DFINITY’s Dominic Williams’ Keynote at “Hashed Night” Korea Blockchain Week


Okay. So that actually doesn’t wrap it up. We actually have a one last special guest
here. You guys all heard of this project – it’s
called DFINITY. We have Dominic Williams, who is the chief
scientist there. I think it’s his first time in Korea. So he’s a little bit hangover from last night;
we had a little too much fun. So go easy on him but he’s going to be just
explaining at a high level what he’s working on and some of the problems he can be solving
here. Thank you. Dominic. Thank you, Alex. Thank you for organizing this conference and
inviting me and thank you for reminding everyone that I’m a bit hangover. So to make it more difficult for myself, I’m
going to just sort of acappella talk on why I think decentralisation is valuable, specifically
trying to give some pointers to how you can generate value from this kind of technology. Before I start with that, I’ll briefly introduce
DFINITY in case you haven’t heard of it. DFINITY is developing what we call the Internet
Computer. So that’s a blockchain supercomputer and
our mission is to host the next generation of the world’s software and data. On this slide here, you can see a photograph
taken in the Palo Alto Research Lab a few weeks back, it’s the DFINITY test network. It’s a kind of interesting thing, the testnet now has got 400 nodes but what we’re able to do is to combine computers distributed around
the Internet to create a single supercomputer the world can share. That provides all of the properties you’d
expect from a blockchain to the software that runs on it. Software can be written in a thing called
WebAssembly or that’s a WebAssembly virtual machine. So you can write the software in a variety
of languages. It has some special properties – it’s very
fast, it will produce finality in a couple of seconds even though it uses probabilistic
protocols, it’s very resistant to attacks. Of course, as you probably guess from the
name DFINITY, it’s designed to scale out as capacity as necessary because there are a
lot of software and data in the world. Anyway, I’ve been in this industry for a long
time. Actually the DFINITY project came out of another
project that I worked on through 2014 trying to speed up blockchain protocols by repurposing
traditional consensus protocols. Beginning of 2015, I had to abandon all that
work. I found a different way of doing things. So, DFINITY has been cooking since early 2015. In that time, I’ve met a lot of people and
thought a lot about network protocols and also applications that can run on top of these
protocols. Actually in Silicon Valley, there’s really
big focus on a thing called product market fit. Product market fit refers to the ability of
a product or service or indeed a decentralized platform to meet the needs of users and to
deliver value in some way that persuades people to use your service or product. So I want to go on a sort of a lightening
tour of decentralized value starting with the protocol itself, what’s valuable in a
decentralized protocol. So for us, when we design the DFINITY Network
we are very keen that the nodes in the network are all equal and that even if we have as
our ambition stands fifty to a hundred million computers in the network, each of those millions
of computers in the network plays a completely equal role. So, there are no single points of failure
and the reason for that is that it makes the platform more robust. The objective of the Internet Computer is
to host the next generation of the world’s software and data. Today the whole world runs on software. Imagine if we succeed and the whole world
did start building on DFINITY and then the DFINITY network – the Internet Computer
– stopped working, it would be a catastrophe. So we have to design a protocol that doesn’t
depend upon a small number of special computers. This is one of the ways that we deliver value;
we created a symmetric network that is extremely robust. I think the reason this is pertinent to think
about today is that a lot of networks have a different design. They’ve forgotten what decentralization means. Decentralization was invented by the designers
of the Internet to create communications network that would continue to function in the event
of a nuclear war. Even if there was a nuclear war they wanted
people to be able to continue communicating. So, the Internet was designed with packet
switching such that no central points of control such that data could be routed across any
functional path in the network. Since then, it is a huge success. We’ll come back to some other properties of
the Internet and decentralisation that made that possible, but certainly when you’re designing
a network protocol and when you’re looking for a network protocol to build on, you should
be looking for real decentralization. This industry is about creating systems that
are unstoppable partly and extremely robust. It shouldn’t be the case if someone’s proposing
a decentralized computer platform that it actually depends on a small number of computers;
there are quite a few big projects out there today that depend upon a very small number
of special computers. You have to ask yourself, “Would the world
want to build on top of those networks?” If a small number of computers fail, every
system on top of that network fails too. It doesn’t sound very likely to me. Anyway, that’s kind of protocol stuff. There was another reason the Internet succeeded. So back in the 1990’s there was the Internet
which was an open protocol, an open network and there were also private networks. So in America you had AmericaOnLine (AOL),
you had CompuServe. Back in the 1990, I was in the UK and British
Telecom wanted to create this thing called the Information Superhighway. They said to the government, “Hey. The Internet’s is going to be very dangerous;
it’s permissionless. Anybody can create a service on the Internet. This is really bad. Anybody can create a service on the Internet. We suggest that British Telecom creates the
Information Superhighway. We’ll kind of connect everybody to a walled
garden and we’ll carefully curate all the content. Now, these Private Networks didn’t succeed. Overwhelmingly the world chose to build on
the Internet. There are a number of things about an open
protocol. So, first of all, because the Internet is
based on TCP/IP, an open protocol – people could independently extend it. So, if you went back to the 1990s there were
in the UK hundreds of Internet service providers; all of whom were independently connecting
to the Internet to extend it. That was one reason it has succeeded: it could scale-out very fast. But the other reason was that nobody needed
permission to connect to the Internet. Now imagine the difference. Imagine in 1996 or 1997 you’re an entrepreneur
and you wanted to create an online service, What are you going to do? You are going to go to AmericaOnLine. I mean some people did, do deal with them so that
your special service is available to AmericanOline’s users or are you are going connect to the
Internet. Originally, of course, probably AmericaOnline
had more users, but the thing is this if you build on AmericaOnline or CompuServe or the
Information Superhighway had it come into being, your existence would depend on the
goodwill of the private network operator. So you might start off create some service
on AmericaOnline and over time AmericaOnline might decide that they want to create their
own version of your service. What’s going to happen? AmericaOnline controls the network but they’ve
decided that they want to run the same service. Is there going to be a level playing field? Have you got just as much right to provide a service
on AmericaOnline as AmericaOnline has? Of course, you haven’t. You haven’t. It’s not an open environment; it’s a private
closed environment. So, the beauty of the Internet was because
it was a truly open protocol firstly it could be built out very quickly. In the UK where I came from, there were hundreds
of ISPs that extended the Internet so that everybody could connect, but also it was possible
for innovation to flourish entrepreneurs flocked to the Internet because they didn’t need Permission. They could provide a service over the Internet
that nobody could turn off. That was a completely different proposition
to connecting a service to a private network like AmericaOnline, or CompuServe or British
Telecom’s thing. That’s why the Internet took off and
we had the dot-com boom and there was this explosion in Innovation. So the openness of platforms is absolutely
critical and that’s another thing that you need to think about. If you look at a decentralized network and
you’re thinking of building on top of it, well of course you should look at the decentralisation
because if the protocol is more decentralized, it’s more
robust, it’s more likely to be reliable and support your service better. But you also need to look at how that networks
run. Is it really a consortium? Is it really controlled by business interests? Because if it’s really controlled with business
interests and you build your service on this network and then somebody else builds a competitive
service and the owners – the consortium, the people behind that network – invest
in your competitor, what’s is going to happen to you? Mysterious things might happen. All kinds of things can happen. That I think will emerge but you need to
look at the openness of networks. Okay. So let’s think about building Dapps. Let’s assume that we found a genuinely decentralized
network that’s high-performance and scalable, it’s open and it’s a level playing field. How can we build Dapps that are valuable. So these are the last two things in my time
here on stage that I’m going to run past you. I’m the first thing is a thing called Platform
Risk; this actually applies as a adjunct to what I was just describing. In Silicon Valley, the model has been
for a long time to create monopolies. What happens is you have a new field of business
like P2P ride-sharing. Initially there’s a whole bunch of different
competitors and, for example, that P2P ride-sharing; you had like UBER, LYFT, Sidecar, etc. There are a whole lot of other ones I can’t
remember now, sort of I’ve forgotten. Originally, all of these different ride-sharing
companies got funding and then over time what as winners, leaders emerge, capital is then
sort of poured into the leader with the aim of creating a monopoly. Because once you’ve created a monopoly, you
can absolutely make money and that’s happened in every sector – Facebook, Google, Uber,
you name it. So the Silicon Valley model really has been
today to create monopolies and I’ll give you some examples of why that creates a very hostile
environment for entrepreneurs. This is the thing that we can fix on platforms
like the Internet Computer and other decentralized platforms. A great example of platform risk was provided
by Zynga. Who’s heard of Zynga, the game’s company. Zynga was a company in San Francisco that
invented a game called FarmVille. I forgot how many users FarmVille had. It was a social game so it ran on top of Facebook. I think it got a hundred million users or
something crazy. It grew very fast. Zynga was seen as an extremely successful
company. They were doing very well, they ran an ICO,
they had a multibillion-dollar cap, everything was going swimmingly. And then, one day Mark Zuckerberg decided
to change the rules on Facebook that governed how social games could promote themselves. So poor Mark Pincus – this guy at Zynga – woke
up and he just suddenly discovered that the rules on Facebook had changed. This fundamentally undermined the ability
of Zynga to promote their games. Things weren’t nearly so rosy for them after
Facebook changed the rules. I knew a company in Palo Alto that was a unicorn
and one difficulty they had was that they relied upon the profile data from LinkedIn;
so their product needed the profile data from LinkedIn. They didn’t see the threat. But unfortunately in 2015 LinkedIn decided
that the only people that could access that profile data – the only companies were going
to be Microsoft and Salesforce. So anyways this company ended up having to
sell itself to Salesforce after losing 75 percent of its value. This is what’s known as Platform Risk. Now the reason I’m describing that is one
of the greatest ways that these platforms can add value is by enabling us to reinvent
Internet services in the form of open source businesses using autonomous software. What that means is you can reinvent a business
network like LinkedIn using software that has an in-built governance system that is
used to update it. As you probably are aware from you know Ethereum
smart contracts, these platforms support something called Autonomous Software. Autonomous Software exists independently of
a company or a person – that’s its power. So you can create an Internet service using
Autonomous Software and you can tokenize. This software needs to have an in-built governance
system that is ultimately controlled by the token holders. This can adopt software updates, which can
be collected from the community. So what we’re doing is we’ve already got open
source software, but we’re creating a business – an Internet service that’s also open source. Now the actual operation itself, the service
itself is of open source. In the design of the Autonomous Software that
creates the open LinkedIn there has no company or person behind it, you can provide guarantees
about how people will be granted access to the profile data. The reason this is important is it provides
you a way of de-risking the proposition for other entrepreneurs that want to build on
top of your system. This is why eventually – I don’t know how
long it’s going to take – but things, like Dapps on the Internet Computer, Internet services
running on the Internet Computer will explode because people can re-engineer Internet services
in open source form. These open source versions of Internet Services
can provide people, other entrepreneurs and other systems with much better guarantees
about, for example, the availability of API’s, for example, the LinkedIn profile data. Now think about this. Imagine you’re a young entrepreneur and you
see the traditional centralized monopolistic Internet. Then, you look at these decentralized computation
platforms where the Internet services are open source businesses and you can build against
their API’s with guarantees that they won’t be taken away. Which would you choose to build on? You choose to build on the open platform because
there’s no chance you’re getting rid of your platform risk; there’s no chance that your
use of those APIs, your access to those APIs is going to be taken away from you. So what this will lead to is a kind of mutualized
network effect. Currently on today’s monopolistic Internet,
the aim is to create an Internet service that reaches sufficient scale that it can establish
network effects and become a monopoly that nobody can compete with because they’ve established
a monopolistic position. But the network effects are limited to the
monopoly, to the single monopoly. The great opportunity here is to create a
new Internet, which is decentralized using autonomous software to build open source Internet
services that provide each other with guarantees about how you can access their data and APIs. Therefore, this open decentralized Internet
the participants will benefit from mutualized network effects. If you have the open LinkedIn and it is succeeding
because it’s established Network effects and everybody wants to be on it. Guess what now those other interesting services
that extend the decentralized LinkedIn; they can benefit from its network effects because
they can access APIs without risk they’ll be taken away and likewise the LinkedIn service
can benefit from all these other services, innovators and entrepreneurs are adding on
top. Anyway, that’s it. It’s a big topic; there’s lots of ways that
you can add value in this space. But I think considering openness and the value
of openness is absolutely essential to the planning for any decentralized service; whether
that runs right away from selecting the network that you want to build on top of through to
designing your own architecture and thinking about how you can provide guarantees for other
entrepreneurs, who might want to build on your service. Anyway. That’s it. Thanks.

6 thoughts on “DFINITY’s Dominic Williams’ Keynote at “Hashed Night” Korea Blockchain Week”

  1. the unfair competition that might happen to you when you have your own app on a private system happens ALL the time with the people running on any type app on the google platform. unfortunately 🙁
    I totally understand that argument.

    and also totally understand the idea of decentralized in which each node is equal to the network/mesh its part of.
    EOS relies on a very small number of nodes… that troubles me, although I did love what EOS has to bring to the table of blockchains 2.0.

    this talk reminded me I have to deep dive into dfinity 🙂

  2. It's clear that a new, better "Ethereum" is required. Even if it's not called Ethereum.

    DFinity is a worthy competitor.

    The Global Computer … must happen.

    Software … with a big "S" …. evolution is a never ending progression through ever more powerful and useful levels of abstraction.

    To accomplish this permissionless, distrubuted, un-stoppable computer is more than just a technical achievement.

    This particular abstraction … Blockchain … Consensus … is a very special collection of abstraction capabilities.

    Perhaps unique in the literature.

    Blockchain … Worthy of a Nobel?

Leave a Reply

Your email address will not be published. Required fields are marked *