You know it was.

Good to have you here he said to.

You.

And.

Boy.

And.

Yeah I don't know about
that last part but.

So I put together a presentation.

Last night and.

It's not very polished but
I think it will hit all the sort of

things that are top of mind for me when
it when it comes to you know because and

security so I'm going to start
out with a brief overview

I'm going to soon that most people in
this audience know a little bit about

Bitcoin the block chain that I'm going to
talk about some of the security aspects of

because I know some of the super
security attributes of big one.

And then platform imperatives these
are kind of things that I think of when I

think of the platform that are what.

Is there a.

Is the might.

There not for the audio and here.

OK I'll try to yell.

The third thing I'll talk about our
platform imperatives these are things that

I think of that are important for whatever
platform you're running on top of for

you to think about from a security
perspective making sure private keys and

the like are secure and

then finally some sort of observations
that I've made over the years about.

Corn and its relevance to security.

So a little bit about pay as when
he mentioned I'm a co-founder and

C.E.O. of BIT A We were founded in two
thousand and eleven which makes a.

One of the oldest and
most experienced teams in the big space.

We're headquartered here in Atlanta
we have thirty seven employees.

Our products are primary product
is payment acceptance So

these are merchants that want to accept
that point as a form of payment but

we also have billing tools this gets into
the business to business transactions

payroll and mass pay so people that
need to pay a lot of employees or or

whatever can use our platform for that
pay tends to be more like think of your

app stores where you need to pay lots of
developers scattered all around the world.

Then we also have a wallet it's
called co-pay you can download it

in any of the app stores it also runs on.

Pretty much every platform that you can
run on your mac or Windows machines.

Or Linux.

And then we have the big core platform
which is your cloud infrastructure and

it's the backend for co-pay so
co-pay communicates and

interfaces with the block
chain through Bit core.

Bit pay does approximately four
thousand transactions per day.

Payment transactions that works out to a
little over one hundred thousand per month

so just to give you an idea of how
much activity we have on our platform.

Yeah.

Yeah I'll give you got
a few slides on that.

So you have four thousand networks that's
about what I think that's roughly what

visa does per second so
just to give you an idea.

So moving on to the sort of overview
of declawing little bit of history

I actually got into big coing.

Because of some previous experience
with the platform called digit cash and

that was David challenge in going back
into the early one nine hundred ninety S.

I got very excited about.

That platform it was a it was I think
probably the world's what we could

refer to as the world's first
cryptographic payment system.

The Did you cash never you know quite
gain traction they went bankrupt and

the payment system went along with it and
so that got a lot of people thinking about

how can we build a cryptographic
payment system that didn't have this

vulnerability of being dependent on one
company and it staying in business and

so sort of that background is what
over the course of years me kind of

searching going to lead me to bit coy and
I was very fortunate that when I read this

a Toshiba white paper that I had thought
about this problem space beforehand and

I kind of recognized the the magnitude
of the problem that it had solved.

So you know because it really
is a computer science.

Sort of breakthrough and

it does something that a lot of
people thought was impossible

before because it existed.

Just some other history here the code
was written in two thousand and

eight primarily we think.

By an anonymous author called Nakamoto.

The first Genesis block was created
in two thousand and nine and

it had a timestamp in it to it referred
to a headline in a British newspaper.

To sort of you know prove that
the block wasn't created before

January two thousand and nine.

And then like I said there was a really
a computer science breakthrough which

was the use of proof of work consensus.

To allow thousands of computers around the
world to reach a consensus about history.

So.

This slide is talking a little
bit about the protocol.

I mentioned proof of work based
consensus essentially you have

rules that define what is a valid
transaction with a valid block and

you have all these independent
computers that are applying these rules

to the transactions that
they hear about on the net.

Work.

You know it's a gossip based protocol so
that can the mesh network is

a gossip based use of the gossip based
protocol to disseminate transactions

nodes are validating those transactions
some nodes are mining them and

putting it them into a box and we'll
talk about more of that in a little bit.

So because in the asset there's a lot of
over the course of the past year there's

been a lot of hype around
block change technology and

a lot of people talk about separating
because from block chain technology and

this to me doesn't make a lot of sense.

When you know the whole
the whole point of a block is

to create a chunk of data that you
can compute proof of work over and

the whole point of proof of work is is to
achieve this consensus in the whole point

of a token like decline is to create the
incentive that people will actually want

to compete you know apply the computing
power to create a secure history and so

if you take the coin and proof of work
out of that you don't need blocks and

you don't need a chain so why would
you call it block chain technology.

But.

The asset you can think of it as.

You know it's value is derived
from its utility in the US and

the utility is the bill
the ability to exchange value so

transfer ownership of value from one
person to another one entity to another.

And interestingly enough that's what
gives all forms of money their value

it's that ability to exchange value
that utility that you get out of it so

that's what gives because it's value
a marketplace sets the price of the exact

exchange rate of that coin to other
assets so people are buying and

selling that coins are trading on the
market you can go look at the markets to

find out the current exchange rate.

The trade the actual
trading the mechanics of it

look very much like trading in
a commodity so it's got a fixed or

you know a fixed supply function
that cannot be manipulated.

And because of that fixed supply function
it can be very volatile so it trades very

similarly to other commodities and I do
think of it like a commodity I think of it

like a commodity that makes
a payment system work.

And then the the asset is what creates
that incentive for miners to actually mind

that coin and
secure the block chain I mean if.

You raise your hands if all of
this is like brand new to you

that you never thought about this or

if this is all you know kind of
you've studied all of this OK so most

of you get the basics the basic mechanics
of decline in the block chain All right.

A few quick facts we get one block
every ten minutes the difficulty

rate of the mining algorithm or the
difficulty of the mining problem adjusts

every two thousand and sixteen Blocks
which is roughly once every two weeks so

depending on how fast you're finding
blocks that difficulty rate goes up or

it goes down so that on average you
always get one block every tent every ten

minutes and of course you can get a couple
blocks in a in a span of a few minutes or

you might wait forty minutes for
a block but

on average it's once every ten
minutes that's just statistics.

In January the.

The network reached capacity of one X.
a hash per second so.

You know since two thousand and

seven when I started we were measuring
you know the network hash rate and

get a hash is present in the mega had you
know make a hash and get a hash is and

pet a hash is and now we're up to one
you know we surpassed one X. a hash in

about a year ago somebody did an estimate
and so this is when the hash rate

was much lower somebody did an estimate of
you know just overall computing power and

how to think about it compared to other
computing networks that we've seen and

they they estimated that the they coined
the computing power securing the big

block chain was about one hundred times
more powerful than all of Google.

And now it's probably a mag order
of magnitude higher than that.

I'm sorry.

I've got a slide on that.

Show actually chart here in a minute.

So currently the blocker award is
twenty five declines you get rewarded

with twenty five declines every time you
find a block the next having is in July so

every four years the block reward cuts
in half so it started out at fifty.

Few years ago it was cut
in half to twenty five

in July it will cut in half
again to twelve and a half and

eventually it ends unlike the year
twenty one forty I think it is.

When all twenty one million that coins
have been brought into existence so

far we've.

Checked this yesterday or maybe the day
before we've mined brought into existence

a little over fifteen million declines so
almost three quarters of all the big coins

have been brought into existence now
another way to think about is every.

Every four years the rate cuts in half so
that every four years your bringing into

existence half of the remaining bit coins.

Currently And currently the block size
each block is limited to one megabyte in

size so that limits the overall throughput
to the theoretical limit of like

seven transactions per second but I think
the practical limit is more like three

transactions per second in only enough
I'm going this weekend to a conference

to to discuss the topic of increasing that
and raising that limit it's an artificial

limit it was set to prevent
people from denial of service.

Launching a D.O.'s attack on the network
by creating these really huge blocks that

everybody has to verify.

So it's kind of an artificial limit
there's a lot of debate in the community

about raising that and

the timing of when to raise that limit and
allow the network to scale even higher.

So somebody was asking about
the price this is a price chart.

So you can see back in two thousand and
nine it was worth very little and then

we had a huge spike toward the end of two
thousand and thirteen a big decline and

then it's kind of started to rise again.

The units so up here we're at.

Peaked out at around twelve hundred.

Dollars per bit going in right now we're
at right around four twenty four twenty

five I think was the exchange rate today.

Now you know the value of.

Is driven by or you know
the exchange rate is driven by its

utility it's purely buyers and
sellers using it for

something maybe they're speculating on the
fewer future value of it they're hoarding

the coins or maybe they need it for
trade in exchange.

And then that dries mining mining
activity so the higher the price

the more profit somebody can make by
operating and securing the network.

So the mining difficulty follows
price not the other way around.

I mean this is the price if nobody found
any utility at all nobody would buy

it nobody would use it
the price would be nothing.

This is the house right over time so
you know it's pretty exponential

this might be more useful to
show in a logarithmic scale.

So that's on a log log or rhythmic scale.

And all the time going back to two
thousand and nine so you can see when I

first started in two thousand and eleven
you used to be able to mind it coins on

graphics cards at the very
beginning you couldn't mine on you

know a regular C.P.U. you could still mine
on a C.P.U. or graphics cards today it's

just that you will spend more money in
electricity than you earn in that point.

And so we went through generations of
different technology we went from C.P.

is to G.P. use G.P.S. to F.P.G.A.
these F.P.G.A. is in the now a six and

the latest generation a six that
are coming onto the market or

using sixteen and I mean our process so
kind of catching up to Intel and

the major chip manufacturers.

This is the number of transactions
per day on the network so

you can see we're at all
time highs right now.

And we're just about we're bumping up
against the capacity limits of the system

I think just yesterday or
day before there was a new record set for

the one day number of
transactions on the network so.

It's getting up close to two hundred
fifty thousand two hundred fifty thousand

transactions per day.

What's a typical transaction
What do you mean by that.

Like size like.

The nature of the transaction
you know people.

Well through us we've got you know
approximately four thousand transactions

per day and on our you know they
transactions are people buying stuff

so I mean we sold everything
from houses to cars to

you know new egg is one of our
customers they sell like Tronics.

A lot of people a lot of our
customers are like gift cards or

ways that people can cash out their big
corn and use it to you know now they've

had these gains they cash it out and
they want to buy something with it so

gift cards people buy those with them.

Services things like host cloud hosting
providers domain name registrars things

like that also or our customers and
more and more gaming.

You know kids under eighteen
can't legally have a.

Like a credit card or a checking account
because they're not legally able to enter

into a fact which is what you know
holding a balance with a bank is.

So kids under eighteen can use it
like like you would if you gave them

a twenty dollar bill like cash and

my kids you know I give them declines
they top up you know there are.

There gaming accounts and whatnot so we're
seeing more and more usage in the gaming

space video gaming.

All right.

Actually I'll show you.

Real quick.

Here you can actually see.

You know the network operating live so
you can see transactions coming through so

you can kind of get a sense
of there's a lot of very tiny

fractions of a big point transaction
going on so the very small

value transactions are probably more like
people paying each other buying goods and

services and one of the bigger
transactions are to be maybe exchanges or

doing doing settlement between different
big businesses and things like that or

people move you know that have saved
a lot and are moving it between wallets

stuff like that we can see the see them
happening as they happen live up here

we have blogs all the blocks these are the
recent blocks so the last block came and

nine minutes ago you can click on that and
then see all of the transactions that

are in that block.

OK All right security attributes
of Bitcoin So basically

because.

Yes that in the chain is a Public Ledger
that anybody can read validate the other

it has to be public the data in the block
chain has to be made public so that.

Nodes can validate it and
the database can be constructed over time.

Now they're soon anonymous addresses so
you don't know you can look at

a transaction say that was Stephen or
John or Phil or whoever.

You know you'd have to have some other
mechanism to connect that transaction back

to a real world activity
in somebody's identity.

So.

A transaction is.

Is.

It consists of inputs and outputs so
I'm spending what are known as

unspent transaction outputs I'm creating
a new transactions I feed inputs in and

I have outputs out going out and so if my
wallet was correct crafting a transaction

I wanted to send one bitcoin to somebody
I would give it an address of that

person that I wanted to send it to and
what it would do is it would go find and

spent out puts in my wallet that I
have the keys to sign and it would.

You know find enough inputs to meet or
exceed one Bitcoin So

let's say I had inputs that added up to
one point one declawing I'm sending one

because then I'm going to send myself
change of point one bit coins so

that change will be a new unspent
transaction I'll put that I could use in

a future transaction that I need to send.

And so transactions have you know basic
rules like the sum of the inputs must be

less than or
equal to the sum of the outputs.

If the sum of the inputs is less than the
sum of the outputs the difference they're

a minor can collect as a fee so that's
how you can attach a fee to a transaction

transactions have a scripting engine so
there's a.

Byte code language that allows you to.

You know craft more complex transactions
than just simply sending from one person

to another the are the kind of things that
capability allows us to do things like

multi signature transactions
transactions that can't be spent for

a certain amount of time.

And.
Lots of other stuff.

The one of the consensus rules is that
whatever script is in the transaction must

successfully or else it's not a valid
transaction so one of the buy codes might

be like check the signature on
this you know spinning of this.

Unspent output and it does a check
of the cryptographic signature and

if it fails then the script fails and
the transaction is not valid.

Addresses addresses are there's primarily
two types of addresses in use today

one is called paid a pub key hash so
you take in a public key you're hashing it

you're actually in a shop two fifty six
in a right and one sixty hash of that to

yield twenty bytes that are then
encoded in base fifty eight and

that is your big coin address and when
you spin to one of those type you know

a page or
poky hash address what the subsequent

transaction that spins it is doing is
it's revealing the public key matches

that hash is verifying that that public
key matches the hash and it is using that

public key to then verify the signature
that's actually on the transaction.

Paid a script hashes a little bit
of a different form of address and

you'll see these on the network is
beginning with a three instead of a one

page a script hashes actually hashing the
whole script that's needed to redeem or

spin the output and
it's actually a more flexible way to

construct addresses on
the network made a script hash.

Addresses are used for things like
multi signature transactions so I can

take a script that says check that I have
a signature from this public key from that

public key and that public he or I can
do two out of three or four out of five.

And that's what the script.

Is and then you take that script and
you hash that and

that's how you derive the public a so

when some I spent or how you derive
the address so when you spin to that.

Address.

You're actually not revealing the script
that actually is capable of redeeming it.

That only gets revealed when
you actually need to spend it.

In elliptic curve elliptic
curve cryptography is used for

all the signing.

This is just a diagram I mention
that you've got inputs and

outputs to a transaction.

So if you look at this input it's got

any one of these transactions
it has some inputs and outputs.

Again stop me if I'm moving too fast
I'm trying to get to some of the more

interesting security discussions.

Blocks blocks are basically a collection
of transactions that are put together.

You take all of the transactions and you.

You compute a hash is the hashes
of those transactions and

you put them into a tree and that's what's
called a Merkel tree in the Merkel root is

part of what we call the block header and
you take the block header and that's what

you're computing a double shot to fifty
six hash of and you're trying to you're.

A miner will put a nonce
into that header and

very that nonce repeatedly
compute the hash and

then if that hash is below a certain
number which is the difficulty target

then they found a solution and they can
publish that that becomes in the network

sees that they can and the rest of network
can quickly validate that and then mine or

start building on to the on to
the next block after that so

in the header what we have is the hash of
the previous block so this is how you form

a chain you have the Merkle root of all
the transactions in that block you have

this nonce that's the miners are varying
and then you also have a timestamp.

Version and then the difficulty target.

The timestamp is interesting.

Because it's the only X.
only data that's external to

the coin that is actually
in the bought chain.

And I'll talk a little bit more
about why that's that's interesting

in a bit toward the end
of the presentation.

So blocks computing a double shot to fifty
six finally finding a solution that meets

a difficulty target is extremely
hard takes a lot of computing power

in minors repeatedly.

Apply the double shot to fifty six looking
for a solution it's very hard to solve and

very easy to verify which is basically
proof of work it was a question of where.

It's up to the minor The only
rule is that they can't

they can't be bigger than one megabyte.

Right and you'll see blocks fairly often
on the network and what's happening there

is a minor is seeing a new block that's
been announced on the network and

there and it takes a non-zero amount of
time to verify that transaction and so

what they do is they go ahead and

assume that that block or takes a non-zero
amount of time to verify that block

what they do is they go ahead
assume that that block is valid and

they create another bot that has no
transactions because they don't know which

transactions that are in their memory pool
might have been included in that block so

they go ahead create a new block that's
an empty block no transactions that

goes ahead and points to this block that
they just heard about from the network and

they start hashing on that so a lot of
times I get lucky and they'll find a block

or they'll find a solution before they
fully validated that previous blocking

constructed a new block and so they can
get rewarded twenty five declines in that

case and it's a way that they don't
spin they don't waste a few seconds of

their of mining time you
know mining on their other

choices they use the last known good block
previous to the one they just heard about.

But chances are the one they just
heard about is actually valid and

what everybody else is going
to build upon so they so

there they have a better chance of the new
block there they find not being orphaned

if they you know assume that the block
they heard about is valid and mining and

pointing to the previous block that they
knew was valid if that makes sense so it's

a fairly common occurrence that you'll
see empty blocks appear and that's why.

It.

Has a whole lot been invalid.

Well I mean minors can create
invalid blocks all the time but

when they send them to their peers
on the peer to peer network there

they validate them
before they repeat them.

And if they don't validate they
won't they won't be repeated so

you you don't usually hear about
invalid blocks on the network.

Yeah yeah that's called a reorganization
and it happens very frequently.

I don't know I don't know whether it
happens like ten twenty times a day or

what but.

You can imagine that if you got all these
miners trying to find a solution and

find trying to find the next block that
occasionally you'll get to that happen

very close to the same time and they'll
announce it so one of those blocks will

be picked up by the network and
built upon in a form the longer chain in

the other one will be an orphan and miners
referred to that is orphan rates and so

if you're in the mining business one
of your objectives is to do things that

reduce the orphan rate because if you
produce an orphan then you don't get

there's no reward you know
earn any money for that.

So orphan rates are a big thing
that the miners pay attention to.

In in this whole proof of workplace
consensus is really trying to solve

the Byzantine generals problem.

So there's a pick.

Oriel illustration of
the watch chain just walk

pretty simple.

So I mention that these are used for
transaction signature I think you got

another slide on this some more.

Maybe not.

So it's using E.C.C. that's why
this is actually the curve the coin

uses It's called fifty six K.
one it's a couple it's curve so

there's an easy see there's two different
suites of curves I guess you could call

them one is COBOL it's the other one is
a set of curves based on a pseudo random

sequence of numbers and there's some
you know conspiracy theory that that

the pseudo random curves might have been
compromised by the N.S.A. in some way and

so it's very interesting that one of
the cocoa bits curves was chosen and

this was not a very widely
used curve in most where where

cryptography was used this particular
curve was almost never used so

it's kind of kind of interesting
that this was chosen.

The There's some Like I said some
conspiracy theory that maybe

the particular pseudo random sequence of
numbers wasn't quite so randomly chosen or

the seeds for that sequence were quite so
randomly chosen.

Maybe some of you know more about that
than I do not an expert in an ellipse

elliptic curve cryptography but.

One interesting aspect of it though
is it's fast in space efficient so

you've got a fairly small keys which
is important if you're building

a big database of transactions.

The piece of P. network I mentioned before
this it has a mesh structure it's a gossip

protocol so when I create a transaction on
my node what my no does is it contacts all

the peers it's connected to and since the
hash of it in what's called an I envy or

an inventory message and if one of.

Nodes that I'm connected to has not
heard about that transaction before

it will turn around and say get data and
I get data for that hash and

then I will send it the full body of
the transaction and then now that

that node now has the full body of the
transaction that goes with that hash and

this is out transactions and blocks spread
across the network by repeating these

shorter messages about what inventory they
have and then then nodes fetching from

each other inventory that they don't
have and that they haven't fetched yet.

So it's a gossip protocol.

Only valid transactions and
blocks get repeated so

this makes this makes the mesh network
actually very hard to deal with us because

you actually have to insert valid you have
to put valid transactions out there and

if you try to inject something invalid
the network just it just in zero six

send it to your peers and they'll ignore
you and if you're really badly behaved

the disconnect from you and you want
in your messages will propagate through

the network there are also some tests
on transactions is standard tests

that test whether it's a standard formed
transaction there's also some criteria

like you must have a certain fee
attached to the transaction and

if you're below a certain threshold the
nodes won't propagate your transaction so

all of this means that the mesh
network is pretty hard to us.

It's nodes independently by
observation of the network conditions.

Set those thresholds so.

I mean you can look at a block chain and
see like what the prevailing fees

are per kilo by on a transaction and
get a pretty good idea of what kind of fee

you need to attach on a transaction for
to get accepted into the network and

there's a formula that the software
uses to compute you know what.

You know what what is above
the threshold for to propagate.

And you can just.

I mean every node can make that
decision independently but

most of the software is using the same
formula to compute whether or

not it's going to propagate a transaction.

So this means it costs you real money if
you want to just flood the network with

lots of transactions and trying to us it.

So the another way to think about this
mesh network too is there's there's

also there's two policies you can
choose from one one is deployed and

has been deployed for a long time one is.

Kind of getting phased in now.

So if I broadcast a transaction and
then I say I'm going to try to dump.

Broadcast a second transaction
that spins the same coins and

I broadcasted it to different parts of the
network I might be able to fool somebody

into thinking my first transaction
was good when in fact my second one

which sends a money back to me actually
gets confirmed in the bot chain and

parts of the mesh network might think one
is valid in the other part of the mesh

network thing the different one
is valid the other one is valid.

Although the nodes have a first scene
policy which is one policy you can choose

from that policy says that if I see one
transaction and it's valid and then

I see a second one that conflicts with
that first one the second one I discard no

matter what the fee is no matter what you
know what it you know what it looks like.

So that's that's been the way
the mesh network has worked for

a long time a second policy that's being
phased in is a replaced by fee policy so

I can broadcast a transaction if I for

whatever reason decide I'm not happy with
that transaction I want to replace it

with something new I can
broadcast a second transaction but

it has to meet certain criteria it has
to have a higher fee and it can't be

by a just a nominal amount it has to
be by more than just a couple So she's.

And I can broadcast that
second transaction and

the network will then replace the first
transaction with the second one

is very important you know and then this
is set by a flag in the transaction.

You know if you're receiving funds it's
very important for you to know which kind

it is whether it's replaced by fee or
whether it's first first seen.

Because you might be at more risk
to double spend if you accept

a replaced by fee transaction
than the sender can

replace it with a new transaction that
sends money back to themselves and so

you have to be your wallets have
to make you aware of of that.

But there are actually applications that
where this is going to be very useful.

Moving on to the platform imperative
section so that's kind of the rundown of

Bitcoin overview of decline and sort of
sort of its kind of security properties.

In this section I'm trying to talk about
like what's important from a platform

perspective when you're
operating in this network right.

So.

You have this block chain that's out there
that has all this public information but

the private information is stored
on edge devices phones laptops and

those are your private keys that are used
to sign transactions it's your private key

that gives you the ability to create
a transaction that modifies this database

called Big going in and anybody can create
transactions and modify this database

called Big going as long as they have some
transaction out but somebody sent them

some because somebody sent me some
that point I've got a private key that

matches that I can create a transaction
that sends it on to somebody else.

So key handling is really important.

And you know the subject
of a lot of focus of

you know people building
wallets in this ecosystem so

things that are really important are that
the keys get generated on edge devices so

I start up a wallet on my phone the key
needs to get generated on this phone not

generated somewhere else an important
keys need to be stored on this device and

really never to leave this device
in an unencrypted form and

even if it is encrypted you
need to trust the device a.

If you transfer it to because even
an encrypted key could be attacked if

somehow you messed up or the encryption
is somehow flawed or used a weak key.

Generally speaking keys are encrypted
with a two fifty six so.

Devices go try to generate.

You know very strong keys to
perform that encryption and.

Most of all it's today
are generating random seeds to.

Generate keys from.

And then the other important thing is
you know backup needs to be robust

easy to use and safe.

So working on the US issues with you know
getting people walking people through

the process of backing up
keys is really important.

Yeah.

I can probably give you a demo right here.

I can add a new wallet create
a new wallet can you see this.

But this right the middle of
fortune I can't make it any

bigger.

So I just created a wallet it just
generated a brand new you know set of keys

you know using the random number generator
on here that's another important thing for

you to have a good secure
random number generator.

So if I go to receive it's going to
tell me I need to back this up and

if I click on that it's going to
give me the wallet see now since

this is being recorded I'm
going to destroy this while.

After this so it gives you there are
twelve words that you can go right down so

we try to make this as simple as possible
we tell people write this down somewhere.

Put it in a safe.

Keep a couple copies of it P.G.P.

encrypted if you if
you're really paranoid.

And you should be really paranoid.

So that's that's the backup
process right so this.

Twelve words gives you
enough entropy that.

You know it's sufficiently secure.

At least we think so.

And that's that's basically how we
do the back up in this particular

wallet.

Yes except that we most of the industry
is moved to something called H.D.

wallets which stands for
hierarchical deterministic wallets so

what though that public private keep
areas is actually a Ruki Ruki pair from.

It's a route to keep air so from that we
derive other key pairs from that route so

you can actually generate multiple wallets
from that same you know route key pair and

you only have to back it up once right.

The public key is in the wallet publish

it.

Let me get back here.

He wants me to type the words right.

When you click on receive that.

Well when you quit click on receive it's
going to show you a cure after I've done

the back up it's going to show you a Q.R.
code and that your Q.R. code.

Will be a newly generated address and
so you know if you.

Want somebody to send you the Queens
you would have them scan that.

We don't currently share the room public
key between wallets it does share it with

the big course server so the big course
server C'mon monitor the block chain and

see what's relevant to
the wall that's coming in so

that could be a privacy concern you want
to make sure that whatever server has

your route publicly they're going to be
able to see what activities going on and

while they can't spend your funds or
steal your phones but

they can see what's
going on in that wallet.

There's something that we're working
on called reasonable payment codes that

is a code that you can share that
you can publish in the public and

that nobody can discern
from that payment code

any any transactions on the botching
that are related to that payment so

that's something that you could share
publicly you could post it on a website or

in a D.M.'s like system and then.

And then safely know that people can use
that to send you funds or even communicate

with you but that any transactions
created from that reusable payment code.

Can't be reversed back to the payment
code by another observer seeing

the transaction on the block chain that
makes sense so it's a privacy thing.

And questions on that.

Probably not explaining
as well as I could.

Yeah so that's key handling.

Device integrity you want to
make sure that the devices

you know you take steps
to prevent intrusion

I mentioned the secure random number
generator that's that's important.

Trusted execution environments
a number of chips now have these and

we're starting to use those to
do key handling to generate

the key in the Trusted execution
environment and to do the key signing and

in make it so that private key never
leaves that trusted execution environment.

So a number of chips have this
technology ARM chip Intel and were.

Integrating co-pay with those to.

To make sure that those
keys can't be accessed.

You know nothing's perfect but running
in that trust execution environment

the theory is that that environment
can be accessed by viruses and

malware that gets installed on them on a

machine.

To do yeah it is and.

This is also an opportunity to get that
to that in a bit but things that you

would do with your device you might airgap
it so a lot of people when they generate

their private keys what they'll do is
they'll take a device go get it set up and

then disconnected from all networks turn
off the wife I put it in a Farraday cage.

And then generate the generate the key
on that device and then never allowed to

leave that device and then they they're
very careful about how they train and

if they need a transaction signed
they're very careful about and

some software helps you with this
they're very careful about the mechanism

by which they try and transfer
a transaction over to the device so

they can be signed and
then transferred back off of that device.

So air gapping is a good
security technique.

And then level security
I mention that because.

This is a particular interest of mine I
spend a good number of years working on

compilers virtual machines
operating systems languages and

there's been a lot of a lot of research
that's been done a long time ago

on how to build secure operating
system from the ground up and.

The operating systems that we
work with today were never built

with this threat environment in mind and
so

I think there's a real opportunity
actually because we may.

Really create some incentives to go to
create a new operating system that is

built from the ground up to be secure and

then of course you've got to extend
that into the to the hardware level.

Right.

One good mitigation technique is
multi signature transaction so

I showed you co-pay co-pay
was named co-pay to stand for

collaborative payments originally
we called it cosign but we got a.

I don't so I have a trademark on that.

But the idea is.

I can mitigate this device getting
compromised if I have if I

require two signatures or two out of three
signatures on a transaction in a wallet so

I can create a wallet I could share I can
have this computer join this wallet in

a third computer join the wall and now and
I can require two out of three of those

devices to put a signature on
a transaction and so what happens

is we create a transaction proposal
that gets routed to all the wallets and

then independently on each device
you sign that transaction and

only when you've got two out of three
signatures is it actually valid because

transaction that can be broadcast and
put into the bought chain and so

you never have to bring the keys together
physically on the same device so for

an attacker they've got to compromise
two out of three of those machines and

know that you've got a shared wallet on
those machines and how to access them.

So this has been very
effective we see you know.

We're seeing increase and increasing
percentage of all the big goings being

stored in multi signature addresses.

Actually I say multi signature addresses
but they're technically paid a scrip hash

addresses but probably ninety nine percent
of those are multi signature scripts.

Right.

And I should also mention that that's
a very basic form of a smart contract.

Of what.

Yeah a lot of them.

All kinds of attacks.

Famous One was the Mt Gox one thing got
us really concerned in thinking about

this particular vulnerability
the vulnerability is basically now.

Was running an exchange you send your
Because it's the exchange of senior

dollars the exchange and they keep track
of trading and then people can withdraw

now got to set some limits would draw
limits so that you can withdraw like you

know I don't know what it was more than
ten thousand dollars worth of decor or

yeah you can withdraw more than like ten
thousand ten thousand dollars worth of

because no one account could draw more
than ten thousand dollars worth of

declines So somebody got the idea
well if I can break into that system

credit my account with tons of an issue
a market seller I can crash the price to

practically zero in withdraw all the big
coins because ten thousand they would be

worth less than ten thousand dollars
according to their exchange rate and

they broke into the into that system they
changed you know they they did exactly

what I described they credited their
account with a lot of that coins and then

they did a market sell order in Crash the
price on that exchange to like one penny

and fortunately for Mt Gox they.

They didn't they didn't update that limit
in real time like it was a manual setting

that they had to go in and change and so
they got what were basically you know ten

or twenty or thirty thousand dollars I
don't know how much it was exactly but

they got a relatively small amount
that they were successfully able to

draw back off of that system but
they didn't steal all their bitcoins

course Mt Gox had other problems and
collapse later.

So that's what that was one famous hack
there been tons of you know stolen bit

coins people forgetting not putting
good passwords on their systems and

you have one case where in the early days
back in two thousand and eleven private

keys were encrypted so you just had this
file sitting on your desk in the private

keys were in there and they weren't
encrypted I know of one case where.

A wallet was they had
created the wallet before

wallets were encrypted and then they
upgraded and upgraded to a new version

of software that did encrypt the wallets
and what they didn't realize.

Is that they were using a backup
service cloud backup service and

that cloud backup service happen to keep
the very first version of any file and

then the most recent so many files and

then you know if there was a beyond
a certain age they they didn't keep it and

what in the way we kind of
figured this out was that the.

Some I got ahold of an earth that first
back up that had an encrypted private

keys in it some I probably that worked
at that back up company found the wallet

dat file and
they stole a bunch of there because but

not all of them and the ones that they
did and then when we started looking

at the ones they didn't steal were
the ones that were for addresses or

were created after the they upgraded
to the encrypted version of it and

so the big point I happen to be in those
addresses that were encrypted were

the keys were encrypted they were fine but
the other ones were stolen.

Yes.

Yeah we've got much better security I
haven't heard of like major theft you hear

that most recently it's been social
engineering type of attacks where

some matrix somebody in
the spending money one way or

another we have one case where we found
that there was a tour exit node that

had compromised somebody had compromised
one of our wild card certificates for

a domain which we normally have the Eevee
certificate at that pay dot com But

we also had a wild card certificate and
we've stopped using them since this but

we had a wild card certificate it got
compromised somehow we don't know.

And and so
I tore exit node was mimicking and

making it look like you had
a secure connection to bit pay but

it wasn't the easy certificate is
the Wild Card certificate and then they

were rewriting the addresses on the
invoice to be somebody else's addresses so

they were tricking people that were using
Tor to access a bit pay you know invoice.

Tricking you know tricking those
users into paying the wrong address.

Trying to think.

Which I think of some more interesting

compromises that I've heard
of related to Bickel and.

All right another aspect of security
software disagree distribution this is

something we're focused
on a lot right now.

You know the software would not work
if it wasn't open source because

people look at the code they can read
it they can make sure does what it says

it's doing it I would never trust
a closed source while it's right

because who knows what that wall is doing
with your private keys and you know.

So the first criteria sort of the first
security mitigation is you know make

sure using software that's open source
an audit of all can be peer reviewed

get is you know a great tool I was a fan
of wrote a version of whole system in

the past and about a week after Get was
released I was like This is fantastic but

it's great because it's content based
right it's using hashes of code so

the one bit in that source code is
changed you've got a different hash so

you can verify when you do a clone of
a get repo you can verify that you're

getting exactly what that
what you're supposed to get.

An ancient source code distribution
where possible it is one of

the things that is the scariest thing to
me is that we ship so much precompiled X.

eighty six code around the Internet and
really.

If you compare that with like say web
pages which transmit javascript and

H.T.M.L..

When I download javascript and I execute
it it's only possible to do what that

language you know that whatever
the semantics of that language give you

the ability to do and then that browsers
of course can sandbox that but shipping X.

eighty six code that
somebody downloads and

installs could be doing anything on
your computer and so that's a big risk.

And it's a big risk on devices not
just not just the software that you're

downloading to handle because but any
other software that you might install so

you have to be really careful about
what software you're installing.

On a device that you're also
using because when with.

Deterministic builds.

So when it comes to distribute
distributing software it's not always

convenient to distribute source code and

have somebody wait twenty minutes
while that calling compiles.

And.
To have to install the tool chain and

everything to actually compile it.

So binary to get distributed people do
builds and they distribute the binary.

But it would be nice if you could to say
take this get get commit or this shot hash

of the source code and then you could take
the shot hash of the house by an Aries and

I could run a build and verify that
the software that's being distributed that

shot I could verify independently that
my build produce the exact same results

on software no running that build on
software that I control on a on a machine

that I control and then I can put my
G.P.G. signature on that shot and

that's basically me saying to anybody who
cares I've independently verified that

this shot hash of these binary
is derived from this shot hash

of the Get repast atory following
this process and if enough people.

Verify that that independently
verify that build and

put their signatures on it then
when I go to download it or

when somebody goes to download it chances
are maybe there's somebody they know and

that they trust it's but their signature
on it that says that they verified that

that Bill has is correct and then they
can feel they can trust that binary

that that my narry doesn't have
any nefarious stuff and it.

Turns out this is really hard.

It's hard to do deterministic builds you
know in Linux environment it's hard for

me to say Here take this.

Image of a Linux distribution and
run the compiler and

produce an output there are time
stamps embedded in those builds there

are all kind of stuff imbedded in those
builds that make it non-deterministic So

you need a deterministic build process and

there's actually Debbie in Project
working on this now it's called the.

Reproducible builds project so it's really
important and they're trying to work

backwards and get all the way to the point
where you can deterministically build

you know the Linux kernel and
everything the toolchain everything and

then there's also a project called Gideon.

Project is using this today and it's
trying to deal with some of those trying

to fix all those places where things like
timestamps creep into a build process so

you can have a deterministic build.

Without that were left trusting that
some person that's put their G.P.S.

signature on a building that they
didn't do something nefarious and

put something in there and it's like
one signature or a group's signature

with a deterministic build anybody
can independently run the build and

verify the Shah some and
put their signature on it.

So deterministic builds is
important software distribution and

how you manage software
distribution is really important.

I mentioned using the P.G.P.
web of trust to manage signing or

verifying of minorities.

All right.

Yeah I've talked to some about this on
this get through a couple slides here

because we're running short on time.

Energy I'm going to I'm going to just
say one thing about energy consumption I

always get asked the question are we you
know like destroying the planet by running

all these miners consuming all this
energy in the The short answer is No And

the reason is the the relationship
of money to energy and

think about it in these terms if you
have two people that are bartering for

two products that they each have to build.

And you don't have money you are going
to have to build both products and

both then that takes energy and

consumption if on the other hand
one of those people except money.

Money allows them to defer
consumption into the future so

if they accept money instead of them
building a product that they bought barter

with you're using half of the electricity
or half of the energy for those products

so the relationship of money and energy is
really to me if that's a fascinating thing

then I think enough people of that I think
there needs to be some more thought or

you know sort of philosophy you
know philosophical thought about.

Database technology as a database.

As we think about Bit pay and

we think about solving that Mt Gox problem
at that pay we start to look more and

more of what that pay
does as a block chain and

that watching having different kinds of
transactions with the sides money monetary

transactions the transactions where
merchant updates their account settings

you could instead of that being an A.P.I.
call I can make that a transaction that

the that the customer creates on their
device signs with a private key on their

device and some it's to bit pay and
we incorporated into our chain of history.

Now you need to secure that chain of
history so when we do it is run minors but

we're never going to have the kind of
computing power that the big point at work

is going to have another way we can do
is we can take all those transactions in

periodic lead time stamp them in the big
chain and so we have the pay has its own

set of consensus rules we run a cloud of
computers that are building this chain

of transactions periodically ten times
stamping it into the coin box chain and if

it's hacker gets into our system well they
might be able to see a lot of activity and

maybe they could even change recent
activity anything that's been timestamped

against the big chain they really
are going to be able to change because

they would need cryptographic keys to
create those transactions to forge

those transactions and then we could boot
up a brand new node and independently

synchronize with the big point block chain
and with our own that pay bar chain and

verifying the entire the integrity of
the history from the very beginning and

independently and
trust us Lee repeat replay and Rick.

Cover that entire history so I'm
excited about that because I think it's

kind of the A game changer in terms of
how you think about protecting a company

you know or any kind of you know database
system no longer are you so dependent.

You know on boundary defenses you still
want to keep people out of your systems

but if they do get in and
they in in evitable you will

you're kind of mitigating or
limiting the damage they can do.

The other thing is it's really good
from a proper performance perspective.

To pay we use a database as a primary
secondary kind of architecture and

what that means is that you know your
transaction log is getting created on

the primary placated out to
the secondaries and there's automatic fail

over and all that kind of stuff but it
means that if somebody accessing our site

in Australia there's a round trip that has
to happen back to the primary to get that

thing into the transaction log on the
primary but if instead we architected like

that corn itself that means that we
replicate events all over the world so

that we can have a replica of the Bit
pay state sitting in Australia so

that the customer in Australia only has
to hit that server in Australia and

then we use consensus test like techniques
very similar to bit coing to resolve any

kind of conflicts that could occur among
those independent replicas so from

a performance perspective we're really
interested in moving to this architecture.

And really I think it's a superior
model to locking in transactions and

databases.

I mention A P I's again real quickly.

You know if a merchant was going to change
the state of something in pay rather than

call and make an A.P.I. call and use a
shared secret they can instead have a key

pair of their own and cryptographically
sign a transaction that's modifying

the state of whatever they do and pay and
submit that transaction to bit pay and

we incorporated into the botching that is
our A.P.I. So I think that's again change.

They are as far as software as a service.

So I would expect you know
companies to start moving over

to that kind of an architecture rather
than a traditional you know restful A.P.I.

type of architecture.

And we have an internal project called
chain to be working on a lot of this we

haven't fully moved over to this
architecture yet but we're working on it.

And there's tremendous opportunity
in terms of identity management

because when is an identity system
those private keys are stored and

while it's our identities This is
how other other nodes on the network

verified that you know the person
signing this transaction was the person

that was originally sent the because in
the first place that that's their identity

that private key is their identity and so
when we think about identity of a human

I think about it like let's say I
have a private key on a device and

it's encrypted with a password that
only I know that I've memorized.

I think of a human's identity as being
the particular unique composition of

neural connections in your brain and
in a very real sense your

memorization of that password is forming
those neural connections in the brain and

that password gives you the ability to
access that private key in your device and

inside transactions so I think of it like
my identity extending into the cyber

you know realm by the devices
that I directly use.

So that's the way we think of our
wallet of co-pay is really this

the primary function of co-pay is to
securely store and manage these private

keys that are used to sign things and
transactions happen to be one but

other kinds of transactions may be
interactions with the bit pay service or

any kind of software as a service might
be other than they're all kind of

transactions that you could
sign with those keys.

I mentioned the unit.

Cation of P.K.I. and web of trust.

Really P.K.I. in the D.N.A.
system and web of trust or

really the same thing if you look at
the root CA's there really concentrations

of trust so a lot of people trust those
CA's you could accommodate that model in

a in a fully in a web of trust
system where you know it's.

You know.

How what.

By by what to cover the.

Same way you tax any
other transaction and.

Not sure I understand the question.

If you're a minor Yeah if you're OK so
minor you have costs.

Sorry.

The question is how how are minors
taxed or how they treated.

So taxation of minors.

You know it's actually different
in different countries but but

very simply there it's a business
that earns revenue and

that revenue has you know they are big but
those big ones have a market value so

you have capital expenses you pay however
much you pay for mining equipment and

then you earn revenue as you run that and
you have electricity costs and so

on so it's just a business that are in the
revenue it's not really it's not really

any different from any other business.

I guess of.

Yeah of course not.

But that's true of any anonymous business.

And then And
then once you when you are in decline.

I've.

Talked to lots of mining companies big
you know big professional operations and

the best way to model it from a financial
perspective is you you recognize that

revenue as soon as you are in it so soon
as you find that block you're recognizing

that revenue at the current market value
of declining and if you want to keep that

that coin you put that into another
company where it's an investment company

and so that those big points transfer
to that investment company and

you hold that you have either capital
gains or losses write a lot of mine or

sell everything they just sell every
single bit coin that they get immediately

they've got to pay suppliers they've
got to pay the electricity costs and

they don't speculate at all
on the value of Bitcoin.

It's.

Now.

Used by a miner and
start running it out of your house.

You know I think I'm not a tax expert but
you know the I.R.S. does have

something called like hobby income that I
mean if you're if you're making less than

a certain amount they just don't
care I mean it's just hobby income.

But once you get to a certain scale
then you've got to start reporting and

what not.

I know it's hard to know
that I know some miners

they just they they don't want to
speculate in the degree to which they

do speculate in because it is independent
of their being business right so

they they probably are speculators
in their calling but

they're not necessarily viewing their
mining offer and you really shouldn't view

your mining operation you can conflate
the volatility or the appreciation in

value of big coin with the profitability
of your mining operation right because

you're you're buying equipment you're
paying cost for all of that you're paying

electricity you're getting rewarded
with coins that have a market value.

You there's your revenue your
profitability right there take

the speculation of the value of big oil
and completely out of it or separate it.

Because a lot of times people get into
mining thinking that they can make all

these big point but
they realize that electricity isn't free.

And they and they may be better off
actually speculating in the in the price

of that coin that was a case of me in the
early days in two thousand and eleven and

I was doing G.P.U. mining and it was fun
and I I was able to find some declines but

I actually would have been better off
had I just taken all that money and

bought declawing.

Let's see.

Last slide next to last slide platforms.

So big ups the ante from a security
perspective I've talked a lot of

security reaches researchers and they they
actually are excited about that coming

from the perspective of you know this
is just yet another reason to really

you know it creates economic incentives
direct economic incentives for

building more secure you know platforms.

You know we see new threats emerging
things like crypto locker you've probably

heard of the encrypt people's
you know devices and

charge declines for
them to get them unlocked.

The credit card for this is a credit
card fraud is really interesting.

So I in the early days a lot of people had
the idea hey I can set up a website and

I can sell because I can take credit
card for those big Koreans or

can take a pay pal transaction and then
they put up the website and all this and

they get all the stolen Pay Pal accounts
all the stolen credit cards come in their

website trying to buy coins and so
they very quickly realize they're going to

be out there because and out there you
know Visa or Master Card transactions but

it's an interesting case where you have a
more secure environment fundamentally more

secure platform mixing with
less secure system and

the more secure system is devouring the
less secure system in a very direct way

so you see you know a lot of credit card
fraud a lot of bank account hijacks there.

Instead of using cash and
runners their start to use because when to

get the value out of those find ways of
getting value out of those so that this

is why it's very difficult to buy big ones
all the exchanges are very you know put

a lot of fraud mitigation in place because
you know if they get a fraudulent a C.-H.

transfer or whatever then they could
they could lose a lot of money

in a very short amount of time.

So that that's kind of an impediment
to adoption when it's so

hard to actually buy an acquired calling.

Yet I mentioned you know they calling
creates economic incentives for

secure platforms I think we'll
see lots of new security tools

sort of being designed you
know with protecting and

private keys come about I
mentioned new operating systems

you know linux you know these are all
forty plus year old operating systems

again operating in a are built in
a in an environment where the current

threat environment was not
really you know top of mind.

In the new hardware ultimately you can
do all that all the work in the world to

create secure operating systems
from the ground up but don't

forget about the hardware right we really
kind of need open source hardware and

maybe one day we'll be able to manufacture
own chips you know three D. printing maybe

that will twenty years from now who knows
maybe some maybe people can independently

build their own hardware according
to specs that have been vetted and

verified and not have to trust Intel or
A.M.D. or chip manufacturers.

And I mention that a lot of
the research on it particularly on

operating systems and languages
that get first class treatment to

to security a lot of that research has
been done and done for a long time.

But they're just hasn't been the economic
incentives or the right alignment of

circumstances for those operating systems
to kind of get off the ground but I'm

optimistic that will we'll see movement
in that area in the next decade or two.

I didn't really talk too much about
distributed autonomous corporations or

smart contracts prediction markets but
these are all very exciting

you know areas for the future not
directly related to security but.

Cool stuff in it needless to say
prediction markets are going to be

a really controversial thing prediction
markets are where somebody could say

make a bet on the outcome of the next
election and they can place bets on either

side of that and then the market will give
you odds on who's going to get elected.

And the theory behind a prediction
market is that you can bring

together the collective wisdom of the
market to make predictions about what's

going to happen in the future but there
are downsides to I mean you can you know

an executive at a company if they can make
an these anonymous bets they could do

effectively insider trading by betting
on the future price of their stock.

So there's downsides to it but there's
an interesting side effect I mentioned

earlier that time stamps are the only
external data in the blocking the only

data that's outside of Bitcoin that's in
being embedded in the botching prediction

markets give you a way to actually get
external real world data into the bog

chain in a trust lists in verifiable
manner and so if I can get so

if I predict who's going to win the next
election I create a market election

happens then you have this verification
period and then then actual the bets

get settled but
that verifiable independently

verifiable data of who won that election
gets embedded into a block chain and

then you can use it in smart contracts and
I can build a smart contract that says if

you know Hillary when's the next election
then pay this person this amount of money

and if somebody something else
happens to do X. so getting extra late

a story into the bot chain is where I see
what's really exciting about predictions.

It's big and they're going to be.

I mean if you thought that point was
controversial I think prediction markets

are going to be another
level of controversy.

But exciting at the same time.

Yeah but how do you know they aren't
gaming it you know how do you how do you

yeah they definitely are I guess
I guess you there's no question.

Yeah.

Yeah.

Yeah.

Yeah I mean there's a project called Truth
coin where there's a lot of papers written

on this subject that goes into all
those details about you know how

how do you form the market Saudi you.

Form the bets in a block chain How
do you then once the event happens

how do you verify it.

But it's using sort of crowdsourcing
of information to get data into

a block chain once it's in a blog chain
it's in a form that you can use in a smart

contract which is again
the most exciting part of it.

So all right that's all I had any other

questions.

We have

to.

Yeah so the question is about
the block size and decentralization So

there's a lot of debate about this issue a
lot a lot of you know things to consider.

The main concern that people have
is if you increase the if you.

If you produce really big blocks then.

It'll be more expensive for an individual
to fully verify the block chain so

to run their own full no that's doing
full verification trust verification and

if that's the case fewer people will do
that trust us verification will start

trusting a big server or a company that
does the verification for them right and

then that that harms decentralisation.

So the miners aren't against it
the Chinese miners are against

going too far mostly miners are actually
in favor of increasing the block

size they want more throughput they want
more people to be able to use the system

they see more throughput
as being more customers.

But.

In China they have the fire wall and
there's a concern about network

propagation in it increasing their orphan
rates that I talked about so they create

a block and they can't get it through the
firewall fast enough to the rest of world

and then they have a war from blogs so
they most of them now are in favor of

going up to two megabytes of basically
doubling the capacity of the network but

they don't want to go to like eight or
ten or twenty as have been proposed.

So there's lots of things to consider we
put out a proposal about an adaptive block

size so the original reason for the box
sizes and make sure it wasn't artificially

kept the capacity the system is actually
to prevent us on the network so

it was to prevent a minor from
creating a really gigantic block and

by the way this wasn't in the very
first you know iteration of because.

It was added you know a few months
after because it was actually launched.

Put it in there kind of quietly
didn't really say much about it.

But it was it was to prevent like people
from creating really huge blocks and D. O.

S. in the network but we put out
a proposal called an adaptive block size

the like uses a recent median block
size to enforce a limit so that if

you look back over the last three months
of block size you pick the median and then

that becomes your and then you double that
and then that becomes your cap on blogs.

Rises we think that has the same effect
and it allows the miners to put keep

producing bigger and bigger box
until they start hitting up against

real scale ability limits once they
start seeing increased orphan rates and

real scalability constraints
in the back off and

you'll it allow the network to grow
along with innovations in scale

ability.

What you want to know them.

Trying to remember how many bits of
interplay it is I want to say it's like

one hundred twenty eight bits of
entropy would you say how many.

Yeah OK maybe more each one of those
where there's a there's a pre-selected

pool of words to choose from and so the.

I don't know it's a it's a standard
it's not something we invented.

And a lot of wallets
are using this today but

basically the idea is you create a random
number and then you turn it into these

twelve words that people can easily write
down and you pick words that can easily be

confused with each other in
the order matters of those words.

So.

It's just meant to be.

It's used to derive
the private key the E.C.C.

private key so if you have that you
can derive the private key from it.

That's why I'm going to delete
that wallet after this talk.

Now.

Thanks.