This futuristic, anti-establishment thriller is one of Elon Musk’s favorite books. While Heinlein’s novel can drag on with little action, The Moon is a Harsh Mistress presents an interesting war story and predicts several technological revolutions.
Tech Themes
Mike, the self-aware computer and IBM. Mycroft Holmes, Heinlein’s self-aware, artificially intelligent computer is a friendly, funny and focused companion to Manny, Wyoh and Prof throughout the novel. Mike’s massive hardware construction is analogous to the way companies are viewing Artificial Intelligence today. Mike’s AI is more closely related to Artificial General Intelligence, which imagines a machine that can go beyond the standard Turing Test, with further abilities to plan, learn, communicate in natural language and act on objects. The 1960s were filled with predictions of futuristic robots and machines. Ideas were popularized not only in books like The Moon is a Harsh Mistress but also in films like 2001: A Space Odyssey, where the intelligent computer, HAL 9000, attempts to overthrow the crew. In 1965, Herbert Simon, a noble prize winner, exclaimed: “machines will be capable, within twenty years, of doing any work a man can do.” As surprising as it may seem today, the dominant technology company of the 1960’s was IBM, known for its System/360 model. Heinlein even mentions Thomas Watson and IBM at Mike’s introduction: “Mike was not official name; I had nicknamed him for Mycroft Holmes, in a story written by Dr. Watson before he founded IBM. This story character would just sit and think--and that's what Mike did. Mike was a fair dinkum thinkum, sharpest computer you'll ever meet.” Mike’s construction is similar to that of present day IBM Watson, who’s computer was able to win Jeopardy, but has struggled to gain traction in the market. IBM and Heinlein approached the computer development in a similar way, Heinlein foresaw a massive computer with tons of hardware linked into it: “They kept hooking hardware into him--decision-action boxes to let him boss other computers, bank on bank of additional memories, more banks of associational neural nets, another tubful of twelve-digit random numbers, a greatly augmented temporary memory. Human brain has around ten-to-the tenth neurons. By third year Mike had better than one and a half times that number of neuristors.” This is the classic IBM approach – leverage all of the hardware possible and create a massive database of query-able information. This actually does work well for information retrieval like Jeopardy, but stumbles precariously on new information and lack of data, which is why IBM has struggled with Watson applications to date.
Artificial General Intelligence. Mike is clearly equipped with artificial general intelligence (AGI); he has the ability to securely communicate in plain language, retrieve any of the world’s information, see via cameras and hear via microphones. As discussed above, Heinlein’s construction of Mike is clearly hardware focused, which makes sense considering the book was published in the sixties, before software was considered important. In contrast to the 1960s, today, AGI is primarily addressed from an algorithmic, software angle. One of the leading research institutions (excluding the massive tech companies) is OpenAI, an organization who’s mission is: “To ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity.” OpenAI was started by several people including Elon Musk and Sam Altman, founder of Y Combinator, a famous startup incubator based in Silicon Valley. OpenAI just raised $1 billion from Microsoft to pursue its artificial algorithms and is likely making the most progress when it comes to AGI. The organization has released numerous modules that allow developers to explore the wide-ranging capabilities of AI, from music creation, to color modulation. But software alone is not going to be enough to achieve full AGI. OpenAI has acknowledged that the largest machine learning training runs have been run on increasingly more hardware: “Of course, the use of massive compute sometimes just exposes the shortcomings of our current algorithms.” As we discussed before (companies are building their own hardware for this purpose, link to building their own hardware), and the degradation of Moore’s Law imposes a serious threat to achieving full Artificial General Intelligence.
Deep Learning, Adam Selene, and Deep Fakes. Heinlein successfully predicted machine’s ability to create novel images. As the group plans to take the rebellion public, Mike is able to create a depiction of Adam Selene that can appear on television and be the face of the revolution: “We waited in silence. Then screen showed neutral gray with a hint of scan lines. Went black again, then a faint light filled middle and congealed into cloudy areas light and dark, ellipsoid. Not a face, but suggestion of face that one sees in cloud patterns covering Terra. It cleared a little and reminded me of pictures alleged to be ectoplasm. A ghost of a face. Suddenly firmed and we saw "Adam Selene." Was a still picture of a mature man. No background, just a face as if trimmed out of a print. Yet was, to me, "Adam Selene." Could not he anybody else.” Image generation and manipulation has long been a hot topic among AI researchers. The research frequently leverages a technique called Deep Learning, which is a play on classically used Artificial Neural Networks. A 2012 landmark paper from the University of Toronto student Ilya Sutskever, who went on to be a founder at OpenAI, applied deep learning to the problem of image classification with incredible success. Deep learning and computer vision have been inseparable ever since. One part of research focuses on a video focused image superimposition technique called Deep Fakes, which became popular earlier this year. As shown here, these videos are essentially merging existing images and footage with a changing facial structure, which is remarkable and scary at the same time. Deep fakes are gaining so much attention that even the government is focused on learning more about them. Heinlein was early to the game, imaging a computer could create a novel image. I can only imagine how he’d feel about Deep Fakes.
Business Themes
Video Conferencing. Manny and the rest of the members of the revolution communicate through encrypted phone conversations and video conferences. While this was certainly ahead of its time, video conferencing was first imagined in the late 1800s. Despite a clear demand for the technology, it took until the late 2000s arguably, to reach appoint where mass video communication was easily accessible for businesses (Zoom Video) and individuals (FaceTime, Skype, etc.) This industry has constantly evolved and there are platforms today that offer both secure chat and video such as Microsoft Teams and Cisco Webex. The entire industry is a lesson in execution. The idea was dreamed up so long ago, but it took hundreds of years and multiple product iterations to get to a de-facto standard in the market. Microsoft purchased Skype in 2011 for $8.5B, the same year that Eric Yuan founded Zoom. This wasn’t Microsoft’s first inroads into video either, in 2003, Microsoft bought Placeware and was supposed to overtake the market. But they didn’t and Webex continued to be a major industry player before getting acquired by Cisco. Over time Skype popularity has waned, and now, Microsoft Teams has a fully functioning video platform separate from Skype – something that Webex did years ago. Markets are constantly in a state of evolution, and its important to see what has worked well. Skype and Zoom both succeeded by appealing to free users, Skype initially focused on free consumers, and Zoom focused on free users within businesses. WebEx has always been enterprise focused but they had to be, because bandwidth costs were too high to support a video platform. Teams will go to market as a next-generation alternate/augmentation of Outlook; it will be interesting to see what happens going forward.
Privacy and Secure Communication. As part of the revolution’s communication, a secure, isolated message system is created whereby not only are conversations fully encrypted and undetected by authorities but also individuals are unable to speak with more than two others in their revolution tree. Today, there are significant concerns about secure communication – people want it, but they also do not. Facebook has declared that they will implement end to end encryption despite warnings from the government not to do so. Other mobile applications like Telegram and Signal promote secure messaging and are frequently used by reporters for anonymous tips. While encryption is beneficial for those messaging, it does raise concerns about who has access to what information. Should a company have access to secure messages? Should the government have access to secure messages? Apple has always stayed strong in its privacy declaration, but has had its own missteps. This is a difficult question and the solution must be well thought out, taking into account unintended consequences of sweeping regulation in any direction.
Conglomerates. LuNoHo Co is the conglomerate that the revolution utilized to build a massive catapult and embezzle funds. While Mike’s microtransaction financial fraud is interesting (“But bear in mind that an auditor must assume that machines are honest.”), the design of LuNoHo Co. which is described as part bank, part engineering firm, and part oil and gas exploitation firm, interestingly addresses the conventional business wisdom of the times. In the 1960s, coming out of World War II, conglomerates began to really take hold across many developing nations. The 1960s were a period of low interest rates, which allowed firms to perform leveraged buyouts of other companies (using low interest loans), sometimes in a completely unrelated set of industries. Activision was once part of Vivendi, a former waste management, energy, construction, water and property conglomerate. The rationale for these moves was often that a much bigger organization could centralize general costs like accounting, finance, legal and other costs that touched every aspect of the business. However, when interest rates rose in the late 70s and early 80s, several conglomerate profits fell, and the synergies promised at the outset of the deal turned out to be more difficult to realize than initially assumed. Conglomerates are incredibly popular in Asia, often times supported by the government. In 2013, McKinsey estimated: “Over the past decade, conglomerates in South Korea accounted for about 80 percent of the largest 50 companies by revenues. In India, the figure is a whopping 90 percent. Meanwhile, China’s conglomerates (excluding state-owned enterprises) represented about 40 percent of its largest 50 companies in 2010, up from less than 20 percent a decade before.” Softbank, the famous Japanese conglomerate and creator of the vision fund, was originally a shrink-wrap software distributor but now is part VC and part Telecommunications provider. We’ve discussed the current state of Chinese internet conglomerates, Alibaba and Tencent who each own several different business lines. Over the coming years, as internet access in Asia grows more pervasive and the potential for economic downturn increases, it will be interesting to see if these conglomerates break apart and focus on their core businesses.