Helo World. I’m back blogging!
A small step for mankind, a big step for Dominic… I started my last blog in 2010 but had mostly stopped blogging by end of 2011. You can find my old blog here https://ria101.wordpress.com/.
I used to describe myself as a JAHEBIL, or Just Another Hacker Entrepreneur Based in London, but due to relocation need another moniker. This blog marks my imminent return with a new kind of decentralized venture and my general desire to share more thoughts. But before getting to what’s new, some historical background.
The ria101 blog was created in the UK before I moved to Palo Alto in 2012. I was a serial geek entrepreneur from long before but in a naive, British way; in the Valley entrepreneurialism and its techniques are distilled, and there are lots of venture capitalists and many other geek entrepreneurs. Back then in the UK, I was isolated from my current world. I had no profile to speak of and even avoided having one — I only cared about my the brands of the ventures I created. I worked away on my ventures and was known through them. I now live in a more entrepreneur centric world.
Back in London each venture I created, or at least a decent proportion, needed to generate returns that would enable me to pursue another. This was different to the venture backed approach of The Valley where you invest money aggressively aiming to make it to the moon or flame out. To some extent given the difficulties involved in startup ventures my approach was the tech equivalent of squeezing blood from a stone: grinding away in a european environment which, at least back then, wasn’t a friendly place for new technology businesses. Despite surprising success, in 2010 I returned to pursuing a venture where success and failure would be more binary, more like I had been pursuing before the 2000 Dot Com Crash. This was Fight My Monster, a massively multiplayer online (MMO) game and fun social network for kids. My plan was to get every online kid in my target demographic — worldwide — playing.
Bored with a trajectory driven by the need to safely achieve returns, this would need to scale dramatically or burn out, and would hopefully indulge my technical passions for distributed system design by supplying me with millions of users. After some trial runs, my online game and network launched properly New Year 2011, gained thirty thousand users within two weeks, three hundred thousand users within a few months, and went on to gain a million users in just more than a year.
I had started blogging January 2010, a year before launch, to record my experiences with the different technologies we were using. We had run through the gamut of the first distributed “scale out” databases. After putting HBase, Cassandra and others back to back we chose the early Cassandra beta, the database which for example is now used to power Netflix. I built various tools for Cassandra, including the first systems that made it possible to run atomic transactions over scalable eventually consistent stores, and eventually we had the world’s first really complex Cassandra based system in production. The ria101 blog posts hint at some of the challenges we faced, and without revealing all I will say that in late 2011 in production with the accounts of hundreds of thousands of users at stake we even faced data corruption issues due to bugs in the beta code — which were only saved by the very most senior people at Cassandra and our own Cassandra-committer team member working through the night on several occasions. This was hair raising stuff but the system did its work: as the business scaled like crazy, we were able to simply add new servers to our gossip network to scale up.
Judged in terms of financial return on potential, Fight My Monster ultimately failed, although the experience was invaluable. The best part was working with some truly amazing people and indulging my love for distributed systems and scalability. For example, Fight My Monster would have been impossible without the help of British designer Jon Ball, whom I believe to be one of the world’s most talented and creative designers (see his FMM work here http://www.pokedstudio.com/fightmymonster/ and http://www.pokedstudio.com/fight-my-monster-advert/), Aaron Morton the Cassandra committer and long time collaborator Matt Grogan. We created so many innovations. For example one of the first components I coded was system that could automatically generate new monster designs on demand from “meta design” libraries, and Jon was quickly able to master the tooling enabling us to create a limitless supply of extraordinary creatures. We went on to create TV adverts that broke response rate records, and many other things.
One of the challenges in Fight My Monster was scope and quantity of work involved, which I had underestimated. Major online games require teams that are more multidisciplinary than other ventures (for example, they involve coding, databases, networking, payment systems, database administration, complex analytics, graphic design, audio work, animation, advertising…), much more so than any venture I had ever created before, and it quickly munched through investments my wife and I made and those of some friends who had stumped up additional money. Furthermore, and perhaps consequently, I had built a team that depended pretty heavily on me in every way. My workload became insane in a way that is not imaginable here in The Valley where investment is easier to come by. Today it’s a cliche to talk about working 12-18 hour days in technology startups, but for me this was normality. I had to combine business administration with critical systems administration and distributed systems programming and still carve out time for my personal life. I owe my wife, who was also a director and played a role in project conception, a debt of gratitude for keeping our lives on track.
For all the reasons then, in the summer of 2011, I started seeking venture capital, something I had not been involved in since the Dot Com days! We had perfect, textbook, hockey stick growth charts, and so this would be easy I reasoned. Actually, though, in the United Kingdom back then, no, it wasn’t easy at all. The insane growth of the site was a hinderance in some ways because the statistics were so overwhelming for London investors who just didn’t know how to process them. We had meetings were the VCs said they wanted us to stop growing and create cohorts of users to prove our ability to monetize, which were only slightly more little more idiotic than the meeting where we were immediately offered investment but with a 5X liquidation preference. The better VCs (and there weren’t many) had invested in a competitor (there weren’t many of those either) who forbade them to invest in us. This made things difficult.
Some of the seeds of our failure to become a $100MM exit were sowed then. In the process of seeking venture capital we got involved with a corporate finance hand later given a role who became disruptive. We did eventually manage to get a large deal on the table from a european VC, but when they delayed at the end of 2011 I turned back to VCs in the US I had been introduced to through a chance encounter in Palo Alto. Funding was quickly closed in January 2012 with Californian backers, almost a year after launch. Within a couple of weeks of raising funding we hit 1 million users, and within months I had relocated the company to San Mateo (a town halfway between San Francisco and Palo Alto). The game eventually scaled to 3 million users, but some serious speed bumps were in store. There were strategic difference with the corporate finance guy we brought in while in Europe, who was given a senior role after funding, and despite our amazing growth timing was not on our side. In hindsight, it is obvious we were a late to the game with our online Web-based experience. At that very moment a massive shift was occurring as people moved from Web to tablet and mobile based gaming. We could have succeeded but needed to move faster: if I had my time again, I would have relocated to The Valley very soon after the company started growing to raise money faster and gain access to a bigger pool of experienced gaming executives.
So, fast forward to today and my work in decentralization. Back in 1998 my then venture in London was working on an online storage system called Smartdrivez, and I used Wei Dai’s Crypto++ library (http://www.weidai.com/) for a bunch of things including creating a certificate scheme. Wei Dai was obviously brilliant, and I loved his C++ code. Among his works I found his “bmoney” paper, which of course is one of the precursors of Bitcoin, and I became fascinated with who exactly this mysterious “extropian” was. Time and events quickly overcame my ability to indulge and investigate, and I had to press on, but not before an indelible impression had been made that something extraordinarily interesting was going on. After I began to move on from Fight My Monster in 2013, the early exposure to his work was one of the reasons I immediately naturally gravitated towards Bitcoin.
By late 2013 I was working on several ventures but was being drawn to the decentralization community as though by magnetic pull. To be clear this was for reasons greatly exceeding entrepreneurial logic. At the time, I was quite some way down the road to launching an unrelated new business — a kind of automated Groupon — that was testing extraordinarily well and into which I had sunk significant funds. Partly, of course, it was technical attraction. I love algorithms and distributed computing, and in the distant past at Computer Science undergraduate level won many prizes for this kind of thing. But more than that, I had never encountered a technical field that combined considerations of finance, law, politics, economics and philosophy while holding the promise of driving truly impactful changes in the world. For me, this emerging area is a dream come true. I made a bigger life decision to rededicate my career to this field.
During 2014 I largely went back to basics. Unusually for an entrepreneur, I spent most of my time studying technical theory full time. For example, I studied traditional Byzantine Fault Tolerant consensus theory in depth, and after a few months began devising derivative math and designing decentralized networks that can “scale out” like Cassandra — that is, decentralized networks that can process more and more transactions the more computers join. During this time I wrote a paper describing a scalable cryptocurrency with features of microtransactions called Pebble. The paper was only ever circulated within narrow crypto circles mainly because I moved on to other ideas, but was the first system to describe decentralized sharding. Within that system, individual shards reached agreement using asynchronous Byzantine Consensus algorithms. HoneyBadger uses similar approaches to BFT consensus and discussions are mentioned in that paper (https://eprint.iacr.org/2016/199.pdf).
While I have moved back into highly committed entrepreneur mode — more in a moment — my theoretical network work went in a different direction and I am trying to find time to get technical papers out. My work has focused entirely on protocols and cryptography that can underpin what I believe to the be the future — “world computer” networks such as Ethereum that 1. can be open 2. can scale out to process any number of transactions as new computers join and 3. provide much stronger consistency properties. The project loosely proceeds under the banner of the Dfinity project, and updates are made at http://dfinity.io.
TLDR; the non-technical will want to skip this paragraph, but the main components of the pending Dfinity work I’m planning to release include:
1. A “macro”/master chain for such scale-out networks that runs by applying a unique deterministic BLS threshold signature scheme as a globally shared Verifiable Random Function (VRF) that has superior performance properties, is highly resistant to forking (is strongly consistent relatively speaking) and produces an unpredictable secure random beacon that can drive other components of a scalable network.
2. A system of “validation towers” and “validation trees” driven by the random beacon that produce a scalable global framework for validating state changes.
3. Two systems based upon VRFs run independently by processes: a new kind of blockchain called B-VERF, multiple instances of which can be run securely in parallel, and an O(n) probabilistic consensus protocol called C-VERF, which allows vast numbers of processes in large public decentralized networks to decide (classical consensus theory says that such a protocol is impossible, and therefore you will deduce that I have relaxed the conditions of consensus to include that a processes may only know they have decided correctly with overwhelming probability).
4. Systems for creating Sybil resistant identities that can be applied in the above systems.
Many of the key ideas have been shared with industry insiders over the years, and for example application of e.g. regarding application of VRFs will likely appear in some well known crypto projects shortly.
So apart from Dfinity research work, what and where now?
I am super excited to say that last year I partnered with Tom Ding, another decentralization entrepreneur with his own amazing story, and we founded the String venture, which is pursuing what we call “autonomous finance” projects (information will be made available at http://string.technology). Other well known participants include Gustav Simonsson the Ethereum core developer who joined as a founding engineer. We are primarily interested in financial systems that are inherently international and open in nature that run on open blockchains.
Although there is currently much interest in “b2Bank” companies and applications of blockchain within traditional finance — and we have even already had acquisition interest from major companies in this space — we are dedicated to pursuing open chain finance projects. As far as I know, we are the only such project with serious financial backing. Others ventures recently proudly announced they were the first venture-backed Ethereum companies, but we actually both significantly predated them and have raised multiples of their funds from IDG Ventures, Amino Capital, Fenbushi Ventures, Danhua Capital and others. In common with many Ethereum ventures we have distinctly Asian focus. The reason is simply that The Valley threw its chips into the ring with Bitcoin early on. We are building systems that require a “world computer” to run.
We are working on several distinct financial systems, many of which have been alluded to in the many talks I have given around the world in the past year. We haven’t decided which systems will lead, but expect announcements soon.
Generally speaking, our focuses include, but are not limited to, systems that provide decentralized “mirror asset” tokens that act like cryptocurrency but securely track the values of assets traded in mainstream financial systems, systems for extending loans and related systems that generate settlement currencies that are price stable but which exist independently of mainstream financial systems, and systems that provide users with predictable interest in the manner of savings accounts but without banks acting as intermediaries. All of our systems require Turing complete “world computer” systems such as Ethereum, and we have developed many building blocks such as decentralized financial exchanges that would have been impossible previously. We also aim to leverage “spooky financial action at a distance” where decentralized systems induce actions by independent and anonymous players in mainstream finance, helping produce useful properties for the open systems.
The purpose of this personal blog shall be to relay my personal opinions, tidbits and ideas from the field of decentralized computing and finance from now forwards.
My current ambition is to create decentralized platforms and systems that scale exponentially.
This is my opening note, and I’ll have more to say :)