Computers, Society, and Nature

The NYTimes today has an article on the future of war fought with artificial intelligence, “a 21st-century fighting force of automated tanks, helicopters and planes, remote missile launchers and even troops of robot soldiers – all coordinated by a self-configuring network of satellites, sensors and supercomputers.”

This is little different from the myriad computer models and satellite photos with which we make life-and-death decisions about the environment. Protect this habitat here. Put that dam in there. The NYTimes article reminds us of the conundrum inherent in leaving everything to the computer:

The whole point of automation is to rise above human fallibility – knee-jerk decisions, misunderstood orders, cowardly retreats. Machines are faster, more focused, impermeable to propaganda and, at least for now, they don’t talk back.

As the thinking machinery continues to evolve, the strategists will keep asking themselves the same question: Is there still a good reason to trust ourselves or should we defer to a computer’s calculations?

In other words, is it better to model the environment or better to walk in the woods? Base judgements on our spiritual connections to nature or let the computer determine the optimal solution? If nature is socially constructed, that is an individual’s mental vision built up of past experiences, biases and culture, then whom/what should we believe? Would a computer be less biased?

In politics all it requires is a little tweak of the computer model, perhaps a change in the units of analysis, and you get the result that the politicians want.

ECOLOGL@LISTSERV.UMD.EDU
From: Ecological Society of America: grants, jobs, news
Subject: Survey: political intervention in science pervasive at USFWS

Hello everyone,
The Union of Concerned Scientists (UCS) held a press conference to announce the disturbing results of a survey of U.S. Fish and Wildlife Service field scientists: political intervention to alter scientific findings has become pervasive within the agency. At field offices around the country, USFWS scientists tell of being asked to change scientific information, remove scientific facts or come to conclusions that are not supported by the science. As a result, the scientists say, endangered and threatened wildlife are not being protected as intended by the Endangered Species Act.

Despite agency directives to scientists not to reply to the survey even on their own time nearly 30% of the scientists responded. You can find a summary of the survey, its methodology, and a summary of results broken down by region here or by clicking here.

RESULTS SUMMARY
The survey paints a vivid picture of the systemic abuse of science and the need for change. Results show that:

Large numbers of agency scientists reported political interference in scientific determinations. Nearly half of all respondents whose work is related to endangered species (44%) report that they have been directed for nonscientific reasons to refrain from making findings that protect species. One in five have been instructed to compromise their scientific integrity, reporting that they have been “directed to inappropriately exclude or alter technical information from a USFWS scientific document.” In the Southwest region, that number was even higher -closer to one in three.

Agency scientists reported being afraid to speak frankly about issues and felt constrained in their role as scientists. 42% said they could not publicly express “concerns about the biological needs of species and habitats without fear of retaliation,” while 30% were afraid to do so even within the agency. A third felt they are not allowed to do their jobs as scientists.

There has been a significant strain on staff morale. Half of all scientists reported that morale is poor to extremely poor; only 12% believed morale to be good or excellent. And 64% did not feel the agency is moving in the right direction.

Political intrusion has undermined the USFWS’s ability to fulfill its mission of protecting wildlife from extinction. Three out of four staff scientists felt that the USFWS is not “acting effectively to maintain or enhance species and their habitats.”

In one of numerous essays submitted on the topic of improving scientific integrity at USFWS, one biologist wrote: “We are not allowed to be honest and forthright…I have 20 years of federal service in this and this is the worst it has ever been.” Another scientist reported that Department of Interior officials “have forced upper-level managers to say things that are incorrect.” A manager wrote: “There is a culture of fear of retaliation in mid-management.”

Encouragingly, it is clear from the survey that USFWS scientists are committed to and proud of their work and believe in the potential of the agency to conserve, protect and enhance fish, wildlife and plants and their habitats. However, political intervention is having a chilling effect on the ability of USFWS scientists to carry out the agency’s mission.

UCS has joined with Public Employees for Environmental Responsibility (PEER) to design and conduct surveys of several government agencies to document the abuse of science and determine the pervasiveness off the problem. The surveys will assist the scientific community in documenting that the abuse of science is an ongoing, serious concern. We are looking into ways that the results of the USFWS survey can be used to further a more thorough investigation of this problem.

It has taken decades to build worldclass scientific staff at the USFWS and other government science agencies. The future ability of the agency to fulfill its mission will be severely hampered if this political interference is allowed to continue. To restore scientific integrity at the USFWS, at least two reforms are needed: there must be protections for scientists who are asked to take actions that violate their scientific integrity and the Bush administration must recognize at its highest levels that manipulating or suppressing science for political reasons is unethical.

signed: Michael Halpern, Outreach Coordinator Restoring Scientific Integrity in Federal Policy Making Campaign Union of Concerned Scientists, Dean A. Hendrickson, Ph.D. Curator of Ichthyology, University of Texas, Texas Memorial Museum, Texas Natural History Collections, others.

See also last week’s LA Times, U.S. Scientists Say They Are Told to Alter Findings

A recent report in Nature covers the first findings from the distributed computing project, climateprediction.net. The report finds that global temperatures could rise by up to 11C. This is two times the amount of other studies.

The project is remarkable because it utilized distributed computing. Distributed computing spreads out the computer processing and analysis among multiple CPUs (the chips in your computer that do the number crunching). Often these chips are located in separate computers. What’s even cooler is that the distributed computer can be your home computer. Instead of using an incredibly fast computer with multiple processors or a supercomputer, you download the software to your own PC, which runs the program while your computer is idle. The application generally shows up as a nice screensaver.

Distributed computing was first used by the SETI@Home (SETI stands for the Search for Extraterrestrial Life) project, in which separate PCs sift through packets of radio-telescope data, looking for evidence of extraterrestrial communication. It has since moved on to analyzing data for pulsars and studying protein-related diseases.

An excellent synopsis of the climate change project can be found at the BBC. They report that “More than 95,000 people have registered [to download the software], from more than 150 countries; their PCs have between them run more than 60,000 simulations of future climate. Each PC runs a slightly different computer simulation examining what happens to the global climate if levels of carbon dioxide in the atmosphere double from pre-industrial levels – which may happen by the middle of the century. ”

Of course, the accuracy of any model is just as dependent on its underlying assumptions as its computational power. So it will be a while before we can assess the accuracy of the dramatic temperature rise predicted by the project. However, distributed computing holds great promise for climate change, especially as the models increase in complexity and are reaching the upper limit of what can be feasibly and affordably accomplished in standard computing frameworks.

(An irreverent note. To date, distributed computing has been devoted to very serious projects. However, I’m waiting for the first silly project. How about a Shakespeare project that simulates millions of monkeys? I’d download the software.)