Friday, March 28, 2008

Census Project Going South

Those that have been reading this blog know that I'm big on two things: 1) Not calling major problems "glitches" and 2) Understanding and overcoming the human-computer interface concerns.

Well, this is a story that blends both of these. I was reading our local paper this week and came across an interesting article, "Census glitches may cost billions". " So, I thought "Hmmm. This looks interesting. That's a pretty expensive glitch! What's up with that?"

The skinny is that it looks like the 2010 census project is in trouble. The government figured that the manual method was just too old fashioned for 2010, so they (the census bureau) decided to fund a $596 million project to use handheld computers. (The project cost is now at $647 million.) Now, problems are emerging that will likely add another 2 billion dollars to the original tally of 11 billion dollars for the 2010 census. This works out to about $43 for every man, woman and child in the USA, using the current estimated U.S. population of 303,729,132 at

At risk is both the accuracy and feasibility of conducting the assessment. (Think "voting machines" and you get the idea.) The problems are so serious that census officials are considering (gasp) pencil and paper!

Further information obtained from congressional testimony reveals that the census bureau was "unprepared to manage a $600 million contract for the handheld computers that will be vital."

Unprepared?? I realize I'm on the outside barely looking in, but good grief, why embark on the project if you don't have the ability to run it? Of course, knowing how the government procurement process works, I'm sure someone was told the project would be no problem at all.

Project management may sound boring, but good PMs know how to bring a project in on time, within budget and at the agreed upon scope. This knowledge is not kept in tightly locked vaults, but is openly available for anyone who wishes to learn. The problem is, the PMs who really need to learn project managent, don't think they need it!

Most projects like this, however, have lots of blame to go around. Things like poorly defined requirements and contracts, lack of communication, difficult project sponsors, expectations that are too high, technical solutions for non-technical problems, etc.

The problems boil down to:

1) The handheld computers are too complex for some temporary workers.
2) The original programming wasn't efficient enough to transfer the high volumes of data generated.

I believe communication is the underpinning of all IT. Interestingly enough, census director Steven Murdock admitted that"communication problems" between census officials and the contractor (Harris Corp) have caused "serious issues."

This is why I'm a big fan of pilot projects. This is a huge scope of implementation. It's good to try things in the small world before ramping up to the big world. Sure, technology changes a lot in 10 years, but at least you have a lot of time to phase in a technology. Plus, you can pull the plug at $1 million instead of $600 million.

So, don't be surprised if you get a paper form to fill out in 2010, or see census workers using "Plan B" - pencils and paper. The sad thing is, it will most likely be a $2 billion "lesson learned" with our money that could be used for more important things. (Unfortunately, there are so many project failures that I doubt the lessons will ever be learned!)

I hope I'm wrong and they can get it sorted out - correctly, that is! Census bureau - We're counting on you! (Sorry about that.)

What project management or testing lessons do you see in this story?


Steve said...

We are having the same problem with our voting solution. Been told it is to cheap. It is also less prone to errors and too simple. It uses a Digital Pen and Anoto dot enabled paper.


Randy Rice said...

Hi Steve,

Thanks for your comment. For some reason, if something is seen as "simple" it is not seen as "robust enough". I think there is much value in simplicity, especially where technology meets the average person.

In Oklahoma, we have a voting system that I wish other states could see and emulate. Actually, it reminds me some of your solution, Steve. Just not as modern.

The ballots are paper and a person makes a choice with a felt-tip pen by connecting two sides of an arrow. Then, the ballot is fed into a scanning machine where the ballot is either counted or rejected. It can be rejected if multiple selections are made, write-ins are made, etc. Then, a paper tape is printed upon, which is the audit trail.

All I can say is that the poll workers understand it, the voters use it with no problem and that it is highly accurate. Of course, we're just Okies, so what do we know? Right!

Hey, Steve, thanks again for your comment and the best of success with your product!


Anonymous said...

Hey Steve,

I came across this Federal News Radio story you might be interested in. The story gives a lot of insight into the census handheld device.