Designing a Human Language via SoftwareAugust 18, 2015
As I've studied language, one thing that starts to be appreciated is that languages are, in a sense, like an engineered/designed thing. And in that, there are all sorts of "parameters" of the design.
Let's think of written language as a means to render a semantic structure (which we'll model as a graph of nodes and edges) into a 2D sequence of characters. If you want to do that, you need to come up with a scheme for how you do that "translation"/encoding from the graph to the sequence of characters.
There are all sorts of choices you can make.
Take for example the basic clause, consisting of verb, subject, and object:
Rendering this clause, we could choose the order Subject-Verb-Object, we could choose Verb-Subject-Object, etc, etc. Or, do we not restrict order and use some kind of system of affixes to indicate what is the subject, which is the object, etc.
How might we indicate that the event occurred in the past? Add a suffix to the verb? A prefix? Via a new word that comes before the verb? After the verb?
What letters should we use to denote the actual words? The English correspondences are only one option. If we were to invent our own lexicon, what's the best design there?
What comes to mind is that I wonder if one could design a computer program to invent a language. It would use a variety of scoring criteria to explore the possibility space. For example, we want to avoid ambiguity, so choices that lead to ambiguity are unwanted. We want to keep things compact, since that relates to the cost of producing and consuming language. Etc, etc.
An interesting area...The Intelligence RevolutionAugust 16, 2015
The world has been revolutionized many times by the winds of change. In the 1700s and 1800s, the industrial revolution represented a massive shift. In the 1900s, we witnessed the computer revolution. By 1994, the Internet revolution was upon us. So what’s next?
In 2005, the DARPA grand challenge saw its first relatively successful autonomous car race, marking a symbolic if not significant upward shift in the proven ability of computers to interact with the physical world in a profound way. This inflection point was followed in 2007 with an urban version of the DARPA grand challenge, in 2009 with the launch of Wolfram Alpha, in 2010 with the launch of Siri, and in 2011 with Watson winning at Jeopardy. Since then, so-called “deep learning” has raised eyebrows at its impressive abilities to push the boundaries in all sorts of challenging areas, from facial recognition, to learning to play video games.
What is unfolding in front of our eyes is the beginnings of the intelligence revolution. Of note is that it has been a long time in coming. Some people in the 1960s thought it was just around the corner, but as it turns out, more computational power was needed, and more time was needed for researches to start to hone in on intelligence’s secrets.
The next 50-100 years are shaping up to be a fascinating period of history. At its core, the intelligence revolution will make it possible for computational systems to do things that previously only human beings could do. Beyond the philosophical intrigue, this is functionally significant for a three primary reasons.
The first reason is that human workers are extremely expensive, and so society currently must limit the application of intelligence. But imagine a world where there were the equivalent of 1 trillion people’s intellects to work on problems and yet still only 10 billion people’s needs to be met. Suddenly all of those areas of your life where you just don’t have enough time and energy to properly manage become neat and orderly. Anything from meal planning to weed picking to vacation planning.
The second reason the intelligence revolution will be profound is that the fusion of computers with intelligence will produce something with abilities quite unique from our own. Recall that computers can do math millions of times faster than we can, they can remember almost perfectly, they can do an operation millions of times exactly the same, they never get tired, and they can be perfectly copied extremely quickly to produce millions or billions of copies. In the 2030s, tens of thousands of lives will be saved by autonomous vehicles that are watching every angle around your vehicle with perfect concentration, and behaving using best practices. Such an example illustrates how the intelligence revolution will not only make intelligence more ubiquitous, but also apply it in ways that produce a significantly better end product.
Before we look at reason number three for why the intelligence revolution will be so profound, let’s pause for a moment and consider what the above two reasons imply about life beyond 2070. To do this, we’ll consider two more areas where the intelligence revolution will make significant impacts to society.
Consider the world’s energy issues in 2015. Now imagine that solar panels can be produced and deployed without any human labor. A computer system identifies geographic areas ideal for panels, gets human approval, and then goes ahead covering the land with solar panels and connecting them to the electricity grid. If and when a panel fails, it is fixed or replaced autonomously. Even the mining operations that find and harvest the raw materials needed from the ground are almost entirely autonomous. In such a world, it’s hard to imagine that there would be a scarcity of energy. Stop and think about that for a moment. Energy is perhaps humanity’s most basic and profound material requirement, and the intelligence revolution seems likely to eliminate its relative scarcity.
Next, consider the impacts on the physical construction and maintenance of infrastructure. We live in a world where our infrastructure is crumbling. Meanwhile, we’d love to have high speed rail connecting major cities, but the costs are astronomical. Now imagine that roads can be repaved by autonomous systems that can monitor the roads and repair them as needed, 24 hours a day. Want high speed rail between cities? Done. Want a hyperloop system between cities? Done. Even the air transportation industry is set to be transformed. Autonomous air taxis one day will shuttle people around in a more granular manner than air travel today. All of these things are possible in theory today, but constrained by the relative scarcity (and cost) of human intelligence and human labor. Even housing will be dramatically affected, allowing new houses and buildings to be constructed for a fraction of the current cost.
The above societal changes are significant, and this leads us to now consider the third reason why the intelligence revolution will be profound. And that reason is that a world with a trillion intelligent agents will be able to more rapidly progress towards constructing what might be called “superintelligence”. Unlike other thinkers, I am not convinced that a dramatic superintelligence is a for sure thing. It may happen, it may not. If it does happen, I think it’s very unlikely that it will happen quickly. But I do highly suspect that, given enough time, the intelligence revolution will birth intelligent systems that leave us humans in the dust. As we have already considered, the fusion of computers with intelligence will already come with valuable synergies. To review, recall that computers can do math millions of times faster than we can, they can remember almost perfectly, they can do an operation millions of times exactly the same, they never get tired, and they can be perfectly copied extremely quickly to produce millions or billions of copies. If it comes to pass that we build intelligent systems that far surpass our intelligence abilities, those synergies, especially the ability to reproduce perfectly and very quickly, are reason to pause and think.
Imagine a world with a trillion intelligent agents, each of which makes a human being in comparison seem like a serious dimwit. That is a profound thing, but what might it actually look like? Here are a few possibilities…
The first possibility is a revolution in science. In a few hundred years, a relatively small collection of scientists have made a wealth of discoveries in the areas of physics, chemistry, biology, etc. On one hand, we feel amazed on how much is now known about the universe and how it works. And yet we also sense that there are significant pieces to the puzzle that are missing. We wouldn’t be shocked if 300 years from now, people look back at 2015 and smirk at how little we understood. It’s quite possible that superintelligence will be an important part of that continuing story of discovery, and coincidentally, those discoveries are likely to feed right back into allowing yet more intelligent systems.
The second possibility is a revolution in management, governance and justice. Let’s look at the governance piece. Currently, we elect leaders, hoping that their intelligence, experience, and character will be good predictors of their ability to govern our city/state/country. But in a world of superintelligence, we would need to face the question of who should be leading the country. And this is where we get into some uncomfortable territory. But I’d argue that our discomfort, while perfectly rational, is based on our vantage point. We’ve never seen a superintelligence. We have no experience with one. Any system of trust must be earned. The real question is this: If and when superintelligences earn a strong degree of trust, will be put them in positions of leadership, subject to democratic vote? I’m not sure, but I suspect that at least some humans would, and I wouldn’t be surprised if doing so led to stronger decision making as judged by what we typically call rationality. What would the societal, philosophical, and religious fallout of being led at a national, state, and city level, by a superintelligent system?
A third possibility is the expansion of human beings onto other planets and solar systems. If labor were “free” relative to what it is today, and we had superintelligences to design very complex systems very rapidly, it seems highly unlikely that enterprising people wouldn’t want to seek the adventure of populating Mars. Even without superintelligence, this seems likely. But with the scientific gains that a superintelligence would possibly bring, it becomes more likely/possible that it would be uncovered how to travel faster than light. Or failing that, it might become possible to send a probe on a hundreds-of-years journey to another solar system with human DNA on board and the ability to gestate and care for humans on the new planet. While very fantastical, there’s nothing about a long distance probe and gestational system that seems absurdly difficult given a world where intelligence is super-ubiquitous. (I’m pretty sure I’ve already seen pictures on the Internet of animals being gestated outside of a biological womb)
A fourth possibility, and the one that makes me most squeamish, is the altering of human DNA. It seems almost for certain that at some point, people will want to eliminate genetic disease by altering DNA, and I’m all for that, but beyond that, there are many possibilities. Quite possibly, having good outcomes here is beyond human intelligence, but it might not be beyond the limits of a superintelligence to made modifications, simulate the outcome computationally, and then once a high enough certainty is obtained, gestate people with those modifications, or modify already-existing people’s genomes. And what might be modified? All sorts of things… aging characteristics, intelligence, appearance, personality, etc. My spiritual sensibilities are very uncomfortable with this possibility, but it seems like something humanity is bound to crash into.
One final note in conclusion: It should be recognized that the revolutionary changes we’ve been observing in the last few hundred years tend not to “end” so much as they allow subsequent revolutions to begin. It could be argued that the industrial revolution is very much still alive and progressing, just as the computer revolution is very much alive and progressing, just as the Internet revolution is very much alive and progressing. Likewise, it seems likely that the intelligence revolution will be a long process, spanning many decades, and if we’re around long enough, many centuries.
So what comes after the intelligence revolution? The answers may lay in the above paragraphs… the energy revolution, the infrastructure revolution, the DNA revolution, the superintelligence revolution, the governance revolution, the superintelligence scientific revolution, the Mars revolution, and the interstellar revolution. There’s lots of uncertainty in all of this, but make no mistake, this computational intelligence thing is going to make for an interesting story.
Perhaps most fascinating of all is to consider how all of this intersects God's plan for planet Earth and for the universe as a whole. If the intelligence revolution ramps up quickly in this century, I will be fascinated to see how Christians navigate such a rapidly changing world.Elon MuskMay 29, 2015
One of my earliest memories of Elon Musk dates back to an interview he did where he is quoted as saying that he planned to retire on Mars. I remember thinking at the time how bizarre it was that at times one hears seemingly brilliant people say really dumb things.
In the last 12 months, I've seen Elon in more and more headlines, and in the last 6 months or so, I've become quite captivated. I watched the "Revenge of the Electric Car", which I hadn't realized focused on Tesla, watched a Bloomberg documentary on him on Netflix, and more recently read the Ashlee Vance book on him. I have been especially mesmerized by the rocket landing attempts.
A few months ago I came to realize that Elon and his companies are like a super-concentrated formulation of the kinds of things that I find interesting and inspiring. Where to begin...
I think what I find most inspiring is that there are strong indications that he is being successful at making rocket re-usability a reality. When you see it happening, it's a kind of forehead slapping moment. Like duh, why haven't people been working hard on this for decades? When someone can provide the kind of leadership that turns an industry on its head -- and rocket science at that -- you have to sit up and take notice. It's inspiring.
The second aspect that I find inspiring is that he has succeeded at so many different things. Not just SpaceX, but Tesla as well. Not just those two, but SolarCity. And not just those three, but PayPal. It's mind boggling. When you see a pattern of success like that, there aren't really any modern comparisons. As even Elon points out, people tend to notice precedence and superlatives, and perhaps not coincidentally, Elon is the poster boy for both of those things in the tech world.
Of course, when people see patterns, they are tempted to extrapolate into the future. If he has done this much in 15 years, what might the next 15 years look like? Obviously no one knows, but if the pattern continues, it could make for an interesting ride.
There is a long laundry list of aspects of his approach and products that I admire, so here is somewhat of a rambling through them:Best Car Ever
In a world that is increasingly "winner takes all", I am extremely impressed with a company that produces a vehicle that is rated by respected reviewers as "the best car ever" as well as "the safest car ever". If Tesla can achieve scale and bring prices down, who knows where that could lead. (again, the increasingly "winner takes all" nature of the world these days is important there)Fastest Sedan Ever
Watching videos of the sheer terror or glee of people being accelerated from 0-60 mph in about 3 seconds is some great entertainment. As a guy, I can relate to the general feeling -- even if it's just the joy of a V6 engine. Comparatively, I remember the feeling of a 1998 V6 Honda Accord as being "wow" when you accelerated onto an expressway... and it apparently has a 0-60 time of almost 8 seconds compared to the 3.1 seconds of the Tesla P85D. While almost functionally useless, we run into the phenomenon of superlatives again. What is it about superlatives that captures our imaginations?Cheap Space Prices
The US government is working on a super heavy lift vehicle and a space capsule right now with a price tag of something like $18 billion. I think some people are estimating that development + flight costs could total $40 billion. Kind of staggering. When you compare that to the development and flight costs of SpaceX hardware, it's mind boggling. Even compared to ULA, SpaceX's prices are hugely cheaper. In a world that has little patience with government waste and huge corporations that are bureaucratic, slow, and wasteful, the SpaceX prices are a huge breath of fresh air, and they set an example for what can be done with good vision and execution.Long Term Thinking For The Win
Most people probably thought that SpaceX was unlikely to succeed at putting something into orbit. But not only did they do that, they are on the cusp of demonstrating rocket reusability. How did that happen? How can a company go from being an under dog to a savant over achiever? I think a big part of that is that the vision is long term. Not just long term, super-long-term. And when you think about it, of course long term thinking is best. The problem is that we sometimes associate long term thinking with failure, because it can be hard not getting bogged down in the complexity that long term thinking can incur. But Elon seems to be demonstrating that long term thinking (+ execution) is perhaps one of his greatest gifts. And if you really are brilliant at that, then watch out.
Part of what makes it fun to watch is that you know that in any given year, he is thinking about things 5 years, 10 years, 15 years on the horizon. And likewise, in any given year, details are released on something inspiring, and you realize that it has been 5 or 10 years in the making. (in terms of vision + planning + execution) Every time it happens, it's kind of exhilarating. Perhaps it's like watching a grand master chess player, and as the game progresses, you start to sense of much careful forethought has gone into moving the game state to where it is now, for this move.Cars + Computers = ?
As I've reflected on before, kids like me who grew up in the 80s saw computers, realized how much potential they had, and then saw cars, which were kind of "dumb" but also very cool, and wondered in the back of our heads what it would be like if you could combine the two. And better than any other effort, Tesla has achieved that. The Model S is essentially a computer on wheels. Everything from its millisecond response times to traction control and braking to its huge touch screen panel to its automated system for routing you from point A to B to make use of supercharger stations. (not yet to mention that it runs on electricity) And of course, the cherry on top: That Tesla has recently emerged as one of the leaders in the race to autonomous vehicles, showing off some really impressive capabilities this year. For guys like me, my eyes bulge at this stuff. Any time you take the wistful imagination of a once 15 year old and turn that into reality, it's a rather powerful experience.Automation
I've always been a huge fan of automation, scripting, efficiency, etc, but this year especially I've been thinking about automation and its long term consequences. When I hear Elon speak, I get the sense that he's on the same page, and I think that bodes extremely well for the future of his companies. (I think it applies profoundly to all three of Tesla, SpaceX, and SolarCity)
To use one example, imagine a fully automated factory for producing high efficiency solar panels, paired with a fully automated solar panel installation robot/system, paired with an automated order placing system. While that's a ways off (especially the installation robot), it could be as little as 2-3 decades before it is a reality... and I'm curious what that would mean for the future of electricity generation.Beautiful Factories
This seems somewhat meaningless in a way, but there's something about the sparkling, clean, white and red Tesla factory that captures my imagination. I've toured a car factory before, and it was somewhat dark, somewhat dirty. It is something deeply psychological that I can't quite put my finger on, but perhaps there is a certain genius behind it.The Machine That Makes The Machine
Elon often says that manufacturing is hard. I like how he says "it's the machine that makes the machine". He talks about the Gigafactory as a "product", as a "machine". I resonate strongly with that mentality. It may seem trivial, but there's something to it that feels important. Perhaps it's as simple as taking pride in or having passion in not just your product, but also how you make your product, which is also hugely important. He gets it.Hyperloop
When the Hyperloop was first revealed, I didn't pay much attention. Perhaps it was yet one more insane idea that, yawn, would never see the light of day. But as Elon's accomplishments have stacked up in the last couple of years, and as he has established a pattern of saying crazy things that wind up being true, suddenly the Hyperloop is something that my psyche has needed to reconsider.
I think the Hyperloop is important, in part, because it is the "outer most" node in Elon's idea sphere that he has put forth as a near-medium term possibility. It's so on the outside that he hasn't taken it on as his project to manage. So why is that important? It's important because if it is successful, it acts as a new data point for his ability to ideate something crazy (really crazy), not even manage it, and still have it manifest into reality. Which cycles back to our tendancy to extrapolate... at which point the Hyperloop works, we're left to wonder what his limits are for ideating things and having them turn into reality. (again, without him even needing to do the work) Taken to the extreme, you could imagine a genius who establishes such a track record that any time he spurts a hyperloop-like idea, there's a 90%+ chance that it is in fact a great idea that will be turned into reality. Perhaps we shouldn't get too far ahead of ourselves, but you can see the entertainment value in that. Already, the mere mention of just about anything far-out by Elon Musk is an instant news story. How far will it go?Reinvest
I'm not a fan of rich capitalists spending insane amounts of money of themselves. Ugh. For the most part, Elon's companies have not been about making people rich, but instead reinvesting all of the capital into furthering the companies visions. When that happens, it feels somewhat "ideal", that is if the vision and execution are excellent.
That said, Elon does live in a completely excessive house now, so he doesn't fit the ideal as well as he once did.
Something I loved was hearing about how during university he figured out he could live off of a dollar a day by eating hot dogs and oranges. This fits my personality so well -- figuring out how simply one can live for various reasons.Closing Thoughts
I'm obviously very inspired by Elon's successes and potential. Who knows where it all goes.
I have to say though, I'm still somewhat appalled by the thought of 80,000 people moving to Mars this century. That still sounds completely insane, and I'm sticking with that assessment even though Elon's track record is becoming increasingly impressive. We shall see where all of that goes...older >>