b. Chula Vista
The OIG interviewed 54 witnesses who testified that some sensors at the Chula Vista Station had been inhibited: 27 stated generally that sensors had been inhibited; 18 testified that sensors in the "valley" area had been inhibited at some point; and 13 said they recalled that sensors in the "docks" area had been inhibited. These witnesses included line agents, supervisory agents, station managers, Sector Communications personnel, and Sector Electronics personnel. Like Imperial Beach, it was clear that sensors had been inhibited but the question was when and why.
We received varying reports regarding when these sensors had been inhibited. Three witnesses from Sector Communications testified that some unspecified sensors had been inhibited for approximately three months during 1995, but they did not specify which three months. One witness from Sector Electronics testified that the valley sensors had been inhibited for three months in 1996. Various witnesses from Sector Communications and Electronics testified that some Chula Vista sensors had been inhibited for one month in 1995, one month in 1996, and for two weeks in the spring of 1995. Accounts from agents at Chula Vista were equally varied. One said he had requested that supervisors inhibit the valley sensors shortly after Gatekeeper began because there was no manpower to work them; thereafter, he stopped hearing these sensors being called out. Two witnesses testified that the valley sensors had been inhibited in the spring of 1995. One said the valley sensors had been inhibited for part of a shift one night in July 1995 after they had been registering a lot of traffic. Two said the valley sensors had been inhibited in the fall of 1995. One agent said the docks sensors had been inhibited in the fall of 1995. Another agent testified that the valley sensors had been inhibited in May 1996. One witness said the valley sensors had been inhibited for six months in the fall of 1995. The person whom everyone identified as responsible for ordering the sensors inhibited, No. 1101, admitted having inhibited sensors on his shift throughout 1995, 1996, and 1997, whenever he felt it was appropriate. He declared, however, that he had never inhibited the docks sensors because his shift worked that area.
As with the Imperial Beach sensors, we found that we could not rely on recollections as to timing. The three witnesses who testified that the valley sensors had been inhibited in 1996 were probably confused. Movement records indicated that at least one of these sensors was removed in July 1995 and the rest were removed by mid-October 1995, leaving no sensors in the valley area from October 1995 until September 1996 when new sensors were planted in that area. It is unlikely that sensors would be planted in an area only to be inhibited immediately thereafter. Given that the testimony of these witnesses is generally consistent with that of others, save for the year, we believe they were simply mistaken on this point.
In addition to the testimony of agents, we found documentary evidence that sensors were in fact inhibited at Chula Vista. A memorandum to Sector Communications from No. 751, a supervisor, requested that 11 sensors be inhibited from 3 p.m. to 7 a.m. from April 21, 1995, through May 10, 1995, at which time the policy was to be reviewed. Five of these were the valley sensors, and five were relatively close to the border. We could not determine the location of the eleventh. Interestingly, none of the witnesses from Sector Communications said they recalled receiving this memorandum, and none mentioned No. 751 as having requested that any sensors be inhibited. Each of them, however, recalled No. 1101 calling and requesting that various sensors be inhibited.
No. 751 testified that his memorandum was prompted by a line agent concerned that these sensors were so close to agents' positions that agents were setting them off. For the five listed sensors near the border this explanation makes sense. Indeed, No. 1101 offered the same rationale for inhibiting them. This explanation could not have related to the five valley sensors, however, because agents were not deployed anywhere nearby. According to No. 1101, these sensors needed to be inhibited because livestock was frequently setting them off and it made no sense to have agents respond to hits not caused by alien traffic.
No. 751 said he forwarded his memorandum to Sector Communications but did not recall what occurred after that. We found no documentation of whether the continuous inhibition of these 11 sensors on the swing and midnight shifts continued past May 11, 1995. No. 1101 testified, however, that he inhibited them on various shifts off and on, month after month, depending on the weather, the level of manpower, and other constraints. Although he could not recall specific dates, he said he had inhibited sensors before Gatekeeper and continued to inhibit some sensors currently as he deemed appropriate.
It was harder to determine when the dock sensors were inhibited. Only five of the 13 witnesses who made this claim purported to have personal knowledge. One said late 1996, but because he claimed the valley sensors - which in fact had been removed the previous year - had been inhibited at the same time, his testimony was not reliable as to timing. One witness said these sensors were inhibited sometime in the beginning of 1995. Another claimed sometime in September 1995. The remaining two did not provide a time frame, but said they had been inhibited because there was insufficient manpower to work these sensors. Two of the persons who had merely heard rumors said they had heard that line agents had asked Dispatch not to call out the dock sensors because there was no one to work them. Two witnesses mentioned a memorandum sent to Dispatch listing a number of sensors to be inhibited. Based on the description of the note, however, we believe they must be referring to the memorandum already described, which did not deal with the dock sensors.
The sensor report for February 11 through March 12, 1996, reporting the events and tickets for each of the Chula Vista sensors, indicates that none of the sensors listed in the April 1995 memorandum was inhibited during the time covered by the report. But it also indicates that five different sensors were inhibited at some point during the month. Four of these sensors could not have been inhibited for long because the difference between the number of events and the number of tickets is small.189 The 202 sensor, however, recorded 1,082 events and 520 tickets, suggesting that it was inhibited regularly, probably on a timer. Due to the limited information in this report, we could not determine the dates or times the events without tickets occurred. In recent times, however, this sensor has been put on a timer from 8 a.m. to 4 p.m. each weekday when its site has significant levels of legitimate traffic.
In our review of the year's worth of sensor data beginning in mid-1996, we selected five random months (which included typically busy and slow months) and found that nine sensors at Chula Vista had been inhibited during that time. The 201 sensor was on a timer during the weekdays as described above. Five of the other sensors previously inhibited registered events without tickets on only a single day. The remaining sensors registered events without tickets on 2, 5, and 16 days, respectively. Although the data does not record exactly how long each of these sensors was inhibited, it does indicate that the time between the first event without a ticket on a given day and the last event without a ticket on that same day was on average less than two hours and in no case more than six hours. This suggests that either these sensors were not inhibited for an entire shift or there was very little traffic. In either case, these periods are not significant.190
In sum, we believe, based on the testimony of No. 1101 and others, that the valley sensors were inhibited for extended periods throughout 1995 until they were removed in July and October 1995. The five sensors located close to the border that were listed in No. 751's memorandum also were inhibited frequently and for lengthy periods of time during 1995 and perhaps in early 1996. The dock sensors were inhibited on various occasions, but we could not identify by whom or how often. Witnesses indicated, however, that these sensors were inhibited when agents were not working that area. Other sensors appear to have been inhibited on a sporadic basis. Significantly, by early 1996, the one sensor report we obtained indicated that sensors were not being inhibited regularly or for significant periods of time. This is particularly important because alien traffic is generally high in February and March.
The next question is why these sensors were inhibited. Two agents alleged that the sensors were inhibited to demonstrate that traffic had decreased. However, they offered nothing to support this claim. Given that, as we have already noted, inhibited sensors still register hits, and sensor hits were not used to measure traffic levels, we find the claim both unsubstantiated and highly implausible. Four agents believed that the sensors were inhibited to avoid detecting aliens and to lower apprehensions. They, too, offered nothing but their suspicions, based on their belief that management wanted lower apprehensions at all costs. In previous sections of this report, we found no evidence to support the assumption that management wanted to avoid apprehending illegal aliens, and we found no such evidence in connection with sensors.
In the end, we found the explanations offered by No. 1101 and by No. 751 in his memorandum to be credible. The memorandum said the sensors were being inhibited due to the deployment of agents closer to the border under Gatekeeper. No. 1101 also explained his actions by noting that agents were deployed closer to the border, leaving no one to work sensors so far from the border, and that sensors were frequently set off by livestock. Other witnesses, including several line agents, concurred that these sensors were just too far from the border to be worked when agents were assigned to positions along the border. And this was the testimony of Sector Communications personnel as well. Given that the valley sensors were in fact two to three miles from the border, where manpower was concentrated, this explanation rings true.
The claim that five sensors close to the border were periodically inhibited because agents kept setting them off also is credible. These sensors were located in prime locations where agents frequently would be working. And the need to rely on sensors in these areas was reduced because the sites could be surveilled by scope operators who could direct agents to alien traffic.
It is telling that No. 751's memorandum explicitly ordered that these sensors remain uninhibited when it rained, a period when sensors exhibit their highest level of activity. Under rain conditions, agents are unable to work close to the border fence. If No. 751's goal was to hide alien traffic and to prevent agents from being aware of traffic entering the country, he would have been particularly interested in inhibiting these sensors when it rained. His decision to do exactly the opposite substantially undercuts any inference of ill motive behind the inhibiting of sensors at Chula Vista and corroborates the stated resource-related explanations.
c. Summary of findings regarding inhibited sensors
Sensors at Imperial Beach and Chula Vista were inhibited, primarily in the first year of Gatekeeper. At Imperial Beach, several sensors in the third tier away from where the agents were usually positioned were inhibited on various occasions until late 1995. Sensors at Chula Vista located far from the border were inhibited for extensive periods until they were removed in late 1995. Other sensors close to the border that were being set off by agents also were inhibited on numerous occasions in 1995. There was little evidence of continued inhibiting of these sensors in 1996.
The decisions to inhibit these sensors were made at the Field Operation Supervisor (FOS) level or below, shift by shift. There was no evidence of Sector input. This supports the claim that the decisions arose out of particular operational needs and were not a management attempt to alter the appearance of Gatekeeper results. And, in each case, we found the claim that sensors were inhibited to hide the existence of alien traffic to be unfounded, and particularly implausible, given that the ICAD system continues to register hits when sensors are inhibited. Furthermore, because sensor hits were neither used to measure traffic levels nor were publicly reported, any effort to manipulate sensor hits would have had no impact on the reported success of Gatekeeper. We also found no evidence that sensors were inhibited to keep apprehension levels lower.
189 The 202 sensor had 374 events and 357 tickets; the 203 sensor had 406 events and 396 tickets; the 220 sensor had 156 events and 152 tickets; and the 429 sensor had 79 events and 56 tickets.
190 Although only 9 sensors registered events without tickets (indicating inhibition), 15 sensors registered events for which tickets were originally created but subsequently deleted. Unfortunately, the data does not record when, why, or by whom the tickets were deleted. One of these deletions, for example, involves the 201 sensor at 5:35 p.m., a time when legitimate traffic may have been present but the sensor was no longer on a timer. Another possible explanation for such deletions is that Dispatch already knew that an agent was going to be in the vicinity, making it unnecessary to call out an event. Because of the extremely limited number of these deletions, we have no reason to believe there was any inappropriate motive behind them. This lack of an audit trail in the system, however, could permit abuse of the system to go undetected and is a weakness that should be remedied.