My Two Census

Formerly the non-partisan watchdog of the 2010 US Census, and currently an opinion blog that covers all things political, media, foreign policy, globalization, and culture…but sometimes returning to its census/demographics roots.

Posts Tagged ‘David Allburn’

MyTwoCensus Editorial: Fingerprinting changes are long overdue because the media failed to report on the potential problems

Thursday, June 3rd, 2010

For nearly a year, MyTwoCensus.com was the only media outlet reporting about the problems that the Census Bureau faced in terms of fingerprinting the 1.4 million people who were set to work for the 2010 Census. And we continue that fight today.

In December 2009, I reported that a convicted felon in Alaska was working in a supervisory position for the Census Bureau. This was discovered only after the man killed his mother and then himself. Clearly, this incident should have made calls for improved fingerprinting procedures at the Census Bureau obvious. However, the Census Bureau maintained the status quo and did nothing — fending off my questions and ignoring my concerns.

This incident occurred two months AFTER I originally posted the flaws of the 2010 Census fingerprinting process that were written by child advocate and fingerprinting expert David Allburn, who offered solutions to the Census Bureau that were ultimately refused. Allburn wrote:

(1) The Bureau should announce that trainees are responsible for the “readability” of their own fingerprints, and that fingerprint “failure” due to un-readability (or to discovery of disqualifying criminal history), terminates the canvasser’s employment. This stops attracting ex-felons who would intentionally blur their prints, but it is manifestly unfair to honest workers whose fingerprints are blurred by the inexperienced print-takers. This is fixed by step two.

(2) The Bureau should augment its fingerprint capture by adopting part of our patented “self-capture” technique. Invented by a war veteran, the method has applicants use an extra minute or two to make their own set of “backup prints”, observed and authenticated by the print-taker. Barcoded and enclosed with the cards forwarded to the scanning center, those self-captured prints are readily available for fixing any individual print impressions found “bad.” Well tested, this gets the cards through the FBI with the same dependability as live-scanning offers, typically twenty times better than the old rubber-stamp method now in use.

Only after a handicapped woman was raped by a 2010 Census employee and a sex offender was caught going door-to-door did the Census Bureau decide to change their policies. Is that what it takes to create “change” in America?

Letter To The Editor: Census Bureau Ignores Fingerprinting Problems And Focuses on Name Checks

Monday, December 7th, 2009

The following letter, from David Allburn of National Fingerprinting, comes in response to our recent post that features questions about why felon/presumed murderer Thom Gruenig was working in a supervisory role for the 2010 Census answered by the Census Bureau:

To The Editor:

Did you notice that your questions about FINGERPRINT comparisons were answered with statements about NAME-CHECK comparisons?

The Census Bureau made the two statements that “No criminal record was found,” and “He was not in their criminal database.” Those statements ask us to assume that he was not ANYWHERE in their criminal database, most especially not in the FBI “fingerprints” database. It is not evident at all from the investigation report you published, whether the FBI had actually compared Mr. Gruenig’s fingerprints to the fingerprints of felons, no matter what the names were.  The questions to ask should have included these, for which I proposed what the carefully considered Census answers might be:

1.  ”What would normally have happened at the FBI side if Mr. Gruenig’s fingerprints were determined by the FBI automated equipment to lack sufficient image quality to enable print-to-print comparison?” [Answer: A name-check is done instead, and Census relies upon that.]

2. “Is there any record entry maintained at Census or at the FBI, by them or by their contractors, that shows whether the aforementioned image quality test was passed or failed, either by a direct data description or by a reliable indirect indicator?” (…such as an indication that the fingerprint query defaulted into the name-check process by returning a TCR number.) [Answer: If a TCR is returned, that indicator is probably retained by either FBI or Census or their contractors somewhere.]

3. “If due to ‘normal procedure’ Mr. Gruenig’s fingerprints may not have actually been compared with others in the FBI file, is there any process by which new prints can be taken of assured-adequate quality and re-submitted to assure AFIS acceptance and comparison?” [Answer: If Mr. Gruenig were to be booked after our background check, presumably pursuant to a new criminal allegation, his prints would likely be routinely sent by the booking law enforcement agency to the FBI for comparison, and re-sent however many times necessary to assure the fingerprint check was actually accomplished to reveal whether any previous forensic-purpose prints on file matched his.]

4. “If a disqualifying record were thereby exposed and reported, would Census have the same confidence in the fingerprint portion of its background check process as previously asserted?” [Answer: Yes, but our confidence would be higher for those prints that passed the quality check at the FBI side.]

5. “if there were a way to assure that fingerprints submitted with insufficient quality to support an actual FINGERPRINT COMPARISON did not result in a default-hire as may have occurred in the Gruenig case, and such a way could be instantly and simply incorporated into the current logistical process, is there any reason why Census would not adopt it?” [Answer: Census routinely considers all helpful proposals according to the Federal Acquisition Regulations.]

6. “Would Census reveal whether an internal investigation was done to determine if Mr. Gruenig’s prints were rejected for quality reasons, and whether or not there actually were matching prints in the FBI file after all? [Answer: The Census Bureau considers personnel records confidential and does not reveal their contents.]

7. “Would a Freedom-of-Information Act request limited to whether Mr. Gruenig’s prints got a TCR result from the FBI allow a FOIA response?” [Answer: Consult the answer to #5 above.]

8. “If it were to be revealed by other legal means that there was a TCR returned by the FBI in Mr. Gruenig’s case, and that he indeed did have matching prints on file with the FBI under a fake name different from the one he gave on his Census employment application, …. (question left to be finished by MyTwoCensus.)

Of course, the above is an interrogation, not an interview. And it may turn out that Mr. Gruenig’s prints indeed got compared with the FBI print collection and turned up with no matches. Such a result would impugn Alaska’s reporting system, not Census Bureau procedure. But such close questions is necessary when jousting with a skilled PR department that carefully chooses its words such as providing NAME-CHECK answers to FINGERPRINT-CHECK questions.

I am glad that MyTwoCensus will “soon get to the bottom of this.” Can’t wait.

David Allburn

The Census Bureau’s Unintentional Jobs-For-Felons Program

Thursday, October 15th, 2009

The following piece was written by David Allburn, who owns National Fingerprints, LLC and does not represent the views of Stephen Robert Morse or MyTwoCensus.com.  It should be noted that the Census Bureau rejected David’s unsolicited proposal to integrate his company’s services into 2010 Census security procedures.

To assure integrity and to comfort the public, Congress insisted on a fingerprint background check of canvassers for the 2000 census. The Bureau begged-off that time on grounds of insufficient time and funds, but promised they would do both a name-check and a fingerprint check for the 2010 census.

Next March the Bureau will mass-hire more than 500,000 canvassers to visit homes that did not fill out the form. Evident to the canvassers will be: Which homes have burglar alarms, disabled children present, an overworked single mom, expensive décor and vehicles, etc. Such “data” is not sought by the 2010 census. And it probably IS unthinkable that a census canvasser would assault or steal from homeowners during their census visit. But if a census worker was previously arrested for such crimes there is increased risk they might seek to list such unofficial “data” for use in criminal activity later. That is probably what Congress was concerned about when they insisted that former felons not be hired as canvassers.

It might have been tolerable in the 2000 census that only a couple of assaults were committed by canvassers, but the 2010 census will be different. Not just because there are a lot more felons on the street, and a lot fewer recession-jobs for them, but because the Census Bureau screening method for canvasser candidates will attract them.

Felons will automatically resort to half-century-old methods for evading the criminal history name check and fingerprint check. These obvious methods will work this time because the Bureau has chosen fingerprint procedures and policies which are a full century-old. The situation was fully described to the Census Bureau in a classified section of our August 2008 proposal. The first part, about felons getting hired, was publicized in the recent GAO report and the Senate hearing. The second part, about attracting felons to apply in the first place, was not. We begged Census program office officials to consider the impact of this on public confidence and the Bureau’s PR expenditures should it leak out. Sadly, it doesn’t have to leak. It’s evident to felons already, and will probably be left to YouTube and Jay Leno to “further advise” America about it. I hate to think of Jay Leno “interviewing” (comedians) Gilbert Gottfried as the print-taker engaged in fingerprinting applicant Fred Willard.

Knowing that bad prints generate a “you’re hired” outcome, felons will do what they already do to get a job: Use the internet to obtain fake names and buy convincing credentials that pass the name check. Now trainees, they will exploit the 100-year-old “grapple method” of fingerprint capture selected by the Bureau. In this method the Bureau’s “print-takers” grasp each of the trainee’s inked fingers one at a time and roll it onto a card like it was a rubber stamp. If several prints are blurry the print-taker has to start over. But time is limited, and the process depends on GOOD COOPERATION BY THE TRAINEE.

There is cooperation all right, but it not good. While the print-taker grasps each of the fingers, the applicant feigns helplessness, and causes the finger to squirm, tremble, or press down too hard on the card. Since there is limited time for re-takes the trainee just runs-out-the-clock. This forces the unreadable prints to be routinely shipped to the central card-scan facility where they are scanned into, and rejected by, the FBI. Since re-takes are logistically impossible, the felon gets hired as if he passed anyway, by reverting to reliance on the (fake) name check.

Our proposal warned strenuously about this vulnerability, not only for the predictable 20% rejection rate, but also for the liability: If poor print quality were to cost honest trainees their jobs, it could create a cause-of-action because the blurry prints were arguably the fault of the print-taker, not the applicant. (It appears that consideration of this risk may have caused the absurd “you’re hired” policy when prints are unreadable.) We considered this information so sensitive at the time that we packaged it into a classified section of our proposal. It showed exactly how to plug this gaping security weakness with two simple steps:

(1) The Bureau should announce that trainees are responsible for the “readability” of their own fingerprints, and that fingerprint “failure” due to un-readability (or to discovery of disqualifying criminal history), terminates the canvasser’s employment. This stops attracting ex-felons who would intentionally blur their prints, but it is manifestly unfair to honest workers whose fingerprints are blurred by the inexperienced print-takers. This is fixed by step two.

(2) The Bureau should augment its fingerprint capture by adopting part of our patented “self-capture” technique. Invented by a war veteran, the method has applicants use an extra minute or two to make their own set of “backup prints”, observed and authenticated by the print-taker. Barcoded and enclosed with the cards forwarded to the scanning center, those self-captured prints are readily available for fixing any individual print impressions found “bad.” Well tested, this gets the cards through the FBI with the same dependability as live-scanning offers, typically twenty times better than the old rubber-stamp method now in use.

For the few cases where prints are still unreadable the fault lies clearly with the applicant and not the Census Bureau’s print-taker. This forestalls thoughts of lawsuits and class actions. The method fits easily into the current logistics, gets everyone’s prints promptly evaluated by the FBI as intended, doesn’t require logistically impossible re-takes, and discourages ex-felons from trying to exploit the process.

All that’s needed is for the Bureau to invite an amendment to the proposal. A better/faster/cheaper method, simpler than the full-blown method originally proposed, is described in the patent and is readily available and easily deployed to fit the existing logistics. Fortunately it’s neither too late nor too expensive to fix the problem.

One last thing: The Census Bureau is getting a bad rap on print-taker training. They must have trained them well, and the print-takers must be good at it, because those folks are apparently achieving the same 20% FBI reject-ratio that experienced law enforcement officers get, those few who still use that old manual card-rolling method.

Interview with Robert Goldenkoff of the Government Accountability Office

Tuesday, October 13th, 2009

On Friday, October 9, 2009 I interviewed Robert Goldenkoff, who has worked for more than 20 years at the Government Accountability Office and currently serves as the GAO’s Director of Strategic Issues. One of his many areas of focus is the 2010 Census, which the GAO defined as a high risk operation in a March 2008 report. On Thursday October 8, Mr. Goldenkoff faced questions at a Senate hearing investigating the 2010 Census. In the following interview he discusses the recent fingerprinting problems that he shared with Congress and many other long-term issues with the decennial census.

SRM: What led to the discovery that there could have been criminals hired by the Census Bureau?

RG: We’re looking at all aspects of the Census Bureau’s readiness for the 2010 Census. The decennial census is so huge that we’re focusing a lot of our work on areas where the Census Bureau doesn’t have a lot of experience, where they haven’t done that particular operation before. One of those areas is fingerprinting. In the past, at least for the 2000 Census, they relied only on a name background check. That was why we included fingerprinting as part of our review, because it was a new operation. They’ve been doing the census pretty much the same way  – obviously technology changes – but, the fundamental approach to the 2010 Census is very similar to say the 1970 Census. So if there’s going to be an issue, it’s more likely in something that they’ve never done before.

SRM: Why is your office investigating this rather than the Commerce Department Inspector General’s office? Or were you working together on this?

RG: We are two independent agencies, two different reporting authorities. We do work together, collaborate and coordinate our work just so the right hand knows what the left hand is doing. Sometimes we work in the same areas and other times we work in different areas, depending on facts and circumstances.

SRM: Where did you get the figure that you reported to Congress that 200 criminals could have been hired by the 2010 Census? And can you clarify what “could have been hired” means?

RG: It’s strictly based on the percentages. There were 162,000 people in total hired for address canvassing. 1,800 passed the name background check but their fingerprints revealed that they had criminal records. Of those, 750 were disqualified for census employment, because their criminal records were such that they were ineligible for census employment. All we did was project those same ratios for the 35,700 people who went through the name background check but whose fingerprints could not be read. So it’s strictly a projection. It’s unfortunate that the reporting of this was not always accurate or perhaps sensationalized it. We’re not saying that 200 criminals did work on the census, but we’re saying that based on that projection it’s possible.

SRM: During the summer, I was contacted by a man named David Allburn who runs a company called National Fingerprints, LLC, which can be found at NationalFingerprints.com. His firm placed a bid with the Census Bureau to receive a contract to handle the fingerprinting of employees, because right now prospective employees are fingerprinted by other Census Bureau employees who are not well trained in fingerprinting. David informed me that someone who is an experienced criminal would know that it’s very simple to smudge your fingerprints and make them unreadable by simply pressing your hand too hard on the paper when your fingers are being rolled in the ink. The Census Bureau chose not to use David’s company but rather to conduct the operation on their own without outside help. Of course part of the reason David called me originally because he was upset that his company wasn’t chosen for the contract, but he was also concerned that the 2010 Census operations would be infiltrated by criminals. At first I figured David could be overexaggerating this scenario, but now I know that he was absolutely right. So I’m wondering, do you have any idea why David’s method was rejected?

RG: No idea.

SRM: I’ve also heard from many sources that after people have been hired by the Census Bureau and started to work, their criminal background check reports came in later, and only then, after they already had access to a significant amount of data, were they fired. Why did this happen?

RG: I don’t know. Our point to all of this was not to scare people or anything like that. Our point was to make it clear to the Bureau that they need to have a better policy, or at least have a better policy for those people whose fingerprints can’t be read. With so many people working on the census, even if only a small percentage of fingerprints are flawed, you’re still talking about a substantial number of people.

SRM: Has the Census Bureau done anything to try to fix this flawed system?

RG: It is important to point out though that the Bureau has acknowledged that they have a problem with this and they are taking steps, improving training for example, to improve how the fingerprints are actually captured. Moisture is an issue with the quality of prints. The remaining issue is what to do about people whose fingerprints can’t be read.

SRM: I’m also wondering, can social security numbers be used as an element of background checks? Having sat for the employment exams for the 2010 Census, I know that it is mandatory to provide your social security number at that early stage.

RG: That already might be used, but I’m not sure. But people can change their social security numbers or use fraudulent social security numbers. That’s why it’s not as reliable. As we saw, just  the name background check can’t be the only tool used as criminals can get past that system.

SRM: Who do you hold accountable for these errors?

RG: This is something that the Census Bureau had no experience with. It’s clearly something that the Census Bureau and its parent agency, the Commerce Department, need to deal with. We’re not out to get anyone or point fingers. We want to see a successful census. I think the Census Bureau has acknowledged there’s a problem and they are going to work on it – and we are going to keep tabs on them. There are some smaller field operations coming up, but the big one is non-response follow up in the spring, to follow up with non-respondents. That’s going to be around 600,000 people hired. So we’re going to watch the Bureau’s progress in improving fingerprinting abilities.

SRM: On a related issue, I wrote about how the Census Bureau’s three Data Capture Centers may have similar human resources issues. Because, for example, in Baltimore, the Data Capture Center is run by contractor Lockheed Martin, who subcontracted the hiring efforts to Computer Sciences Corporation, I am wondering if the same rigid hiring standards that Census Bureau employees are subject to apply in these cases? I was told by Stephen Buckner, spokesman for the Census Bureau, that these employees are subject to the same standards, but a couple of loopholes that I noticed are that employees at these centers are not subject to drug tests or that because of time lags, people who undergo background investigations now might not start work for six months, meaning that they could potentially develop criminal records in the interim period. Can you address these issues?

RG: I’m not familiar with the specifics when contractors are involved.

SRM: What are the greatest challenges for the 2010 Census from your perspective?

RG: I’m glad you asked that because what we’ve been reporting on is much bigger than fingerprints. That’s certainly an issue, but the Bureau has other things they need to be concerned about as well. Speaking positively, the GAO has a high risk list, and we put the Census Bureau on this list in March 2008 because of weaknesses in the Census Bureau’s IT management, problems with the handheld computers, the difficulties they were having in coming up with the total cost of the decennial census, the fact that they did not conduct a full dress rehearsal, and on top of all that time was running out. And we put the decennial census on our high risk list because it’s a critical statistical program for the nation. Using March 2008 as an anchor point, we have seen that the Bureau has made a lot of progress in terms of risk mitigation. There is certainly a lot more work to be done but we are also encouraged by a lot of the improvements that we’ve been seeing. Certainly it was important to have a president appointed and senate confirmed Director (Robert M. Groves), so it’s certainly important that the top leadership is now firmly in place. We’re encouraged by some of the advisors that Dr. Groves has brought in who have experience from the 2000 Census. And we’re also encouraged by the fact that the Census Bureau acknowledges that they have a problem. The first step in solving a problem is acknowledging that you have one. But some of the areas that still concern us: IT management, requirements and testing plans have not been finalized, it’s difficult to track progress because of vague metrics, and some of the IT systems face tight implementation time-frames. Of all the IT systems, the one that we’re most concerned about is the paper-based operational control system (PBOCS).

SRM: Can you elaborate on that?

RG: That was the program that was put in place when they abandoned the handheld computers for non-response follow-up. So PBOCS basically controls the office workflow. There’s a lot of work to be done in terms of nailing down requirements and testing in the short time remaining. Basically, they have a lot of work to do and not a lot of time to do it before it needs to go live.

SRM: There was a Census employee named Bill Sparkman who was murdered about a month ago. Is your office involved in that investigation?

RG: No, not at all.

SRM: Do you have any comments on the recent decision for the Census Bureau to sever its ties with ACORN?

RG: The Bureau just needs to make sure that it has adequate guidance so that it can make a determination as to who they should partner with and who the shouldn’t.