Random Thoughts

Recalculating Worldwide Computing Power

 I previously speculated on why economic growth in the 21st century will be spectacular..My calculations then were based on sales of personal computers and the expected power of each personal computer. It wasn't a very sophisticated analysis, so I was happy to recently learn of this great paper that estimates the total power of graphical processing units (and central processing units) in the world.

That paper calculated the total processing power of all GPUs in 2007 to be 6.4 x 10^18 instructions per second, and that the rate was increasing by 86 percent per year. I translated that into human brain equivalents (HBEs) by assuming that one HBE was equal to 20 quadrillion instructions per second (20 petaflops, or 20,000 teraflops) or 100 trillion instructions per second (0.1 petaflops, or 100 teraflops). The value of 20 petaflops comes from Ray Kurzweil's work, will the value of 100 teraflops was proposed by Hans Moravec.

Anyway, the graph below has two straight lines for those two cases, and one curved line, which represents the results of my earlier calculations. The bottom line is that the curved line representing my earlier calculation is very, very close to the result from the paper, assuming that one HBE is equal to 20 petaflops (20 quadrillion instructions per second). Another bottom line is that we're talking about some pretty stunning numbers very soon. For example, by 2025, the number of HBEs from computers (even using the 20 petaflops value) is approximately 1 billion. And 

A bet offer to Anders (of "ATTP")

I was having a discussion with Anders of the blog And Then There's Physics. He claims that the statement "that future anthropogenic emissions would need to be eliminated in order to stabilize global-mean temperatures" is true.

I'm challenging him to put his money where his mouth is. I bet him the equivalent of 30 U.S. dollars that I can honestly answer the following multiple choice question, and he cannot:

Which of the following statements must be true, in order for global temperatures to be stabilized:

1) Anthropogenic emissions must be eliminated.
2) If anthropogenic emissions are not eliminated, then we must somehow remove as much CO2 from the atmosphere, and sequester it somewhere, as we’re emitting into the atmosphere.
3) Both of the above.

My honest answer to this question is "#2."

If Anders can honestly answer this multiple choice question, I will give him $30.

Jobs vulnerable to artificial intelligence, part 2

I previously posted on the jobs that I think are most vulnerable to artificial intelligence (computers, robots). Here's a table with the number of workers (in millions) in each of the top 15 job categories in 2012, and my predictions for what the employment will be in each of those categories in 2024, 2034, and 2044. I think the two jobs most vulnerable to artificial intelligence are cashiers and truck drivers; I expect about a 90% reduction in their numbers from 2012 to 2044. Other jobs I see as particularly vulnerable are bookkeepers/accountants and material movers (e.g. loading dock workers). But I don't see even one of the top 15 job categories in 2012 as gaining workers by 2044. And overall, I expect about a 50%(!) drop in the number of people in the U.S. employed in the top jobs of 2012 by the year 2044.

Robin Hanson recently had a blog post with the title, "This time isn't different"...relating how previous predictions of job losses due to computers and other machines didn't come to pass. I beg to differ. I think this time most certainly is different. That's because this time we're talking about computers/robots very shortly (in the next 1-2 decades) generally equaling then surpassing the capabilities of a human brain and body.  

Popularity, 2012	Job	Employment relative to 2012 (i.e., 2012 = 1.0)					Percent Change, 2012 - 2044
		2012	2024	2034	2044
1	Retail   Salespeople	1	0.95	0.48	0.29		-71
2	Cashiers	1	0.88	0.53	0.12		-88
3	Office   Clerks	1	0.93	0.71	0.50		-50
4	Fast   food prep and service	1	1.04	0.89	0.59		-41
5	RNs	1	1.00	1.00	0.88		-12
6	Waiters/waitresses	1	1.04	0.78	0.52		-48
7	Customer   Reps	1	1.05	1.00	0.91		-9
8	Material   movers (e.g., loading docks)	1	1.00	0.62	0.14		-86
9	Janitors	1	1.05	0.57	0.19		-81
10	Stock   clerk, order filler	1	1.16	1.26	0.84		-16
11	Secretaries	1	1.06	0.94	0.83		-17
12	Bookkeeping,   accounting	1	1.00	0.56	0.22		-78
13	General   managers	1	1.00	0.94	0.88		-12
14	Tractor   trailer truck drivers	1	1.06	0.56	0.06		-94
15	Elementary   school teachers	1	1.00	0.87	0.53		-47
Totals	Top 15 jobs in   2012	1	1.00	0.76	0.48		-52

Postscript (11/10/15): I should have made it clear that all of these predictions are population-weighted. For example, I have employment in 2044, relative to 2012, for all top 15 jobs, as 0.48 (i.e., a 52% reduction in the number of total jobs in all 15 sectors). But that's a population-weighted number. For example, suppose the total U.S. population increases by 20 percent from 2012 to 2044. Therefore, a population-weighted value of 0.48 would actually look like 1.2 x 0.48 = 0.58. So if the population increases by 20 percent, the absolutely number of jobs would be 0.58 times the 2012 value, rather than 0.48 times the 2012 value. Of course, if the population didn't change at all from 2012 to 2044, the 0.48 value would be appropriate to calculate the total number of jobs relative to 2012.

"Citizen Scientists" of the world: Seriously, the emperor is naked

Most folks know the story of The Emperor's New Clothes. It ends when the innocent little boy says the truth, "But he isn't wearing anything at all!" Then all the people are no longer scared to admit the truth, and they acknowledge the emperor is indeed naked, and everyone lives happily ever after. (Except the emperor's two weavers, who die mysteriously of radiation poisoning a few months later.)

But I propose a different, "Fractured Fairy Tales" ending, in which the people shout down the boy: "You don't know what you're talking about! Look at the weave! It's extraordinary!" The boy then tries several more times to demonstrate why the emperor is indeed naked (such as describing moles in private places). The people then shout that the boy is a Clothing Denier. And the boy eventually shrugs and gives up.

I've had a similar experience recently, trying to point out John Cook's (and others') misrepresentation of the paper, "Quantifying the Consensus on Global Warming." Specifically John Cook wrote, in a later published paper, "Of the 4,014 abstracts that expressed a position on the issue of human-induced climate change, Cook et al. (2013) found that over 97% endorsed the view that the Earth is warming up and human emissions of greenhouse gases are the main cause."

This is an absolutely blatant misrepresentation of the paper. To put it more bluntly, John Cook is lying. There is simply no way on earth that, "Of the 4,014 abstracts that expressed a position on the issue of human-induced climate change, Cook et al. (2013) found that over 97% endorsed the view that the Earth is warming up and human emissions of greenhouse gases are the main cause."

To get into details, the paper has 7 categories, and here are the number of abstracts in the various categories:

(1) Explicit endorsement with quantification = 64 abstracts.

(2) Explicit endorsement without quantification = 922 abstracts.

(3) Implicit endorsement = 2910 abstracts.

(5) Implicit rejection = 54 abstracts.

(6) Explicit rejection without quantification = 15 abstracts.

(7) Explicit rejection with quantification = 9 abstracts.

Category (4) isn't presented above, because only Categories 1-3 and 5-7 are important to the paper. There are 3896 abstracts in Categories 1-3 and 78 abstracts in Categories 5-7, adding up to a total of 3974 abstracts. Then there were 40 abstracts rated as "uncertain" with respect to anthropogenic global warming ("AGW").

The main point is that Cook et al. get the "over 97%" of abstracts by summing all the papers in Categories 1 to 3. However, of those first 3 categories, only "Category 1" truly fits the characterization as "...the Earth is warming up and human emissions of greenhouse gases are the main cause." So only 64 out of the 4,014 abstracts meet that characterization.

To see how much abstracts in Category 3 do not meet the characterization of, "...the Earth is warming up and human emissions of greenhouse gases are the main cause" one simply needs to read a few abstracts in Category 3. For example, there's the abstract: Is the extent of glaciation limited by marine gas-hydrates?

"Methane may have been released to the atmosphere during the Quaternary from Arctic shelf gas-hydrates as a result of thermal decomposition caused by climatic warming and rising sea-level; this release of methane (a greenhouse gas) may represent a positive feedback on global warming [Revelle, 1983; Kvenvolden, 1988a; Nisbet, 1990]. We consider the response to sea-level changes by the immense amount of gas-hydrate that exists in continental rise sediments, and suggest that the reverse situation may apply—that release of methane trapped in the deep-sea sediments as gas-hydrates may provide a negative feedback to advancing glaciation. Methane is likely to be released from deep-sea gas-hydrates as sea-level falls because methane gas-hydrates decompose with pressure decrease. Methane would be released to sediment pore space at shallow sub-bottom depths (100's of meters beneath the seafloor, commonly at water depths of 500 to 4,000 m) producing zones of markedly decreased sediment strength, leading to slumping [Carpenter, 1981; Kayen, 1988] and abrupt release of the gas. Methane is likely to be released to the atmosphere in spikes that become larger and more frequent as glaciation progresses. Because addition of methane to the atmosphere warms the planet, this process provides a negative feedback to glaciation, and could trigger deglaciation."

There is nothing in this abstract that says, "The earth is warming up"...let alone that "human emissions of greenhouse gases are the main cause." The whole abstract (and paper) are about events that happened long before the Industrial Revolution!

Similarly, there is the abstract: On the nature of methane gas-hydrate dissociation during the Toarcian and Aptian Oceanic anoxic events.

The magnitude and timing of a major rapid negative carbon-isotope excursion recorded in marine and terrestrial matter through the Early Toarcian (Early Jurassic) and Early Aptian (Early Cretaceous) oceanic anoxic events (OAEs) have been proposed to be the result of large methane gas-hydrate dissociation events. Here, we develop and evaluate a global carbon-isotope mass-balance approach for determining the responses of each component of the exogenic carbon cycle (terrestrial biosphere, atmosphere and ocean). The approach includes a dynamic response of the terrestrial carbon cycle to methane-related CO₂ increases and climatic warming. Our analyses support the idea that both the Early Toarcian and Early Aptian isotopic curves were indicative of large episodic methane releases (∼5000 and ∼3000 Gt respectively) promoting warm ‘greenhouse’ conditions in the Mesozoic. These events are calculated to have increased the atmospheric CO₂ concentration by ∼900 and ∼600 ppmv respectively and land surface temperatures by 2.5° to 3.0°C. However, we show that much of the methane released from oceanic sediments is rapidly sequestered by terrestrial and marine components in the global carbon cycle, and this effect strongly attenuated the potential for ancient methane gas-hydrate dissociation events to act as major amplifiers in global warming. An increase in oceanic carbon sequestration is consistent with the deposition of globally distributed black shales during these OAEs. Our analyses point to the urgent need for high-resolution marine and terrestrial carbon-isotope records to better characterize the nature of the Toarcian and Aptian events and improve our interpretation of their consequences for the global carbon cycle.

Not only is this abstract about times before the Industrial Revolution, it's about times long before humanity even existed! But somehow, according to John Cook (and others), this abstract demonstrates that, "the Earth is warming up and human emissions of greenhouse gases are the main cause." On the blog, "And Then There's Physics," I attempted to explain how the Cook et al. "Consensus" paper is being misrepresented by Cook et al. But the crowd there simply would not believe that the emperor had no clothes. They insisted they could see the clothes. Eventually, a boy has to say, "Whatever."

But the emperor is still naked, and John Cook et al. are still misrepresenting the results of their paper. Their paper most emphatically did not find that, "Of the 4,014 abstracts that expressed a position on the issue of human-induced climate change, ...over 97% endorsed the view that the Earth is warming up and human emissions of greenhouse gases are the main cause."

Do people of 2100 have a "right" to be "pretty angry" about global warming?

I made some comments about a post by Kerry Emanuel at the National Climate Change Forum. Andrew Dessler first commented:

John: here’s how I would look at the issue. Imagine you’re a citizen year 2100. If it turns out that the climate sensitivity is high and human society does nothing about it, then you would rightfully be pretty angry.

I responded as follows:

Andrew Dessler writes: “John: here’s how I would look at the issue. Imagine you’re a citizen year 2100.”

OK. Let’s imagine that. In the 20th century, worldwide GDP/income growth averaged about 2.3% per year. Let’s assume that continues. Therefore, as a citizen of the world in 2100, my income is approximately 7 times what it was for the citizens of the world in 2014. Let’s further assume that there are 10 billion of us in 2100 (a nice round and reasonable number). Therefore, our global income/GDP in 2100 is approximately 10 times higher than at present (7 times more GDP/income per person, and approximately 1.4 times as many people). Therefore, our global annual GDP is about $900 trillion (~10 times the 2014 value of $90, purchasing power parity basis), and our global income is about 80% of that, or $720 trillion.

Let’s further ...suppose that the global CO2 concentration is about 700 ppm, and this has resulted in a warming of 5 degrees Celsius relative to the year 2014. (Note that such warming seems extremely improbable, because *average* warming rate over the next 8 decades that’s more than 3 times the current rate.) Let’s further suppose that this warming of 5 degrees Celsius cuts the income/GDP-per-person for the citizen of 2100 from 7 times the value of the citizen of 2014 to 6 times the value of the citizen of 2014 (i.e., about a 14% reduction in income).

Suppose we citizens of 2100, suffering from a 14% reduction in our income due to global warming, decide to spend 20% our income on pulling CO2 from the atmosphere and injecting it underground (ambient air capture and sequestration). That 20% of our income would be $720 trillion times 0.86 (from a 14 percent reduction caused by the hypothesized 5 degrees Celsius global warming) times 0.20 (20 percent) = $124 trillion a year. One (pessimistic) estimate of the cost of ambient air capture and sequestration in 2012 was $1000 per metric ton of CO2 (http://www.pnas.org/content/108/51/20428.long). (Note that this paper also mentions other estimates for the current cost at “no more than a few hundred dollars per tonne of CO2 avoided.”) Since there are 2.13 gigatonnes of carbon per ppm of CO2 in the air, and 3.67 tonnnes of CO2 per tonne of carbon, the total cost to remove 1 ppm of CO2 from the atmosphere at $1000 per metric ton would be $7.8 trillion. So if we citizens of 2100 spent 20% of our income (i.e., $124 trillion per year) on ambient air CO2 removal and sequestration, we could remove about 16 ppm per year. Let’s say were adding 2 ppm a year at the time; therefore, we could only lower CO2 by 14 ppm per year. So how many years would it take us to get from 700 ppm down to the pre-industrial concentration of 280 ppm? It would take us (700 – 280)/14 = 30 years.

At the end of that 30 years, assuming that GDP per capita continued to grow at 2.3% per year, and even assuming that we continued to lose 14% of our income to climate change damage and 20% of our income was spent on ambient CO2 removal and sequestration, our income would still be 6.8 times the income of the average citizen of 2014 (i.e. 1.4 times the average income of the citizen of 2100, which was 4.8 times the average income of the citizen of 2014).

Dr. Dessler continues, “If it turns out that the climate sensitivity is high and human society does nothing about it, then you would rightfully be pretty angry.”

How do we citizens of 2100 have any “right” to be “pretty angry?” We have an income that is 4.8 times higher than the average citizen of 2014. And that’s even *assuming* that there is 5 degrees Celsius of warming between now and 2100 (which is very unlikely to happen) and that such warming causes a 14% loss in income from climate damage, AND that we citizens of 2100 spend 20% of our incomes removing CO2 from the ambient air and sequestering it. Furthermore, by 2130, we average citizens will have an incomes that are 6.8 times higher than the average citizen of 2014, and will have lowered the ambient air CO2 concentration to the pre-industrial value of 280 ppm.

My comments are essentially a follow-up on posts I've made here on my blog that question the morality of sacrificing for the benefit of the people of 2100, since I think they'll be so much better off than the people of 2014. (Or they'll be waging and likely losing a war to super-intelligent terminators, and so will have much bigger worries than global warming.)

Related articles

Global warming is not "irreversible"

What is the Morality of the Less-Well-Off Sacrificing for the Better-Off?

 

Jobs Vulnerable to Artificial Intelligence

 According to news on the street, the table below shows the 15 most common jobs in the United States (with number of workers in millions, and salary in thousands of dollars). A potential question is, "How many of those jobs are vulnerable to artificial intelligence progress?"

To me, it looks like almost all of them are. The only two exceptions I see are general managers and (possibly) registered nurses (RNs). I see tractor trailer truck drivers and cashiers as being particularly vulnerable. But in 2-3+ decades, I think we're all vulnerable. 

P.S. I previously had "janitors" listed twice, but #15 should have been "elementary school teachers." I think they'd be less vulnerable than truck drivers and cashiers...but more vulnerable than "general managers." It seems to me that the elementary school teachers would be vulnerable to the artificial intelligence teaching the lessons, while the people who were formerly cashiers would simply look after the kids.

Popularity	Job	Number of Workers	Salary
1	Retail Salespeople	4.2m	25k
2	Cashiers	3.4m	28k
3	Office Clerks	2.8m	28k
4	Fast food prep and service	2.7m	18k
5	RNs	2.6m	67k
6	Waiters/waitresses	2.3m	20k
7	Customer Reps	2.2m	32k
8	Material movers (e.g., loading docks)	2.1m	25k
9	Janitors	2.1m	24k
10	Stock clerk, order filler	1.9m	23k
11	Secretaries	1.8m	31k
12	Bookkeeping, accounting	1.8m	35k
13	General managers	1.7m	111k
14	Tractor trailer truck drivers	1.6m	25k
15	Elementary school teachers	1.5m	53k

 

Population growth rate vs per-capita GDP growth rate

Regular readers of this blog will know that I've pointed out on several occasions that the late Julian Simon's principal message was that the human brain is the "ultimate resource." Here are two graphs that tend to illustrate that point. The first is a graph is of the annual world population growth rate from 1800 to 2005 (and extrapolated to 2025).

The second is a graph of world per-capita GDP growth rates from 1900 to 2000 (from Brad DeLong's paper on GDP from 1 million BC to present). The growth rate then has five predictions up to the year 2100. 

T

The way I think these two graphs show the importance of human brains is that both the human population growth rate and the world per-capita GDP growth rate rose from circa 1940 to 1970, peaked circa 1970, and then declined from 1970 to 2000.

Arnold Kling and I both predict that the per-capita GDP growth rate will increase again (in fact, far exceeding any historical rates) as computer brains become much more plentiful.

Global warming is not "irreversible"

Science Magazine has a paper recently claiming that global warming was "Irreversible". It's an indication of how strongly global warming alarmism is a religious matter that the paper was published with such a title.

Global warming is of course not "irreversible." If CO2 is removed from the atmosphere, temperature will go down. And there is no magic behind removing CO2 from the atmosphere. Plenty of people have studied it. The question is one of cost/affordability.

Below is an analysis of the cost/;affordability removal of CO2 from the atmosphere. What I did was to obtain global GDP and atmospheric CO2 projections for four IPCC scenarios. Then I took the GDP for the years 2080 to 2100, and projected that growth rate out to 2200 for each scenario (since GDP projections to 2200 were not readily available). I did the same thing for projections for CO2 emissions.

Then I calculated what would happen if the world spent 10 percent of its GDP on removing CO2 from ambient air in the year 2100, and continued to do that until the ambient CO2 concentration reached the pre-industrial level of about 300 ppm. My assumption was of a cost for CO2 removal of $1000 per metric ton. That works out to about $7.8 trillion per ppm of CO2 removed. 

The results below indicate that for all scenarios but one, the CO2 is reduced to the pre-industrial concentration of about 300 ppm before the year 2200. I'll probably modify this post a bit as I have time. But the bottom line is that global warming, contrary to what you may have read in Science magazine, is clearly not "irreversible."

Some notes:

1)  Global GDP values can be found at this site: SRES scenarios.

2)  Atmospheric CO2 concentrations can be found here: CO2 concentrations and emissions

  

A few more predictions for the 21st century

On William Connolley's "Stoat" blog, Neven asked some questions. Responses are placed here (because William Connolley censors comments):

1)  Sea Ice Extent in 2030, 2070 and 2100: My estimate for September/October arctic mininum sea ice is close to zero by 2030, and at zero in 2070 and 2100. My estimate for March/April maximum sea ice extent is 12 million square kilometers in 2030, 10 million square kilometers in 2070, and 8 million square kilometers in 2100.

2) Effects of Declining Sea Ice on the Jet Stream: Every single effect is obviously unknown, but I do not expect winters in the U.S. and Europe to become more severe (as some have speculated).

3) Sea Level Rise: As posted just awhile ago, I expect 0.14 meter of sea level rise by 2050, and 0.4 meters by 2100.

Postscript: Sea ice extent predictions can be compared with March/April average maximum values in the 2000s of a little over 14 million square kilometers, and 2000s average September/October minimum values of approximately 5.6 million square kilometers.

What is the Morality of the Less-Well-Off Sacrificing for the Better-Off?

Technology Review recently had an editorial addressed to President Obama (Dear Mr. President: Time to Deal with Climate Change) asking him to make global warming his top priority. They acknowledge: "It means immediate spending and economic sacrifice by present-day voters in order to achieve benefits that will be realized decades from now." 

I don't understand the morality of "immediate spending and economic sacrifice by present-day voters in order to achieve benefits that will be realized decades from now." I wrote to Jason Pontin at Technology Review (who seems like a nice fellow) and told him that I thought that the people of 2050 and 2100 will be better off than the people of the present "under any realistic scenario of global warming." He questioned whether this would indeed be the case. So I put some potential "quality of life" indicators in a table, with my estimates for their values circa 2012, and my estimates for what they would be in 2050 and 2100.

I'd be happy for him or any other Technology Review (TR) author of that letter to present their estimates. I'd also be happy for him, them, or anyone else to provide some potential other parameters that should be on the list.