« Google Earth Adding Immersive Travel Content | Main | The Age of Democracy »

Comments

usnjay

At the same time I feel (distantly) sorry for my ancestors unable to enjoy all the benefits of modern living, your article makes me feel sorry for MYSELF that I won’t be around for future miracles. I would love to experience the history as humanity sets foot for the first time on an Earth-like alien world. Can you imagine starting over with the Earth with all of our current knowledge to avoid the mistakes of the past!? I’ll have to take comfort in the fact that my efforts today will be some contribution to that future, even though I won’t be there to enjoy it.

Chuck the Lucky

Another example of an accelerating trend that I had never heard about. Everyone knows that modern observatory technology is better that what came before but it is surprising (though it should not be) to realize that the technology is accelerating. I wish it were easier to get other information on trends over time. Looking for, say demographic information will usually result only in snapshots about who lives where TODAY or who believes what TODAY. It is almost as if no one has heard of rates changing over time. Thanks for being an exception.

GK

usnjay,

Well, your lifespan might be longer than you are currently estimating. Technology is extending lifespans as well.

GK

Chuck the Lucky,

Thanks. Very few people grasp the accelerating nature of time and progress. So many supposed great thinkers miss this, making predictions that assume the rate of change is constant.

I have a whole set of articles on accelerating change. It is the single most important ingredient of any future predictions.

Chuck the Lucky

"Very few people grasp the accelerating nature of time and progress."

I agree. Even when a trend is not shown to be accelerating (at least not conclusively) it would still be very useful to know if there has been a recent change in the rate of change. For instance, we are often provided with information about a change such as literacy in a given area or demographic over a certain time period but we are given no impression about whether the rate of change is constant or speeding up / slowing down. The same goes for the rate of urbanization or the spread of memes or migrants... just about everything is handed to us in snapshots. Pity. It is much like wanting to understand calculus in a pre-calculus world. Time is a whole other dimension and yet it is treated as if it were an irrelevant factor in life.

Even when we personally understand the implications of accelerating change in certain fields (as much as humanly possible - given our lack or experience with accelerating processes in our historical and personal past), it is very difficult to understand how different fields interact as they change. Still, tricky as it may be, when one sees an implication that one has not considered before it is quite a eureka moment.

I have been wondering lately (as many have) if there is a way to maximize accelerating change. I realize that technology and, to an extent science in general are driving most aspects of acceleration so increasing the number of scientists and engineers in the developing and developed world is one goal and harnessing the simple AI available today and other information technologies like databases, uniform standards and open source projects is another. I suppose that the question is one of priorities. If I had huge gobs of money to invest in R&D projects and my priority was to increase the rate of acceleration rather than return on investment, what sectors should I dump the cash into?

Chuck the Lucky

The answer to my own question might be to support technologies that augment our powers of making observation (microscopes, MRIs, telescopes, cameras, sensors etc.), storing, communicating and processing data. Processing being perhaps the most important since our ability to gather data is somewhat lacking our ability to make use of it.

GK

Chuck the Lucky,

I would say that processing power and AI are strong candidates.

Technology of miniaturization is another, as per here.

Open sourcing of everything (software, wikipedia, Google travel blogs) is a natural progression and a means to an end.

Economic growth is also exponential and accelerating. The curve is rather smooth. The world will grow at about 5% a year, and if the US, Europe and Japan are below this (2-3.5%), then the developing world will necessarily be above this (6-10%) to get the weighted average to the trendline.

kamil

One good way to maximize accelerating change is invest in Grid computing\ tools and applications.

When fully realized this will change the actual use of computers form 1-15% to close to full utilization.

More importantly the ability to create a world wide pool of computer resources and resources according to demand.

On demand ad-hock supercomputing pay by the hour,
or invest in a data center turnkey business and
have people pay you.

This results in much greater returns on investment in computing, and therefore demand for computers creating a fast virtuous cycle.

Unfortunately we have to research, develop and test the
most basic of tools now, in order to develop more complex tools in order to then rewrite most of the applications to take advantage of these technologies.

kamil

Dan H.

You have to be careful about extrapolating these trends into the future. In the 1970's, it was fashionable among science-fiction and science afficionados to trot around various charts that 'proved' we would be travelling faster than light by 2010. By plotting the maximum speed attained by humans over history, you come up with a startling exponential growth curve - for thousands of years, humans could only travel as fast as they could run. Then they domesticated horses, and went a little faster. Then came steam trains, cars, and then within the last hundred years we went from the car to the airplane to the spaceship. Plot that trend, and it shows that we should be travelling at immense relativistic speeds now.

The flaw, of course, is plotting the progression as a smooth line rather than a series of discontinuous jumps caused by breakthroughs in technology. When you think of it that way, the future becomes a lot murkier.

In the 70's, other extrapolations of current trends 'proved' that we would all be starving by 2000, and that we would run out of raw resources by then. Based on the same faulty logic.

A better way to analyze what we'll be looking at in 20-100 years is to consider the theoretical limits of the techniques that have driven us to the point we are at, and then to consider the proposed telescopes of the next century and their theoretical limits. For example, we've made big jumps in resolving power for Earth-based telescopes with the invention of adaptive optics. But there will be a limit to how much that can do for us, and we need another innovation to continue the trend. In the near term, interferometry arrays on Earth and in space offer promise, but the big space-based interferometers like the Terrestrial Planet Finder are still decades away.

GK

Dan H,

Noted, but these are telescopes that are already under development.

On the issue of transportation speed, those projections in the 70s were faulty, which was apparent even then. We have a better one here, that shows a wide range of times in which we may reach light-speed, given a variety of conditions.

Dan H.

And that graph is just as useless. All you're doing is changing the baseline period to make the curve come out longer.

There are plenty of examples of rapid change due to new technology that flattened out once the technology was mature. Bursts of economic growth due to transportation or manufacturing advances, for example.

Try this experiment - re-draw your curve of mankind's maximum attainable speed - only start from 50 years ago. Then extrapolate that curve out until you hit the speed of light. Tens of thousands of years, huh?

Curves like this are useless if they don't reflect a process that is continuous and predictable. Currently, we have no idea how to accelerate large cargos anywhere near the speed of light. To get there requires a breakthrough. You cannot predict breakthroughs by extrapolating on a graph.

Here's an example: Aircraft speed. I fully remember people using trend extrapolation to claim that we'd all be traveling around the earth in hypersonic transports by now. After all, it took us 30 years of manned flight to go from 100 mph to maybe 350. Then in the next five years we got to 550 or so, but within 10 years after that we were going Mach 2. For the people living in that era, it was not an unreasonable assumption that we'd be going Mach 10 in a few more years.


But guess what? We hit some physical limits, and progress stopped. Today's fighter jets are no faster than the F-104 Starfighter, designed in the 1950's. With the ending of the Concord, airliners are actually slower now than they were 30 years ago.

As another example, consider power density - if you graph that over the industrial revolution, you'll get the same kind of exponential growth curve as we moved from water wheels to steam to internal combustion to nuclear. A curve like that might show that we'll very soon we'll be able to generate all the power the earth requires with a device that fits on the head of a pin. Does that sound likely? In fact, our power density curve has probably started moving backwards now, as we move away from nuclear and towards lower-density power sources like solar, wind, and coal.

Another example is Moore's law. Moore's law is often graphed to show that we will have immense computational density in the near future. The problem is that Moore's law only applies within a narrow portion of the curve - the one we happen to be in right now. Sooner or later, we will run into physical limits that will break the law down. We may have some more breakthroughs along the way that help us extend this a little further along, but someday it will end. Then we either discover something new to continue improving, or we'll flatline at our new physical limit.

Whenever futurists attempt to predict the future by doing extrapolation of current trends over a long period of time, they generally miss their mark, badly. It's better to think like an economist - to consider all the forces affecting the current world, how they are changing, and how that might change the future. For example, taking breakthroughs in nanotech and extrapolating what will happen if a material with those properties becomes widely available to car manufacturers. That makes sense to think about. Trend extrapolation is a waste of time, unless it's a trend that describes an ongoing, well understood process, and not just an aggregation of inventions and processes that already happened with the assumption that future will mirror the past.

Moore's law made sense, within certain limits, because it was an extrapolation of known techniques as they become more refined. Gordon Moore would be the first to agree with me that mapping that curve out over the next 500 years makes no sense.

GK

Dan H,

Not all trends are accelerating at all times, but many that are go unnoticed, as people fail to see the big trend under the series of S-curves, while fixating on one S-curve.

Moore's Law was specific to semiconductors, and hence will have a life from 1958 to about 2018. However, exponential doubling in computing power existed before semiconductors (with vacuum tubes and relays before that). After semiconductors, computing will move to carbon nanotubes, molecular computing, or even quantum computing.

Regarding telescopy, the trend is certainly a staircase. But the evidence is clear that progress is much faster than it was in the 18th or 19th centuries, where a doubling of power took 50-80 years. The first exoplanet was not found until 1990, and now we see hundreds. The size of planets discovered continues to shrink too. Look at the chart.

There are many powerful new telescopes under development already, with power much greater than the Hubble. They are both orbital and land-based.

Dan H.

Sure. And based on what we know we are building over the next 100 years, we can certainly make some predictions about what we'll be observing. The way we do that is not to extend a curve of resolving power over time - it means nothing. For example, the curve of past resolving power since Galileo was really a curve of improving manufacturing processes for large mirrors. That curve flattened out when mirrors got so big they ran into other limits.

Then the curve took off again with computing power. We learned to build segmented mirrors and adaptive optics, and CCD cameras and computer processing allowed us to spot all kinds of new celestial objects. That curve didn't look anything like the curve in single mirror optic improvements.

The next curve will be marked by advances in orbital and terrestrial interferometry. We know where that is taking us in the short term - we can make very good predictions about how much resolving power we'll have 50 years from now (and it's a LOT more than we have today). After that, it gets murky again. Theoretically, we could build very long baseline arrays of massive sizes in orbit, and resolve incredible details around our nearest stars. But today we have no idea to solve some of the problems inherent in that, nor do we know what the economics of it look like.

Let me give you another example of a curve that everyone used to predict a grand future but which totally failed: human spaceflight. We went from launching a can 50 miles into space to putting men on the moon in less than 15 years. That curve had many, many people convinced that by 2000 we would have spread throughout the solar system, that there would be huge moon bases and commercial space stations. None of it happened. No one predicted that Apollo would be our last grand hurrah in space and it would take us at least 40 years just to get back to that point. That is, no one who simply plotted our advance in exploration on a curve and used it to extrapolate the future. You would have had a much better shot at predicting what would happen in space by examining the forces that drove Apollo, the limits of the technology it used, and the political and economic scene on Earth.

I see you're doing the same thing with economic growth, and using that to predict a rosy, rich future. The same problem exists. One of the reasons the growth looks so smooth is that the real growth comes when you move from an agrarian to an industrial economy. Once you reach '1st world' status, growth begins to slow dramatically. But it continues on a global basis because once one economy fully matured (Europe, for example), other ones who were further behind pick up the slack. China, India, and a few other countries are driving global growth right now - Europe is almost stagnant. What happens when China gets to that point? Or what happens if China fails to make to that point because of its authoritarian government? Your curve extrapolation is miserable at predicting that, because it's based on events that have happened in the past, not in the future.

GK

Dan H,

Don't fixate on just one technology. Consider each level of technology (as you mention) as S-curves that combine to create a megacurve that is both smooth and accelerating.

I come back to computing. Before Moore's Law for Semiconductors, there were relays, then vacuum tubes. Then Moore's Law from 1958 to 2018 (first noticed in 1965). After semiconductors get saturated, new technologies will emerge. But the cost of MIPS per dollar has been a smooth exponential curve for over 100 years, and will continue until a technological singularity.

Regarding space, people were merely projecting the wrong trend. Do you know that putting a man on the Moon today would cost only 1/30th of what it did in 1969, in proportion to GDP? Why do you think SpaceShipOne, Virgin Galactic, etc. are now starting to become serious business ventures? There HAS been accelerating progress here - some people just didn't know how to detect and observe the trend, and thus chose the wrong one.

Regarding economics, your criticisms are too short term. The smooth, exponential curve existed for thousands of years, long before implications of 'Europe vs. the US vs. China' were even applicable.

The World GDP trend is now solidly at 4.5% a year. In the 19th century, it was only 1% a year, and in the 15th century, only 0.1% a year. Some regions will be above the trend and some below, but the weighted average is what matters. If China and India moderate, the US will accelerate to keep the weighted average at the trendline. The US is already growing faster than it was before 1995 - the productivity growth rate shifting into higher gear with the IT revolution is well-documented.

An even bigger example of accelerating change has been evolution. Over 4 billion years, the time intervals between increasingly advanced, intelligent creatures emerging has shrunk exponentially.

Thus, along with thinking about accelerating MIPS per dollar, Artificial Intelligence surpassing human intelligence by 2050 is quite possible, and cannot be ruled out.

Don't use a few trends (projected by people with a poor understanding of how to assess this) to ignore the many genuine accelerating trends that do exist, and will continue for a long time.

The comments to this entry are closed.