Study claims scientific proof for "cycle of violence"

From The Boston Globe, an op-ed by Nancy Kanwisher and Anat Biletzki:

As the Israeli-Palestinian peace process once again crashes on the hard rocks of Middle East reality, it is worth stepping back to reconsider the conventional wisdom on this apparently intractable situation. In a paper we recently published in the “Proceedings of the National Academy of Sciences,’’ we found that, in contrast to the perception on each side of the conflict that the other side is the aggressor while it only retaliates, in fact, both sides act in response to the other’s aggression.

Although anecdote and speculation are popular in discussions of the Middle East conflict, we used data and quantitative analysis to determine whether these perceptions are true. One data set was the timeline of Qassam rocket firings, compiled by the Israeli Defense Forces. Another was the day-by-day timeline of killings of Israelis by Palestinians and of killings of Palestinians by Israelis, compiled by the Israeli Human Rights organization B’Tselem. We tested whether the violent behavior of each side occurs in response to violence committed by the other side — or whether it is simply arbitrary.

We found that the violence on each side is not arbitrary. Instead, a few days after Palestinians kill Israelis, Israel retaliates by killing Palestinians, and in the few days after Israel kills Palestinians, the number of rockets fired into Israel increases. Thus, both Palestinians and Israelis are more likely to attack after they themselves have been attacked.

These findings refute the common view that because the conflict results from the immutably hostile character of the foe, there is nothing either side can do to stop it. Our data suggest that the conflict is not the inevitable result of the fundamentally violent character of either Israelis or Palestinians. Instead, the violence of each side is at least in part contingent on the behavior of the other side. So there is, in fact, something each side can do to reduce the violence directed against it.Our result may seem obvious: if both sides retaliate, then the Middle East is yet another part of the world where retaliatory “tit for tat’’ dynamics perpetuate conflict.

The authors of the study are a professor of neuroscience at MIT. Anat Biletzki is a professor of philosophy at Tel Aviv University.

Looking further at their methodology they use something called Vector Autoregression, the math of which I cannot hope to understand.

But read carefully what they say again in the parts I put in bold. They are saying that Israel retaliates to killings with killings - incursions, targeted attacks, whatever - and the Palestinian militants respond to Israeli killings with rockets.

There is a very important unstated fact that supports both these findings and the opposite conclusion:

When Palestinian Arabs mount a terror attack meant to have a high probability of killing Israelis - meaning, remote controlled bombs, or suicide bombings, or ambushes - it takes time for them to plan it. Usually one can expect a number of weeks between conceiving of an attack and the actual operation.

Clearly, fatal terror attacks are not responses to specific events where the IDF killed people, because it simply takes too much time for the PalArabs to mount such an operation. Therefore, their retaliations have been chiefly rocket attacks, something that takes little time to plan and implement.

Which means that, contrary to what the authors are implying, Palestinian Arab terrorists by definition instigate every non-rocket terror attack against Israeli civilians, and Israel indeed does retaliate. Yes, the terrorists fire rockets back for further retaliation, but the cycle dies down because most of them do not kill or injure anyone. If they are "lucky" and a rocket hits someone then the cycle can go another round or two.

Yet the clear piece of information that the authors miss is that the major terror attacks are not retaliatory, unless you expand the definition of "retaliation" to include the entire existence of Israel or of "occupation."  Bombings are certainly not planned and implemented in the "few days" that the authors used to feed their mathematical models. Their methodology is fatally flawed because they do not take the time it takes to plan terror attacks into account.

Their airy conclusions, that if only Israel would stop retaliating then the terrorists would stop as well, are unfounded.

And a significant number of Hamas rocket attacks, at any rate, were not retaliatory for specific events. This can be seen from their own press releases. I don't know the percentages, but while many rocket attacks are stated as to be in response to specific events, others are more general.

For example, here is how they announce rocket attacks that are in response to specific events, from June 14th 2006:

Occupation forces continue to perpetrate war crimes against Palestinians. Last Friday, they killed 7 members of the same family and injured dozens of civilians on the Beit Lahya beach in the north of the Gaza Strip. And on Tuesday, nine civilians, including 2 children and 4 paramedics, were intentionally killed by occupation planes.

In response to these crimes, Ezzedeen Al-Qassam Brigades continues the response to these crimes. At 04:00 today, Al-Qassam Brigades fired one Qassam rocket at the Sofa checkpoint, east of Rafah.

But for a different set of rockets a couple of weeks later, the press releases were far more general. From July 4, 2006:

Izzedeen Al-Qassam Brigades fired one Qassam Rocket at the occupied city of Asqalan north of the border of the Gaza Strip. The bombardment took place at 19:00 on Tuesday, July 4, 2006. The operation is a new development of the ongoing "Faithfulness of the Free" resistance campaign against the occupation assault on the Gaza Strip, in which occupation forces continue to attack civilian targets.

Even the rocket attacks cannot be said to be generally retaliatory for specific events, because Hamas is nice enough to tell us the reasons for the attacks!

For these two reasons, it appears that this study is fatally flawed in its approach, its methodology and in its conclusions.