FCW Insider: Beyond the Census fiasco... spurring innovation
The news in recent weeks that the Census was essentially abandoning its e-collection efforts and turning to paper
probably didn't come as a huge shock to... well, just about anybody. (See this FCW story from 2006
-- yes, almost two years ago. In fact, I've pulled all sorts of Census links together at del.icio.us/cdorobek/census
. In addition to FCW's ongoing coverage of the 2010 Census -- there were red flags back in 2006
, if not earlier. On Delicious
, I've included a number of GAO reports, with warnings going back several years. In fact, just a month ago, GAO added the 2010 Census to its high-risk list
. [GAO testimony; .pdf])
There is no doubt about it -- this is a black eye on Census... and on the federal government, for that matter. Most people in the country don't know the difference between federal, state and local governments, let along among different agencies. To them, it is all just "The Government." So it is interesting to look at how this story has been reported. The San Jose Mercury News blog, Good Morning Silicon Valley
, reported it this way:
Census Bureau decides it can’t count on computers
This is just pitiful. As a taxpayer, and especially as a Silicon Valley taxpayer, it rankles me no end to see yet another government technology initiative botched up because the people in charge are out of their depth. Now it’s happened again
, this time with the Census Bureau’s plans to bring its tallying tech at least into the 20th century, if not the 21st.
The bureau contracted with Harris Corp. in 2006 to pay more about $600 million for a system that included 500,000 handheld devices to be given to census takers in 2010 to gather information from the millions of people who don’t return their mailed forms. The devices were intended to verify all residential addresses in the nation using GPS, collect and transmit the data from the uncounted, and manage the workflow in the field. Unfortunately, in a scenario familiar to most valley engineers, there was a disconnect
between contractor and client over the specs. In a test last year in North Carolina, the computer’s interface ended up leaving workers baffled and confused, and initially the units couldn’t transmit the large amounts of data required. At one point, said Commerce Secretary Carlos Gutierrez, the government had some 400 new or clarified technical requirements on its feedback list for Harris. The costs kept climbing, helped along by gross miscalculations. Initially the bureau and Harris agreed that $36 million would cover a help desk for field workers using the handhelds; on further reflection, that figure is now up to $217 million. At a March hearing, Rep. Carolyn Maloney, D-N.Y., said, “What we’re facing is a statistical Katrina on the part of the administration.”
Yesterday came the final blow. Gutierrez told Congress that the whole computer initiative has been called off
for the coming census and that the bureau would hire 600,000 temporary workers to do the job with pencil and paper. The expensive handhelds will be used for address verification, but that’s all. She blamed the problem on “a lack of effective communication with one of our key contractors,” while Harris cited the bureau’s ever-shifting specifications. Bottom line, the government will still buy 151,000 of the computers and associated service from Harris, but now the cost has risen to $1.3 billion. That and the related changes will push the cost of the census from what was already a record $11 billion to more than $14 billion. And the census count — with billions in government spending dependent on its accuracy — will still be conducted in crude manual fashion. Pitiful.
But just as interesting are the comments on the blog post. Here are a few:
*Yes, this is pitiful. $14,000,000,000 divided by approximately 320,000,000 people means the census will cost about $14 per person. Does this make sense? A few of my techie friends could put this system together in a few weeks. Why can’t the government learn how to handle large tech projects?
* The folks responsible need to be replaced with competent personnel. We need to write our Congressmen and voice our displeasure. Otherwise we will see history repeat itself. (My own note here but... Congress is as much responsible. If you want something done, then fund it. Not only are lawmakers seemingly incapable of passing a budget on time, but they have cut Census funds when they are needed.)
* Let’s remember projects like these when we talk about corruption in other countries. When a handful of companies qualify to bid with carefully crafted RFPs that keep others out, these are the results you get!! Corruption done legally, with revolving doors and fixing RFPs gets results like these. And guess what it’ no different than corruption in a third world country!!!
Clearly some of these are over the top. Those in Silicon Valley probably don't want to throw stones given that just about every IT vendor failed to get systems out the door, and most of those systems are not nearly as complex as what the Census does every decade.
There was one post seemed more informed... GaryM of the Cranks R Us
blog, who posted a comment on the SJMN's blog post
As someone with experience working for a Federal agency, the reasons are all-too familiar:
- Middle and upper-level management more concerned with their careers and their empires than with completing projects.
- Turnover in management (especially political appointees), with the new hires having their own agendas (causing project churn).
- Lack of incentive and empowerment (e.g. responsibility with no authority) among the actual implementers to get the job done.
- Poor communication among all parties.
- Development based on “Big Bang” delivery (everything all at once) instead of small incremental deliverables that can be tested and validated.
If those in government in charge of these projects had their compensation tied to their performance, that by itself would improve the process. Fat chance of that ever happening.
All of the bravado aside, the interesting question to me is how can one provide incentives for agencies. Right now, there is no incentive whatsoever to risk doing something new or different or taking a risk for the public good. In the case of the Census, one can argue that officials took a risk -- and it didn't get them anywhere.
Earlier this year, I was speaking to a large government contractor and we had a really excellent conversation. (When I go out to speak, I always prefer to have conversations rather then speeches so everybody gets something out of it. It is more fun... and more interesting.) One of this contractor's employees was a former fed and he made a point that we all know but is so easy to forget: The government doesn't run like a business. We say that all the time, but it is true. A business has a relatively simple focus: To make a profit. If only agencies had that clear of a mission. This person used the example of the Homeland Security Department's port initiatives. What is the goal? One goal is to prevent a terrorist attack on the United States. And everybody knows that the ports are porous. There are millions of containers that come through U.S. ports. If the goal is to prevent a terrorist attack, the United States could prevent containers from coming into the country. That would also prevent commerce, of course, so... it's clearly not a viable option. So there is no black or white. The question is what shade of gray.
The question is how do you get agencies be innovative when they can always be second guessed?
Posted by Christopher J. Dorobek on Apr 20, 2008 at 12:17 PM