A paradox

A paradox, a paradox, a most intriguing paradox

      The Pirates of Penzance, Gilbert and Sullivan

---

Straight forward logic would seem to dictate that if you improve the efficiency of a given work process within a system then the whole system should enjoy the greater efficiency. Right?

Or we might reasonably assume that developing new ways to produce energy would on the whole increase the availability of energy.

These are slightly different ways of saying that a local optimal solution to a problem should contribute to the overall optimality of the global system. It turns out that this is mostly magical thinking.

In my last blog I said something to this effect:

As in all network of processes relationships, a local optimum does not imply a global optimum. What looks good at one node in the network, the choice to go to that other node instead of any other choices just because it looks like the best outcome can have disastrous impacts if all of the subsequent choices lead to extremely high costs.

I finally got around to reading Leonardo Maugeri's October, 2009 article in Scientific American called "Squeezing more oil from the ground" and I couldn't help but realize that this kind of thinking dominates the technological cornucopian perspective. In the article Mr. Maugeri promises us that there are much greater reserves of oil left in the ground than had previously been believed to be recoverable. And, that "new technology" will make that oil recoverable (real soon now). Hence, all this fuss about peak oil is something we can ignore for at least another couple of decades. While he does point out that these technologies are costly, what he never gets around to pointing out is what the actual total cost, in energy units, are realized in developing and deploying them. He leaves the reader to assume that because these technologies provide greater extraction efficiency (as defined by the recovered volume of oil) that we are all better off as a result. What he doesn't ask is: What is the net energy gain after all energy costs are taken into account. This wasn't an oversight or a belief that the average SA reader wouldn't be able to understand the nuances of energy return on energy invested. He simply doesn't think like a systems scientist.

It is extremely typical to apply the greedy method of thinking to what you believe is profit making decisions. You are at a node in a decision network. You can't see anything except the next possible moves you could make (decision taking) and have an estimate of the costs of each move. So you choose the one with the least cost. But suppose this is just one step in a complex web of possible (and possibly interconnected) steps along a path to a specific destination or specific result. If there is a longish chain of steps between you and your destination, and all you know is the cost of the available steps from where you currently sit, you could be tempted to apply the greedy method at each node in the network. But if you do, you are in for a nasty surprise.

This figure is a simple representation of a decision network. You are on the node with the 'X' and you have to figure out which way to go next; what the next node you should go to would be. All you have in the way of information about this network is the cost of each outward bound step. In the figure these are 5, 10, 50 and 20. What you don't know is what the costs out of the node that you end up choosing. Once you choose, you cannot go back. Hell, you don't even necessarily know if your choice will lead you to your desired end point (notice the upper left node?)

Notice that there is an actual cost associated with every link in all of the possible paths from where you are to where you want to go. But when you are standing at the 'X' node you can only see the costs of going to the next step. You have no god's eye view of this 'problem'. Unfortunately this is how most of our real world problems actually work. We have limited knowledge to work with, and have to make decisions with limited information. What do we do?

Clearly in this example picking the path with a cost of 5 is a disastrous choice. Picking the one with a cost of 10 is no better. The path starting with a cost of 20 looks undesirable from the location at 'X' but as can be seen from our special god's eye view, it turns out to be the best choice, cost wise.

A lot of energy efficiency problems have a similar structure. I mentioned my experience in the 80s in solar energy where the energy required to make the flat plate collectors could have easily been more than the energy gain for the local householder. The latter enjoyed a lower heating bill but at the expense of the rest of us in terms of total available energy. The problem there was the cost-price distortions imposed by governmental and market convolutions that decoupled energy costs from dollar costs. The latter did not accurately reflect the former, otherwise, no one would have bought flatplate collectors to heat their houses.

Just to give an example of what information distortions can do to decision making take a look at this next figure.

The bolder lines and numbers represent an information feedback chain that accumulates perceived costs according to the number of steps in the path. Think of these as 'prices'. Note that they accumulate at each step and that they are artificially maintained at 10 added per step. Suppose you now had the information shown (only two examples) in bold numbers at the 'X' node. Which path would you choose? Once again you would be fooled into taking the wrong path even though it is cheaper in terms of feedback numbers, i.e. price. As far as you can discern, that price is really less than the one for 60 and since this represents prices accumulated all the way back to the destination, you are perfectly rational to assume this is the best, cheapest path. But the underlying costs are what really count. Believe it or not, there is a lot of this dollar price distortion versus energy costs so that our local buying decisions do not fairly reflect the true costs being borne.

In the case of increasing, or at least maintaining, oil flows by adopting new technologies we have a similar problem. Secondary extraction techniques, for example, involve pumping CO₂ or CH₄ under high pressure down into the wells to increase flows. Something similar is done by pumping water (sea water in Saudi Arabia) into the well. Oil floats on water so this boosts production. What the oil engineers report as the costs of doing this is the dollar amounts used to pay for the gasses and electricity to pump. This works out OK as long as the gasses don't have to be shipped long distances. None of the oil pumping operations puts a price on externalities, like toxic spills. In fact much more energy (hidden in the assets section of the balance sheet!) was used up just to set up the infrastructure needed to perform these operations. It would take a small army of accountants backtracking through the capital equipment assets to derive what energy was actually consumed, but I assure you it would amount to a great deal more than shows up in the reported costs of operations of secondary extraction.

Also consider the energy costs of exploration. How much of that shows up in the real costs of keeping the oil flowing? Or the drilling in deep water or frozen tundra? They say it is expensive, sure, but we keep the same number of barrels of oil flowing. We never get an accounting of the net energy gained.

Another case in point is that of the use of cadmium telluride in a 'newer' technology for photovoltaics. What the enthusiastic purveyors of this technology will gush about is the increased efficiency at collection and conversion, at, now get this, lower prices than conventional polycrystaline silicon cells. What is left out of the equation is the fact that tellurium is rare and will eventually prove to be extremely energy costly to produce, or that cadmium is an extreme poison that needs very careful processes for recycling to prevent toxic accidents. All of this takes a lot of energy but is not currently reflected in the traditional economic sense in the prices.

A large part of the seeming cost savings come from the process efficiencies of manufacturing the cells (supposedly). But those efficiencies were gained by developing newer plants and equipment. All very good on the surface, but what lies beneath. How much additional energy went into the capital goods so that we could get a cheaper production process?

Supposedly, over a very long time scale, when all of the capital has been properly amortized and the competitive pressures on pricing have been stabilized, eventually the long-term price of a product like this should reflect all but the true externalities (the costs of environmental damage resulting from operations), that is the costs the companies can avoid paying [There is a move afoot in accounting to report what is called in financial statements. The purpose of this is to account for potential liabilities companies accrue and might end up having to pay for, like fines to the EPA. Well, that will be the day!]

Unfortunately it isn't just energy problems that are of this local optimum, global sub-optimum ilk. They seem to come up over and over again when someone attempts to solve a local problem but ends up having global consequences. Food aid might fit in this category according to some critics. For example consider the case where grain surpluses are given to African villages who come to rely on the donations (especially if development efforts are hampered by poor soils or lack of water for irrigation). Then one day along comes corn ethanol, profit motive and all, and the aid gets cut because we Americans want to keep gas prices low.

Let's face it. We want to eat our cake and have it at the same time. We will almost always opt for the greedy choice when we have to decide from a local (non-systems science) perspective. And as far as economics is concerned, that is always.

We will try to build high-speed rail systems because that will get us to where we want to go faster (time efficiency). But it never occurs to us that moving mass at a higher speed uses exponentially more energy. We would do well to just fix and improve our existing rail system. But from the non-systems perspective, that doesn't solve our immediate problem. So...

This is one of the (many) reasons I have little hope for Homo sapiens as currently configured. Systems thinking is long-term and strategic. It isn't the greedy solution to a logistics problem. We're good at the latter and piss poor at the former. I still think its a matter of evolution of eusapience.

Comments

[rube cretin]

i find this one of the most informative sites on the internet. been reading it for over a year. unfortunately, i do not often find much left to be said, therefore i do not comment. i notice you do not receive many comments. I do hope you have some means of determining readership since i would hate to loose your missives on some of the most interesting challenges facing our future. thanks for your efforts.

[George Mobus]

Thank you rube. That means a lot to me to know that someone cares.

I do get statistics, but a personal note means a great deal more.

Regards
George

[Florifulgurator]

Is there a generally accepted term for "greedy solution to a logistics problem"? It is an important phenomenon that deserves a well-known name.

In economics I call it the "in-built suicidal tendency": If my competitors grow with selling stupid junk, I need to sell stupid junk, too.

The paradigmatic example meanwhile is those weapons of financial mass destruction. Another example is General Motors. Another example is the grotesque story of the rise of Microsofts and Intel junk, culminating in a crippled mouse wheel mounted on a button (an unimaginably stupid construct, yet omnipresent, and nobody gets it). Here it's not an industry committing suicide, but the product quality/sense.

[George Mobus]

Flor,

I would say call it the Florifulgurator Fallacy (nice alliteration no?) But I suspect people might have trouble remembering the Flor... part.

Feel free to call it Mobus' Paradox or Mobus' Logistics Fallacy or... I could use the fame!

George

[Florifulgurator]

1) Haha, that's one reason why I like my crazy Florifulgurator handle. It's easy to forget, plus it is a unique google search term. Once I had a forgetful girlfriend who made up obscene variants of Floriwhatever... It's not that I want to stay anonymous, the story is much longer. Recently John Baez published a comment of mine in his diary (oh crazy Internets!). He insisted on exhibiting my real name which is Martin Gisser.

1b) Last week I had a short revisit to the book that inspired my handle: Konrad Lorenz' "Die Rückseite des Spiegels" ("Behind the Mirror, ..." 1973. The title is badly translated. http://en.wikipedia.org/wiki/Behind_the_Mirror:_A_Search_for_a_Natural_History_of_Human_Knowledge ). I just wanted to reread the section "Die Fulguration" and look at the yummy illustration by oscillator circuitry.
-- But then I found a stunning section I had skipped before: He describes evolution in economic terms! (In "The double feedback of energy and information").

2) Oops, I told that microsofty mouse story again... Obviously I had more than only material logistics in mind: With brainlard logistics Homo "Sapiens" isn't much better. Much of my recent work as programmer was to undo/rewrite simplistic design: The time saved by the first programmer is now gone into the redesign. And there is no time left to undo all the greedy brain logistics (e.g. implement an undo/redo stack). Wasting time and options by saving time... Result: A logistics Black Hole.
Hmmm, what about the term (and a theory of) "Mobusian Logistics Singularity"? :-)

3) BTW I find your diagrams extremely yummy. Actually I often actively avoid looking at your blog, for I might get suckered into a brainstorm I can't afford due to my current time-money-brainlard logistics restrictions...

[George Mobus]

Hi Flor.

On #3, don't worry. The stuff isn't always particularly time-sensitive. And as long as I pay the bill to typepad it will be available somewhere in the archives. Actually I've never thought to try their search options; something I'll have to test.

On #1, I'll take a look at the Wikipedia article. Lorenz was one of my favorite authors in my early days studying behavior and ethology. I don't think I had read this particular book though. May need to have a go.

George