The march of time is relentless, and since we’re nearing the end of 2018, now is a good time to share some updates on what’s happening at DFINITY and our work on the Internet Computer serverless cloud. We have been very busy, and seem to get busier every day.
Our Destination Is Unchanged
The DFINITY project involves complex Computer Science and software engineering challenges. Our aim was never to build a “better blockchain” even though our Threshold Relay and Probabilistic Slot Consensus technologies may one day become a cornerstone for creating fast and secure chains. The project was conceived in 2015 with the specific aim of creating a public decentralized “serverless” cloud with special properties and features that change how we build and host software systems — as described on the project’s first Web page created that year. The original cornerstone objective was to create a public cloud that could provide the foundations for a new generation of open mass market Internet services that exist in the form of autonomous software instances, but there are many other objectives, including the provision of a new and better way to build and host a variety of software systems that might compete with the cloud incumbents. Nothing has changed, and in fact we have doubled down.
If successful, it will lay the foundations for the realization of what many of us involved in this big crypto project have been striving for for years. To deliver the vision DFINITY has to hit many key points. Our public cloud has to host software and data securely, including autonomous software that makes it possible to create open internet services, to run software with sufficient speed and efficiency and scale out capacity with demand and not stall under load. Moreover, it must provide a new serverless software framework that enables developers to create systems in a more productive and reliable way, and, perhaps most of all, our public cloud must be formed from a real decentralized network comprised of vast numbers of machines operated by independent participants, data centers, and ISPs.
We are not planning on compromising on our goals at all, because we feel confident in our vision, and in our abilities as researchers, engineers and operators to deliver, and believe in the value we are creating.
Our Route Is Unchanged
The nature of engineering is that the task grows somewhat exponentially with complexity. Given the requirements, our project is necessarily one of the more complex around today. Our task is difficult, but we are succeeding in our mission and progress further every day. Our view has always been that there’s no point talking the talk if you won’t walk the walk. That is why some time ago we announced our aim to build our organization into a “NASA for decentralization” that can deliver our ambitious vision and indefinitely support a competitive cryptography-based public cloud infrastructure the world can trust. The DFINITY project has been demonstrating we meant what we said.
We began with advanced research, but much more is involved with delivering and later supporting a technology of this magnitude. At its core, our strategy involves building out the organization by hiring the very best people in the world and creating a research, development and operations resource large enough to handle the challenge.
Recently, we announced a new Zurich research center had gone operational, led by Jan Camenisch, that is already staffed by a brilliant and rapidly growing team, with more centers to be announced soon. We are now persuading some of the world’s best researchers and engineers to join apace, aided by our reputation, growing support for our vision and mission, our integrity and authenticity, and our increasingly effective operations, in which regard I’d like to make a shoutout to our outstanding talent acquisition team. We are now more than halfway towards our first 100 brilliant people, with the aim of scaling well beyond that by this time next year. While others are laying off talent, DFINITY is continuing to hire.
But no matter how brilliant your team of scientists, engineers and operations staff harnessing so much talent and executing on such a complex vision requires management and structure. At the head of this effort is Mack McCauley, who joined as VP of Engineering from Facebook, where he was Director of Core Systems. Mack also held senior technical leadership roles at Amazon Web Services, and before that, at Microsoft during their exponential growth phase, and has enormous experience leading complex engineering operations. He is also hiring engineering directors to scale himself so our organization, structure and processes are growing in synchrony (join us!).
Working with Outside Parties
Although we have provided descriptions of Threshold Relay, Probabilistic Slot Consensus and other novel designs and techniques, and released cryptography libraries and other resources that are already being used by several other projects, of course, these things represent only the tip of a very large iceberg of work inside the project that has not yet been shared. This will of course eventually change but for the moment we have chosen to work in private for maximum speed, leaning on our highly structured internal engineering processes (although we will continue to collaborate in areas where peer review is required, such as security testing).
A few months ago, we shared an early SDK with a limited set of partners. A new version of the SDK is currently under development reflecting a change to how software hosted on the Internet Computer persists its data, and we also have an all-new language under development called ActorScript that will maximize developer productivity within the framework. In the meantime, we have been focusing on ways to help selected partners effectively develop systems that will be ready at launch. This has involved formulating a unique approach that enables partners to rapidly develop systems for the Internet Computer using stable tools and traditional platforms in such a way that their systems can be easily ported to the new SDK, allowing DFINITY and its partners to progress at maximum speed in parallel.
Chosen Release Strategy
We have come to believe that users will be best served by the earliest possible release of an MVP version of the full Internet Computer vision, which necessarily includes the ability to scale out capacity, among other things, and have chosen to focus on that rather than making piecemeal releases which would have to be supported and slow us down. The new SDK will however be released before network launch in a staged manner, which will involve increasing the number of partners with access on each round.
Current Expected Timelines
We were originally planning on releasing a cut-down version of the Internet Computer in Q1 2019, which would not have been able to scale out its capacity, but would still have been a force to reckoned with thanks to its relatively high capacity and revolutionary speed, and strong, decentralized underpinnings. However, supporting such a version in production would greatly slow down our overall progress towards delivering the full vision, we have stepped back from this objective. We are currently providing some selected key projects and partners with assistance in building systems in parallel that can easily be ported to the Internet Computer. Again, it is too early to release details now, but exciting things are happening both with respect to the kind of projects being pursued and the organizations now coalescing around the DFINITY vision.
We expect that our new SDK, which will reveal critical aspects of what we are doing, and for which we have pushed timelines, will be reasonably mature by the end of Q1.
Meanwhile, we are working on timelines for our MVP version of the Internet Computer. The requirements we have set include that it must clearly be a true public cloud rather than just a better blockchain computer. That means that it will be able to scale out its capacity with demand as required, since otherwise even though it will be inherently fast it will still stall once its bounded capacity has been consumed, rather like Ethereum during the CryptoKitties craze or during major ICOs, or alternatively computation and storage will have to be exorbitantly priced to reduce usage (all of which, of course, makes it difficult to develop and host real production applications).
Furthermore, we want to ensure that developers who create systems for the first release of the cloud do not have to make adjustments to their systems as we move forwards, and that conversely, continuing to support the software installed on the earliest versions of the network does not create a drag on our continuing development. Finally, from launch the network must be adaptive and controlled by a version of the Blockchain Nervous System governance mechanism.
Our engineering leaders are busily working with our researchers, engineers, talent acquisition and operations people to pin down and optimize our schedule with respect to the initial MVP release. I can share the following best guess internal estimates. These are that the SDK and new software framework will be released in a stage manner beginning Q1. A key internal engineering milestone, which involves the release of a version of the network in which the lower layers (our highly specialized novel P2P networking layer, consensus systems, message routing and other components that are key to performance and security) is of production quality, shall also be ready late Q1, so that we can allow white hats and others to start attacking the system to help prove out the security of the implementation.
A full test network, including our unique software framework, algorithmic governance, and other features, will then be available (again, with the level of access provided still undecided) in late Q2, with a production version of the network following.
Our progress towards the MVP network is of course only part of the story. Many other interesting events will be happening next year related to DFINITY, I’m looking forward to keeping you updated. I’d like to thank everyone for their interest in the project and their patience as we continue to strive toward our shared goals. Finally, I’d like to thank our ecosystem and supporters, and most importantly of all the DFINITY team, without whom the talk of an Internet Computer would only remain crypto talk. As always, onwards and upwards.