'Confirmation bias' led coalition to mistakenly bomb Syrian troops, investigation concludes

A series of errors sparked by the misidentification of a vehicle as being driven by Islamic State fighters led coalition aircraft to mistakenly bomb Syrian troops in September, an Air Force investigation released Tuesday concluded.


Air Forces Central Command found that the Sept. 17 airstrikes near Dayr az Zawr in Syria did not intentionally target what it referred to as “forces aligned with the government of Syria,” though officials stopped short of calling them Syrian troops because they were not wearing uniforms or displaying flags or insignia.


Instead, Brig. Gen. Richard Coe, the Air Combat Command inspector general who led the investigation into the airstrike, said that “confirmation bias” stemming from that first misidentification colored the subsequent decisions made by others involved in the strike. And a series of missteps involved in communicating with the Russians via a hotline meant the strikes continued longer than they needed to.


But Coe said the mistakes, while “unintentional” and “regrettable,” did not stem from negligence or malice, and that the misidentification of Syrian forces as ISIS fighters was made “in good faith” based on the information that was available at the time.


“There were good people trying to do the right thing,” Coe said. “These people get it right far more often than not, but this time, they came up short. Warfare is an ugly business, it’s a serious business, and we can’t afford to settle for anything short of excellence each and every day and each and every time.”


Forces from Denmark, Australia, and the United Kingdom also took part in the airstrikes, which were carried out by F-16s, A-10s, F/A-18s and remotely-piloted aircraft. Coe said that his investigation confirmed at least 15 Syrians were killed, though the investigation believes the actual number is much higher. Coe said the Syrian Observatory for Human Rights reported 83 Syrians were killed.


Coe said the coalition aircraft released 34 precision-guided weapons and fired 380 30mm rounds.
AFCENT said in a follow-up email that there were no disciplinary actions taken as a result of the strikes, since they were made in good faith and in accordance with the law of armed conflict, and were based on the information available at the time.


Coe said that intelligence, surveillance and reconnaissance aircraft followed a vehicle for hours before the strike, and its occupants behaved in ways that led the strike’s planners to decide it was an ISIS vehicle early on. That was the initial decision that colored subsequent choices.


That vehicle eventually met up with the Syrian troops that were eventually struck, and coalition planners observed the vehicle’s occupants were friendly with the other forces. The ISR aircraft also observed those troops – armed with military weapons and vehicles – were not wearing recognizable uniforms, and there were no unit flags, insignia or other markings visible. That was another factor that led coalition aircraft to conclude they were likely ISIS fighters.


“In many ways, they looked and acted like the forces the coalition has been targeting for the past few years,” Coe said.

But at least one warning that the fighters weren’t ISIS was overlooked and not pushed to a larger group or the final decision-makers. Coe said that an intelligence analyst, after watching video from the scene, typed in a chat room, “What we’re looking at can’t possibly be ISIL,” using an alternative acronym for the militant group.


That warning fell on deaf ears, because officials were already convinced the troops had been confirmed as ISIS fighters.


Coe said that the analyst was watching a full-motion video of a tank, which he had never seen ISIS use. However, other people in the Combined Air Operations Center had recently seen ISIS commandeer a tank from the Syrian army in the same general area, he said, which meant it was “not unreasonable at the time” to decide the tank was being used by ISIS.


Over a several-hour period, the strike fighters used a “hybrid” targeting process, which was described as a mixture of the “deliberate targeting” process that can take days or weeks to validate targets, and the fast-moving, accelerated process used for quickly emerging “dynamic targets,” Coe said.


When dynamic targeting is used to hit “targets of opportunity,” Coe said, there is a greater emphasis on making sure the target has been validated and alternative interpretations have been fully considered in real time.


But blending the two processes in the Dayr az Zawr strike allowed the planners “to not fully and critically second-guess their conclusions as the situation unfolded,” Coe said.


One of the main recommendations as a result of the investigation is to review that hybrid process, which Lt. Gen. Jeff Harrigian, commander of U.S. Air Forces Central Command, has already ordered, Coe said.


The coalition’s unfamiliarity with the Dayr az Zawr region also led to the mistakes. It knew that pockets of ISIS fighters and Syrian regime troops were operating in the area, but had not operated much around there and lacked knowledge of the specific terrain.


For the first time since a safety hotline was set up with the Russians to help manage the complicated war zone in Syria, the Combined Air Operations Center, or CAOC, called its Russian counterparts in advance to warn them of the strike, Coe said. That hotline eventually led to the coalition halting its airstrikes – but, once again, the process was hampered by multiple mistakes.


The coalition’s first call to the Russians accidentally passed on incorrect information about the location of the airstrikes, leading the Russians to believe the strikes were actually three kilometers away from their actual location. That misidentification may have been why the Russians didn’t express any concerns about the strikes in advance, Coe said, although it’s not certain.


But after the airstrikes had begun and the Russians realized Syrian troops were being struck, another problem stood in the way. The Russians called the CAOC back, but could not reach their normal point of contact. While the normal point of contact, who was on the other side of the base, was being called to the phone, the Russians declined to pass the warning on to the colonel who did answer the phone.


As a result, 27 minutes passed with the Russians on hold before the cease-fire message was received, the investigation report said, during which time 15 of the 37 strikes took place. He said the airstrikes stopped immediately as soon as the message was received and understood.


“This was obviously a missed opportunity, to be able to limit the damage of the mistake,” Coe said. “We just think that critical safety information could be passed immediately rather than waiting for the traditional” points of contact.


In addition to the review of the hybrid targeting system, the investigation is also recommending analysts improve information sharing to safeguard against these kinds of human errors and the CAOC establish a more effective lessons-learned process.


And the U.S. and Russia should make a greater use of their safety de-confliction hotline to make sure critical information is more quickly communicated to the personnel who are available.