[wordup] How Coder Cornered Milosevic

[Adam Shand]

From: http://www.wired.com/news/politics/0,1283,51106,00.html

How Coder Cornered Milosevic
By Farhad Manjoo

2:00 a.m. March 19, 2002 PST

Patrick Ball, a 36-year-old "hacktivist," was armed with mathematics,
logic and a set of stark graphs in his appearance as a prosecution
witness last week at the trial of Slobodan Milosevic, the former
Yugoslav president accused of atrocious war crimes.

That may sound like a meager arsenal, but the conclusions of Ball's
analysis on killings and refugee movement in Kosovo were difficult for
Milosevic to refute. 

With the methodical manner of a forensic examiner who uses statistics
and open-source software instead of scalpels and x-rays, Ball undermined
Milosevic's claim that NATO and rebel forces -- and not Milosevic's own
Serb army -- caused ethnic Albanian deaths and displacement in Kosovo.

At the trial in The Hague, Ball -- a sociologist, statistician and coder
-- told the court that when he mapped the locations of the deaths and
the patterns of the refugee flows out of Kosovo in 1999, he saw an
unmistakable method in how they occurred over time and space.

High death rates occurred at the same time as high refugee outflows; low
death rates occurred at the same time as low refugee outflows. In every
region of Kosovo, when deaths plummeted, refugee flight plummeted.

The peaks and valleys on the two graphs -- one marking death and the
other departure -- dance in unison.

Ball said this shows that the refugee crisis and the killings were
caused by the same thing. They were, in other words, systematic and not
the result of random occurrences.

Furthermore, the changes in the graphs do not correspond, Ball said, to
actions by NATO or the Kosovo Liberation Army, the ethnic Albanian rebel
force that had been fighting the Serb army. When KLA or NATO activity
went up or down, the graphs didn't necessarily respond; but when Serb
activity changed, the graphs changed accordingly.

"For example," Ball explained in an interview on Monday, "on the night
of April 6 (1999), the Yugoslav government declared a unilateral cease
fire, and we observed that immediately those numbers (of killings and
refugee flows) plummet, they go down almost to zero."

During that same period, Ball said, NATO air strikes doubled in
intensity and KLA activity tripled.

Ball concluded from this that Milosevic's Serb forces were to blame for
the refugee crisis and killings in Kosovo. (His report, co-authored with
four other researchers, is available in PDF format.)

Ball, who traveled to Kosovo many times during the conflict, obtained
his data from several sources: interviews conducted by human rights
groups, Albanian border guards, exhumation records and Yugoslav
government sources. Using statistical techniques, his team determined
that 10,356 Kosovar Albanians died during the war.

In court last Thursday, Milosevic, who is acting as his own attorney,
called into question the rigor of Ball's analysis, saying that it did
not take into account the inherent chaos of war. Ball "simplified in the
extreme this complex phenomenon which is war," and Ball was "serving the
American policies of enslavement," Milosevic said, according to wire
service reports.

A video archive of the trial shows the defendant and Ball engaged in a
spirited back-and-forth over the report. At one point, Milosevic accuses
Ball of "inventing" -- rather than "estimating" -- his numbers, seeming
to call into question Ball's statistical techniques.

Ball replies that his numbers are derived from "well-known scientific
methods," which he says his report explains in "significant detail."

That may be the case, Milosevic responds, but since they're not "actual"
numbers, Ball's report can only be used "for educational purposes."

On Monday, Ball recalled his testimony without much stress. "I'm not a
lawyer," he said, "so I don't know how it went from a legal point of
view. I do know that I was able to answer all of Milosevic's questions;
and I don't think he raised any issues that called into question the
accuracy of the work."

This was Ball's first courtroom appearance to support his work, but for
the past 10 years he has applied his scientific and technical knowledge
to human rights work all over the world, including working in the truth
commissions in Haiti and South Africa. He is now the deputy director of
the science and human rights program at the American Association for the
Advancement of Science.

He has filled a crucial, although probably little-recognized, position
in his various posts around the world: information management.

"Before you can say human rights need to be improved or are problematic
in a certain region," he explained, "you have to show what the problem
is. I worked to design (the human rights groups') information management
system, and I would hire and train the teams to do that."

Ball added it was important to note that human rights work "is best
served by free software, so that any group who wanted to reproduce my
work could do so without a huge investment. I did the data processing
using Python, we used MySQL, and all the data coding teams used Linux
and Apache.

"For me, software might not be a huge expense, and I could have used any
software I wanted to -- but for a group in the developing world software
expenses might be insurmountable."