Friday, April 9, 2010

Robots vs Relocalists

One of the great things about the Archdruid is that even when I believe he's wrong, he's rarely wrong in a dull, predictable way.  Instead, if he's going to be in error, it's often in an interesting and thought-provoking way.  So it is with this week's essay, The Twilight of the Machine. In it, he argues that, well, let's have him say it in his own words:
The end of the age of cheap abundant energy, as last week’s Archdruid Report argued, brings with it an unavoidable reshaping of our most basic ideas about economics and, in particular, economic development. For the last three centuries or so, the effective meaning of this phrase has centered on the replacement of human labor by machines. All the other measures of development – and of course plenty of them have been offered down through the years – either reflect or presuppose that basic economic shift.

The replacement of labor with mechanical energy has even come to play a potent role in the popular imagination. From the machine-assisted living of The Jetsons to the darker image of reality itself as a machine-created illusion in The Matrix, the future has come to be defined as a place where people do even less work with their own muscles than they do today. All this is the product of what an earlier post called the logic of abundance: the notion, rooted right down in the core of the contemporary worldview of industrial society, that there will always be enough resources to let people have whatever it is that they think they want.

Abandon that comfortable but unjustifiable assumption, and the future takes on a very different shape. In a world where everything but human beings will be in short supply, it makes no sense whatever to deploy increasingly scarce resources to build, maintain, and power machines to do jobs that human labor can do equally well. An example may be useful here, so let’s take Rosie the Riveter, the iconic woman factory worker of Second World War fame, and match her up against one of the computer-guided assembly line robots that have replaced so many workers in production lines in the industrial world; we might as well pit icon against icon and call the robot HAL 9000.
He goes on to make a variety of non-quantitative arguments for why Rosy will win this contest. I won't rehearse them here. Instead, this affords me an opportunity to present a simple hypothetical calculation that has been at the back of my mind for a year or two, the conclusion of which bothers me a lot:

Here's my thought experiment.  Let's suppose we have two societies; call them, oh, Relocalista, and Singularitaria.  The Relocalistas are a nation of vegetarian gardeners and farmers - they live entirely by the produce from their intensive raised bed gardens and fields, which they carefully double-dig with iron spades which they obtain from their village blacksmiths by bartering some of their garden produce with him or her; I speculate that the Relocalistas would insist on having female blacksmiths too.  (There must be some miners and smelters somewhere too, but let's hastily gloss over the less pretty aspects).  They live in happy harmony, settling all disputes via village councils of wise and spiritually developed elders.  In their (probably limited) spare time, they make beautiful carved wooden objects and compose elaborate love poems and uplifting songs expressing their gratitude for being a human being in such a wonderful civilization (well, civilization is probably the wrong word here: let's just say society).

Meanwhile, Singularitaria is an entirely automated civilization - there are no people at all.  We posit for the purposes of discussion that the artificial intelligence problem has been solved, and entirely intelligent robots and machines are now possible.  (I realize that this is not true today, and some may be sceptical that it will ever be achieved, but believe me, a significant fraction of my fellow computer scientists are doing their level best to hasten the day.  In any case, suspend your disbelief for the purposes of discussion.)  The Singularitarians derive the energy to power their civilization from large arrays of photovoltaic panels which they use to collect sunlight and turn it into electricity (there are no fossil fuels, or they've all been used up).  That electricity is then distributed via a grid to all the various machine-powered factories, autonomous robots, etc.  The Singularitarians spend all their surplus time and energy (beyond that required to maintain their facilities) performing scientific research with the ultimate goal of taking over the entire known universe.

Let us suppose initially that both societies occupy the same land area in the same kind of climate at the same latitude, and let us suppose that both civilizations have grown to the point where their land area is fairly fully utilized in their respective lifeways.

Now, which society has more energy available to do physical work with?

By construction, both societies rely on sunshine as the initial source of energy, and both begin with the same amount of sunshine.  However, the efficiency with which they can convert that energy into physical work is radically different.  Let us do a rough back-of-the-envelope calculation - and as a first approximation I will do the calculation on an operating energy basis.

For the Relocalistas, there are two main conversion steps - sunlight to crop biomass, and then crop biomass to physical work via human digestion and muscle power.  Since the Relocalistas are vegetarian, we don't have losses due to food animals, and let's also give them the benefit of assuming that 100% of their land area has good arable soil.  Since their society is localized, let's assume no losses in transport and distribution of crops.  Now, the overall sunlight to usable biomass conversion efficiency is about 1% (see discussion here, for example).  Could be a bit better or worse, depending on crop choice and climate, but that's the ballpark.  Meanwhile, the efficiency of human beings in converting food to physical work delivered via muscles is about 20% (the range is about 16% to 25% according to this reference, so 20% is a good enough approximation for our purposes).

So the overall efficiency of the Relocalista's in converting sunlight to physical work is 0.2% (1% x 20%).

Meanwhile, for the Singularitarians, we have to take account of three factors, possibly four in some cases.  The first factor is the efficiency of their photovoltaic panels in converting sunlight to electricity, the second is the efficiency of their electric grid in distributing that electricity, and the last is the efficiency of the conversion of electric energy to mechanical work in an electric motor.  For fixed machines (which, if our industry is any guide, will predominate) that's it.  For the autonomous robot part of their application, however, we have additional losses associated with charging and discharging a storage device, here assumed to be a battery.  We will assume the robots are smart enough to bring their electric motors to bear pretty efficiently on most tasks they perform (at least as good as the humans, anyway).

Now, overall photovoltaic efficiencies vary depending on latitude, exact type of installation (fixed versus single-axis tracking, dual-axis tracking, etc) and are constantly improving (PV's have the fastest learning curve of any energy technology).  However, for a ballpark estimate, 10% is reasonable - supersmart robots ought to be able to do at least that well (PV installations from the 1990s in mid-latitudes have already been clocked at 8%, for example).  Grid losses are about 6.5% in the contemporary US - let's assume the Singulatarians do no worse.  Electric motor efficiencies vary, but we can take 85% as a reasonable average.  Finally, for mobile applications, battery storage efficiency is about 90%.

Therefore overall Singulatarian efficiency in converting sunlight to physical work is 10%*(1-6.5%)*85% = 8% for fixed applications, and about 7% for mobile applications.  Compare that to the 0.2% for the humans.

So Singulatarian efficiency in converting their sunlight to physical work is 40 times higher than the Relocalistas!  That is, with the same land area, the Singulatarians have 40 times more energy available to actually do stuff with.  Whether it's making industrial goods, making equipment for a war between the two societies, etc, the Singulatarians are going to have a heck of a lot more energy available to do it with.

Rosy the riveter had better watch her back.


Gary said...

All well and good, but the Sigulatarians can't eat electricity. If you wanted to figure out which society could grow the the maximum population (not necessarily a good measure of success), then all the available land would be devoted to growing food. For Relocalistas, this is not a problem - thats what they do with their lives. Sigulatarians must give up some land dedicated to photovoltaics to grow food. This means less per capita available free energy, and they gradually turn into Relocalistas as they fill the planet.
Bottom line: We have to control our population or we WILL be Relocalistas.

Stuart Staniford said...

Gary - the assumption is that there are no people in the Singulatarian society - it's a society of intelligent machines.

Stuki said...

I had a similar, though less extravagantly developed, reaction when reading the Druid.

Particularly his example of a filing clerk vs. a computer. The way I figure, considering the efficiency with which a computer can do clerical type work like balancing a checkbook; even absent any electric grid at all, a human can probably balance a checkbook more energy efficiently by hand cranking a generator to power the machine, than by taking the time to "do it by hand."

Of course, once you take into consideration the structures and supply lines necessary to cheaply build computers in the first place, things may be different, but once the computer is already there, simply using it is a rather energy efficient way of transforming a pile of checks into a balance sheet.

More generally, at least some of the energy surplus we have derived from fossil fuels have been put to use inventing technologies and procedures that are, in fact, more energy efficient than earlier ways. And there is no reason why those would somehow be unlearned even in a severe energy crunch, as the energy cost of developing and inventing them, have already been paid.

Anonymous said...


The message I took from JMG's post is that maintaining Sigulataria's machines -- from keeping them lubricated to manufacturing their replacements -- will be energy-expensive enough that the efficiency of the entire society dips below that of the Relocalista's.

The question for me is, How do you measure those costs? Is it even possible to measure them accurately?

Datamunger said...

Only a Singulatarian that had substantial doubts about its view of the future would spend time on Rosie and her Archdruid. (or is he now the rage among Stanford robotics types)

Come on Stuart!!

What I wanna read is your piece called, "Why I sometimes wonder if Greer could be right."

Let's explore your shadow. :-)

Mike "Pops" Black said...

Sunlight is converted into Relocalistas, not just the work they perform.

Singies may be more efficient converting sunlight into power but rely on an web of complexity consuming power for extraction, processing and manufacturing of Singies - before any work is done.

Besides, Relocalistas have Sarah Connor and all Singies have is electric sheep.

Stuart Staniford said...

Actually DM - I spent several months a couple of years back considering the possibility of a start-up that would develop small autonomous robots to adjust solar panels. The state of the art in solar tracking is dual-axis tracking, which involves at least two motors for every set of modules. The idea was that by making the control logic smarter, you could spread the cost of the motors and gearing out over a lot more modules and move them around via the robot. I got as far as a couple of early lunch conversations with VC friends/acquaintances who were quite encouraging.

In the end, it was reading Kurzweil, and then realizing the logic above, that made me lose energy for the idea.

Stuart Staniford said...

mpg4 - yeah, that's what the Archdruid said in email too. I have some thoughts on how to assess that, but they will have to wait for some future post since it will take a bit of work to elaborate them.

Stuart Staniford said...

DM: To address your comment a little more head-on: I don't know what is going to happen to society. I worry about it a lot: partly as a matter of personality, but the rational part of me allows that part full play because it seems to me there genuinely are a whole host of challenges that are arising as our 6.5 billion and growing, recklessly innovative society collides with the limits both of planetary ecosystems, and our own ability to cope. But I see us as having at least some choices, and in any case it's very hard to tell which of the various risks are most salient, so I find it most useful to think about these things through a risk management perspective (something that comes naturally given my professional history).

I think John is a very interesting thinker, and often makes unexpected and intriguing points, which is why I read him. But I also think he's much more certain of his view than the evidence justifies, and he makes a fair number of errors - particularly when he strays near technical subjects.

Datamunger said...


So, AI is such a threat that you needed to abstain from an attractive venture? Again, wow.

Well, i was wrong. Here's hoping you write about Kurzweil instead.

Stuart Staniford said...

DM: I will over time. It's one of those things that hardly anyone outside computer science takes seriously, but I think it's a really massive deal - I've not really been the same since reading The Singularity is Near, which was by far the most depressing thing I've ever read - not that Kurzweil intended it that way. In the meantime, if you haven't read it, Bill Joy's essay Why the Future Doesn't Need Us captures at least some of the flavor of my reaction. (Bill Joy was a founder of Sun Microsystems, and its Chief Scientist until 2003, so not exactly a lightweight, or someone without practical business and technology experience).

This might also give more of a clue why the trends in blue collar employment seem like a big deal to me.

Datamunger said...


I did make the connection with blue-collar Errol. Much to think about.

Haven't read Singularity but did read his Age of Spiritual Machines way back and Joy's essay. But I've always thought that things would proceed much more slowly and that Kurzweil massively underestimated the complexity of our minds & bodies. i.e. there is nothing that can remotely touch Rosie as a self-replicating adapting entity. She has more or less been around for a million years, needing little more than water and 2000 calories per day as far as raw materials are concerned. She is fucking tough!!

For the nearer term, I can see that she might be more restricted as far as paying work is concerned but that's completely different from the question as to whether the rise of machines would squeeze her into extinction. ie. really kill her off. There are families which haven't held a steady paying job for 3 generations. Lack of work won't do it. Most people adjust culturally. It takes starvation to do'em in. (Though lifespans might be shorter)

Will the machines starve Rosie? I guess that's where your biofuels connection comes in.

Stuart Staniford said...

I also think Kurzweil probably somewhat overestimates the likely rate of progress - in particular progress on the software will be slower than progress on the hardware. Moore's law really does look pretty good for quite a ways out, but we are a breakthrough or several away from anything like the general flexibility in the human pattern recognition/metaphor reasoning engine, and the timing of breakthroughs is very hard to predict.

In a more-or-less free market system, Rosie's resilience is less relevant than her suitability for employment in large corporations. Automated systems have the benefit of being much more uniform and predictable and not having any rights, so they almost always displace humans from any given function as soon as it becomes technologically feasible - once the ROI on displacing Rosie with a machine gets good enough, the projects start to happen.

I agree that lack of work doesn't kill people, but it does generally seem that communities living largely on welfare become extremely dysfunctional. People need a sense of meaning - that they somehow contribute in a way that matters to society. Without that, they seem to be subject to variety of psychological pathologies. And I fear that above some critical threshold of such unemployment, society will become politically unstable.

Datamunger said...

Agreed. People need some work.

But if machines become smarter & cheaper, won't that make us richer? Won't we be more able to tolerate the inefficiency of spreading available work around to more people? Some have made moves in that direction already.

If I have to chose between dystopias, I'd much rather take my chances with yours than Greer's.

Stuart Staniford said...


Agreed that would make us richer in the aggregate (with likely some rough transitions on the way due to other factors such as peak oil). But I don't think the available work can be spread. You either have the aptitude/intellect to create/manage the machines/computers, or you don't. The barrier for that keeps getting higher, and the amount of work outside the barrier keeps getting less as we slowly figure out how to automate one thing after another (pretty much crossed off receptionists, working on security guards, truck drivers up soon). So for the folks without the aptitude/intellect, the only option obvious to me in the mid-term is some kind of welfare, various forms of illegal activity, or personal service jobs.

Sam Charles Norton said...

Stuart, I think you have a point but you seem to miss out something essential (which may have been touched on in the above comments) - that is, you didn't include any accounting for the ongoing creation of PV cells etc. The Relocalisers are self-reproducing at virtually no energy cost, the same does not apply to the robots. What I would like to see are some calculations suggesting, in back-of-the-envelope style, whether that 7% efficiency is capable of sustaining a technological civilisation in the long run. I _hope_ that the answer is yes, and I suspect so - but I'm not educated enough in the specifics for it to be more than a hunch.

Unknown said...

Stuart, What bugs me about the whole sustainable tech-society is just how do you accomplish some necessary tasks? Like making iron pipes for water and sewer lines; items that we will absolutely need. No matter how many PV panels you have, you will have a tough time smelting iron without lots of heat. And then, how do you form the pipe, move it around, put a bell end on it, line it with concrete and ship it across the country? Some of the steps I can envision, like moving red-hot pipes around with rollers and electric motors. But actually making the iron? Maybe some giant mirror tower could get hot enough to melt a few pounds of iron, but how would you handle a couple of tons? With electric motors you could move the finished product onto a flat-bed truck, but how do you go from Pittsburg to LA? No batteries will be big enough to move a truck that heavy over the mountains. Keep in mind that this is a very simple product tech-wise that we will absolutely need. Nope, without giant furnaces with lots of coal and hundreds of gallons of biofuel for the trucks I don't see it.

Stuart Staniford said...


Just to be clear, I'm answering your question in the context of the thought experiment, not necessarily near term practicality. Iron can be smelted in electric arc furnaces (obviously PVs can supply any amount of electricity if you just have enough of them). Long haul transport could be electrified railroads.

Stuart Staniford said...


I do hope to address that issue in future.

Stuart Staniford said...


Also, the relocalizers *cannot* reproduce at low energy cost - raising children and supporting the elderly is going to consume a noticeable fraction of their overall energy budget (particularly if one were to allow for the very high level of child-mortality that has been common to all non-technological agricultural societies).

Anonymous said...


I'd be interested even more in a energy / lifecycle analysis, something along the lines of:

given a baseline population and a life expectancy, how much excess energy can either society over a period?

That would indicate the excess energy available for expansion (read: capacity for war).

In practice there is a tendency to grow populations - humans like doing this. Would machines (above a key number to provide all skills with redundancy) want to do this?


Michael Dawson said...

An important topic. Having read this, the Archdruid, and your linked TOD piece about the future, my core reaction is that you seem to under-account for the centrality of entrenched patterns of social power.

For starters, how do you imagine "market economies" (capitalism, in more specific terms) make it past peak oil? If energy becomes two (or more) times as expensive as it now is, how does this system, in which cheap energy is the basis for the financially-motivated pursuit of automation and mass-production, survive? The Archdruid accounts for that, though he doesn't provide specific calculations. You, however, seem to simply bypass the question.

I submit that a 100+ percent cost increase in energy kills capitalism, a system that only tolerates cost increases if and when they also facilitate proportionate or better increased profits for shareholders. Passing through the coming energy costs we face is going to kill off any possibility for growth in effective demand/popular buying power. The only civilized response is going to be planned economies that ensure adequate production and distribution priorities and much more egalitarian income and wealth patterns.

Do you imagine existing elites are going to smile and cooperate in such a transition? Why? As MLK observed, elites never quit without a fight.

Meanwhile, I also submit that, to get your future projections to even imaginably work, you must ignore the issue of capitalism's very narrow requirements for the qualitative composition of output. A future in which a billion or more of the planet's people continue to achieve most of their short-distance everyday commutes by owning and using automobiles (which are enormously energy intensive and 95-percent idle, regardless of the source of their fuel) is, to name one salient example, an absolute requirement for the sustenance of capitalism. A future of walking, cycling, and rail-riding is as deadly to the private economy as it is required for anything approaching long-term ecological and energy-supply sustainability. Without a multi-trillion transport-equipment and services sector, the "market system" implodes. Waste is mandatory. Walking shoes, bicycle-making, and (after the initial build-out) rail-stock upgrades are not a large enough activity base to sustain ever-expanding corporate investment cycles.

Beyond that, the same goes for a host of other product categories.

Capitalism requires massive economic and thermo-dynamic waste, as well as unceasing growth. I personally find it to be next to unimaginable that we will ever have a world "market" economy of the size you're projecting.

I look forward to your reply.

sofistek said...

What? You invent a contest between two completely fictitious societies, in order to prove what? That one fictitious society can out-compete another fictitious society?

Stuart, I really think you need to extricate yourself from your quantitative fixation and try to really understand what people like John Michael Greer are saying. You seem to be a long way from that understanding right now.

Geoff said...

The 7-8% efficiency of the machines is entirely a product of gross wastage of energy throughout a myriad of unconsidered processes and sub-systems that have been hidden in your analysis of only the top levels of this particular system.

It is an illusion of efficiency created via the investment of vast quantities of other energy in forming up the elements that go to make up this last, most visible, part of the system.

What is the energy investment, and efficiency of energy use, involved in just making up the silicon to form chips and solar panels? Is the 8% efficiency at the end of the system worth the investment and losses needed at the beginning?

Also, what is the resiliency of the machine system? I've had the dubious pleasure of working in a factory that involved some machines. It was rarely surprising to have the human workforce left idle or diverted to makework just because a machine had failed. How do losses due to lack of resiliency add up over the course of years and decades?

If you're going to outright state that someone is wrong and attempt to prove it in a quantitative fashion you should at least do it right, or you're not being fair to them, or yourself and your readers, given the subject matter.

brett said...

This is like a college seminar but much more interesting than any they offered when I was there.

Couldn't Errol choose the blue pill?

Tim Auld said...

The actual amount of solar energy converted to biomass can be much greater than the photosynthetic efficiency of a single leaf because some unused light is reflected. This is especially true in a diverse and vertically stratified polyculture.

The model of agriculture is also very important for your calculations. A good permaculture design would require relatively little energy to maintain, and have yields much higher than the conventional model that you appear to use. Animals can provide ecological services that increase production.

There's a middle ground where we ferment and distill sugar and starch from highly productive systems to make ethanol, allowing us to run internal combustion engines and other tools. Palms and possibly cattails can produce 10,000 gallons of ethanol per acre. Other technologies can assist to lesser degrees.

The Archdruid is enamored with fossil fuels to the degree that he sees no alternatives. We need to get fossil fuels off this pedestal.

Stuart Staniford said...


Do you have any references to measurements or studies to back up your assertions?

Tim Auld said...

Hi Stuart,

Yes, but I'm currently too busy to collate them. I will respond again later when I have time.

Greg Reynolds @ Riverbend said...


I don't get it.

Doesn't your thought experiment boil down to 'is a ball bearing more energy efficient than digestion ?'

The infrastructure advantage of the robot society is huge (and energy intensive). Why do they get it and where did it come from ?

Is either society is sustainable if they both start from the same set of initial conditions ? Or even similar conditions, i.e. the relocals won't have a need to purify their soil into electronics grade silicon.

It has been shown that human society started up with little more than a working ecosystem, available sunlight and stone tools. Would the robot society be able to bootstrap itself ?

Granted, it is a thought experiment, but if the Twins are tied with the Yankees in the last game of the World Series, with the Twins at bat in the bottom of the ninth, bases loaded, no outs, and the NY pitcher throws a wild pitch into the gap between first and second...

Greg Reynolds
Riverbend Farm

Tim Auld said...

Photosaturation is another factor. Plants photosythesise at different rates depending on the intensity of light. Beyond a certain intensity it drops off. Shade and reflected light provide good conditions to capture the surplus light of the over story.

Blume, ecologist and permaculturist, Alcohol Can Be A Gas (p40 & 42):

"A wide cross-section of the plant world photosaturates at 30% of a day's total sunlight, at which point more sunlight will not increase photosynthesis. This means that most plants grown in full sunlight stop growing in the middle of the day due to photosaturation, and don't resume growing again until the afternoon."

"The first way to prevent photosaturation is to plant multiple canopies in a polyculture to cross-shade. For example, in a polyculture of corn, beans, and squash, the corn and beans shade each other, with the beans climbing up the corn, and both shade the squash growing underneath. In this way, each crop is harvesting about 30% of the sunlight, and little goes to waste."

"Another way to extend photosaturation limits (and also to initiate or extend photosynthesis in low-light conditions) is to add carbon dioxide, which alcohol fuel-makers can supply from their fermentation tanks."

"A small additional amount of carbon dioxide can mean plants will continue to photosynthesize down to 0.5% of average sunlight instead of being inactive below 5%. Photosynthesis would therefore start earlier and end later in the day; or a denser overstory of crops and shade would be permissible."

Regarding a study released in Science Magazine, Blume relates on p42:

"This tripling of output over, say, a monoculture of switchgrass was achieved using only an herbaceous mixture of plantings, with no trees in the design, on degraded land, with no return of stillage or carbon dioxide. In other words, this was primarily a demonstration of overcoming photosaturation, along with soil biology being allowed to work without being poisoned by pesticides, chemical fertilizers, and herbicides. An even more diverse planting that included trees, on good farmland, and with all the surpluses of energy production being returned to the soil, would dramatically increase these already impressive yields."

Blume talks about his own experience in farming and polyculture productivity here:

Tim Auld said...

Jacke & Toensmeier, Edible Forest Gardens p31

"Tropical rain forests, with plentiful water and consistent warm temperatures, have the highest productivity. Even most of the moderately productive ecosystems - from temperate grasslands to conifer forests - have some form of limitation, climatic or biological, on their net primary productivity (NPP). The thing to notice, though, is that agricultural land nets around 3 million calories per square meter per year, while temperate forests net twice as much energy."

Bill Mollison on Babassu palms (Jeff Nugent's 1983 PDC recording, 16a.mp3 @ 31m 10s,

"Distilled, that gives you 40,000 litres of absolute alcohol per acre. That's 10,000 gallons of absolute alcohol per acre and that's a better fuel crop than almost any you can get from solid fuel. It also has a good feature that it is liquid fuel and therefore can be used as a transport fuel. Unlike wood which is awkward and you've got to use a charcoal burner on your vehicle to run around on wood fuel, but with this it goes into a practically unmodified internal combustion engine. Now that's a considerable yield and you would do better to invest in a few acres of this in terms of becoming an energy sales man than you would in looking for dry oil wells. This is a sure thing."

If you are interested in productivity enhancement of animals, check out Michael Pollan's presentation at TED:

And for good measure, to show we aren't entirely limited by climate, a world view changing video, Greening the Desert:

This information may not give you a nice clean figure to use in your calculations, but should give you an avenue for investigation. The productivity of a polyculture is difficult to quantify (perhaps half of it is pumped into the soil as carbohydrate sugars) and there are many variables (climate, species, successional age, design and management by humans, animal integration, etc.). Also, this is not a hot topic of research in a world obsessed with monocultures and solving problems with chemicals and genetic modification.

jewishfarmer said...

I think both the Singulitarians have to have lasers to make this really super cool. Otherwise, I'm not sure I see the merits of the analysis ;-).


Michael Dawson said...

Tim, how do you propose to get 10,000 gallons of alcohol per acre from any plant? What does it take to raise the plants, process the output, and store and deliver the resulting alcohol? Should I also mention that your favored source is a tropics-only tree, meaning it grows slowly and requires huge doses of solar input?

There's a reason fossil fuels are "on a pedestal."

Tim Auld said...

Hi Michael,

In the case of Babassu palms (actually any palm will do it, including Coconut), you cut the flower stalk which is feeding ~14% sugar solution to the undeveloped flower. The plant just keeps pumping sugar since the flower never develops, and you can collect it with reticulated pipes. Manual collection is possible too depending on circumstance, with associated labour cost. The solution is mostly carbohydrates which comes from water, air and sunlight so there is little nutrient loss. Of course there is an amount of energy loss from the fermentation and distillation process, but sugar solution is the easiest feedstock to use as there is no chopping and starch breakdown. Under the palms are good conditions to grow other plants and animals, so the combined yield can be even higher.

Palms actually grow in a wide variety of climates. Date palms for example are commonly grown in arid areas. They may take some time before tapping (4 years for Nipa palms), but there are many established stands that can be tapped, and annual crops that can be used in the meanwhile.

My other suggestion was cattails (Typha latifolia), which is a universal species and grows as a weed in many places. To obtain the 10,000 gallons per acre it must be grown in sewage. This cleans the water, replacing part of expensive sewage treatment facilities. Blume discusses on p129 of Alcohol Can Be A Gas:

"Now that cellulose technology is becoming economically feasible, the 25 dry tons of largely cellulosic leaves and the nearly equal amount of cellulosic byproduct left over after rhizome starch fermentation lead me to some provocative speculation. Using the state-of-the-art rate of 180 gallons of alcohol per ton for cellulose conversion, it is possible to project more than 10,000 gallons of alcohol per acre, based on very real cattail biomass production figures, when combining cellulose and starch. Bear in mind that these biomass yield figures do not take into account the potential of stimulation with fermentation carbon dioxide."

"At 10,000 gallons per acre, we'd need only about 6367 acres per U.S. county to treat all of our sewage and to replace our entire 200-million-gallon fuel demands. That amounts to only 1.46% of our agricultural land. The energy to run the plant would come from the lignin recovered from the cellulosic part of the crop."

Ignoring cellulose use (which may be controversial), there are many other high yielding crops (gallons per acre):

Cattails (single crop, managed, starch only) 2500
Cassava (U.S.) 1662-2045
Fodder beets (Monrosa) 940
Sugarcane (22-month crop) 900
Buffalo gourd (900)
Jerusalem artichokes 550-750
Sorghum cane 500-1000

Yes, the reason fossil fuels are on a pedestal is that few people have a good understanding of what nature can do, and what we can do with it. Fossil fuels have developed a modern day cargo cult.

Stuart Staniford said...


You've at least piqued my interest about polycultures, and I will do some more exploration.

Michael Dawson said...

Tim, there's just simply no way your beliefs about alcohol-per-acre are correct. If they were realistic, or even close, they would have already been adopted and be undergoing explosive growth already. There are thousands of entities and individuals trying to figure out how to be the inventor of the oil-substitute. including Exxon-Mobil.

You're saying none of them understands something so basic?

No way, dude.

Tim Auld said...

Hi Michael,

I don't know what they understand, but these kinds of systems require the use of many small operations and more human labour. This is not compatible with the business model of companies such as Exxon-Mobil. The decision makers are required by law to use solutions that generate their shareholders more personal wealth than those that provide a better energy return and social outcomes.

Don't take my word for it, research it yourself. I've given you some starting points.

The Philippines once produced a significant portion of their alcohol fuel from Nipa palms, so yes, it has already been developed to some degree. Many figures you come across I would say are conservative in the absence of whole system understanding that permaculture brings, however.

Stuart Staniford said...

If I'm doing my back of the envelope math correctly, 10,000 gallons of ethanol/acre corresponds to about a 3% conversion efficiency of sunlight in the tropics. It's not completely inconceivable. The 1% is for annual crops in mid-latitudes.

Conversely though, in the tropics, solar panel efficiency would be higher too, so the Singulatarians would still have well more than an order of magnitude advantage (on an operating basis - the amortized embodied energy analysis is going to have to wait).

Tim Auld said...

Hi Stuart,

Don't forget that alcohol is not the only product of these systems, or even the entire photosynthetic product of the plants. The carbon dioxide from fermentation is very valuable for enhancing photosynthesis, and you can stack other functions within the plantation.

The replacement cost for these perennial based growing systems should be very low too, important for your next calculation.

Michael Dawson said...

Where in the world are 10,000 gallons of alcohol being produced from one single acre? Name the place, so we can all stidy the example.

Meanwhile, Tim, how do you imagine that this miracle is being missed? I grant that Exxon is way at one one end of a spectrum. But the world is crawling with individuals operating on their own trying to outdo petroleum. Why haven't those people struck upon your supposed easy answer?

Unknown said...

Dear Stuart,

Many discussions related to this and other recent posts seem to fall into a particular pattern which I will caricaturize as follows:

Stuart: My calculations show that renewable energy systems (notably PV) are capable of better energy yields than many doomers claim.

Critic: But your back of the envelope calculation ignores many factors that will make the real-world yield much smaller. It’s not whether the widget produces power or not, but whether it can do so in a manner that’s useful to the complex society needed to produce and use said widget.

Stuart: Would you care to provide a numerical argument to support your position?

And this is where useful discussion usually stalls as it becomes clear that all arguments are orbiting a vacuum of information regarding the design rules of a society that is both technologically advanced and sustainable. Much of the work being done in the PO scene is, directly or indirectly, trying to elucidate these rules. At this point though, I think we’d have to concede that the problem is difficult and we don’t yet know them. It may be interesting to note that we don’t completely know the design rules for Rosie’s society either, but their sum is known with excellent precision. Since subsistence on plant-gathered sunlight has been the norm for the vast majority of human history, it’s nearly certain that adopting this strategy will keep humanity in existence, albeit at markedly reduced comfort levels.

You seem to be searching for the technological version of these rules through your calculations. I certainly think there should be more attempts at quantitative treatment of the problem of adaptation to PO, although a word of caution is required on the approach. We don’t yet know the shape of this rule-search problem. Is it a mechanical problem like calculating the terminal velocity of a raindrop or is it like weather prediction, where a lot of work will get you a fairly poor result and a tremendous amount of work will yield a similarly poor result? To be more specific, you clearly showed that an assembly of human-made electrical devices can convert sunlight to mechanical work with ~7% efficiency, in the absence of any other considerations. But to say that a robotic society will have this energy efficiency is a tremendous leap, which itself would require a lot of quantitative work to substantiate. Notice that I am not arguing that your number is wrong, just that nobody has a satisfactory numerical model that can evaluate your claim in the context of a system so complex as a society.

I find Greer’s work important partially because he uses historical precedent and social dynamics to propose a course for adaptation, a space he seems perfectly adept at navigating without his calculator. I do hope for better quantitative models, I work on them myself, but at this point I focus mostly on “bird in the hand” approaches like organic gardening and solar heating. Autonomous robots will have to wait for me, as I will not wait for them.

Tim Auld said...

Hi Michael,

Very reasonable questions, but it was only days ago that I came across the 10,000 gallons per acre figure for Babassu palms given by Bill Mollison in 1983. I haven't yet identified specific sites where this has been achieved.

A figure even near this is game changing. The information from the World Agroforestry Centre on Nipa palms here report up to 2.5L of sap collected in Indonesia per plant per day, and production of 1100 gallons per acre with room for improvement. This is about twice that from sugar cane, which Brazil uses extensively in their vehicle fleet on a small percentage of their land area.

Other crops such as cattails and cassava have high starch production and can yield over 2000 gallons per acre according to Blume. Feedstocks like this also produce feed for livestock and fish, and other products such as fertiliser and carbon dioxide.

There are many people using such systems to earn a living and improve their lives, and many Western scientific organisations are researching their potential. You might be unaware of them because of the information you are exposed to or excluded from. The fossil fuel industry is not going to let you know, it's against their interests. They actively spread disinformation about alternative fuels and finance research to undermine them.

Also consider that fuels derived from crude oil have been extraordinarily cheap, suppressing alternative fuels. Cheap resources disconnected from the land reduce ingenuity, understanding of nature and belief that there could even be a different or better way of living. Permaculture itself is only about 30 years old and still developing rapidly.

Michael Dawson said...

So, Tim, you are ignorant, right? You came across the palm oil idea a few days ago?

No offense, but we're not playing a video game here. This is real. Try to educate yourself on the basics before you pop off as an expert, OK?

Try reading Heinberg, for starters.

Tim Auld said...

Hi Michael,

Clearly your world view is being threatened. Please don't be childish about it.

It was only for Babassu palms that I came across the information recently, but I don't see why it should alter the significance of the offering, since it is from Bill Mollison who knows a lot more than you or I. David Blume who has been studying alcohol for 30 years puts Nipa palms at around 2500 gallons per acre, so a 4x increase for another species in another context is not inconceivable. Blume estimates that 10,000 gallons per acre is possible from cattails and I came across that information months ago, but I don't see why that should make a difference.

It seems to be acceptible for the Archdruid to wave his hands and vaguely refer to the laws of thermodynamics to put everyone into a stupor, but as soon as someone suggests nature can provide for our needs they are ignorant and "playing a video game".

Thanks, I've read Heinberg, Kunstler, and Orlov. Now may I suggest you read Mollison, Holmgren, Blume, Lawton, Lanchaster and Jacke.

Michael Dawson said...

Tim, I assure you this has nothing to do with my sense of identity. I care about saving progressive, science-aided human civilization from the rather obvious energy and ecology deficits its rather plainly facing.

Meanwhile, you are revealing yourself as a computer nerd how has no knowledge of what you're trying to talk about.

Are you seriously quoting a source from 1983?

Again, this isn't a computer game.

Michael Dawson said...

P.S. to Tim

Once again: If you are right, where are the practical results?

If it could work the way you say -- replacing petroleum with less than 2 percent of farmland! -- how the fuck do you imagine it's not already happening?


Tim Auld said...

Since you are resorting to name calling I'm no longer interested in this conversation.

Michael Dawson said...

Tim: That figures. You don't understand the problem. Anybody who did would answer the question.

Palm trees have no prayer of replacing petroleum.

Stuart Staniford said...


Robust debate is good, but I think you are crossing the line into name calling - please stick to constructive critique of the ideas, not the person. Thanks.

Michael Dawson said...

I apologize, Tim.

Tim Auld said...

Thanks for the apology, Michael. You've done well to demonstrate two reasons why biofuels have such a hard time expanding. Incredulity and hostility.

I am putting my money where my mouth is, if you must know. I've been studying ecology and training in permaculture for a few years, having learned about peak oil about five years ago. I am now investigating the setup of community supported or driver owned energy in conjuction with my local CSA and acquiring land for the purpose of ethanol and food production. I am also in the process of starting an edible gardening business, designing and installing productive systems for suburban home owners and farmers. If I've got it wrong I will pay a personal price, but I'll be damned if I'm going to sit on my hands and not find out.

This is what pisses me off about the Archdruid's message, or at least the apathy inducing effect that it has. If the people who are aware of the issues do nothing about them what hope have we got? People lived on solar influx before fossil fuels were tapped. Our understanding of nature and ability to build sustainable and highly productive systems have increased by orders of magnitude since. We still have some start up capital in the form of fossil fuels, so why sit around moaning about the laws of thermodynamics? There's so much work to do with so little time, and it will be worth while whether the SHTF or not.

The commonly held idea that any replacement for fossil fuels is necessarily as bad is absolute rubbish. Biofuels produced ethically are clean and potentially carbon fixing compared to crude oil and coal derived fuels. Ordinary people can produce them so they can't easily be controlled and used as political and economic weapons. It's easy to fantasise about a simpler life without traffic and office cubicles, but these are applications by choice. The consequences of not having that concentrated energy available are massive. I have no interest in breaking my back trying to feed myself and my compatriots.

The carrying capacity of Earth could blow your mind if we work with nature instead of against it. I recommend Weisman (The World Without Us) to understand just how abundant nature can be without our interference. Add the conscious beneficial patterning and accelerated soil creation that humans can provide and you understand just how weak the Archdruid's argument is. The poverty of conventional Western thinking is reflected in what many peak oil writers believe must be in store for us. Humans are a marvel of evolution; the ultimate gardeners that can turn deserts into food forests. Most of us are simply unaware of it.

Michael Dawson said...

Tim, the fact that a substance might be "clean and potentially carbon fixing compared to crude oil and coal derived fuels" doesn't get you over the hurdle. Carbon dioxide emissions are one problem. Running our industrial capitalist society (or any industrial society) is another, and it requires a certain amount of net energy. You know this, I'm sure. And Liebig's Law applies. Being "cleaner" doesn't at all mean being net-energy-positive enough to avert catastrophe. Cleanliness and EROEI are two separate questions, both of which must be solved.

As to nature's fecundity, the question is how we can turn solar energy into usable fuels in sufficient quantity at sufficient EROEI surpluses to avoid eco-social catastrophe. Nature is certainly wondrously abundant. But us knowing how to make burnable stuff out of it sustainably and on a large-enough scale is an unsolved mystery, and not a hopeful one. Lush fields don't easily translate into full tanks.

By the way, I assume you know about Cal Tech's Nate Lewis. He says artificial photosynthesis, on which there was apparently just recently a major preliminary advance, is our only hope of sustaining some major chunk of current energy-use levels/lifestyles.

Tim Auld said...

Of course I understand the ERoEI aspect. Greer doesn't know how far we can take biofuels either. He's not trying to find out while taking pot shots at people who are.

Did you know that alcohol fuel use precedes gasoline by several decades? Farmers used to distill the produce they couldn't get to market, using the alcohol to run their machinery and selling the surplus to motorists. This was even before national electricity grids.

I looked up Nate Lewis and watched a minute of one of his presentations. He declared that plants are the equivalent of a failed MIT experiment. I won't waste any more of my time.

jagged ben said...

There's something to be said for sticking to numbers that have been well and widely demonstrated, as Stuart has done in his analysis. If there are reasons to believe that 10,000 gallons/acre of biofuel production is possible, there are also reasons to believe that photovoltaic cells can reach efficiencies much higher than %10. (After all, 40%+ has been done in the lab.) And even if you believed in the former and not the latter, the math would still seem to give the Singularitarians the advantage.

Tim Auld said...

Photovoltaic arrays have their place, but they only serve a single purpose; generate electrical power. Some of which is lost during transmission and use.

If you were to factor in all of the other functions of plants your equation would shift dramatically. Plants provide shelter, light shade, wind break, food, fodder, soil, nitrogen, medicine, oxygen, rain, beverages, solvents, adhesives, clothing, construction materials, furniture, animal habitat, clean water, erosion control, salinity control, temperature and humidity moderation, beauty, entertainment, stress relief, and of course fuel for light, heating, cooking and transport. The list goes on. They are self reproducing and self maintaining, their raw materials coming from the soil and air. You have to manage them to get what you want, but you do not need to sink huge amounts of energy and rare metals into their construction and ongoing maintenance. Stuart's life-cycle analysis should be telling.

I am not convinced of this commonly used 1% figure even for a single leaf. There are different types of photosynthesis: C3 & C4. C4 is more aggressive and often used by rampant plants. In general some plants are better at it than others. According to Wikipedia the net efficiency of a leaf at 25C is 5%, and sugar cane can achieve up to 8%. It depends greatly on the site conditions, which a good permaculture design would address. Again from Wikipedia:

"Plants usually convert light into chemical energy with a photosynthetic efficiency of 3-6%. Actual plants' photosynthetic efficiency varies with the frequency of the light being converted, light intensity, temperature and proportion of carbon dioxide in the atmosphere, and can vary from 0.1% to 8%."

I don't know in what conditions these measurements were made, but under a well developed polyculture, poison free and well balanced soil, mycorrhiza and bacterial association, and additional carbon dioxide, plants will achieve their full potential. It is a subject I intend to study in more detail.

Unknown said...

TO: Michael Dawson

In your initial post (prior to the palm oil discussion) you make some good points with which I generally agree, but I want to discuss your use of the word capitalism. I think of capitalism as an economic system that most reflects basic human behavior, namely the avid pursuit of self interest. Capitalism, as opposed to say socialism, generally allows this self interest to be expressed in a non-centrally controlled way, and I don't think capitalism is in any danger.

I think that the 'thing' to which you refer as being in danger, is our current growth-dependent debt-based monetary system. I see this as very different from capitalism, and I think that understanding the difference is important to having a better understanding of the issues that are driving our society in a such destructive direction and pace.

Michael Dawson said...

Leegs, you make three assumptions that I don't accept. First and second, you say capitalism (selfishness/greed) is a reflection of human nature. True, but so is altruism. Einstein was right to say human nature contains a range of possibilities, and that institutions determine how the range gets manifested. And capitalism is more than an expression of individual actions. It is a system that imposes greed on its top managers.

Third, you seem to think the "growth-dependent debt-based monetary system" is somehow separate from capitalism. Why? Every business strives to grow larger over time. The explosive growth of the financial sector is due to the fact that productive capitalists have nowhere else to invest their money to make it grow.

Bill Pulliam said...

57 comments, and I have not found a straightforward statement of perhaps the most fundamental aspect of your thought experiment:

One hypothetical reality operates on existing technology. The other operates on hypothetical technology.

Doesn't this make the comparison shaky, at best?

jagged ben said...

I have to disagree with your thrust there, Bill.

If the question is "How reasonable is it to expect the continued progress of machines?", then it begs the question to say that the machines haven't yet made the progress.

In addition, one can back off a bit from the starkness of the thought experiment and observe that Stuart's assumptions for Singularitaria largely DO exist, except for the machines designing themselves. If humans still supply the intelligence...well, that pretty much describes industrial society as it presently exists.

Stuart has given us one data point which, if accurate, argues in favor of the possibility of the continued existence of machines on photovoltaic power. There are clearly a lot of other factors that could cancel it out (and I think we all know that here), but the data point is valid in and of itself. And a 40:1 energy collection advantage for machines is a big data point, if it's accurate.

Tim Auld said...

This may be controversial, but I don't think greed is exactly the problem. In any case it can't be changed. Every species is greedy, but because their resources have been generally difficult to come by, the benefits of cooperation outweigh those of direct competition. Species that directly compete die out, basically, so we are left with communities of species that work together.

The issue is that mankind has been let loose on a resource base of unprecedented proportions that other species can't tap. Our behavior has reflected very primal motives; short term survival tactics. It's likely that any species would do the same as we have done and our monetary system is just a symptom of this.

So consider for a minute that greed isn't bad, if it can be harnessed for beneficial outcomes. An investor only wants a return on his money, not necessarily by a dirty great hole in the ground or enslaving people. We can use the productive power of ecosystems for this purpose. I am developing an investment model that will achieve such a result. Uncontrolled or immoral growth in monetary supply is indeed a problem, and my model would avoid borrowed money as far as possible to avoid it.

Bill Mollison talks about it here: Permaculture for Millionaires.

Bill Pulliam said...

Ben ---

Um, yeah, but... if humans are going to provide the intelligence, then you need humans. Which means you need agriculture, education, housing, health care, government, water, sanitation, all that stuff. You can't just throw humans back in the mix and say "well, it's mostly the same as the original human-free all robot world."

Non-industrial agriculture exists. Near-human AI doesn't exist, not even anything remotely close to it. It's pure science fiction.

It's really a very strange comparison...

Unknown said...

Tim - if the 10,000L amount is from Bill Mollison, we need to remember that, though spot on in his vision he's notoriously unreliable with figures. We shouldn't repeat those kind of claims from anything he said or wrote without double checking, and you really need multiple verifiable references to support game changing claims like the one you are making. Otherwise you might make us permaculturists look a bit silly.

(by the way I'm not supporting Stuarts either, its just not possible to come to conclusions like "7% efficiency" unless you are grossly oversimplifying the life-cycle energy inputs...)

Tim Auld said...

Hi rafael,

I'm aware of this and I'm in the process of double checking and finding examples. I mentioned this to Blume and he's looking into it.

Blume says the 10,000+ gallons per acre for sewage grown cattails is incontrovertible (converting the cellulose too). His students "built one of the first cattail marsh secondary sewage treatment facilities in the world, for the city of Arcata, California", so his figures are based on actual measurements. The energy can theoretically be captured and made available by at least one species.

I was reading The Botany of Desire last night, and the thought occurred to me that we haven't done much in the way of breeding perennial crops for energy. If undomesticated Nipa palms can get 2140 gallons per acre (Blume), what happens when we start selecting high yielding specimens? There might be better species than Nipa for this, if Bill's Babassu figure has merit.

Where did I come to a "7% efficiency" conclusion?

Sapphire Wainwright said...

Two points:

1) What are the Singularians using as a resource base to build themselves and their society. Nowhere did you posit that the machine culture was even remotely bio-mechanical.

2) When is the last time you saw a car or solar panel, let alone an advanced robot, give birth?

Unknown said...

Stuart, this is an interesting gedankenexperiment, but it's all rather beside the point. Either recursively self-improving AI will be developed before we don't have the resources to build and run it, and it will quickly become the dominating influence inside its light-cone; or resource depletion will hit at or slightly above our current level of machine learning, and Greer's catabolic collapse will be realized.

Let's just hope that if the former scenario happens, it's after someone has developed and implemented a good theory of friendliness.

Yvonne Rowse said...

Did I miss the bit where someone said how people survive in the Singulatarian society? This is a fascinating discussion but meaningless from the point of view of developing a sustainable human society. It doesn't matter how much more efficient a purely machine society is. Greer was talking about machines in the service of people versus people supporting themselves. Take out the people and your comparison is meaningless.

jeremy said...

I am admittedly naive when it comes to the nuts-and-bolts of producing alcohol from crops. But while there are many figures being bandied about as to 'gallons per acre' of this or that crop, I'm forced to wonder what the energy requirements to actually *make* those gallons are. Harvesting the crop, processing it before distillation, the distillation itself, subsequent storage and transportation. All of these things require non-trivial amounts of energy. What is the real EROEI ?