Family Archival Storage

One of the things my paternal grandmother left -- in addition to a marvelous music box -- was an overstuffed three-ring binder full of notes about the family tree.  I've finished scanning all 705 pages that she left.  Most of the pages are either pencil-on-paper or typed with a fabric ribbon; even the ones that are approaching 50 years old are in pretty good shape.  There are a few pictures included that go back much farther that are also well preserved.  The old black-and-white photos are, in fact, in better shape than the few newer color photos.  Copies of the scanned pages are being distributed to several family members to reduce the risk of losing everything in a single fire/tornado/other disaster here.

I used to have lunchtime discussions about archival storage, intentional or otherwise, with the woman who managed the company's technical library.  She maintained that we have a much better idea of what the common man thought and wrote about the US Civil War than people living 100 years from now will have about what we think today.  Her fundamental reason was that during the Civil War, people left their thoughts on media with an inherently long shelf life: silver-based black-and-white photography and pigment-based inks on relatively low-acid paper.  Letters or a diary written on such material can be tucked away in a trunk in the attic and still last for a long time.

Today, of course, we send e-mail and post stuff on Facebook and tweet.  Most of which never reaches paper, and all of which is subject to the vagaries of computers, both our own and those "in the cloud."  I have CD-Rs that are approaching 20 years old and seem to work fine on those occasions when I need something from one of them, but digital media are often an all-or-nothing proposition: they work perfectly until they fail, but once they fail they're unusable.  Then there are file system formats, and formats for the file contents themselves.  In 100 years, even if the bits on a CD-R are still good, will there be a drive that can read it, or software that understands the ISO file system, or applications that still handle JPEG image files?

Even if you assume conscientious descendents that periodically copy the material from the old physical medium to a new one, and transcode images from one format to another, there's a more subtle problem.  Serial encoding of images with lossy algorithms (eg, JPEG as it is almost always used) results in steady deterioration.  The first encoding introduces small errors in the reconstructed image.  Many of these errors produce visible artifacts in the image if you know what you're looking for.  The image to the left illustrates ringing errors in an MRI image (in this case, visible "echos" of the sharp dark/light boundary) [1].  The serial encoding problem occurs because a future encoding with a different algorithm will waste bits trying to accurately reproduce the artifacts.

I find myself struggling to find some sort of middle road, one that allows me to take advantage of the ability of contemporary tech to impose indexing and organization on Grandma's work and additions to it, and at the same time, to allow both Grandma's stuff and any material I add to survive something as extreme as skipping a generation along the way.


[1] Image taken from the American Journal of Roentgenology, An Introduction to the Fourier Transform.

Food Prices and Political Crises

The New England Complex Systems Institute has updated its briefing on the relationship between food crises and political crises, with particular attention paid to the region of the Middle East and North Africa.  The bottom line, overly-simplified is this: when global food prices go too high, the MENA countries experience an increase in the frequency of political unrest.  All of the MENA countries are net importers of food calories, much of it in the form of bulk grain.  Several MENA countries import more than a million metric tons of corn, rice, and wheat per year: Algeria, Egypt, Iran, Iraq, Israel, Morocco, Saudi Arabia, Syria, Turkey, and Yemen.  Egypt alone imports over 15 million tons, more than 400 pounds per person.


The NECSI results are summarized in this figure.  The black line shows nominal food prices; the vertical red lines are incidents of political unrest, many of them concentrated in the "Arab Spring" of 2011.  The paper develops a model and a price threshold at which political unrest is likely to occur.  Based on current predictions of price increases (US corn prices reached record highs in August, but have retreated somewhat since then), the model predicts that another round of incidents will begin in the region in October, 2012.

What's driving the price increases?  One large factor is the drought that spread across much of the US grain-growing regions this year (weekly maps showing the extent can be found here).  The higher prices reflected a belief that US grain output would be significantly smaller (although today the USDA announced that the corn crop may not be quite as bad as has been previously feared).  A second factor is the use of corn as a feedstock for producing ethanol for use as a transportation fuel in the US.  The US has a renewable fuel standard that requires an increasing total amount of ethanol to be blended into gasoline supplies.  While the overall situation regarding ethanol is complex [1], the ethanol industry will still purchase large amounts of the corn crop this year.

The NECSI work provides a nice formal mechanism quantifying something that I find myself saying too often: "Drought in the US means that Africa starves."  To a lesser extent, you can replace the US in that statement with Argentina or Australia.  The reduced US grain crop follows relatively poor harvests in both Argentina and Australia, two other major grain exporters.  You can also replace drought with increased domestic demand and get the same result.  US ethanol mandates may be fine domestic policy -- although there are lots of analysts who disagree with that -- but terrible global policy.  And not just in terms of millions of starving people.  The MENA countries produce and export a lot of oil.  Political unrest generally disrupts activities like oil production (oil pipelines are easy targets for dissidents).  High prices for global grain exports could conceivably lead to higher prices for global oil exports.

Aren't complex dynamic systems, filled with feedback loops and time delays, fun?


[1] The penalties for failure to meet renewable fuel standard requirements are fines; companies may choose to pay some fines rather than very high prices for ethanol.  In previous years the oil industry blended more ethanol than required, accumulating credits that could be used to offset a shortfall and avoid fines this year.  Blended ethanol raises the octane of the final gasoline product; there are minimum octane requirements; the alternatives to ethanol may be more expensive, which would cause the oil industry to continue to use ethanol even at higher prices.  Based on the years I spent working for large corporations, I can guarantee that someone -- probably many someones -- is busily trying to solve an optimization problem that minimizes the costs of all the factors in aggregate.

Saudi Arabia, Oil Importer?

One of the frequently cited pieces of oil financial analysis the past few days is a Citigroup report suggesting that Saudi Arabia could become an oil importer by 2030 (eg, Bloomberg's story on the report).  The notion that the Saudis would be consuming all of their production, and then some, is based on extrapolating several trends.  The two fundamental trends are illustrated in the graph below (from Mazama Science's oil export database visualization tool): relatively flat production over the last two decades and steadily increasing internal consumption.

There are a number of reasons to believe the direction of the basic trends will continue.  Books have been written on when the Saudis will reach terminal decline in their production; Simmons' Twilight in the Desert is probably the best known, as well as being one of the most pessimistic.  On the demand side, the Saudis generate almost half of their electricity using oil (a practice the developed countries of the world gave up in the 1970s during the two oil crises); the country's population is young and growing rapidly; and domestic prices for oil products are set well below global market prices.  All of those factors point to growing domestic demand.

The best argument against the notion of the Saudis becoming net oil importers is the enormous role that revenues from oil exports play in the country.  75% of the government's revenues are from oil, and that money pays for many programs including defense, infrastructure development, education, and health care.  This is not spending that the government can simply turn off without serious consequences.  In the absence of increased production, the government needs to maintain a substantial level of oil exports, or rapidly build the non-oil economy, or both.  An additional problem the country faces is the make-up of its work force: a staggering 80% of workers are foreign nationals, not Saudis.

The Saudis appear to be taking some steps to delay the decline in oil-based revenues.  They have signed nuclear deals with all of France, China, Argentina and South Korea.  Commercial reactors would be used in place of oil to generate electricity and desalinate sea water.  The government is also considering construction of a large amounts of solar generating capacity for the same reason.  And for many years the Saudis have been working to move "up the value chain" in order to get more revenue from their petroleum: selling plastics derived from petroleum pays better than selling the petroleum.  Similarly, selling refined products yields more revenue than selling crude oil, and the Saudis have invested in refining capacity at home and overseas.  "Saudiization" of the work force is an official government policy, which might, in theory at least, reduce the population and domestic demand.

Can the Saudis pull all of this off and continue to export at current levels in 2030?  My own opinion is "not a chance."  Will their exports go to zero by 2030?  My thinking is that that also is not going to happen, although the exports may be plastics, chemicals, and finished petroleum products rather than crude oil.  I also think the path to that point is going to be erratic and possibly violent.  A lot has to change, and 23 years isn't a lot of time to accomplish those changes.

A Disappointing Study of US Renewable Electricity Potential

Back in July the National Renewable Energy Labs (NREL) published a report titled U.S. Renewable Energy Technical Potentials: A GIS-Based Analysis.  Graphical Information System (GIS) data has become really popular.  Thanks to satellites and the ability to handle massive amounts of image data, the world today is better mapped than previous generations could even dream about.  Using available GIS data, the authors evaluated the potential for each of several different renewable sources of electricity in the US.  Consider one resource -- rural utility-scale photovoltaic panels.  The authors identified all of the land which met certain criteria (rural, sufficiently flat, not water, not in a national park, etc) and calculated how much power could be generated using today's PV technology if all of the identified land were used for that purpose.

 The table to the left summarizes the results from the power for ten different renewable resources.  The potential is shown in petawatt-hours; in 2010, total US retail sales of electricity were about 3.7 PWh.  The total figure is not meaningful.  The land considered suitable for concentrated solar and rural utility PV is the same land, and it can't be used for both.  Even some of the individual numbers aren't meaningful.  Much of the land suitable for rural utility-scale PV is also suitable for growing food crops, an application that can't be realistically ignored.  Most renewable resources are intermittent on some time scale.  Rural utility-scale PV may be able to provide many times the US current power consumption, but without some form of storage, it still won't keep the lights on at night. 


The authors presented their results at the state level and used the data to produce maps like the one shown to the right.  This particular map shows the potential for hydrothermal power, which is concentrated in the western states.  But state level aggregation doesn't seem particularly helpful either.  Western states are large, so energy sources can still be far from the population centers (as well as separated by the odd mountain range here and there).  Western states vary enormously in terms of their population.  California and Nevada may be in the same category in terms of hydrothermal potential; but because of the differences in their populations, Nevada's resource may be sufficient to meet all of Nevada's needs, while the same resource in California can provide only a small fraction of what is needed.

I'm seriously disappointed by the study, which I think adds very little value.  We already knew that rural utility-scale PV could potentially produce far more power than we currently consume.  I can think of a half-dozen things to do with the data that would have been much more useful.  For example, for some (or all) of the 50 largest metro areas, how far and in what pattern would rural PV need to be deployed to meet the metro areas' power needs?  Avoid mountain ranges; avoid areas that are heavily forested; avoid existing towns.  Where do the patterns overlap?  What happens if current crop land is excluded?  There are enormous amounts of GIS data of various types available, and some of the ways that it can be used would be valuable.  The results in this paper aren't one of those, unfortunately.