Home ⇒ misc ⇒ tech ⇒ Current Article

Technology Forecast - 2016:

Notes and 'Dotes

Dave Tutelman - July 23, 2016

Notes and Anecdotes:
Some things in the original article call for footnotes that would interrupt the flow of the article. And some call up related anecdotes, well worth writing down but not essential to the argument. Here they are, on their own page. Finally, some of the notes are added-after-the-fact "progress reports", either denials or confirmations of the predictions.

Technology forecasting

...And why would anybody even care what I have to say about a technology forecast? One of my jobs during my 40 years with Bell Labs was technology forecasting. In addition to a number of formal forecasts at five-year intervals, much of my career was spent on questions of the sort "what if this actually works?" So I have some experience in forecasting.

I have written an article on methods used by technology forecasters, in case you're interested in how it's done.

Phone service and wideband data via cable TV

When I started working for Bell Labs, the "Bell System" was in fact the nation's phone company. My fellow engineers at The Labs had to understand how the entire phone system worked. As part of new-employee orientation, we were even loaned out to a local phone company somewhere in the USA for six weeks to experience operations first-hand. (I went to New England Telephone in Boston and New Hampshire.)

One of the things you learn with the phone company is that most of the costs are in "the last mile" -- the wires from the phone company's local office to your home. Not the telephone on your desk, and not the big, computerized switching system, and not the transcontinental transmission systems. Nope, just copper wires hanging from poles or buried in the ground. Why? It's the multiplier. A switching system serves thousands or even tens of thousands of households. Long-distance transmission may serve even more. But we need a pair of copper wires -- dedicated -- from the last switch in the TelCo building to the telephone that is going to use it. (Remember, this is before the days of cell phones. If you had a phone, you needed wires.)

The last mile (or three or ten miles, depending on population density) is called "local distribution", and it's where the money is. Find a way to save money there and you're talking big dollars. As a forward-looking technologist (my job for most of my career), one of my jobs was doing assessments of novel ways to get phone signals to the end user for less cost.

In 1979, the question came up of whether the cable TV industry, growing quickly from just about nothing a few years earlier, could be part of the local distribution solution. They had plenty of bandwidth, because TV signals are far more demanding than telephone voice signals. Is there some way to share a little of that bandwidth to piggyback telephone signals on the cable? My group was assigned to answer the question. The lead members of technical staff were Debi Bennett for technology and Tony Cooper for business issues. The way we attacked it was a several-month role-playing exercise: we imagined ourselves as the Sandy Springs Cable Company. Why Sandy Springs? We had good technical relations with Southern Bell Telephone Co. They told us that Sandy Springs, GA, was a very typical fast-growing suburb, the perfect example for a cable-TV market. And they were willing to share a lot of data with us, including street-by-street and manhole-by-manhole utility maps, so we could do detailed engineering.

We quickly became experts on the technology available, and invented our own where needed. (Very little new technology was needed; the market already had niche products for adding phone channels to the TV channels.) Here were some key results of our study:

In 1979, cable company coverage had grown to the point that 50% of all homes in the USA were passed by cables. Of those homes, 50% were cable TV subscribers, and the other 50% could be easily and cheaply connected to the cable that ran by their house. So doing local distribution for phones was a likelihood for about a quarter of homes, and not much more of a stretch for another quarter of them. We couldn't replace phone cables with TV cables; the phone cables were already bought and paid for -- in the ground. But we could defer (perhaps indefinitely) adding any new phone cables. That is big bucks.
If I were a cable company wanting to offer local phone service, my last mile costs would be less than the phone company's copper wires. I already had cable in the ground; my only costs would be the incremental cost of adding phone capability to the cable. I would have to make an investment in some electronics, basically replacing some amplifier modules in my cable. And I would have to turn a few of my TV channels into phone channels. (But most cable TV operators in 1979 weren't using all the TV channels their systems could carry.)
Given these findings, why did it take another 20+ years for telephone service via cable TV to become a large business? The answers are financial, not technical nor economic. (The difference between "economic" and "financial" is the difference between "it can make some money" and "it is the best disposition of our resources to make the most money.") In 1979, and for years after that, cable companies had more lucrative things they could do than offer phone service.

Adding more customers on the routes they already served (houses they already passed) was remarkably low cost, and had a big revenue return.
Pay-TV services (premium channels and pay-per-view) required almost no capital investment at all, and promised substantial revenue. That's a huge margin. Not only a more lucrative return on investment, but it was a better use of those spare channels that they might have used for phone service.

So it was feasible, but there were good reasons it didn't get done for many years.

Three years later, my group (not the same group nor the same project) was looking for ways to deliver broadband data service to homes and small businesses. It had now been ten years since we had figured out how to deliver 56kb/sec data on ordinary phone lines (Dataphone Digital Service), but more than that still required special lines -- making the last mile even more expensive. Perhaps we could use some of the capacity of cable TV for high-speed data distribution. This is not phone service; the premium data service might change the financials for the cable company, and the technology could probably be made to support it.

The cable industry was growing by leaps and bounds; lots of new technology since our 1979 study. Also, my lead engineer on this project, Ken Huber, wasn't involved in the earlier study and had none of the knowledge we had acquired then. So Ken and I went back to college. George Washington University offered a three-day intensive course in up-to-date cable TV technology. So we drove down to Washington DC to get educated.

The answer came out pretty much the same. There were business reasons, not technical nor economic, that limited the approach. For one thing, there were still investments with better return that looked more like cable TV. For another, the industry was still mostly fragmented; much of the nation's coverage was small "mom'n'pop" companies, not today's consolidated industry of big companies. Getting each small company on board and technically equipped and trained was daunting.

That brings me to an anecdote that depends on the small-company nature of the industry, and the fear among those small companies of big companies and consolidation...

There were about 50 people taking the course. Almost everybody there was from a small cable operation, the one or two technical people in the whole company. Only one large company was represented, and it wasn't a cable company! It was AT&T (of course, including Bell Labs); there were a dozen attendees from my company. AT&T was a huge company (about a million employees if you include the subsidiary local phone companies), and Ken and I didn't know the other AT&T students, nor did they know each other. None of us even knew that other AT&T employees would be attending.

But to the small cable company people, it looked like a huge conspiracy. They were convinced that this presence from the Bell System (about a quarter of the students) meant AT&T intended to take over the cable business -- and throw them out of business. We got the evil eye from them for the entire course.

Whenever I was in Washington on business (and that was a lot in the 1970s), I would make time to have dinner with my friends David and Brenda, who lived in Georgetown. So I called them to see which day they would be available. They couldn't make it. Seems that Brenda had a new job, general counsel to the Cable Industry Association, a commercial association (Washington talk for lobbying group) funded by all those small cable companies. She was working overtime to get up to speed on her new industry, and didn't have time for anything else, not even dinner. I said I was there to take a cable TV technology course, and would love to discuss her new job. But no dice.

The middle day of the course, we got out of class to be greeted by huge news -- interesting to everyone in the telecommunications industry. Divestiture was announced that day!!! AT&T and the FCC announced that they were settling their antitrust lawsuit by breaking up the Bell System. The local telephone companies would be spun off as independent companies. AT&T would just be left with long distance, which was a competitive business already. That meant that AT&T would be free to go into other related businesses. Like the cable TV business? That was the first thing on the minds of our classmates when they heard it. If we were facing paranoia from them before, it was doubled now.

And, when we got back to our hotel that evening, I had a bunch of messages from Brenda asking when we could get together. Looks like the paranoid conclusion was drawn by everybody in the industry -- including their general counsel.

Finding meaning in a non-productive life

My brother Bob pointed out,

You do not discuss – but I see as a very significant issue – the problem of finding meaning in a non-productive life. Certainly there will be people whose hobbies will give them satisfaction and a sense of usefulness. But many – perhaps most – of us now find justification for our existence in our work. The person who retires to a life of idleness and dies in a year is cliché.

Bob is absolutely correct. But I don't know what technology nor the overall economic system can do about that -- not that I'm any kind of expert. But let me at least give some independent data in support of Bob's assertion.

During the first half of my Bell Labs career, there was a weekly company newspaper, the Bell Labs News, distributed to every desk. The entire back page was devoted to "milestones": promotions, service anniversaries, retirements... and deaths of [mostly retired] employees. In the case of deaths, the year of retirement was mentioned. Over the course of years, I noticed that there was a strangely bimodal distribution of the time between retirement and death:

The largest number of people lived at least 15 years after retiring.
A smaller but substantial number of people died within 5 years after retiring.
Almost nobody died between 5 and 15 years.

I wondered about that. My first guess was retirements due to illness that soon took the ex-employee's life. But there were too many for that to be the only reason. Eventually I had been working there long enough so that I knew the people In Memoriam. It struck me that those who were dying early were almost always people who had been "married to the job", people with no hobbies who just lived for their work. The people who made it 15 years or more had outside interests when they were working, that they could cultivate to full-time activities after they retired.

So I am in complete agreement with Bob about the significance of the issue. But I don't know how to deal with it, in the context of technology nor the economic system. Education, perhaps. But I don't want to get into my opinions on education here.

Food to feed the world

Again, from brother Bob,

In fact, we are now producing more than enough food to feed the world. The reason that people are starving in some places has to do with distribution and politics. In the Horn of Africa, for example, charitable relief food is deliberately blocked from getting to starving segments of the population by government, insurgent, or paramilitary forces. Even without considering the lab-grown meat discussed in Gollub’s review, today we have huge potential to expand the food supply simply by reducing the amount of meat that we eat. By converting the land given over to livestock and growing feed for livestock to growing food for people, we could feed a much larger world population. The problem is that people enjoy eating meat, and as the economies of poorer countries grow, more and more of their people can afford to eat meat more often – and want their turn at the table. Supply and demand forces have resulted in the current meat-vegetable balance in the world’s food supply, and presumably will continue to do so.

I disclaim any expertise in agriculture and food distribution. I have seen articles supporting what Bob says here, and I have seen [fewer] articles claiming the opposite. I don't know enough to weigh in on one side or the other. Some of the controversial issues that will determine how it will work itself out include:

Politics (as Bob notes).
Global climate change.
Acceptance or demonization of genetically modified foods.
Acceptance or demonization of chemical aids to agriculture.
Animal rights activism.

More on electric cars

I read MIT Technology Review almost daily. On August 16, 2016, there was an article suggesting that the range limitation of electric cars is not such a big deal. Of course, that generated a lot of discussion, some thoughtful and some simply Luddite or self-interested. But one comment was particularly interesting. Martya (that's a forum nickname; I don't know who it actually is) posted an interesting note on possible unintended consequences of a transportation market with a significant percentage of electric cars.

martya

Before one can assert that this will have positive, or negative environmental effects, one must analyze all the globally significant supply and recycling chains that would feed these vehicles. The big questions:

1. Where does this much electricity come from?

At this scale, and with the vehicles running in dense urban areas, the nation will require a large shift in base load demand to night time. This has implications for almost all the electricity infrastructure of the US and the sources of power generation and transmission.

I'm researching solar and wind systems - on the ground - as deployed in the US. Land use implications are far from well understood, and include water use, disruption of "desert crust" which has been recently shown to sequester carbon, and many other unintended effects of seemingly benign infrastructure changes.

2. What are the environmental implications of this much energy storage? (Batteries).

Travel to the mines. At this scale of vehicle penetration, the effects on Earth's physical resources are quite large. Mining uses significant water, in water-scarce places, for example. At an 80% electric vehicle penetration rate in a fleet of more than 230 million vehicles, the impact of batteries required to handle such energy density will be enormous. Recycling batteries is far from a known phenomenon, and based upon early data from around the world, may represent a significant environmental challenge.

3. What are the environmental implications of the dramatic expansion in electronics and data infrastructure required to manage this much power and these many "mobile transportation computing devices"?

Current global e-waste patterns for small items such as the several billion mobile phones recycled every year reveal a significant effect. An electric vehicle can generate 1,000 to 7,000 times more e-waste than a mobile phone, so the global implications of 150,000,000 operating electric vehicles in just the US will be a significant environmental event.

The old joke about New Yorkers who did not know milk came from cows is applicable here....except that it may not be a joke. Every fundamental innovation in transportation, housing, food supply, water use has affected the global environment in many unintended ways.

This is an interesting study, but does not address the core environmental issues and tradeoffs.

This mirrors and expands on some of my own concerns. All of those concerns might be handled eventually. (Might! Some won't be easy, and some solutions will be controversial.) But the horizon of the Singularity forecast is much too close to deal effectively with the issues.

Let me draw an analogy to the settling of the western United States in the 1800s. Chop down trees to build your houses. As long as there were only a few people per square mile, it was cheap and without noticeable side effects. At a thousand people per square mile, it deforests the area. And a thousand people per square mile is just light suburban population density. I live in a suburban town of no more than average density, and that is 3000 people per square mile. It doesn't take that much density or usage to turn a nearly-free resource into a scarce resource.

The electric car is similar. As long as it is just 1% or less of the automobile market, the benefits are obvious. So are the drawbacks, of course. When it starts to exceed 10-20%, it places non-obvious strains on the economy and the environment (non-today, very obvious when it happens).

Progress reports

I read quite a bit of technology news, including the MIT Technology Review every morning. Thus I am bombarded with articles confirming or denying either my predictions or the Singularity University's. Here is what I have seen so far. I am trying to sharply limit the number of references; there is a virtual flood of articles about topics in the forecast. I'm especially avoiding articles pushing specific products, looking more for dispassionate overview articles.

Artificial Intelligence

The press has a nearly continuous flow of stories on AI. I'm not going to list every one I see here, just those that seem to me indicative of a trend, or a broad observation about AI.

2017 Oct 6 - Rodney Brooks, an MIT researcher in AI, presents a set of common misconceptions about AI, that you'll need to avoid to make sensible predictions about it. In fact, you should be aware of these if you're going to be intelligent about even reading AI predictions.

2019 July 14 - Since my 2016 article, there are been one headline after another hailing yet another AI achievement. I am thinking a little differently today:

AI techniques, specifically learning from massive amounts of data, are solving well-defined problems and solving them pretty well to very well. Things like face recognition and the ability to beat humans in increasingly complex games are very impressive. Just this week, an AI program managed to conclusively beat some of the world's best poker players, including wagering and bluffing.
Each of these successes has been completely isolated. In fact, one characteristic is that even a small broadening of the problem is "back to square one" with defining the problem and training from data. So it is more a great leap of applying computation to a problem, and less an obvious step towards "the singularity". Generalized intelligence has not been part of the AI track record.
AI is facing a lot of pushback, and not just from Luddites. AIs related to people, such as face recognition, exhibit racial and gender bias, and it is not obvious how to change the data to completely eliminate such bias. AI programs can be (and possibly have been) used by repressive regimes to further subjugate their citizens. And AI "algorithms" (I hate that application of the word; they are at best heuristics) are completely non-transparent; even the programmers/trainers don't understand the rules under which they operate. This sort of problem has given rise to a whole field of "AI ethics", which is increasingly saying, "Hold on there! Just a minute!" This trend is likely to push "the singularity" back significantly, if it happens at all.

2019 Nov 5 - I have long been skeptical of generalized AI, which is what would be needed to achieve "the singularity". We have found ways to automate very specific tasks. Even small changes in the task require expensive retraining, as noted in the previous reference. And retraining is expensive. A modern AI application based on neural networks can require more than a million dollars worth of cloud computing to train from data. And the carbon footprint is scary big; that same training uses as much energy as five automobiles in their lifetimes. So I saw AI as addressing only specific applications or tasks that had a high value to automate.

Can generalized AI be evolving, in a place or a way that I'm not seeing? Perhaps so. It appears someone is working on the problem of making these tasks work together based on context. This contextual orientation is, IMHO, essential to generalized AI, probably even key to it. And it is not being driven by the research community. Rather, it is the future that the world's third-largest company sees for itself. So funding is not a problem.

What is this AI breakthrough to watch? It is Amazon's big plan for Alexa, the voice-operated personal assistant at the core of its smart speaker product. It may have started as a smart speaker, but Alexa has spread to more of Amazon's product infrastructure, and is collecting data on how those products' owners live. The picture at the right is an illustration of what the Alexa team is looking to do in the short term. In order to develop this level of sophistication, it will have to crunch the day-to-day data of millions of Alexa owners' use of the existing tasks Alexa can do. That's a huge amount of training. But now somebody has (or can and will get) the data. And that somebody has the financial wherewithal and incentive to bear the expense to crunch the data.

That is simultaneously encouraging and scary. No not just scary for people who fear the singularity rather than welcoming it. Getting there involves a huge invasion of privacy by a profit-driven company. Amazon is in business to sell you stuff. This big AI push is not just to sell Alexa-enabled products. It is to be in a position to know when you will need something, and try to sell it to you before you start shopping with one of their competitors in retailing. That's the scary part.

2022 Mar 10 - An article in Nautilus by Gary Marcus pointed out that "deep learning" was running into a brick wall in the biggest applications of AI. Deep learning is the "training" of a simulated (or even real hardware) neural network to learn to perceive things by very extensive trial and error. It is pattern recognition rather than what Marcus refers to (correctly IMHO) as "symbol manipulation".

The article points out some recent high-profile and very critical failures of deep learning applications, most dramatically some fatal accidents by self-driving vehicles. He points out the limitations of deep learning (so far, anyway, thought he doesn't concede that) to pattern recognition, and to the exclusion of logical reasoning. I happen to agree with that assessment, and I can see no path around some of the problems if sticking exclusively to deep learning.

Marcus' conclusion is that the only way to real and reliable AI application is a hybrid approach: deep learning for pattern recognition and symbol manipulation for abstract combinations of those recognized patterns. I would agree. Although I think Marcus is biased and perhaps I share his bias, the evidence seems to support him.

Autonomous vehicles

I could be wrong on this one. Time will tell. My specific skepticism was that by 2020 it would replace the need for owning automobiles. The Singularity University prediction was that an autonomous car would be available on call almost immediately, and could deliver itself autonomously. Think Uber, but without a driver -- resulting in a cost that is competitive with owning your own car. I am still skeptical anything like that will be widespread in 2020. But the autonomous vehicle technology seems to be coming along much faster than I expected. I see articles all the time of another company committed to it. Many articles do not necessarily report technical progress, but at least dollars committed to making it happen. I'll cite some of those articles here -- but they are way too numerous to mention them all.

2016 Autumn - Not only car companies, the leading software and computer giants are tossing their hats into the ring. Apple and Google are on board. And autonomous trucks are being investigated and even tried.

2016 Dec 5 - I guess we should not be surprised that Uber itself intends to make use of this. They would not want to be put out of business by autonomous car sharing. If you can't beat 'em, join 'em. Of course, all those people who are earning their living as Uber drivers would be jobless.

2017 Dec 14 - A very level-headed article from Science Magazine. It doesn't focus so much on the technology, but rather how fast it could be available and how fast it could be accepted -- two different questions. It looks at capabilities during six stages of making a car driverless. The early stages are tagged as "now" and "soon" (with dates). It rates the last stage as "somewhere over the rainbow". Unfortunately, the last stage is where Uber and trucking companies need to be in order to fire all their drivers. Well, maybe the next to last stage if they don't mind shutting down in adverse weather, at night, specific locations -- in other words, when you want a taxi. (As usual, I looked at the author's bio; he's a staff writer with no obvious axe to grind.)

2019 July 10 - An article in Fast Company, a generally insightful business publication, takes a very pessimistic view of of both autonomous vehicles (AVs) and electric vehicles (EVs). It cites many industry insiders revising outward their projections of when these technologies will "happen" -- and wonders what could change to make them happen at all. My skeptical comments are positively rose-colored compared to their assessment. Not only is the technology way behind, but the companies pushing them are putting themselves in danger of going broke in the process. It's a very tough business to be in, and they see nothing in the near future to remedy that.

2022 Mar 10 - See the Gary Marcus article, "AI is Hitting a Wall". One of the key examples in his argument is autonomous vehicles.

Electric power and electric cars

2017 Mar 16 - One way to deal with the 1/3 duty cycle of solar power (due to the earth's rotation) is a worldwide grid. I cited the technical and political difficulties in making it happen. But someone is trying, or at least working on part of the problem. Europe is geographically large enough to have diverse non-fossil energy resources (solar in the south, wind in the north, and geothermal in Iceland). And to a significant extent, the European Union is one political entity (though strains are beginning to show). The EU is upgrading its grid to get closer to completely renewable energy. From the article:

An international power grid is gradually developing, using power interconnectors to trade surplus energy across national electricity networks, allowing big wind power producers in northern Europe, for example, to trade electricity with large solar energy generators in southern Europe.

2017 Mar 29 - The most often touted solution to the 1/3 duty cycle of solar power is batteries. In 2017, almost all the R&D is going to the lithium ion battery techology. Let's assume that all this is successful, and we figure out how to make very efficient LiIon batteries: weight and volume efficient, fast charging, and cheap to manufacture. That will put a huge demand on the world's supply of lithium. Even if the market were only electric cars and electronic gadgets, the lithium market is starting to spike, based on world supply. If we were to add the much larger market of reservoirs for solar-generated power, Li-Ion batteries may not be cost-effective.

2017 Jun 8 - Apparently not just lithium. Cobalt is also important in the manufacture of high-energy-density batteries. And the demand for those batteries is pushing up cobalt prices. From the article:

There’s no way that current supply is going to keep up. Some of the forecasts for cobalt supply are pretty dire.

2017 Oct 31 - Nickel, too. It's about to become difficult to get enough of it, if the battery market keeps growing. The article is about new opportunities for the nickel producers, but makes it pretty clear that enough battery growth could disrupt the nickel market.

2018 July 27 - My criticism has been about the daily variations in availability of solar (or wind) power. Never even thought about annual variation. Turns out variation over the course of the year would be even more demanding of batteries. Several studies show that 100% conversion to renewable power in California would be prohibitively expensive, even if the cost of batteries came down by a factor of three.

2019 July 10 - See my comment in Autonomous Vehicles, about the article in Fast Company. It also addresses electric vehicles.

2020 Jan 23 - Germany has a mandate to close all its nuclear power plants. That is proceeding, but the cost in CO2 emissions and other pollution is high. Wired presents an argument that both the cost and the health and safety implications of shutting down nuclear power represents a big problem for the US, based on this German experience.

2020 Nov 5 - A common theme from these follow-up items is what happens when you try to scale a new technology, and run up against straining the supply of some normally ordinary material. And here it is again -- with a vengeance. The manufacture of solar panels requires a lot of glass. Glass doesn't sound like a big deal, but the solar panel market has created a shortage of glass. This article from Bloomberg Green points out how competing demands for increasingly scarce glass (because the world's production facilities are being outstripped by demand) may push the price of solar up so it is no longer competitive.

Future of Work

The technical press has lots of articles about robots either replacing or cooperating with human workers. Each tends to look at some specific new robotic offering, and most believe what the company making the offering tells the reporters. It may be factual but, self-serving as it is, I find it hard to put a great deal of credence in it. I'll spare my readers the time of reading anything not well thought through. That said, here are articles that address the problem systematically, rather than a selling point for some robot product.

2017 Oct 4 - This is an MIT economist's view of the middle-term future. He makes a distinction between "enabling" and "replacing" kinds of technology. Replacing technology obviously puts workers out of work. Enabling technology helps workers do their work better, makes them more productive. He says that the history of technology shows enabling technology to far outweigh replacing technology, so the destruction of human work is not imminent. But bear in mind that the American Enterprise Institute, which published this article, is a decidedly conservative think tank. Some of this bias comes through, and you have to read it with an eye out for political point of view. For instance, he calls for an end to public policy that subsidizes the replacement of labor with technology. Reading between the lines, he is proposing weakening labor laws that make it more expensive to employ human beings. That may or may not be the right answer, but it is certainly a politically conservative answer.

2017 Dec 12 - A good article, with more input from the legal profession than software companies. The carry-away I got:

Paralegals are in trouble first -- document discovery is being successfully automated.
Temporarily, it may help lawyers themselves, especially young lawyers learning the trade. Working with software rather than delegating to a paralegal is teaching them, making them better lawyers -- and more productive lawyers -- faster.
Of course, more productive lawyers means fewer jobs for lawyers; many of them are thus endangered, too. Especially true for new law graduates looking for a first job.
Other white-collar professions better be looking over their shoulders. I don't see this pattern as something that can't be duplicated elsewhere.

2018 Mar 27 - "AI won't replace doctors anytime soon." That is the conclusion reported about a Google paper on AI diagnosis. It seems to indicate for doctors what the article above says about lawyers. The software reportedly allows doctors to be very effective and more efficient, suggesting a new paradigm for the effect of AI on knowledge-worker employment. The old paradigm was replacement; AI is going to take my job. The paradigm suggested by this article is twofold:

If I use AI as a tool, I will be more effective. I will be able to do a better job.
If I use AI as a tool, I will be more efficient. But this is replacement by slow degrees. If I am more efficient, that means I can handle more patients, which means fewer doctors are needed than would be needed without the AI.

2018 Apr 11 - Continuing in the vein of replacement by slow degrees, Ajay Rajadhyaksha of Barclays reports the result of their study: "Technology frequently ends up lowering the skill-set needed to do a job, in turn expanding the pool of potential workers, which then acts as a drag on wage growth."

That's pretty simple supply and demand. It has been the result of technology for centuries. The trick to continued job and wage growth is for new jobs, requiring more skills, to emerge. I find it hard to see those jobs at the moment, but that may just be my lack of foresight. More disturbing, I don't see a work force with higher skills developing; our system seems to be dumbing down the population.

Bitcoin and other cryptocurrencies

2019 Jun 12 - Since I wrote the forecast, much has been made of the "carbon footprint of bitcoin". Bitcoin, and other cryptocurrencies as well, require computation to create the crypto pairs. In fact this "mining", as it is called, is essential to the entire notion of a cryptocurrency. And computation takes energy. Most of our energy currently comes from burning fossil fuels, so creation of new bitcoin money also creates greenhouse gases. How big is this carbon footprint.

The article cited is a more accurate (and considerably lower) estimate of bitcoin's carbon footprint, than previous estimates. It places it at 0.2% of the world's use of electricity. That doesn't sound like a lot, but it means that creating bitcoin "money" uses as much electricity as all of Kansas City. Still, bitcoin enthusiasts think that is worth it. Is it? Not just for the few enthusiasts, but as a full, functioning economy?

The best estimate I can get of bitcoin's standing in the overall world of money puts it at about 0.05% of the world's money. Let's look at a few more numbers to give us some scale.

Bitcoin = $41 billion
All cryptocurrencies = $100 billion
All money = $83.6 trillion = $83,600 billion

So what does this say about bitcoin? Or even cryptocurrencies in general, because we know bitcoin has a limited size of the economy it can handle? Bitcoin already costs more in electricity than its overall size in the economy: 0.05% of the worlds economy, but it costs 0.2% of the world's electricity. If cryptocurrencies grow as a share of the economy, so will its power consumption. So if the bitcoin enthusiasts are right about its being the future of the world economy, we better have a far superior way to generate electricity, because just maintaining the money supply will eat up a significant chunk of the power we generate. If it approaches a quarter of the world's economy (without a substantial improvement in power use), it will require as much power as we use for all other things combined.

Since I am talking about technology here, let's assume technology can improve on the power consumption of cryptocurrencies. (That is a big assumption, because the value of a cryptocurrency is derived from the work necessary to mine it. Make that more efficient, and you probably devalue the currency.) Then we must also assume that the world's electric power needs are going to increase markedly as well, for all the technological reasons elsewhere in this article -- like electric transportation rather than gasoline or diesel.

That does not sound like a ringing endorsement of cryptocurrencies to me. They are grossly wasteful of resources and, until the electric power problem is solved, bad for the planet as well.

Online education

2016 Dec 14 - My assessment of the performance of online education is still skeptical. I wrote, " I don't know anybody without skin in the game who will argue that the smartphone is currently making our society smarter, on average." Now someone with skin in the game seems to agree with me. Sebastian Thrun, who invented the "massively open online college" (MOOC) course and founded Udacity to deliver it, has apparently changed the concept drastically. According to the article,

Udacity’s completion rates were as low as 2 percent, and the people who made it through were mostly the kind of well-motivated students already served well by conventional institutions.

He has changed Udacity's modus operandi from college degrees to vocational "nanodegrees" focused on specific technology skills (mostly in software development). That seems to be a success so far, confirming my assertion that it will help a small percentage of people.

Last modified - Jan 24, 2020