TRANSCRIPT: MAY 10 BACKGROUND BRIEFING ON CHINA EMBASSY BOMBING
(Outdated information on embassy location led to bombing)

Washington -- NATO's bombing of the Chinese embassy in Belgrade was a mistake partially caused by a failure to update the location of the embassy in intelligence databases, senior intelligence officials said in a May 10 background briefing at the Department of Defense.

"The map that was the principal map that most people have in the process was a 1997 map," one intelligence official said. "The 1997 map had a building unidentified at the (target) location and had the Chinese embassy depicted in the same location in old Belgrade (as it was listed in 1992)."

Other information sources also failed to accurately locate the Chinese embassy, according to the officials.

"The error was that the target development process got the wrong function at the wrong building ... and that the checks and balances that we go through in trying to determine consequences of hitting the target that has now been incorrectly identified did not come up with the Chinese embassy being located nearby," one of the officials said.

According to the officials, the bombing of the Chinese embassy marks the first time collateral damage has occurred due to faulty target selection.

"Nobody was going out looking for the Chinese embassy as they developed this target," an official said.

"We have struck about 270 targets, I believe. We've struck many of them multiple times. There have been over 4,000 strike sorties. This is not an excuse, but this is the first time I have been aware of where there were inadvertent unintended casualties because of a target mistake at the wrong facility," one of the officials said.

In general, potential targets are carefully analyzed to avoid collateral damage, the officials stressed.

"We have on numerous occasions retargeted, or removed targets, conducted different aim points, or ways to strike the target because we found collateral damage issues, and they were successfully taken care of," a senior intelligence official said.

Following is a transcript of the briefing:

(begin transcript)

Background Briefing

Monday, May 10, 1999 - 4:42 p.m.
Subject: Monday, May 10, 1999 - 4:42 p.m.
Presenter: Senior Intelligence Officials

Chinese Embassy Strike

BACON: Let me start with a brief announcement that there will not be a regular briefing tomorrow because the Secretary and General Shelton are going to be on the Hill testifying before the Senate Appropriations Committee.

Second, the next part of this briefing is a continuation of Secretary Cohen's briefing, but it's on background. We have a senior official from the Central Intelligence Agency and a senior official from the Defense Intelligence Agency. You can refer to them both as senior intelligence officials. And they will come up here and answer your questions to the best of their ability that may be left over from the Secretary's briefing.

Gentlemen.

Senior Official: Good afternoon. We heard the Secretary's briefing about this regrettable incident or accident, and we are here to answer your questions to the best of our ability.

QUESTION: When -- we've been told about the map, produced in 1992, reviewed in 1997, reviewed again in 1998. When in sequence of time did the embassy actually move?

ANSWER: To the best of our knowledge, the embassy moved in 1996. The map that was the principal map that most people have in the process was a 1997 map. I think the way the '92 map gets involved in this is the '92 map had the Chinese embassy in old Belgrade. And it had no building at the location of the new Chinese embassy. The 1997 map had a building unidentified at the location and had the Chinese embassy depicted in the same location in old Belgrade. But fundamentally, a map is out of date the day or the week after it's printed or produced. And what we rely on is databases, intelligence databases, which are all-source and automated, and regrettably in this case, the database not reflecting a move of the Chinese embassy did not help us uncover the original error in targeting the wrong building.

Q: (Sir), we were told that those databases, for instance, were not updated by people both from our State Department, diplomats and people from Langley, be they diplomats or something else, who were actually in that complex in that building during the past three years. Is that true, and secondly, if so, why weren't they updated?

A: Well, I am sure there were people from our embassy in that building in the last three years. But what I am not sure of (is whether) any report got sent in (so) that a person who was working on the database was able to enter it. I also would stress this has been looked at for two and a half days now. It's a very large group of dedicated people who are working on these targeting issues, very long hours doing the best they can, and we need to go back and look all the raw material as well as all the databases. I mean databases sometimes are out of date because reports don't get sent in on new information. Sometimes they are out of date because they don't get entered as quickly as they should. The bottom line is: these that were used to do verifications of the targeting were not up to date, and we need to find out why.

Q: (Sir), there are two separate issues. One is how the wrong city block became to be identified on the map. And then, that's separate from the fact that the wrong city block you got happened to be the Chinese embassy. Could you go back to the initial identification of the wrong city block? Could you tell us rather than asking these scatter-shot questions? Could you walk us through the process by which you derive targeting information from looking at a mixture of addresses and city block photographs taken?

A: Let me say something about the generic targeting process, and then turn it over to my colleague to talk specifically about this target development in selection. Targets are nominated to a war-fighting commander from two directions. One is his components, who may nominate targets up to him -- the air, ground or Navy component. And the other is National Agency or National Intelligence Committee nominations down to the JTF commander or the CINC. This was a nomination originally from an agency in Washington. And so, when the nominations are made, then they go through a process of validation and vetting and collateral damage analysis and things like that. And so, that's the generic process, and I will turn it over to my colleague to talk specifically about the initiation of this target.

Q: Which agency (sir)? We're on background; which agency nominated the target?

A: There is an effort to, first of all, to identify the target by its function and to evaluate whether or not it is a legitimate target for this purpose, and then secondly, to locate it geographically. In this instance, there was information as to what the appropriate street address was for the facility in question, and then the next step was to locate that on the base map, the 1997 base map that was just discussed. There are no street numbers on those maps; there are little street numbers on other available maps. There was a process that we went through to use what information we had on other facilities in this area as to their appropriate street addresses, to focus in on what was erroneously identified as the target facility. After that, there is a multi-stage check, both within the intelligence committee and DoD, to make sure that there is no -- that this is the correct location. It's a multi-step process. None of those fail-safes worked.

Q: Now that you are using this same map to continue -- the Secretary said there would be no bombing clause in Belgrade. You are using this same map. How are you assuring yourself that there aren't other embassies that have moved?

A: I think that the map can be overstated. Maps are important to us. It's the data; it's the information that you need to try to run to ground to see if your targeting information is valid. Nobody is going out looking for the Chinese embassy. After the Chinese embassy was bombed on Friday night, my immediate assumption, when we looked at our targets that night, is that the nearest target to the Chinese embassy was the Minister of Defense headquarters in old Belgrade. There were several embassies closer to the Minister of Defense than where the Chinese was listed on the map and the databases. And so, we need to update the maps, and then bounce all of that information against every single target that we select. We have struck about 270 targets, I believe. We've struck many of them multiple times. There have been over 4,000 strike sorties. This is not an excuse, but this is the first time I have been aware of where there were inadvertent unintended casualties because of a target mistake at the wrong facility.

Q: I'd like the other gentleman. You just said that it is in fact much more fundamentally apparently than the map. I believe you said all the other fail-safe mechanisms did not work.

A: I should have been more -- once we located this on the map -- that is the vehicle to extract the coordinates of this particular facility, those coordinates are then used to cross-check against other facilities that might be in the area that could possibly receive collateral damage, and to make an evaluation as to whether or not that is an appropriate target.

But, we have a no-strike list.

Q: But, I want to follow-up on this for a minute. In that case, then, was this not an all-source analysis situation? In fact, was this single-source analysis?

A: No.

Q: So, what else did you use to establish this target other than the map?

A: I can't get into the details, but it was not a single source.

Q: Did you have other sources?

A: It was not a single source.

Q: Did you talk about the no-strike list and how that's a seller, when you said you weren't looking for the Chinese embassy. But obviously that is the type of thing you look at when (inaudible).

A: All embassies are not on a strike list. Hospitals, schools, mosques, churches, all those things are on a no-strike list. And we, you can imagine trying to keep a database on all of those things, not just embassies. Schools, churches, all of those things that are on the no-strike list. And, the people who work this do the best they can when a target is selected and validated as a good target for the function it performs, in the area or the building or whatever this located. And then we run that selection against multiple databases, which have other targets and no-strike list to see what shows up, as well as doing imagery analysis of the surrounding area to try to double check. So the words or the Chinese embassy did not show up in this validation of another target that was struck the same night that was very near by.

Q: (Sir), it was the process in this targeting where you actually get down to picking aim points, where you are going to involve photo interpreters, analysts that are trained to look at these kind of targets to pick the ideal spots where to basically point the bomb. Wouldn't that have been a point where an experienced analyst would have seen this, and granted I know from an imagery point of view, they would have looked similar looking down. You are not looking at then from street level. But this an embassy compound. It had a wall around it. It had some buildings. Wouldn't there have been some kind of red flag or issue raised that would have distinguished that building from the target that you think you're going to strike.

A: Not necessarily. It was believed to be a headquarters, an administrative building for this large (inaudible). It would have fit the same general description, and point of fact, the imagery analyst would have found that the footprint described on the map was accurate.

Q: Admiral, you said there were two possible problems in how the database base was not up to date. One, that a report of this information about where the Chinese embassy was never was made, or two, that a report was, but was not entered in the data base. Have you determined which of those things were the problem?

A: It is early, but we have not -- in trying to find any raw data, which in message searches of computer data files on messages -- we have not found any which reported to the databases that the Chinese embassy had been moved.

Q: Who is responsible for that?

A: Who is responsible for reporting? Anybody. Could be in the diplomatic community, the civilian intelligence community, military could report.

Q: No one source is responsible for that?

A: For example, there are two NATO embassies still in Belgrade active.

Q: There was a U.S. mission active right up until mid-March. You would of thought that these would have been places that would know how to get in touch with the Chinese in Belgrade. Would they not have been responsible?

A: If we had suspected that the Chinese embassy was near by, we could have done that.

Q: How long will it take you to upgrade (inaudible)? Will you have to stop and go back to all your targets now that you are planning to hit? Double check and triple check?

A: Worldwide databases -- even the database for city the size Belgrade or the country the size of Yugoslavia is a daunting challenge to maintain currently and accurately. The community has taken its fair share of resource cuts in the last few years, and it's a challenge for the intelligence community to keep the databases completely up to date. So what we will do is to go back and revalidate to the best of our ability the maps, the data bases, the charts for accuracy as this goes on. And as we pick targets, we can revalidate in the exact areas of those targets. I mean, we don't have to do the whole thing all at once.

Q: How do you know your database is correct if you are going to go back now? You have got targets you're hitting now? I mean you are reevaluating, I would assume, as we speak, but how do you make sure your information is not faulty again?

A: We can only do the best that we can do. I mean we have -- there is nobody that can assure you that if the database is revalidated, it's up to date as well as it can be, that everybody is reporting reliably and accurately and on time. And everybody is churning away on computers and updating the data base. Nobody can assure anybody that its going to be 100 percent accurate and up to date. It's what we strive for, but the reality is that its difficult and impossible to achieve. We can only do the best we can, and we can't you know, guarantee that something like this will not happen again. I can only say that of the problems we've had on collateral damage, this is the first one that I'm aware of that had to do with faulty database or target -- initially a target selection didn't get (inaudible), I can also tell you that we have on numerous occasions retargeted, or removed targets, conducted different aim points, or ways to strike the target because we found collateral damage issues, and they were successfully taken care of.

Q: (Inaudible) was the target you were planning to attack?

A: We have not, but I won't comment on future plans.

Q: You said more than imagery was used to target this particular location. Obviously, other means -- signals intelligence we would assume was used -- did that fail you as well, or you just couldn't...

A: It was multiple sources. Multiple sources gave us the correct street address for the facility in question.

Q: ...this be signals intelligence, multiple sources?

Q: Could I follow-up? After you translated the street address, as you would say, the area into the city block that would be photographed by triangulating from other main facilities, these main addresses, and from that you derived geographical coordinates, and then begin the checking process. I'm unclear whether the checking process involves somebody going back and saying, "do you have the right geographical coordinates" or whether that statement is a (inaudible) in the process of going on and it's a question of people looking at the set of (inaudible) military/collateral damage and so on?

Is there a second process in going back and saying, "do you have the right geographical coordinates for this address?"

A: In retrospect there should be a multi-step process to ensure that. In point of fact, the evaluation and the checking was -- first of all, the discussion was more "is this the appropriate target," and two, "what is the collateral damage risk within the area."

Q: How much time?

A: Pardon me?

Q: How much time between choosing the target here and actually the bomb, you know...?

A: Well, this target has been under evaluation for a considerable period of time.

Q: (Inaudible)

A: Well, I was going to say that when we double-check, it's both the geographic coordinates and the no-strike list or collateral damage facilities. So, the -- you know -- the answer is that we check both of those things.

Q: To what extent is this all essentially U.S.-derived information. I mean, this is a NATO operation; presumably, other countries knew where the Chinese embassy was. To what extent were they included in this process?

A: The -- we take for data -- the Chinese embassy -- nobody was going out looking for the Chinese embassy as they developed this target. The Chinese embassy was a factor in terms of is it in the database at the proper location, and we would use any and all sources to update the database. And through no source did the database get updated. That's a mistake. And, we all consult with our allies in terms of target development, but because of sources and methods, I don't think we should discuss which countries, if any, were working on this particular target. That's really more in your area than...

Q: Can I just ask you, this was a place that the U.S. had an embassy, however, until recently. Isn't it a town where you could take advantage of shoe leather and at some point or another, or perhaps in the past, people from your organization or others could have mapped out the town, checked out some of these places? I find -- I think what people -- especially in Belgrade itself -- find it so hard to understand is there's this big building with Chinese flags outside it, Chinese people coming and going. Everyone in Belgrade seems to know where it is. Is this, I guess, one of the prices of conducting war at 15,000 feet?

A: No, this is knowable information. The location of the Chinese embassy is knowable information. Yet that information did not get reporting in the form that would allow us to update the databases and the maps that were used, or the maps in question.

Q: The question was, at some point the U.S. embassy, for instance, in Belgrade, they of course know where the Chinese embassy, everyone knows where the Chinese embassy (is) and because it's sort of considered obvious, nobody bothered to...

A: We don't know. It was sad. You know, we're within what, 48 or 36 hours of the event. There are a lot questions to be asked and answered.

Q: Did the CIA go back after the bombing and find that, in its own materials, that the Chinese embassy in its new location?

A: No.

Q: Let me ask about precision-guided munitions that we're told they were, but I don't know if you want to identify them, although most of us know what they were? The precision-guided munitions fired at the embassy -- did all of them hit their targets?

A: I don't think that we have that ascertained for sure. I believe that all but one hit the target, and I believe that most likely that they all did, because there were no reports of collateral -- other collateral damage, in other words, a nearby building being hit.

Q: ... for a total of how many?

Q: Can you finish a thought that you started a moment ago. You said something about how through the imagery we would have to say "it looks something like a headquarters or administrative building." Are you saying that the analyst in looking at these images may have presumed that the arms agency was a complex of similar looking buildings in the same neighborhood, when in reality it was just one building?

A: Well, if you look at the imagery, it's not -- it's not really just one building. There's kind of a multi-story -- five- or six-story building -- and then there is a lower, flat-type building which looks more like a warehouse or something like that. It's not clear that -- there's no definable signature on this building that would scream embassy. The fact that there's a fence around it doesn't -- there's lots of things in Yugoslavia with fences around them that are targets for this operation. So, that was not a qualifying signature or an eliminating signature.

Q: As far as just going back in detail of updated databases, can you tell us (inaudible) of what you're doing and in your agency as well to fix this problem so it doesn't happen again. Whether, on any kind of target, what are some of the things you're trying to do to fix this?

A: Number one, I don't think anybody can say we can fix (it) so it will never happen again. That's number one. We can try to do the best we can to reduce the probability of something like this will happen. I think the Secretary went through a number of things that would be both short-term and long-term, for example, the State Department, which may not be classically in the intelligence collection business forwarding more routinely information about embassies, where we have foreign embassies. We need to do a -- convince ourselves, rescrub about current embassies and no-strike targets and determine the degree to which we need to walk targets after they have been walked through the database by individuals that have walked the ground, as somebody said, and balance the -- I don't want you to misunderstand me here -- I think this target was vetted and worked against the databases. But there is always a compromise between, in some cases, time and speed and how far you want to check one more map, one more source and things like that.

Q: ... participate to award money (inaudible) ...

A: I think the database issue is a resource-constrained challenge.

Q: This morning Yugoslavia (inaudible) that DIA selected the target or the NATO intelligence authorities since this is a NATO war? Who selected the targets, the DIA or the NATO intelligence reports?

A: We did the nomination. It goes through, then, a process that includes consultations with NATO is my understanding, sir.

Q: ... processing. This is highly risky. You are dealing with an urban area. You have a location you want to hit. You've got an address. You don't where that address is on the map, and then you extrapolate. That seems highly risky. Is this a process you are going to look at again, I mean...

A: We are certainly looking at all processees. That -- because this extrapolation is less than precise, that's why you have a system that has multiple checks, both with the CIA, within the intelligence community and within DoD. And as we said before, there's a two-part problem here. The first part was the initial identification was incorrect, and secondly, all the fail-safes failed to note that that was incorrect.

Q: Sir, all these fail-safes, as the other briefer just seemed to allude to, have been somebody running these lists by people who have actually been in that place before or at least that city, or would that be a new type of check that would be installed now?

A: I think, that as a matter of course, I think we try to do that already. If you were to ask me is that going to be an iron-clad requirement, I think its fair to say that it will.

Q: Sir, as the other briefer said, (inaudible) on the map, but when you started out, you said the building was on the map, but it wasn't identified.

A: That's right.

Q: ...and the other gentleman said that you had an address, which you then checked. How did the installation get into the system that this was the federal administrative department? Because the error was not that you didn't know where the Chinese embassy was; the error was that you hit that building. It could have been a hospital; it could have been something else, right?

A: Well,..

Q: ... it wasn't the map, right?

A: You would have thought, that if the map had the Chinese embassy displayed there for sure and...

Q: But the error was...

A: The error was that the target development process got the wrong function at the wrong building, the wrong building, and that the checks and balances that we go through in trying to determine consequences of hitting the target that has now been incorrectly identified did not come up with the Chinese embassy being located nearby. I mean, it should have -- if the databases had been completely up-to-date and correct -- it should have said "the Chinese embassy is located nearby." We start looking at where is it at -- oh my God, here's the target. So it was incorrectly identified. The target development incorrectly put this function at this building, and when we did the collateral damage assessments, the no-strike, you know, reviews and all that, there was nothing about -- there were other targets; there were other facilities that came into our scope, our view there that we did not want to strike. And we looked at where they were, and we determined a successful strike on this target would not endanger those facilities. I mean, but the Chinese embassy wasn't, regrettably, one of them.

Q: You said a minute ago that this process was resource-constrained. Is there something that you used to do when resources were more flush to check these databases? For example, during Operation DESERT STORM was there some additional step or steps that you are no longer able to do because your resources are constrained?

A: I'm not -- number one, I'm not trying to make any excuses. The question I got, you know, if you had more money to have more people doing databases faster, would they more up-to-date? The answer would be "yes." But that's not an excuse for what happened. It is a complicating or a contributing factor that didn't allow us to correct the misidentification. And so, I forget exactly what your question was, but...

Q: There was nothing in the process that you formally did that you would say that we have to eliminate that because we don't have the money to do it?

A: Databases are always -- have always been a challenge to keep current. It is a big world. It's constantly changing. We have been through some fairly significant targeting evolutions, especially since last November -- Iraq, Belgrade -- lots of embassies in Belgrade, hospitals, schools -- I mean, Baghdad, Belgrade -- all these places and this system, which is challenged by the requirement to keep all of these databases update(d), do all this targeting, do all this cross-checking. A lot of time it's the same people working targeting, whether it's Iraq or Yugoslavia, are doing the best they can. And all those bombing operations (in) downtown Baghdad, here, etc., there's this one example where we hit the wrong target. And we regret that, and it was a mistake. And there was original factors and contributing factors that we're trying to improve and correct on.

Q: Can I ask you to clarify? You said you earlier you had the right address but hit the wrong building. Did you in fact have the right address and applied it to the wrong building, or did you have a wrong address and applied it to the right building?

A: We believe it was the former case. It was the right address applied to the wrong building.

Q: In other words the address that you had is still the one that applies to the agency you meant to hit?

A: I would put a "probably" in front of that.

Q: If you give us the address, we'll send someone to check.

Q: Was it the right block?

Assistant Secretary Bacon: Pam?

Q: It seems like the safest place to be right now in Belgrade is the old Chinese embassy. Do you know what's there, and do you-what before this attack did you think was the actual procurement embassy. So that building, the procurement office that you were trying to hit, what did you think that was at that time, and do you know what now is in the old embassy building?

A: Well, when the target came up, the original analysis assigned this, this building with the description of the procurement and supply agency.

Q: So you didn't know prior to Friday actually where the procurement and supply building was? That's what I'm confused about.

A: Well, we hit it on Friday. We certainly believed; we believed-everybody in this system that had anything to do with targeting or striking this building believed that this was what we said it was, the procurement and supply agency, based on the all the intelligence that had been worked. And that's why we struck it. If we'd have believed anything else-anything else at the time it was struck-if somebody would have come up and said, "we think it's the Yugoslav furniture factory," we wouldn't have struck it.

Q: What did you think the actual correct procurement building was when you did the analysis of nearby buildings that-for potential collatoral damage. That's one question. And two, was it in the same city block?

Assistant Secretary Bacon: Jim, last question.

Q: The fail-safes you had in place, did they ever kick out a target that you had before. And also, this is an awful lot of details, whoever worked with this kind of detail before without fail-safes. Sounds like they should have been able to identify that ...

A: I don't think I used the term "fail-safe." If I did, I don't really think there is a fail-safe in this business. I mean, the-we do the best we-we have a no-strike list. We do the best we can to ensure that targets that are on the no-strike list are not affected by our strikes on a selected target. And we have a number of data bases that we run these checks on.

Q: Was it at all perplexing to you ...

A: And the answer is "yes," that targets have been pulled because the cross-checks have indicated that ...

Q: You had targets that either top-down or bottom-up you looked at and when you got to the fail-safe list, you said, "no, we're not going to hit those because of ..."

A: Because of cross-checks.

Q: OK.

A: For a number of reasons, we pull a target off.

Q: How often does that happen?

A: Regularly.

Q: Gentlemen, is it at all perplexing to you that if you hit a complex like this with at least several large precision-guided munitions, maybe 2,000 pounds, that only-I don't mean to be gory here-but only four or three fatalities occurred? Isn't that a little unusual for that size of an attack?

A: Actually, I think that's been a general trend. I've seen incredible damage done at buildings that you would expect there'd have been more personnel casualties, and thankfully in this case, there were, I mean, as few as there were. But they were too many as it was.

Assistant Secretary Bacon: Last question. Excuse me. Just a minute. Excuse me, there's the last question there.

Q: How do the NATO partners feel about this, now that this has happened. Do they have the feeling that they maybe have to double-check your work and your selection of targets?

A: I'll let them speak for themselves. I don't know.

Q: Thank you.

Q: Have you heard anything from them?

(end transcript)


Return to The United States and China.

Return to IIP Home Page.