The Human Security Index: Potential Roles for the Environmental and Earth Observation Communities

EarthzineArticles, Earth Observation, Original, Politics, Sustainability

Figure showing Log(Median Household Income), part of Human Security Index USA Prototype Version 0.1. Blue indicates relatively higher income => better situations.

By David A. Hastings; U. S. Department of Commerce, National Oceanic and Atmospheric Administration, National Climatic Data Center, Asheville, NC, USA (david.hastings@noaa.gov)

Summary

Before 1990, the development level of a country was typically “observed” by looking at a single economic figure, such as Gross National or Domestic Product per capita. With the advent of the Human Development Report in 1990, many people came to prefer the HDR’s Human Development Index, with its enriched blend of knowledge, health, and finances. The 1994 HDR went conceptually farther, with an essay on people-centric human security – the level of economic and social security in a society. One could say that analysts had moved from monochromatic to multispectral observation, with a concept paper but no actual effort to pursue hyperspectral observation of societies. Finally, in 2008, embryonic hyperspectral observation was facilitated by unveiling of the prototype global Human Security Index (HSI). Improved global HSI Version 2 was recently released around a trinity of Economic, Environmental and Social Fabric covering about 230 national-level societies. A prototype HSI at county level for the USA is in review, and the approach is being extended to developing countries at sub-national levels using publicly available data. How do people actually fare in their homes, communities, country, or the world? How well can such a formulation process characterize such situations? How can such an indicator be useful in research, strategy and policy development, and the pursuit of progress? Such discussions are beginning, with considerable opportunities for people in the environmental Earth observation community.

Why a Human Security Index?

It’s an attempt to push the envelope on characterizing socio-economic situations of people.

Since 1990, the Human Development Report (HDR [1]) of the United Nations Development Programme, and its Human Development Index (HDI [2]), have helped to re-focus thought on development from raw “economic growth” to something more profound. The HDI blends indicators of education, health, and income, showing that some low-income countries can still deliver reasonable knowledge and health environments (while in some cases reducing the need for cash in the pocket) and thus deliver high quality of life without financial riches. The 1994 HDR [3] discussed people-centric human security – summarized as “Freedom from fear; freedom from want.” [4] Surin Pitsuwan [5], formerly of the United Nations Commission on Human Security and current Secretary-General of the Association of Southeast Asian Nations (ASEAN), has observed that human security is a longstanding evolutionary concept, going back to such thinkers as Hobbes and Rousseau – and that human security is arguably the reason for the state in the first place. Yet news and decision-making in most countries appear to remain enshrouded in “economic growth” rather than using relevantly targeted measures on conditions of all the people. More recently, the “Sarkozy Report” [6] was commissioned by the President of France aiming to help advance thought beyond “GDP as an indicator of economic performance and social progress.” Do we need a more comprehensive “index of leading socio-economic indicators”? Could such an index help decision-makers to strategize and pursue improved situations? Might it help researchers, the media and the public to better play their roles toward better understanding of current situations “here and in possibly better-run administrations” and more effective pursuit of better win-win outcomes for everyone? Yes – though the Sarkozy report neither made much progress toward answers [7] nor seems to have resulted in a publicly viewable process toward same.

To professionals in the Earth observation community, where GDP could be Human Development’s monochromatic monoscopic photography, the HDI could be termed “the NDVI of Human Development.” (The Normalized Difference Vegetation Index has been used as a proxy to assess several phenomena). The HDI has widespread and diverse application and support. The global HDI has spawned numerous national HDIs, treating many developing countries at higher administrative/spatial resolution, such as provincial/state or county/district levels. There is a non UNDP HDR for the USA [8].

However, there are also laments about claimed imperfections:

  1. It’s a proxy. What are we measuring? Might the HDI be insufficient improvement over GDP per capita [9] for characterizing human development? [One possible response: “The HDI has been useful. If we look deeper we see that some countries have done better with health care or education than for raw GDP, and have done well for their people in the eyes of some analysts. But we should be able to do better than the HDI by now.”)
  2. It’s not fully global. Can “UNDP’s global HDI” be more geographically global? Pursuing global completeness is germane to usefulness a global HDI. When the HDI was first released, it was a challenge to find relevant data of consistent quality for many countries. The first HDI broke ground by encompassing 130 economies. By 1994 it covered 173 economies but still only 177 in 2007 before ratcheting to 182 in 2009 but then dropping to 169 in 2010. The countries left out of UNDP’s HDI have intellectual and cultural assets [10], which the world can benefit from if such countries are not ignored. Has UNDP’s HDI become a laggard in geographic completeness? Yes. An HDI encompassing ~230 countries is now available [11, 12] to demonstrate the potential for more global completeness of such an indicator.
  3. Can we make a better index now? Can the HDI be thematically complemented, or perhaps replaced, by a more comprehensive measure? Let’s see (below) if there is a good answer to that question.

Creation of a global Human Security Index

Some have argued [e.g. 13] that a human security index could not be created, as we lack (A) agreement on the concept of human security and (B) appropriate data. But it has also been argued [ibid] that “there must be a way to recognize it [human security] and measure it.” Similar concerns did not impede the release, refinement, growing respect and use of an HDI two decades ago. And we are now enriched with several innovative datasets and candidate indicators on a diversity of subjects related to human security, which may be utilized and hopefully strengthened by additional engagement. So why not try to forge ahead? Beyond this, in Louis Emmerij’s report on the United Nations Intellectual History Project [14, just stumbled across in 2011] which celebrates selected vital contributions from the United Nations, he specifically lists “measures of human security, where integrated approaches should be explored going far beyond the traditional compass of either the military or the security forces of countries” as the second of five priorities to be worked on now under the UN umbrella. Thus, when a prototype Human Security Index enumerating 200 countries was shown to colleagues for their thoughts [while I was with the United Nations for most of the past decade], the response was “Publish it.” This was done in 2008-2009 [15, 16].

Table 1 showing USA human dimension Index Since then a discussion and prototyping process has led to the formulation of the HSI around a trinity of Economic, Environmental, and Social Fabric Indices, with each of those indices comprised of datasets, and indicators produced by the research community. Currently, over 30 datasets and indicators [17] are used in global HSI Version 2, released at the end of 2010 [18, 19], which covers about 230 national-level societies.

The HSI process consists of vetting candidate data and indicators, assimilating appropriate candidates into tabular data management tools to work on index formulations, linking the data to a geographic information system for visualization and assessment – and discussing issues and findings with data developers and researchers.

Website http://www.HumanSecurityIndex.org was created to present and discuss data, maps, and issues surrounding the global HSI. The Website is hosted by researchers at Osaka City University.

Toward a higher resolution Human Security Index

Discussions following the release of the global HSI resulted in encouragement to attempt a HSI at a community level such as for the USA. A community-level HSI [and its constituent data] could eventually be useful [20] for awareness and assistance in understanding current situations, and in strategy, design and delivery of improved public services. For example, could one better anticipate the information absorption/response capacities of specific socio-economic regions, which could impact efforts to mitigate or respond to disasters as they unfold?

A table showing the flow of input data through thematic components to the Human Security Index (right), a map of one input data set (just below), the trinity of components and the composite HSI are shown (farther below) in this article.

Spatial resolution: Work on HSI USA was begun in 2009, with prototypes starting to be formulated in July 2010 at the county level. Some counties are quite large in area; some (but not all) source data are available to smaller geographic units than counties (e.g. to Census Blocks and Block Groups, or to grid cells in the case of data crafted from satellite imagery). One fertile R&D area may be the merging of data of diverse input resolutions, for optimal observational capability.

 

Figure showing Log(Median Household Income), part of Human Security Index USA Prototype Version 0.1. Blue indicates relatively higher income => better situations

Thematic resolution: Discussions with diverse colleagues in the UN, at conferences, and in academic circles in the USA and overseas have evolved around how to strengthen the conceptual framework for formulating the HSI at global or local “resolutions” – now revolving around Economic, Environmental, and Social Fabric Indices. Reasonable economic, environmental, and social situations for all people are often considered the point of good governance – no matter what the policy modality for achieving such goals. But it could be argued that fiscal growth has long been the dominant focus of many-most decision-makers and their support staff; environmental issues were merely an undercurrent until nationally somewhat mainstreamed since Earth Day (in 1970) and globally mainstreamed at the United Nations Conference on Environment and Development (in 1992). Social issues might arguably reside in the minds of only some decision-makers [21]. The HSI for the USA currently includes about 35 input “channels” as sketched in the table below. Some of the input data (such as the computation for “creative class”) are actually multi-channel thematic aggregates, so the total number of actual inputs is considerably greater than 35.

As in almost any endeavor of Earth observation, there are challenges in characterizing the target (visualizing the level of economic, environmental and social security for people and communities of diverse circumstances) with the available sensors (data collection systems) and data paths (via data communication, management, and funding systems that may have been designed for different core applications). For example, the Bureau of Labor Statistics computes unemployment at six levels, U1 through U6. It regularly reports level U3, but that is not nearly comprehensive enough for researchers or indicator developers trying to understand entrenched unemployment / underemployment, including people who have given up the search for jobs in their fields of expertise (e.g. auto manufacture or exploration geochemistry), and may have accepted work, say, as taxi drivers or in the fast food industry, perhaps part-time with no benefits. There are Websites which discuss such issues [e.g. 22] and argue that a complete enumeration of unemployment would yield considerably higher numbers; but there is no known public county-level dataset for peer review, possible strengthening or use. Recently, the American Community Survey of the Bureau of the Census released its first five-year temporal composite, which now enumerates a diversity of estimates for all counties, where previous three-year composites covered only about 1800 of about 3140 counties. The new figures include estimated parsing of employment by numbers of hours worked weekly, numbers of weeks worked annually, by gender and other

Map showing Economic Fabric Index USA, Prototype Version 0.1. Blue indicates relatively better situations.

Figure 2. Economic Fabric Index USA, Prototype Version 0.1. Blue indicates relatively better situations.


criteria that offer hope of improved characterization of employment at the county (and, in at least some cases, potentially finer) level. However, the situation is not always one of improved observational capabilities. For example, income data suitable for computing the distribution of income inequality (such as the Gini Coefficient [23, 24]) may not be available to adequate quality, partly as a result of the discontinuation of the Long Form process in the 2010 census, according to Mark Burkey (reference [24] and written communication, 2010).

Input data have been selected which appear to be best available proxies for important situations for a cross-section of people. In general, data are managed in the database in their original units of measure, but also are scaled between 0.000 and 1.000 for processing in the HSI and its component Economic, Environmental, and Social Fabric Indices. In general, the data values are given a linear stretch, placing the “high human security” tail at 1.000 and the “low human security” tail at 0.000. However, in the case of income, two decades of experience with the HDI resulted in a logarithmic scaling of income between tails [25], under the generally accepted understanding that an additional $100 in ones pocket is more than 200 times more valuable for people with incomes of $1000/year than for people making $200,000/year.

Candidate HSIs are currently being formulated in two ways:

(1) Averaging scaled input data with equal weights into the six components shown in the Table, then averaging the six components with equal weighting into the HSI, and
(2) Averaging scaled input data with equal weights into the six components shown in the Table, then averaging those six components into their respective Fabric Indices (by averaging the six (global) or four (USA) components of the Social Fabric Index into the SFI with equal weighting), then averaging the three Fabric Indices with equal weighting into the HSI.

Map showing Environmental Fabric Index USA, Prototype Version 0.1. Blue indicates relatively better situations.

Figure 3. Environmental Fabric Index USA, Prototype Version 0.1. Blue indicates relatively better situations.


The latter approach was used in global HSIv2, and is being used in HSI USA – in order to more evenly balance economic, environmental, and social aspects of the HSI.
Why so many input data sets? Partly for signal enhancement. Just as a seismic exploration crew working in a noisy environment takes multiple readings for a given location, input data to the HSI may have undetected, poorly understood, and/or impregnable noise. Where the signal to noise ratio of some input datasets may not be fully known, combining such data with others addressing a similar broad subject may help the economic, environmental, or social signal to rise above the noise.

Some people have observed that such indicators are most useful when they are simple. Agreed. But what does “simple” mean? GDP is simple to show, but not necessarily simple to assess as an indicator of the financial security of a former auto worker now with a part-time job in a fast food restaurant because of downsizing at the former employer. (County median income, unemployment and underemployment, percentage of people without health “insurance” coverage, mental stress and housing foreclosure rates might all better touch on that person’s human security than might county GDP.) The HDI was relatively simple and straightforward, after users got accustomed to its evolved methods of computation. The HSI, being modeled after the HDI, is considered conceptually straightforward, despite its greater comprehensiveness with over 30 input datasets.

What does the HSI look like at this point? Here are plots of household median income, the prototype Economic, Environmental, and Social Fabric Indices, and the resultant prototype HSI USA.

Map showing Social Fabric Index USA, Prototype Version 0.1. Blue indicates relatively better situations.

Figure 4. Social Fabric Index USA, Prototype Version 0.1. Blue indicates relatively better situations.


This paper is not a socio-economic assessment of the HSI, so such discussion will be omitted here. However, when one looks at constituent datasets (such as median income as shown in Figure 1) as well as the three fabric indices and the composite HSI itself, one sees many overall broad similarities, yet significant differences in detail. It is clear that median household income alone resembles human security in the USA (as illustrated by the HSI, Figure 5), or even Economic Fabric (Figure 2), about as well as a single-channel narrow-band monochrome image might resemble a color composite built from hyperspectral data.

There have been numerous indicators of well-being, from the original Places Rated Almanac of 1981 to The Economist’s Quality of Life Index. But most such indicators appear aimed at the middle class and above. The Human Security Index is broader-based, intending to better characterize situations for all peoples. Combining the summary graphics with the input database can be useful for a number of research arenas and interests. This should be possible after additional documentation and review brings HSI USA to releasable quality.

In one test application, an experimental county-based HSI was drafted for rich, upper middle class, middle class, and poor residents, for each county in the USA. This was done by drafting working hypotheses on what factors might be important, or unimportant, for more or less wealthy people – then re-selecting data inputs for demographic prototype Human Security Index models according to the various working hypotheses. For example, per capita income and percentage of university graduates in a county may be less important than cost of living or availability of special public school programs for working class or unemployed Americans. The percentage of people whose English is less than “good” and the percentage of high school graduates in a county may be less important to people in the top 2% of the income spectrum. Some counties do persistently well or poorly in such experimental demographic models, but others change considerably. In some locations this approach is moot: there are relatively few very poor people in Douglas County, Colorado, and few wealthy people in Starr County, Texas.

Map showing Human Security Index USA, Prototype Version 0.1. Blue indicates relatively better situations.

Figure 5. Human Security Index USA, Prototype Version 0.1. Blue indicates relatively better situations.


In another test application, sub-national level HSIs are being drafted for some developing countries to test the hypothesis that HSIs may be initiated using publicly available data for many such countries. A data base being formulated for Thailand, for example, uses data publicly available from national Human Development Reports for Thailand, plus data from the Website of the (Thai) National Statistical Office. Actually, with the considerable progress in decision-support data development in many countries, such a task is rather more achievable than one might suspect – at least for an introductory working prototype and initial use of such information for improving developmental strategy efforts.

The previous two paragraphs merely sketch possible applications of the Human Security Index approach. Undoubtedly there are many more such approaches to data bases integrating economic, environmental, and social data to community levels. Of course, the Human Security Index is hardly the first such compilation. Thirty years ago the U.S. Navy Fleet Numerical Oceanography Center compiled TERDAT [27, 28] which included data layers such as elevation, “characteristics of terrain” and “percentage of urban development” on a global grid.

What might be next?

Some thoughts at this point in the HSI process:

  1. Assessments by indicator developers might focus on selection and improving wholesale source data sets and documentation, and on possible improvements of formulation to support research and application. There appear to be considerable opportunities for partnering between existing curators of certain data sets and groups interested in strengthening such efforts. There are many opportunities for new authorship groups to innovate and publish new data and indicators. But this all depends on possible interests by data developers. Who might be interested in such a thing?
  2. Assessments by researchers (when the full data, not just graphics displays, are made available) might look at what situations may result from peoples’ behavior (e.g. “voluntary” unhealthy practices like substance issues including over-eating, etc.), or from governance (up to 70+% of relatively poor county residents are without health “insurance”, as many as 50+% are functionally illiterate in some counties, over 10% of black males aged 25-40 residing in prisons/jails in the globally leading country in incarceration rate, etc.).
  3. Assessments by decision-makers might ask questions on how to improve situations in areas of weakness. In the example of disaster management, how might evacuation plans and practices consider the non-negligible numbers of people who
    • Are carless and weakly served by existing transportation infrastructure? But when evacuation opportunities are provided, may (for whatever reason) decline to take advantage of them?
    • May be afraid to leave their at-risk abodes for the community shelter out of concern that they or their children may be subjected to harassment or violence?
    • Are injured or ill, but who have tenuous life-lines that have been disrupted by the disaster and lack resources to find alternatives?
    • Are in other groups that may have not been served well from recent strategies or efforts?

Let’s visualize the financial situations of most members of national and state-level strategic planning teams. How many national or local agency strategic plans mainstream inputs from population groups with household incomes below 50% of the national median? Could the strategic planning process be improved with more fully mainstreamed diversity of participation in the key decision-making teams?

Where might the Earth Observation community fit in?

Potentially, quite squarely in the middle of the process. Many of the approaches to developing the HSI derive from three-plus decades of remote sensing, GIS, and indicator development – mostly involving direct Earth observation. But the HSI needs diverse perspectives, sources of data, etc. to make it workable.

Are there opportunities in the Earth observation community? Certainly. Several possibilities include:

• Of the three Fabric Indices, the Environmental Fabric Index may currently need the most work. This is partly because, as a NOAA staff member, I have not engaged environmental and EO colleagues until I am satisfied that a useful HSI can be crafted. The Environmental Fabric Index of the global HSI currently looks at situations (vulnerability from disasters and from environmental contamination), sustainability (living within our means environmentally), and governance (policies and deliveries protecting people in the broad environmental arena). How should one compile and present data to represent vulnerability to floods and other environmental disasters, contamination of air, water, food [29, 30]? How to improve on the current list of inputs shown in the Table – where data can be represented for counties or global nations? Answers to such questions are best evolved by groups of people, not by fiat.
• Currently, little input from EO capabilities is in the HSI. Considering the potentials (drought risk, for example) for adapting EO data to useful Human Security inputs, will this situation change? What R & D opportunities are out there?
• Opportunities abound with respect to the global HSI, and also for national HSIs – including the one for the USA shown here.

Earthzine encourages comment/discussion. Let’s see what evolves.
To take part in the conversation about Earth observation and HSI, please comment below. You can also contact the author, David Hastings, at David.Hastings@noaa.gov.

About the Author:
David Hastings has worked on indicator development in the field with the Ghana Geological Survey, academically in Ghana and Michigan, and in computer analysis labs for the USGS EROS Data Center and the NOAA Environmental Satellite, Data and Information Service, with an additional tour with the United Nations Economic and Social Council. His current work combines these in an attempt to better observe our current situations – using existing data from publicly accessible sources.

References
[1] UNDP, 1990-2010. Human Development Report. United Nations Development Programme, New York. Online at http://hdr.undp.org/en/reports/
[2] UNDP, 2010. Human Development Report 2010. United Nations Development Programme, New York. Online at http://hdr.undp.org/en/reports/global/hdr2010/
[3] UNDP, 1994. New dimensions of human security. IN Human Development Report 1994. United Nations Development Programme, New York. pp. 22-46. Online at http://hdr.undp.org/en/media/hdr_1994_en_chap2.pdf
[4] This was an unattributed use of Franklin D. Roosevelt’s famous “Four Freedoms” speech.
[5] Pitsuwan, Surin, 2007. Regional cooperation for human Security. Keynote address: Conference on Mainstreaming Human Security: the Asian Contribution. Bangkok, 4-5 October 2007. Online at:
http://humansecurityconf.polsci.chula.ac.th/Documents/Transcriptions/Keynote Speech on Regional Cooperation for Human Security.pdf
[6] Stiglitz, Joseph E., Amartya Sen, and Jean-Paul Fitoussi, 2009. Report by the Commission on the Measurement of Economic Performance and Social Progress. Commission on the Measurement of Economic Performance and Social Progress, Paris, 292pp. Online at http://www.stiglitz-sen-fitoussi.fr/documents/rapport_anglais.pdf
[7] The report had lengthy discussion of problems, but little detail about proposed solutions.
[8] SSRC, 2010. The Measure of America 2010-2011. Social Science Research Council. Online at http://www.measureofamerica.org
[9] Cahill, M. B., 2005. Is the Human Development Index redundant? Eastern Economic Journal, vol. 31, pp. 1-6.
[10] For example, several Pacific island countries find that social disincentives against crime can succeed, virtually eliminating the need for incarceration. Could the USA (the global leader in incarcerating people) learn something from them?
[11] Hastings, David A., 2009A. Filling Gaps in the Human Development Index. United Nations Economic and Social Commission for Asia and the Pacific, Working Paper WP/09/02. Online at http://www.unescap.org/publications/detail.asp?id=1308
[12] Hastings, David A., 2011. A “Classic: Human Development Index with 232 countries. HumanSecurityIndex.org Online at http://www.humansecurityindex.org/?page_id=204
[13] Tadjbakhsh, Shahrbanou (2008). Human security. Human Development Insights, Issue 17. United Nations Development Programme, New York. Online at http://hdr.undp.org/en/nhdr/support/insights/2008-02/
[14] Emmerij, Louis, 2007. Creativity in the United Nations: A history of ideas. Development, v50, pp. 39-46. Online at http://www.unhistory.org/reviews/emmerij_creativity.pdf
[15] Hastings, David A., 2008. Describing the human condition – from human development to human security: a remote sensing and GIS approach. Proceedings, GIS-IDEAS 2008 Conference “Toward Creative and Sustainable Humanosphere” online at http://wgrass.media.osaka-cu.ac.jp/gisideas08/viewabstract.php?id=299
[16] Hastings, David A., 2009B. From Human Development to Human Security: A Prototype Human Security Index. United Nations Economic and Social Commission for Asia and the Pacific, Working Paper WP/09/03. Online at http://www.unescap.org/publications/detail.asp?id=1345
[17] Several composite indicators, such as the Environmental Vulnerability Index, the Environmental Performance Index, the Global Peace Index, and the Food Security Index (the latter developed for the HSI), incorporate 25, 50, 24, and 8 datasets respectively, so the global HSI incorporates over 150 input datasets overall when one counts the source data used in composite indicators.
[18] Hastings, David A., 2010. The Human Security Index: An update and a new release. GIS-IDEAS 2010 Conference “Toward Environmental Security and Sustainable Development” online at
http://wgrass.media.osaka-cu.ac.jp/gisideas10/viewabstract.php?id=381
[19] Hastings, David A., 2011. Documentation for HSI Version 2. Online at http://www.humansecurityindex.org/?page_id=28
[20] Perhaps (very) cautious testing of such applications could now commence.
[21] But these have been a part of corporate social responsibility for over a decade (see writings on the “triple bottom line” by John Elkington, etc.) and were arguably a foundation for works such as Thomas Paine’s “Common Sense” and the Declaration of Independence.
[22] Williams, John. 2010. Alternate unemployment charts. Shadow Stats. Online at http://www.shadowstats.com/alternate_data/unemployment-charts
[23] Wikipedia, 2011. Gini coefficient. Online at http://en.wikipedia.org/wiki/Gini_coefficient
[24] Burkey, Mark L., 2006. Information on the Gini Coefficient and income inequality. Online at http://www.ncat.edu/~burkeym/Gini.htm
[25] HDRO, 2008. Calculating the Human Development Indices. IN Human Development Report 2007/2008, p. 356. United Nations Development Programme, Human Development Report Office. Online at: http://hdr.undp.org/en/media/HDR_20072008_EN_Complete.pdf
[26] C.O.L. = cost of living
[27] Cuming, Michael J. and Barbara A. Hawkins, 1981. “TERDAT: The FNOC System for Terrain Data Extraction and Processing.” Technical report MII Project M¬254 (Second Edition). Prepared for Fleet Numerical Oceanography Center (Monterey, CA). Published by Meteorology International Incorporated.
[28] Clarke, Leo 1992. FNOC Global Elevation, Terrain, and Surface Characteristics. Digital Raster Data on a 10-minute Cartesian Orthonormal Geodetic (lat/long) 1080×2160 grid. In: NOAA-EPA Global Ecosystems Database Project, 1992. Global Ecosystems Database Version 1.0 Disc-A. US Department of Commerce, National Oceanic and Atmospheric Administration, National Geophysical Data Center, Boulder, Colorado. Online at http://www.ngdc.noaa.gov/ecosys/cdroms/ged_iia/datasets/a13/fnoc.htm
[29] Kessler, David, 2009. The End of Overeating: Taking Control of the Insatiable American Appetite. Rodale Books, Emmaus, PA. 336pp.
[30] Kenner, Robert and Eric Schlosser, 2009. Food, Inc. (film) Online at http://www.foodincmovie.com