(gentle music)
- Good evening everyone, thanks for being here.
So let me quickly take you through
the core technology of Algorand.
So why are we here?
Because blockchains have a great promise.
Who doesn't like a tamper-proof record?
Who doesn't like transparency?
Who doesn't like generating trust among people
who barely know each other?
The application and use cases are essentially unlimited.
However, there is a little bit of an open secret.
Blockchains, as defined, don't quite exist.
Because they've been very much aspirational
and very good for humanity to raise our aspiration
to raise the bar.
But if technology does not allow us
to achieve our aspirations, we are kind of in a bad place.
...but every sociologist says,
every anthropologist says, decentralization is here,
decentralization is here.
Decentralization is not yet here at all.
And in fact
if you look at Ethereum's co-founder pronouncement
the famous Blockchain Trilemma,
analyzing the evidence of 2,000+ blockchain projects
what they've figured out is to say
"Hey, it ought to be a rule
"that in a blockchain you can, at most,
"satisfy two of the following three properties:
security, decentralization, scalability."
You choose which one to exclude.
It's like saying, welcome to blockchain by design,
what property don't you want to have?
Security, excellent choice.
Really?
If you don't have security you have nothing.
So how about decentralization.
If you don't like decentralization
why are you building a blockchain?
And scalability, if you don't care about scalability
are you building a blockchain between friends and family?
There are no good choices here, it's like to say
you prefer to be shot in the left knee
or in the right knee, okay?
No good. Well, ladies and gentlemen, let me tell you
The Blockchain Trilemma is not acceptable
and more importantly is actually false.
So what is instead the truth?
The truth is that
blockchains are very technological products
and we need better technology
to realize our aspiration, okay?
So, what is the challenge?
There are two aspects of a blockchain.
Make sure that things are concatenated well
and guess what, at this point, click, check mark,
we already know how to do it.
You take the hash of a previous block
you make it a part of the current block
and nobody can tamper with the chain anymore.
Everybody does that.
But the difficulty is that who chooses the next block?
Because when things are decentralized
and transactions are actually propagated
throughout the network,
what you think ought to be the next page
and what I think is going to be the next page
in this ledger is quite different.
So who has the right to choose the next block?
So there are popular prior approaches.
Right, the oldest one is proof of work, you know,
really a great idea at the time.
And hat off to Mr and Mrs Nakamoto,
or the Nakamotos, how many there are.
And what is the idea?
We create a competition, a computational competition.
Try to solve a crypto riddle,
the first one who arrives there
has the right to append the block.
What are the cons?
Extreme expenditure
and waste of resources.
That is bad.
On top of it, these things are slow.
They produce a block every ten minutes, why?
Because if you tried to make the riddle softer
so that you can solve the riddle once a minute
then the chance that two people solve it
a few seconds of each other increases so much
that you are going to fork and fork and fork all the day.
So personally I understand expensive and fast.
Expensive and slow is harder to understand.
And on top of it, the expenditures needed,
the capital needed for buying the mining equipment
is such that very few people,
fewer and fewer people participate.
And right now, Bitcoin's blockchain
is actually dominated by just three mining pools.
You know what, slow, expensive and actually
de facto centralized
perhaps we need a better idea.
So delegated proof of stake what is this?
Very simple idea, we just put in charge, say,
21 people who look honest, look how honest they look.
Perhaps it will remain honest, they will choose the block
on behalf of all of us.
Is this decentralized? No!
Okay, because even if the people are actually very honest
and remain honest, a denial-of-service attack
against 21 people, you can denial-of-service
even to 1,000 people.
That doesn't look too promising to me.
Okay, next stop, bonded proof of stake.
What is this?
Oh this is another simple idea, you say,
you push some money in the middle of a table
where you cannot touch it and anybody can put
their money in the middle of the table
and the people who willingly put it there, hostage,
are the ones who choose the block on behalf of all of us.
And if they misbehave their money's confiscated.
Wow, this should work, right?
Does it?
Let me ask a simpler question.
How much of your disposable income can you afford to put
in the middle of a table, hostage.
And the answer is a very small fraction.
So in a system like this not only
do you make it possible but you make it easy
for big thieves with deep pockets
to put a disproportionate amount of money
in the middle of a table for the sole purpose
of controlling the blockchain.
But so what?
If they misbehave their money's confiscated.
Let me tell you, speaking about the borderless economy.
The borderless economy should secure
trillions of dollars, okay.
A very secure, decentralized, scalable blockchain
has trillions of assets under management.
And so by misbehaving you can make a billion dollars
or two, very easily.
So do you think these guys care if they confiscate
over 10 or 100 million dollars?
No, that's just the price of doing business.
The cost to do business, right?
So in other words, if you look at all these prior approaches
they have a strange logic.
The logic goes like this.
The whole economy is secure if the majority of the members
of this small corner of the economy are honest.
Who is this small corner?
In Bitcoin, they're miners, right.
In delegated proof of stake, they're 21 delegates.
So Algorand takes a quite different logic.
The whole economy is secure and works
if the majority of the members of the economy are honest
and that is where you want to go, right?
Always dangerous to put a big thing in very few hands
who don't really have a lot of money in the system.
So let me tell you what we are based on.
It is Pure Proof of Stake.
Many versions of proof of stake, but that's our own brand
Pure Proof of Stake.
What does it mean?
That first of all, we don't punish anybody, why?
Because we believe that making cheating impossible
is way better than pursuing somebody,
and imposing a fine on somebody who is running
to the Bahamas with the goods, that's a bad idea.
So where is the money in Algorand?
It's always where it should be.
In your wallet, at your fingertips, ready to be spent,
invested in the various financial tools
that blockchain offers you.
And if you consider all this money,
wherever it happens to be, if the majority of the money
is in honest hands the system is secure, period, okay.
So secure against whom?
Against a bad adversary.
How bad? Very bad!
So this guy is really scary because...
the system remains secure even though the adversary
can immediately corrupt any players he wants,
instantaneously, and can perfectly
coordinate the players.
And by the way, he can attack not only the protocol
but also the communication network
on which the protocol is run.
What does it mean?
People when they discuss the security most of the time
they refer to: are you sending the messages
that the protocol tells you to send?
If I'm an adversary, no I'm not.
I have better things to do.
But as an adversary you have all kinds
of other things to do.
How about attacking the routers?
Disconnecting the network?
And in systems like Bitcoin you can actually collapse
the system if you attack the network.
So we are going to remain secure
against an adversary who can attack
the protocol and the network.
Do we need to be so adversarial or is it me
because I'm a cryptographer?
Well as a cryptographer I'm actually a realist
because I remember if something
is worth a trillion dollars and up
then it's going to be attacked.
In fact, attackers are going to come up
as mushrooms after the rain, right.
So we must be really prepared.
So how we achieve all these protocols,
decentralization, scalability, security,
essentially is the following:
In Algorand blockchain, the blockchain is grown
by a sequence of ever-changing committees.
In fact, each committee, think of it like:
1,000 randomly and independently selected users.
And what do these users do?
Oh, they do a simple thing, they utter a single message,
short and very easy to compute.
And then another different 1,000 people come over, right.
That's the way it works.
So, well you say, "Gee, Silvio, this high level
description, I've got a few questions."
A few?
You should have many questions but we have time for one.
The most popular one which is: Who selects the committee?
Good question.
Let's assume that my answer is, "I do."
Then you say, "That's the most decentralized blockchain
I ever saw, and you are at the center of it."
That's not what we do.
How about if I tell you "well, humanity gets together
we debate and together select the 1,000 people committee
who then..." really?
The humanity cannot agree on anything at all
let alone 1,000 people committee.
So what do we do?
In Algorand we do something a little bit counter-intuitive.
The committee members select themselves.
They go, "What?
"That is a terrible idea!
In fact, it's the first-worst idea
or maybe the second-worst idea ever."
So, because if I'm a bad guy
of course I will select myself
for this committee and the next and the one after this
and so on so forth.
But not so fast.
Why?
Because in order to select yourself
you must actually win an individual lottery
disconnected from the rest of the world
in the privacy of your computer.
So you just run the lottery to see
am I part of the next committee.
In case one, if you win, you obtain a winning ticket.
A short proof that everybody can verify
you are part of this committee.
And if you don't win you can say anything you want
nobody pays attention, okay.
So that is the idea.
This cryptographic lottery is such
that not even a nation state
with huge computational resources can enhance even a bit,
a little bit, the probability of being selected, okay?
So essentially if you want to have
a million users initially, you want to select 1,000.
The lottery sets the threshold of winning
at one in a thousand
because a million divided a thousand is a thousand.
Well later on we can have billions, see if I care.
Automatically the system selects the threshold to be
one in a million because a billion divided a million
is a thousand.
That's the way it goes, right.
All right, and of course, you know
the probability of winning is proportional
to the total amount of money that you have
because otherwise, right, I can distribute
the Silvio key into a million Silvio keys,
Silvio one, Silvio two, and if anyone wins I win.
But in Algorand, if you have a million Algos,
whether you keep them in one key
or one Algo for a million keys the probability
of being selected is absolutely identical.
So no simple attacks.
Now, let me tell tell you how we
make it super decentralized.
Of course, it's super decentralized
because we are not to have a fixed committee
that runs our affair for a month or for a week
or for a day or for an hour.
Not even a minute because in a minute
a denial of service attack catches you very fast.
So your turn, a thousand, then another thousand people
next time and another thousand people next time.
And they only stay in power for one message only.
Okay, check, that is decentralized.
Now let me argue that this system is super scalable.
How long does it take to do the lottery
to figure out if you're a member?
One microsecond, it's very fast.
Okay, but now that you have these thousand people selected
in a microsecond what do they have to do?
Well, all they do is propagate a single short message.
Can we do that? Yes!
That's our network, it's a piece of cake.
So that's why it scales, right.
And so the important thing is that
even with billions of users the committee never needs
to be more than a thousand, but it changes all the time.
And now let me argue finally
why the system is super secure.
Assume that I am the big, scary guy
that I just showed you, right.
I can corrupt anybody I want, right.
Very, very quickly, instantaneously even.
But I have a problem: Whom should I corrupt?
Should I corrupt you or you or this lady down the street?
Or this guy in Paris?
I don't know. Why?
Because only you know if you have won your own lottery.
I don't! You run the lottery inside your own machine.
Right, so that's the whole idea.
But if you win you propagate your winning ticket
and your message, right?
It's an up/down message, think of it that way.
At this point I know you are, and what am I going to do?
I'm going to zap you right away,
to corrupt you instantaneously.
But so what?
Whatever you had to say, you already said it.
And it's virally propagating throughout the network.
And I cannot put it back into the bottle no more
than the US government can put back in the bottle
a message virally propagated by WikiLeaks.
In other words Algorand is secure because beforehand
you have no idea whom you should corrupt;
and exposed, once you know, it's too late to corrupt them.
And in some sense, this is not a mathematical proof
but that is the idea that was missing
to actually solve the trilemma.
Yet to do sometimes the right and straight path
is a tortuous one.
Okay, so let me tell you in action what happens.
The difficulties of the blockchain is to select
the next block, right, that's what I said.
But one block is not hard to select it at all
which is the genesis block.
Block number one, because it's part of
the very definition of the system.
Next to the block I want you to observe a feather,
a universal symbol of lightness and effortlessness.
And, as this feather effortlessly falls to the ground
the Algorand blockchain unfolds.
Well, isn't this too simple?
How about soft forks?
How about proof of work?
Guess what?
In Algorand there are no soft forks.
There is no proof of work.
And these are great advantages. Why?
Because the absence of soft forks,
the fact that Algorand chain never forks,
gives you transaction finality.
You can consider yourself paid if the payment made to you
appears in the block. As soon as the block appears,
ship the goods.
Because this block is never going to leave the blockchain.
And to have such effortlessness means no proof of work.
Anybody can participate, in fact we want you to participate.
That is the path of Algorand.
And so this is the core. And then let me tell you
that one reason to have a great team
is that we have a very deep roadmap.
We are going to unfold all kinds of things
in addition to the core.
And here there are some of them.
One thing I'd like to highlight and maybe two.
One is Dutch Auctions.
What does this mean?
We have developed the network
through equity fundraising, not crowdfunding.
Now, the network is ready, once we launch it
we are going to make our tokens available to you
by a sequence of Dutch Auctions. Why?
Because, assume the typical project: What do they do?
They fix a price, right?
One token, I'm making it up, $2, okay?
You're going to ask, "Sorry, is this a fair price?"
What? This is the wrong question.
So how many tokens do you want?
or get out of here, right?
In fact, in a Dutch Auction what is going to happen
is that you may decide the price for the Algo token, not us.
So, we want this to be...you
consider yourself fairly treated
all the time you are in Algorand.
Starting from your first entering the Algorand
ecosystem when you buy your first token, okay.
And by the way, these Dutch Auctions
have been used now for a long time.
Who uses them?
Governments to offer their bonds right?
But to whom?
To the usual suspects, Goldman Sachs, Bank of America.
And what do they do?
They buy them and then they resell at a higher price
to all of us, right.
So here instead the bids are on chain auction
it's not in a smoky room, right, so you can
see the bids of yours and the bids of everybody else.
You know that you are paying the right price,
you know that you fairly won this number of tokens
and you directly participated,
these are massive online, on chain
public conducted auctions.
And if we offer this tool for selling
our tokens you can use it to sell your building.
You can use it to sell whatever assets you want.
And just think about: if you are a builder
and you have an office building and you allow somebody
from far away through a Dutch Auction to buy just enough
to put the equivalent of $10,000
to get a very small piece of a building.
I never had this opportunity to build an office building
in Shanghai and vice versa.
And vice versa the builder never had such an opportunity
to cast such a wide net to really realize the value
of whatever he or she has built.
And the other thing I want to tell you is self-governance.
Which is really the missing ingredient
the missing ingredient in most blockchains. Why?
Because we do not know
what our needs are going to be tomorrow.
We barely know what we need today.
But tomorrow who knows?
If you look at the way Algorand protocol works
is that you reach agreement, consensus
on each block by block.
But the mechanism we use to reach consensus on each block
is the same mechanism that can be used
to reach agreement on a new monitory policy.
Or a new rule in the protocol.
And so we have the ability to evolve in a consensual manner.
And I really believe that being alive is
essentially about intelligent adaptation.
That's what really is in my opinion
is the number one property of all this.
All these tokens together are
smart, smart contracts, why smart squared?
Because smart contracts are not that smart.
We cannot have an ICO and a crypto key at the same time.
How smart they can be?
So you need all of these things
to really realize what is our dream
and I hope your dream: To have a borderless economy.
So in conclusion we should be really thankful
for all the great projects out there.
They've picked up a very high challenge.
But some of these blockchains
will not to be a win for all of us.
Some of them are going to stay around
and prove useful for a while too.
But only true technology will realize our aspirations.
Speaking of technology here is some technology.
Very old technology...but for 2000 years
the Pont du Gard (an old Roman bridge in Southern France)
has enabled occupants of different sides of a river
to transact and meet each other.
It has been in continuous operation 2000+ years.
That's what technology can do to bring us together.
I believe that the blockchain,
the permission blockchain, is going to be as useful
to cement and bridge each other as humanity.
And as beautiful as any other physical infrastructure
that we ever built.
And it will last for a long time too.
So ladies and gentlemen lets get together
the right blockchain
and enjoy it for a long time to come. Thanks.
(crowd clapping)
Không có nhận xét nào:
Đăng nhận xét