It has been a year since we started writing about what blockchain means for the future economy and our technology investments. One of my more memorable moments of 2018 was hearing UC Berkeley professor Dr. Dawn Song say: “Blockchain frees AI, and AI can help blockchain reach its full potential.” When we think about how quickly AI is being integrated into our lives, and statements like “blockchain changes everything”, how the two might interact is fascinating.
In this piece, we will look at the state of affairs in technology from both a bottom-up and top-down perspective. First, let’s look at the bottom of the technology stack—the pipes, and trends towards even more distributed systems (today’s most popular being blockchain). At the Stanford Blockchain Conference, my very experienced neighbor commented to me: “blockchain and its related innovations changes the foundation of technology, not just the internet, but how computers are used. It permeates up and down the stack, both accelerating and shaping tech evolution.” We will also explore how AI (artificial intelligence) fits into the lowest layer and how it permeates all the way up to politics, privacy, and concerns around large powerful tech companies’ use of our “private” data today and their ability to direct our behavior using analytics and behavioral science in the future.
Blockchain Basics: What Is It Technically?
Blockchain technology is a shared ledger (more commonly referred to as a database). Because it is shared among parties who are at different locations, it is called “distributed” or DLT (distributed ledger technology). We have had databases for decades, so what’s the big deal? The move to sharing one database and distributing it across multiple systems is the big deal. Everyone can theoretically be on equal footing in terms of receiving the same information, thus increasing trust through transparency. A truly distributed system could act as a coordinated world computer.
What are the Benefits of Blockchain?
There are two key benefits of blockchain technology:
- The first benefit is the efficiency of sharing one database among trading partners who know each other (and generally have a good working relationship such that there are business reasons to act honestly). Entries made into a shared database will reduce conflicts over who shipped what to whom. These are known as permissioned blockchains, and will lead to significant cost savings across many industries, including finance, insurance, transportation, manufacturing, and health care. Supply chain is one of the first proposed use cases. Amazon’s AWS recently offered their own blockchain technology service, which incorporates all the components needed in a shared permissioned enterprise blockchain, including one ledger and provenance. It could threaten other permissioned blockchain efforts.
- The second revolutionary benefit: the open network spirit around the creation of Bitcoin. Due to the way that blocks in the blockchain are secured and tied to each other (through cryptographic hashes and systems to determine who wins the right to enter the data that gets hashed), anyone can join or leave this distributed network, anyone can contribute, and many of the participants will be unknown to other participants—some may even be adversarial. This type of “open” blockchain technology is called a permissionless blockchain, with the two largest being Bitcoin and Ethereum. The goal for any database/ledger is to get the correct data entered in the correct order, be it transactions, transfers, etc. For a distributed system, there needs to be a consensus on the data to be entered, the order in which it is entered, and who gets to confirm the transactions (state). The real revolution with permissionless blockchain is that, with some caveats (less than 40% adversarial by most models), you can reach data transfer consensus with people you do not know. This fundamentally changes relationships, opening up all kinds of opportunities for commerce, international transfers, agreements on provenance of items and ownership logs of land, identities of people, assets, etc.The new kind of business models it can enable is changing technology at its very core. This is why people are so excited about blockchain.
Since we last wrote on the topic in February 2018, Blockchain Technology: Looking Beyond Bitcoin, the open-source world development activity has actually picked up, with 5 times more people at the Stanford Blockchain Conference this year versus last year. The conference was still very focused on research, but several important projects (Interstellar, Kadena and Grin) reported strong technical advancements. These projects are all working towards three main goals: make these distributed databases fast with the correct data in the right order.
Anytime there needs to be a consensus, you have to be careful to make sure the data is correct and everyone agrees—and therefore, the slower that process will be. Blockchain technologies are probabilistic, ie, they wait to publish until they reach a certain probability that the data is correct and in the proper order. Some projects spend more time focusing on the speed (Kadena, Stellar and Ripple with their Merkle tree variants) as they have a more permissioned models and do not have to worry as much about the other two variables.
Other projects, like Ethereum and Bitcoin, focus less on speed and more on confirming that consensus on the correct data is reached even while including unknown participants. The probability of the consensus data being correct increases as the chain gets longer—the probabilistic method of deciding the agreed upon state.
This exciting idea of having the data we view in a distributed ledger be trusted, without necessarily knowing or trusting everyone who has access to the system, is so revolutionary that it is forcing a high-speed re-testing of all of game theory. Cryptocurrencies serve to incentivize behavior on the network such that the data entered can be trusted. They also serve as the broadest most rapid-fire test of incentive systems. We are running the most time-compressed, real game theory experiments with high-speed feedback loops ever in history. We are learning about ourselves, how networks of people and indeed all complex systems really operate, and re-writing game theory and organizational psychology in the process. What is rational behavior? Can behavior be predicted and controlled through incentive systems? What are the unintended consequences? This will feed into our thinking on machine learning and AI.
The management team of one of our distributed software solutions holdings, focused on multi-cloud platforms and blockchain, recently commented that the “correct order” problem, knowing that the data is entered in the correct order and something is not left out, is really a leadership problem (the “leader” should select the order, then the group comes to consensus). This is an issue that has been battled out in enterprise systems via the old Paxos algorithm for decades. To reach a consensus on data and have that data entered in the correct order over a globally distributed system, you run into the laws of physics: the speed of light. A computer in China and one in NYC cannot enter data at the same time on a server in Chicago because one is much physically closer than the other. This company solved for this problem by developing an algorithm to pre-determine the correct order even in very geographically distributed systems. Other open-source projects have leaders chosen via lottery from geographically closer regions. In 2018, we saw significant strides forward in terms of R&D in distributed systems and blockchain technology, game theory and new business models, and yet we are still quite early with plenty of growth to come.
What Is AI, and How Does It Fit In?
Artificial Intelligence (AI) is what it sounds like, making machines smart. Within that, there are various sub-categories such as machine learning, which can be supervised and unsupervised (let it look for patterns without giving it any training, not telling it a certain pattern is a cat). Some interesting ones are nearest neighbor where the machine can learn and compare to other like items without a specific model. Then there are neural networks, layering and prioritizing neural processes much how the brain theoretically does, reinforcement learning, and biological network learning. This last one takes us full circle, as machine learning and big data are applied to biological systems to help us learn more about natural systems, AND this learning is then fed back into machine learning algorithms such that these systems become biological, self-learning and unsupervised.
Exciting advances and new business models could be enabled in a world with truly distributed systems of computer networks, with everyone and anyone able to join or leave, and the intelligence of the system akin to a biological learning engine. One friend at the Stanford conference said “it changes more than “just” the internet, it changes how we see computing”. We are in a subscription, on-demand society, and pushing our systems to be more democratic, more flat and distributed, while having these same systems self-learn and adapt to our changing needs is a potentially exciting future.
Thus far, we have discussed the lowest layers, the transportation layer, and how the way we are networked together is becoming more equitably distributed, enabled by advances in cryptography and machine learning. One interesting application, which will have impacts on data privacy, is the concept of where networks and data should reside. For example, should all your financial data be on servers at multiple financial institutions or in your control? For now, it seems that, collectively, we may be too complacent to go through the effort to re-take control of our data—and many regulations would need to change. But, if you think about many micro-cap private assets, such as real estate, the time to shift from analog to digital may be now. For instance, you could attach value to real estate by putting a building in a private REIT structure (like the St Regis Aspen recently did), fractionalizing a non-control share (in this case 20%) and selling tokens backed by the cash flow and ownership rights of that property. It’s a security very similar to equities, yet these tokens are fully programmable, thus they can be programmed to conform to global and local regulations, only transfer to accredited investors, etc. One can also attach to the token all the data about that asset (location, ownership, IoT (internet-of-things) data, occupancy, tenant comments, social media comments, taxes, inspections, maintenance, drone footage, etc.) This exciting concept of attaching data to a security or an identity is known as self-sovereign identity. It enables asset interoperability as each asset trades with all of its data attached. Asset interoperability may revolutionize portfolio management in the future, driven by these blockchain technology-enabled features.
If the data about the asset moves with the asset, this increases the efficiency with which micro-cap private assets trade and are valued. Algorithms and machine learning could efficiently evaluate the data, use less supervised models to find new relationships between seemingly unrelated assets, and aid in the portfolio management process to adjust portfolios by having the system find substitute assets for those with less liquidity at the moment you want to trade. The combination of blockchain and machine learning algorithms allowing for efficient low-cost transfer of value and data could bring the $220 trillion real estate market, only 2% of which is today traded in public REITs, into the digital age.
Interaction of Distributed Systems and AI/ML
Back to the technology stack and the interaction of distributed systems and AI/ML (“machine learning”, a sub-segment of AI). Machine learning can be applied wherever there are large data sets to analyze, improve and eventually automate decisions. ML technology could be applied to building better chips, circuit boards, network architectures, even running game theory scenarios (think the old movie War Games) for better blockchain-consensus algorithms. It can extend all the way up the software stack to business analytics, scenario analysis, risk exposures, and valuation. Our view is that AI has been limited due to siloed data sets. AI “learns” faster, with greater accuracy, with more data that is also accurate “clean” data. Blockchain’s ability to distribute and free data across systems, making that data available to AI to learn from, where the data about the learning can be deposited back on that asset, allows the next system to learn even more from richer data. AI/ML also can act as the intelligence, or neural layer on top of a network of assets, values and data. This forms a continual feedback loop which informs the optimization of the network, analysis of the assets on the blockchain, their data, availability and trading. The potential scenarios are nearly infinite.
The shared economy is driving an incredible feedback loop with distributed systems, workforces, and an increasing interplay of analytics that is fractionalizing significant portions of the economy (AirBnB, Uber, etc.). In real estate, we have the phenom WeWork, which has changed how we think about permanent offices versus flexible spaces. This led others to question the optimal workplace, with pop-up offices in half-vacant malls, tech companies offering retrofits and office management to traditional spaces. With blockchain and analytics, you could separate governance from ownership from residence, much like some equities have dual classes of shares with different voting rights. This opens up all kinds of potential new services, and living and working models. Millennials are embracing the new shared economy with record low home ownership rates and a clear preference for alternative banking. This speaks to the importance of software that enables and works well in distributed systems, such as disruptive and innovative SaaS, subscription enablers and analytics companies where we have technology exposure in our portfolios. This runs across the tech stack from hardware, network architecture, and routing software to platforms, application layers, internet and analytics companies.
Web 3.O: Self-Sovereign Identity
Personal privacy is a big issue in a world where large social media companies mine and sell our data to analyzing us and then use algorithms to manipulate our behavior (known as marketing, advertising and sales but getting more sophisticated at alarming rates). If getting “likes” on Facebook and alerts of a text message deliver dopamine hits, big data and machine learning may offer large corporations the ability to potentially control populations as never before. As with any force, there is a counterforce, that of self-sovereign identity, or wresting back control of our data. In some cases, large firms would rather not have the liability of protecting our data, so might welcome us having control over our own data. With each of us in control of our own data, there would arguably be less concentrated “honey-pots” of data, and we could selectively reveal parts of our data for a fee or in order to transact with certain firms. There are interesting breakthroughs in this space. For example:
- ZK-STARKs (“ZK” stands for zero knowledge) represent a breakthrough privacy solution that enables users to safely secure and control their data. ZK-STARKs works like a lock, where you show you can unlock the lock, without showing the combination. To get into a bar, for example, it just shows whether you are over 21, not your age, address, weight, or other information that the bouncer does not need.
- Doc.ai enables users to download their health records to their own control and then reveal them to biotech firms for their machine learning algorithms to analyze for potential new drugs. These users get paid, may participate in drug trials or just have blood tests as part of a sample population (overweight, etc.) from which biotech companies determine the need for various treatments.
Other companies are working on paying users for their attention, polling, voting, etc., all around bringing back control to the user. This trend is broadly termed Web 3.0.
The Not-too-Distant Future
Quantum computing’s implications are being explored now and may be far-reaching, including changing the way distributed systems and analytics interact. In The Nature of Technology, Dr. W. Brian Arthur argues that we often come across technological advances through better scientific tools to observe natural phenomena. Innovation is a function of what we are able to understand, adapt to and incorporate at any point in time. This feeds into complexity theory: how do things, networks and nature assemble themselves? This is where AI may finally find its pinnacle, in mimicking biological and neurological processes. This is much how alchemy became chemistry. As we understood more, we made tools that allowed us to observe nature in more detail.
Blockchain innovation in technology is driving an expanding secular growth cycle. Our technology investments remain keenly focused on the strong trends of distributed systems and analytics permeating our shared economy.
__________________________
Disclosure: Nicholas Investment Partners, L.P. (“Nicholas”) is an independent investment adviser registered with the SEC. Registration with the SEC does not imply a certain level of skill or training. The firm maintains a complete list and description of performance composites, which is available upon request. Policies for valuing portfolios, calculating performance, and preparing presentations are available upon request. Past performance is no guarantee of future results. No part of this material may be copied or duplicated or distributed to any third party without written consent.
Nicholas does not guarantee the success of any investment product. There are risks associated with all investments and returns will vary over time due to many factors such as changing market conditions, liquidity, economic and other factors. The value of investments can go down as well as up, and a loss of principal may occur. Although Nicholas attempts to limit various risks, risk management does not imply low risk. All risk models are inherently limited and subject to changes in economic, political and market conditions, as well as changes in the strategies’ holdings, among other things, which could affect the risk profile of any portfolio managed by Nicholas. Small- and mid-cap companies may be subject to a higher-degree of risk than larger more established companies’ securities. The liquidity of the markets for these small and mid-cap companies may adversely affect the value of these investments. Concentrated or sector strategies are expected to maintain higher exposures to a limited number of securities or sectors which could increase the volatility, market, liquidity and other risks of the strategy.
Some information herein reflects general market commentary and the current opinions of the author which are subject to change without notice. It is provided for general informational purposes only and does not represent investment, legal, regulatory or tax advice and should not be construed as a recommendation of any security, strategy or investment product. There is no guarantee any opinion, forecast, or objective will be achieved in the future. The information, charts and reports contained herein are unaudited. Although some information contained herein was obtained from recognized and trusted sources believed to be reliable, its accuracy and completeness cannot be guaranteed. Unless otherwise noted, Nicholas is the source of illustrations. References to specific securities, issuers and market sectors are for illustrative purposes only. Nicholas does not undertake to keep the recipients of this report advised of future developments or of changes in any of the matters discussed in this report.
Nicholas used third-party information in the preparation of this report. While Nicholas believes the third-party information was obtained from reliable sources, we cannot guarantee the accuracy, adequacy or completeness of the information obtained from these sources.