
whole
requirements.
What
modules
need
from
the
station?
A
module
should
consume
smallish
amount
of
Computer
Resources
so
that
people
don't
get
annoyed
by
running
station
and
it
should
pay
rewards
in
filecoin,
which
is
quite
tricky
and
oh
yeah.
A
If
you
want
to
build
a
module
yourself,
what
is
important
to
know
that
if
you,
the
module,
is
configured
by
station
using
environment
variables?
This
is
the
old
school
12
Factor
app
design
pattern,
and
it
seems
to
work
well
for
us
as
well,
and
the
communication
between
the
station
and
each
of
the
module
is
through
STD
out
that
this
allows
us
to
capture
all
messages
from
the
module.
A
If,
even
if
the
module
crashes
or
something-
and
we
are
not
able
to
connect
to
the
HTTP
interface
quickly
enough
and
then
there
is
also
a
Json
over
HTTP
interface,
which
allows
us
to
for
get
more
statistics.
More
data,
Maybe
control,
behavior
of
the
module
in
the
future
and
what
is
very
important
right
now
we
are
supporting
only
trusted
modules.
It
means
we
trust
people
building
modules
that
they
are
not
going
to
mind,
bind
Bitcoin
or
steal
your
secrets,
because
we
don't
have
a
fully
isolated
and
sandbox
runtime.
A
Yet
and
that's
oh
and
you
might
be
asking
why,
like?
Why
are
we
using
electron
for
all
the
other
options
there
and
the
reason
is
you
want
to
ship
and
iterate
quickly?
I'm,
an
OJ
as
a
guy,
my
colleague
Julian,
who
is
working
on
the
station,
is
an
OJs
guy
as
well,
and
if
we
had
to
learn
Russ
to
write
all
of
this
on
Towery,
then
we
would
move
much
slower
and
also
because
the
main
station
application
is
very
lightweight.
Most
of
the
heavy
lifting
is
done
by
modules.
A
A
And
then
the
next
big
task
in
front
of
us,
besides
going
public,
is
to
find
a
way
how
to
how
to
handle
the
sandboxing
and
resource
limiting
which
I
mentioned
at
the
beginning
of
my
talk,
so
we
would
like
to
build
a
custom
runtime,
maybe
similar
to
a
VM,
maybe
closer
to
ipvm,
which
will
allows
module
authors
to
run
their
code
in
an
isolated
environment.
So
we
don't
need
to
trust
them
to
behave
well,
we
can
enforce
it
and
we
also
want
to
enforce
limits
on
resource
computer
consumption.
A
Like
network
access
or
CPU
usage-
and
the
idea
is
we-
we
actually
cannot
use
virtualization.
So
no
VMS,
no
containers
because
many
consumer
computers
don't
support
virtualization.
For
example,
Windows
Home
doesn't
support
hyper-v,
so
there
is
no
way
how
to
run
the
car
on
Windows
and
computers
and
fortunately,
recently
there
was
quite
a
lot
of
breakthrough
breakthrough
done
in
how
to
put
it
in
the
isolate
based
architecture.
So
if
you
think
about
cloudflare
workers,
Dino
deploy,
but
also
like
your
V8,
your
Chromeo
browser.
A
All
of
these
to
these
systems
are
using
isolation
where
you
have
a
JavaScript,
runtime
or
webassembly
runtime,
which
provides
isolation
from
the
host
machine
and
it's
secure
enough.
So
we
are
thinking
about
about
leveraging
the
same
approach
and
the
idea
is
the
user
will
supply
the
hardware
and
operating
system
on
top
of
which
we
can
install
station
and
station
will
have
two
parts.
One
part
is
the
runtime
which
allows
modules
to
execute
their
code.
It
will
be
sandbox,
we
will
resource,
we
will
limit
the
resources
and
we
can
provide
them
platform
apis.
A
Maybe
it
will
be
like
posix
Unix
apis
to
work
with
files
and
network.
Maybe
we
can
go
even
higher
level
and
give
them
a
autopilot.
We
have
three
specific
apis.
Like
fetch,
midi,
CID
publish
this
file
to
filecoin
storage
providers.
There
are
many
options
which
we
need
to
explore
and
on
top
of
this,
people
will
be
able
to
build
station
modules
and
leverage
all
of
this
goodness
of
station
runtime,
so
that
their
job
is
much
simpler.
They
can
build
a
module
in
an
easier
way.
A
And
then
there
is
even
more
things
to
it
to
research
and
think
about,
because
the
part
I
described
so
far
is
more
about.
How
do
can
you
run
untrusted
code
on
stations,
but
then
the
another
important
part
is:
how
can
we
actually
pay
people
for
running
that
untrusted
code,
and
that
involves
things
like
proof
of
job
getting
done?
It
involves
billing
and
then
we
also
need
to
figure
out.
A
How
are
we
going
to
schedule
jobs
and
we
are
not
sure
yet
whether
this
is
something
that
should
be
part
of
the
platform
or
the
station
runtime
or
whether
this
is
so
specific
to
each
module
that
it
needs
to
be
implemented
as
part
of
the
module.
So
more
things
to
research
and
I'm
getting
to
the
end
of
my
dog
I
would
like
you
to
give
us
a
to.
Let
us
know
what
you
think
about
the
station.
You
can
download
it
from
that
QR
code.
A
I
hope
it
works
on
in
this
resolution
or
if
you
like,
any
of
the
technical
challenges
I
described
and
if
you
have
any
ideas
how
we
can
solve
them,
then
please
come
and
talk
to
us.
We
would
be
super
happy
to
learn
more
or
if
you
also
have
ideas
how
or
what
kind
of
modules
you
can
or
we
can
build
and
run
on
station
how
we
can
leverage
station
to
do
distributed
computing.
Caching,
anything
else,
then
let
us
know-
and
maybe
we
can
even
work
together
and
build
a
module
four
station.
A
B
So
I
think
that
I
mean
this
is
amazing.
I
think
that
one
of
the
things
that
sort
of
interesting
comparing
this
you
sort
of
compared
it
to
sort
of
ethereum
nodes,
and
things
like
this
and
one
of
the
areas
that
I
guess
this
Falls,
partly
into
the
legal
side
and
partly
into
kind
of
like
how
do
we
encourage
people
to
adopt
this
on
a
mass
scale?
B
One
of
the
things
we've
seen
in
that
space
is,
you
know,
people
love
downloading
tools
that
make
money
for
them,
especially
when
they
can
download
it
onto
other
people's
machines.
B
Right
so
is
that
are
you
thinking
about
some
sort
of
central
control
and
Reporting
mechanism
so
that,
if
somebody
say
decides
to
roll
this
out
across
a
company
that
they're
either
the
system
in
or
not
that
you
will
be
able
to
essentially
control
that,
and
does
that
mean
that
you
are
thinking
about
how
to
make
a
connection
between
the
identity
of
the
person,
who's
receiving
the
file
coin
and
and
the
machines
they
operate
on
and
all
of
the
questions
that
kyc
and
things
like
that
raise?
A
There's
a
great
question:
I
think
we
haven't
been
thinking
about
that
yet,
but
we
definitely
should
and
then
maybe
I
can
ask
you
a
counter
question
if
I'm
running
station
on
my
employer's
computer,
should
it
be
me
earning
the
file
coin
or
should
the
employer
be
paid
for
it?
So
I
think
it's
definitely
something
we
need
to
look
into
yeah,
but
I.
Think
in
theory
the
idea
or
the
design
of
station
should
allow
for
this,
because
station
can
decide
which
modules
it
will
install
and
run.
A
C
A
Now
it's
up
to
the
module
to
Pro,
to
provide
a
consensus
mechanism
and
I
think
our
idea
is
to
build
first
few
modules
in
a
way
that
the
module
is
responsible
for
everything
and
then
we
as
station
authors,
can
look
into
these
individual
modules
and
look
for
opportunities
to
extract
some
of
that
difficult
part
into
stations
so
that
other
modules
can
build
on
top
of
it.
But
we
don't
know
yet
what
kind
of
consensus
mechanisms
will
make
sense
or
maybe
what
kind
of
proofs
models
will
need.