►
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
So
I'm
going
to
do
it
just
do
a
quick
order,
a
five
minute
intro
to
this
track,
just
to
get
us
started
and
what
I'm
going
to
cover
is
I'm
going
to
go
through
the
track
on
the
different
talks,
I'm
going
to
talk
about
computational
network
models
and
then
I'm
going
to
talk
about
computation
and
ibfs
nodes.
A
So
this
is
the
ipfs
on
wassum
track
if
you're
in
the
wrong
place,
now's
the
time
to
go,
we're
gonna
go
through
a
set
of
talks
today,
we'll
start
with
several
things
that
are
ongoing
projects
right
now.
So
things
like
wasn't
my
pld
and
experiments
around
putting
computation
into
our
professionals
already
we're
going
to
hear
from
the
fvm
work
and
also
long-term
visions
for
how
was
on
an
iple
could
work
together.
A
Then
we'll
hear
from
the
vision,
folks
on
their
work
with
advanced
ability
and
wasm
and
brooklyn's
ipvm,
which
is
a
really
exciting
new
project,
we'll
also
hear
a
set
of
ideas
from
a
whole
set
of
folks
alonso.
I
think
you're
gonna
talk
about
hierarchical
consensus
and
all
the
models
there
and
I'm
going
to
walk
through
a
set
of
use
cases
that
we've
talked
about
in
terms
of
how
to
use
ipfs
with
vms
and
so
on,
and
the
rest
of
the
day
will
be
it's
kind
of
open
workshop
time.
A
So
we'll
either
do
some
planning
around
making
decisions
on
what
should
these
vms
do
or
how
to
do
it
like
what
implementation
path
to
take,
what
projects
to
work
on
and
so
on
so
quick
brief
into
competition
network
models
throughout
the
development
of
computing,
that
we'll
see,
we've
seen
a
ton
of
different
competition
models
and
moving
through
many
different
platforms.
A
These
days
most
of
the
machinery
out
there
has
these
layers
of
of
systems
that
end
up
virtualizing
and
layering
vms
of
different
types.
On
top
of
hardware,
you
have
things
like
vms
and
hypervisors,
and
so
on.
You
have
containers,
you
have
now
even
smaller
little
shards
of
computation
like
lambdas
and
so
on,
and
of
course
you
have
things
like
like
blockchains
and
whatnot.
You
also
have
a
ton
of
embedded
device
systems
that
also
run
very
tiny
little
programs.
They
don't
even
have
a
big
operating
system
around
them.
A
There's
been
many
large-scale
computational
network
scheduling
systems,
so
things
like
borg,
which
turned
into
kubernetes
or
ec2
and
lambda,
and
so
on.
Cloudflare
workers
there's
all
kinds
of
operations
there.
That
could
be
extremely
interesting
to
enable
with
things
like
fpfs,
then
we
have
things
like
important
cryptographic,
advances
that
give
us
different
computational
models,
so
think
of
zero
knowledge,
proofs
and
fully
homomorphic
encryption
and
many
other
ways
of
issuing
computational
jobs
in
a
cryptographic
setting
where
you
want
to
be
able
to
create
proofs
about
the
computation.
A
So
not
only
do
you
want
to
run
a
program,
you
want
to
be
able
to
run
a
program
in
a
setting
where
you
can
produce
a
cryptographic,
proof
that
verifies
that
the
computation
was
done
correctly.
A
There's
also
things
like
the
this
is
computational
models
like
the
airline
actor
model
and
many
other
similar
systems,
where
you
have
a
very
flexible
run
time
where
new
machines
can
be
added
and
removed,
and
you
have
a
a
space
to
address
processes
and
and
send
messages
between
processes
that
is
partition,
tolerant
by
design,
and
that
enables
easy,
easy
blending
of
having
the
system
parts
of
the
system
fail
at
a
time
and
come
back
and
resume
and
so
on.
So,
models
like
these
are
for
kind
of
distributed
operating
systems
and
so
on.
A
We
also
have
things
like
the
object
capability
model
that
comes
from
the
e-writes
family
of
of
work,
things
where
we
can
reason
about
the
accesses
and
rights
and
and
so
on.
The
different
objects
may
have
both
in
their
native
local
machine
or
in
remote
machines,
and
how
to
reason
about
the
passing
of
information
and
the
passing
of
rights
and
getting
to
being
able
to
build
safe
systems
despite
the
presence
of
failure.
A
So
in
all
these
concepts
distributed
systems,
we
also
have
a
blockchain
computational
models
where
we
now
have
these
massive
trees
that
are
that
are
accumulating
both
state
of
programs
and
code
of
programs
and
schedule
the
running
computation
around
them.
We
have
tons
of
systems
that
leverage
compiler
tooling
to
enable
much
more
portable
execution
of
different
different
runtimes.
We
have
these
days.
A
Wasm
is
now
one
of
the
main
things
that
are
achieving
this,
so
so
the
world
has
explored
things
like
lvm
ir
and
so
on,
and
now
it
landed
on
was
a
really
good
direction
for
most
of
most
industries.
So
it
was
worth
mentioning
that
these
days,
most
industries
are
investing
deeply
in
wasm
being
the
computational
substrate
that
is
going
to
enable
lots
of
different
operations
in
different
use
cases.
A
So
it's
a
really
good
technology
to
bet
on,
and
so
when
you
think
about
vm
runtimes
and
we
think
of
using
wasm
as
a
kind
of
hypervisor
to
be
able
to
run
very
different
models
and
we
think
of
a
what
content
addressing
gives
us
and
being
able
to
address
code
by
by
hash
and
being
able
to
invoke
our
programs
directly
by
hash
or
being
able
to
treat
the
whole
ipld
space
as
like
a
memory
model
where
you
can
create
new
data.
Struct
data
structures
create
new
objects
and
call
them,
and
so
on.
A
Where
you
can
run
local
programs,
you
can
produce
new
data
and
so
on
and
send
it
around
use
the
underlying
os
tooling
to
move
around
the
state
of
those
computations
so
think
of
a
model
where,
like
your
programming,
literally
a
system
in
one
where
you're
writing
code-
and
you
have
as
the
primitives
in
your
in
your
in
your
type
system
and
in
your
in
the
functions
of
your
objects,
the
ability
to
call
objects
that
are
elsewhere
in
the
ipfs,
runtime,
so
think
of
being
able
to
define
the
actual
objects
themselves,
the
state
that
they're
going
to
store
and
the
computation
that
is
going
to
run
over
those
objects
as
all
within
ipfs
itself.
A
So
think
of
having
you
know
if
you're
familiar
with
like
the
c
memory
model.
Just
imagine
the
the
function
point
like
the
object
description
to
include
a
function
pointer
to
the
code.
That's
going
to
run
the
object
and
think
of
that
as
the
as
all
of
that
being
in
ipld
itself.
And
if
you
have
wasom,
then
you
can
run
whatever
code
you're,
throwing
there.
You
can
compile
whatever
language
down
into
wasm
and
run
it.
A
The
fem
is
already
an
example
of
this
kind
of
system,
so
the
fem
took
it
was
it's
a
it's
a
system
based
on
blockchains
and
four
blockchains
and
so
on,
but
it
already
has
this
this
kind
of
runtime
model
where
you
have
ipeldy
at
the
core.
You
use
that
to
model
the
computation
and
then
you're,
bringing
all
the
compilation,
all
the
code
down
to
wasm
and
executing
in
the
awesome
layer,
and
this
can
already
support
like
many
different
runtimes
being
added
on
top.
A
A
A
So
you
have
a
bunch
of
ipfs
nodes
or
data
and
platform,
or
things
like
that,
and
you
want
to
be
able
to
run
a
lot
of
programs
on
top
of
that
data,
and
you
want
to
be
able
to
express
the
programs
and
and
the
large
scale
systems
in
some
computational
runtime
that
that
lets.
You
kind
of
issue
out
those
this
these
computational
jobs
to
a
lot
of
these
notes
and
so
think
of
being
able
to.
A
A
It's
also
worth
mentioning
that
there's
a
whole
swath
of
computational
networks
that
will
emerge
that
that
are
in
this
cryptographic,
setting
where
you're
either
doing
zero
knowledge,
computation
or
fully
homophobic
encryption
and
so
on,
and
over
the
next
couple
of
years,
they're
gonna
yeah,
end
up
end
up
developing
and
appearing,
and
those
are
different
computational
models,
but
for
the
most
part,
they're
compilable
down
to
wasm.
A
A
Yeah,
so
I
think
now,
let's
talk
very
briefly
about
computational
iphos
nodes.
Naturally
we
have
the
whole
ipld
model.
We
want
to
be
able
to
write
codecs
in
a
runtime
where
you
don't
have
to
install
them
in
every
system,
but
rather
you
can
just
sort
of
download
them
and
run
them
in
kind
of
the
web
style
model
you
that
includes
codecs,
for
how
we
lay
out
the
data,
it
might
be,
how
we
lay
out
the
dag
and-
and
we
might
include,
also
transformations
between
objects.
A
So
you
think
of
one
ipld
graph
as
being
able
to
represent
computation
itself,
so
you
can
think
of
a
function
and
the
argument.
Like
the
code
of
the
function
and
the
arguments
the
function
are
all
represented
by
by
hashes
the
actual
function
invocation
can
be
represented
by
a
hash,
so
you
now
can
have
transformations
of
graphs.
So
you
have
some
ipld
graph.
You
can
apply
a
function
to
that
graph
and
end
up
with
a
separate
graph.
So
that
gives
you
a
full
programming,
runtime
and
install
content,
address
and
hash
length,
and
so
on.
A
A
So,
for
example,
if
you
have
an
encrypted
file-
and
you
want
to
read
it
unencrypted-
what
you
really
would
like
to
do,
instead
of
decrypting
the
whole
file
and
storing
another
copy,
is
to
just
apply
a
function
to
that
file
that
decrypts
the
file
as
you
read
it
so
imagine
having
a
handle
on
an
encrypted
file.
So
you
have
some
tree
that
represents
the
encrypted
file.
You
want
to
be
able
to
add
a
handle
on
that.
A
That
has
the
decryption
key
and
you
want
to
be
able
to
model
that
to
higher
layers
as
an
unencrypted
file,
and
you
want
that
reading
to
just
happen
automatically
similar
to
that
you
could
have
other
layouts
or
other
representations
of
files.
You
can
store
a
compressed
file
and
expose
an
uncompressed
representation
to
layers
above
that
would
just
uncompress
on
the
fly
or
things
like
that.
A
A
Databases
at
the
heart
are
just
a
set
of
indexing,
trees
and
algorithms
that
operate
over
those
trees,
really
really
quickly,
so
think
of
being
able
to
just
generate
those
those
data
structures
themselves
and
being
able
to
generate
the
trees
over
the
data,
as
they
are
all
of
that
in
kind
of
this
ipld
computational
runtime.
So
you
can
create
databases
or
in
in
this
model
and
run
them.
A
Then
we
can
think
of
having
an
application
model
where
you
can
have
a
run.
Ipfs
could
evolve
to
have
a
runtime
where
you
can
deploy
applications
to
to
an
ipfs
node
similar
to
how
you
can
have
you
can
install
an
application
in
an
operating
system
or
you
can
install
a
service
worker
in
a
browser
or
or
just
the
fact
that
you're
like
visiting
a
web
page
and
running
some
code.
App
professionals
can
evolve
to
include
that
application
model.
A
This
of
course,
needs
a
really
good
model
in
terms
of
resource
usage,
so
things
like
extremely
good
sandboxing
for
any
kind
of
untrusted
code,
good
resource
usage
constraints,
things
like
gas
models
and
so
on,
models
for
how
to
share
resources
like
the
network
and
so
on,
and
capabilities
for
being
able
to
access
certain
objects
to
be
able
to
reason
about
what
programs
that
your
remote
programs
understand.
A
Programs
that
you're
running
what
kind
of
resources
are
you
giving
those
programs
to
be
able
to
use
yeah
by
far
one
of
the
most
common
things
I
hear
about
people
wanting?
This
is
being
able
to
do
task
scheduling
so
being
able
to
have
some
set
of
computers
that
they
add
a
vm
on
top
and
they
can
issue
small
jobs
to
run
on
those
computers,
often
in
parallel,
so
think
of
being
able
to
have
a
vm
that
then
runs
millions
of
lambdas.
On
top
of
that
on
that
vm.
A
One
other
use
case
here
is
that
we
can
use
this
for
protocol
development
itself,
so
you
can
think
of
evolving
ipfs
nodes
and
ios
implementations
themselves
by
being
able
to
describe
the
protocols
as
code
and
ship
the
code
to
those
nodes.
So
you
write
a
new
dht
implementation
or
you
write
a
new
possible
implementation.
You
write
some
new
like
subnetting,
algorithm
or
whatever
you.
You
can
write
that
protocol
as
a
wasm
bundle
that
then
would
get
downloaded
into
an
ipfs
node
and
run
that
way.