Slashdot Mirror
FBI's New Info-Sharing Software Project Fails
Spy Handler writes "After 4 years and half a billion dollars, FBI's attempt to create new information sharing software - called Virtual Case File - simply didn't work.
← Back to Stories (view on slashdot.org)
I actually read the FA (Ok, scanned it), and I didn't see anything that the FBI required that isn't commonly available now. Get a robust DB, have information decrypted at the user's computer, do not have any portion of this network on the Internet - instead use VPN/SSH connections physically secure the boxes, etc. Why they went to a third party in SD who blew through 130 MILLION of our tax dollars with nothing to show for it is beyond me.
"As the intrepid kobold companion continues his journey, he begins to wonder... if priests raises dead, why anybody die?
Someone at the FBI definitely should have read this article.
-Rob
Marriage doesn't have to suck!
Perhaps using the Duke Nukem engine as a front end was a bad idea?
Seriously, when you look around it's amazing how many software projects just completely fail with no usable code produced. It's not uncommon for companies to spend several million and just shut the thing down a couple years into it.
I think we're about a century behind our technology. We still try to use industrial age models for 'building' things...and the digital/info/[buzzword] age has major implications that those models just don't take into account.
Sounds just like the TIERS Texas Integrated Eligibility Redesign System software my agency has been trying develop. The Texas Department of Human Services, now the Texas Health and Human Services Commission, has contracted to Deloitte to develop a web based system similar to what is described in the article. $3 million a month (according to some) has been spent on this for a couple of years now and it is a HORRIBLE excuse of a system. I know case workers that are being forced to test the software that say it takes at LEAST twice as long to work a case now than it did with the old system that was developed in the 80's. This has been a boondoggle in the worst sense and any Texas taxpayer should be pissed off about it.
It gets to be depressing working for the government because you see so many contracts like this awarded simply because some higher up gets his palm greased. Another example of this is the fact that I had to pay Banctec (the company that has our hardware support contract) the standard fee of $340 to replace a CPU FAN in an old machine the other day. So sad.
P.S. - I'm having to post this anonymously because anyone that has even begun to criticize the TIERS software, even internally, has been officially reprimanded or worse.
Large organizations are only truely served by in-house developed software. The trick is for said organizations to hire folks who really know what they are doing.
I can generally tell when a project is going to fail. The whole process begins with sending the project out for bid. For specific projects, yes, farm it out. There is no need to write your own relational database. If you don't have a Unix weenie in-house, it is cheap at twice the price to farm out hosting.
But making up a wishlist for projects with 100,000s of thousands of users and 10,000s of thousands of uses is just asking to be ripped off.
"Learning is not compulsory... neither is survival."
--Dr.W.Edwards Deming
"Science Applications has received about $170 million from the FBI for its work on the project. Sources said about $100 million of that would be essentially lost if the FBI were to scrap the software."
Hold on, isn't Science Applications (also known as SAIC) also that company that was so incompetent in creating Pro-U.S. propaganda in Iraq that their contract was pulled from them?
http://www.corpwatch.org
http://www.prwatch.org
Why does the government keep giving them contracts when they suck?
If I sold a car to the government that didn't run at all, I'd be in jail for fraud.
Why don't they do the same for software?
If you sold a car to the government that didn't run at all, it's probably because the order that the government placed with you was for a vehicle with an 80-gallon gas tank, that can use gasoline, diesel, compressed air, or vegetable oil and run equally well with each, that is no larger than a stationwagon and weighs less than 1000 pounds.
The only way to meet those specs 100% is to leave the engine out altogether, no?
Project failures can often be traced back to a failure to provide good project specifications.
I disagree.
No you don't. All you need is a sample dataset that looks like the real data. Creating suitable simulation data and final testing with real data can be done by a small core of cleared personnel.
Bullshit. That's overpaid consultant-speak. If the requirements are clear, they can be met. It would be the job of the core cleared group to find out what those requirements are, and articulate them in a form they can pass on to the OSS developers. They can be your overpriced consultants if you like. All the OSS people need to know is what they're making and how it's supposed to look in the end.
Now I'll grant you this - it would be much harder to communally develop software without an existing model. Linux had minix et al, Firefox had IE et al, etc. All that means is that the inner circle has to be smarter about communicating the requirements - giving as much as possible without compromising security.
Festering bullshit. This is not a commercial project, this is the goddamn FBI's data base. You don't choose between it and some other option. You either use the damn software as is or you don't have access to the FBI's information sharing network. That is all.
Irrespective of the quality of the software, how long do you think a police chief would last if it became known that he refused to use the FBI's information-sharing software because [insert self-serving, job-maintaining excuse here]?
No, the only difficulty I can see in making this an OSS project is communication. With good facilitators, this is quite doable. There may be a few segments of code that cannot be publicly developed, but those could be localized to a single module (which, if done, could even make the software MORE secure, because you could just replace that single module whenever the network is suspected to be compromised).
It's hard to soar like an eagle when you're surrounded by turkeys.
When will people learn: with extremely complicated systems that humans have to interact with you can not specify it 100% correctly the first time?
Experience in building such systems has lead many of us to realize you must use an iterative approach that allows the end users to be part of the feedback loop.
Release early and release often, let your users use and break the application, and come closer to the ideal system with each iteration.
Now, I wouldn't blame the FBI for the problem completely - after all, they are not software developers. A portion of blame should go toward the contractor for failing to realize the issues surrounding development of such a complex system and taking appropriate actions to determine and meet the needs of their clients. Their contract should have been written to a) specify customer satisfaction as the key measurable for success, and payment of the contract b) put in a rider that basically states any functionality needed to bring the application to minimal usability as discovery occurs will be part of the first contract (this is negotiable - some things are really enhancements and new functionality - and some are required, even though not originally discovered in the first iteration - this allows both parties to recognize up front that 100% discovery of requirements does not take place in practice).
This approach has worked extremely well for me as a manager of vendor development (I have been extremely lucky to have vendors who understand what I am talking about), as well as for my own projects that I develop and implement. While there is a bit of risk involved in negotiating key usability issues discovered late in the development cycle - going out of the gate with an iterative approach ameliorates much of that - and is certainly less risky than giving someone $100,000,000 before I see the first line of code...
Lodragan Draoidh
The more you explain it, the more I don't understand it. - Mark Twain
During the 15 years I ran my own computer consulting business it was common to be invited to make a bid, do the analysis and present a proposal, only to have the analysis given to a another to impliment. Sometimes the connection was nepotism, sometimes it was a competitor who under bid, so the putative client thought they'd save money by using the low bidder. They "Cherry Picked" me. That happened only a few times before I realized what was happening and begin charging for the analysis. If they wouldn't agree to pay for the anlaysis I wouldn't submit a proposal.
I am wondering if a similar thing isn't happening here. SAIC is, in effect, being paid to the system analysis, but the most lucrative part of the project will be given to an insider, a crony or for a political payoff.
Running with Linux for over 20 years!
Large organizations are only truely served by in-house developed software. The trick is for said organizations to hire folks who really know what they are doing.
Agreed. However, there is a mindset in a lot of government agencies that COTS and/or using consultants saves money - no matter how many times they get burned. In typical fashion, the article indicates the FBI ("I" for investigation) has hired two more groups of consultants to investigate the problem.
Unfortunately, we cannot compare W. Von Braun and aparatchik who runs FBI's IT department. Von Braun was a man with vision from his youth to send a rocket to another planet (at least his current biographies say so). We may argue about his role in WWII, but we must admit that his WWII project was also quite successful - V2 was at least decade in front of any equvialent project of that time.
If you find me the man who was able successfuly to run expensive technical project in war time, with shortage of all possible suplies, and with a threat of being sent to Eastern front if anything goes wrong, I'll hire that man for any project.
No sig today.