The scope of the project was not adequetly defined. The next step is to fire the programmers.
Reports that the White House is running away from CGI Federal and seeking help elsewhere could signal that, rather than having things under control, the administration is still scrambling for a solution to its faulty multi-billion dollar website.
For picking the milestones there is only one relevant rule. Milestones must be concrete, specific, measurable events, defined with knife-edge sharpness. Coding, for a counterexample, is “90 percent finished” for half of the total coding time. Debugging is “99 percent complete” most of the time. “Planning complete” is an event one can proclaim almost at will.*
The only way to fix it is to start again. I’ve been on a cleanup project where the previous guys were fired and we came in and were to “fix things”. After looking around the client was just looking at the tip of the iceberg in terms of incompetence of the system so we rewrote the whole thing.
what james said . . . been there, done that as well. Called them the “6 pounds of shit in 2 pound bags projects.
Always better to start over.
Having been in software engineering for nearly twenty years now, I can say with some confidence that nobody has learned a damn thing in at least twenty years about software engineering. Software project management is worse than women’s fashion for idiotic fads, and basic metrics are all but unheard of. I chalk this up to the fact that the vast majority of software “engineers” today were not trained as engineers, but rather as computer scientists, mathematicians, or artists.
I’d be willing to bet money that the word “agile” came up a lot during CGI’s tenure on this project.
O’s Buggy Whip Inc. sold.
Frozen O rings on booster rockets blamed for cluster***k fubar failure to orbit.
…-
“CGI to Be Replaced by Accenture on Obamacare Contract”
“Reports that the White House is running away from CGI Federal and seeking help elsewhere could signal that, rather than having things under control, the administration is still scrambling for a solution to its faulty multi-billion dollar website.
Recall that after months of insisting on an in-house fix, the White House conceded defeat in December and handed the website over to experts in the private sector. The reported deal with Accenture could be just another attempt to branch out to find someone to repair the glitch-prone mess of a website.”
http://www.theblaze.com/stories/2014/01/10/obama-white-house-dumps-it-contractor-responsible-for-health-care-website/
Daniel:
Agile methodologies are not completely without merit, but many have translated it to mean “I can just start coding, and don’t have to worry about software design.”
What James said.
Start over. Give consumers a list of the insurance policies and their features costs UP FRONT! (you know, the actual purpose of the website?)
Then direct them to the subsidy section. If they know there is no qualification for subsidies, then the consumer does not need to enter any of that information.
Not being very computer literate I have hesitated to mention the scheduling processes develop, in the beginning I am sure through trial and error, for designing and building ever evolving aircraft.
I wondered if anyone in Washington State ever walked over to Boeing to see how a 737 and/or 747 was effectively designed and proved very scalable to meet ever changing competitive and changing market place requirements.
Appreciate anyone helping my education. Cheers;
The fundamental error in Healthcare.gov was going live without testing it. Even people without software experience
can see that that is really really stupid. It would have been less embarrassing for the O to postpone the launch
for one or two years – embarrassing but less embarrassing than the debacle. Fortunately for the US his hubris
led to the error which exposed his incompetence.
Robert Martin in “Clean code” has a nice description of a software development project decomposing as
productivity falls to zero as increasing complexity leads to confusion, demoralization, and panic – none
conducive to clear thinking. Oh yes, and adding extra programmers does slow the project – or increase the
rate at which the productivity falls to zero.
On the plus side, CGI is now freed up to create Bammy’s long wanted gun registry..
Don’t forget lots of scrums. Been at it for 30 years. All I can say is amen brother. Obamacare will be repealed before this software is made to work.
Look for many more billions being spent in an attempt to fix the unfixable. When a software project is this bad, the best thing one can do with it is give the main server a robotic arm with which it can put a .45 slug through it’s hard drive.
Anyone who’s had any programming experience realizes that some pieces of code are just so awful that the only thing they should be used for is as an example of what not to do. If the reason for total lack of scalability was the underlying architecture of the code then it’s time to start over.
One thing I’ve noticed in medical software is what appears to be a totalitarian impulse in the designers in which they wish to have everything centralized in one massive server farm with all access to and from the data “controlled”. With current encryption protocols such control is necessary. One argument that I have with any person who works in hospital IT is whether data should be centralized or distributed. I’m a big fan of distributed data storage and processing as it greatly speeds up access to the system and one has a backup copy of the last good data in a hospital locally in case the main server goes down (which it does very frequently and since the latest “upgrade” at Interior Health has been at least a 50-100% increase in access times to data) along with 20x more network traffic to perform the same operations. Only a megalomaniac (or a damn good programmer) would assume that they could take an untested system and use it on a nationwide basis in the US. A prudent programmer would have tried the system out first in a small state to see how well it scaled in that environment and then implemented systems state by state working out the bugs as one got to more and more populous states. A far more rational architecture would be 50 distributed databases which would eventually connect to a central database for the whole country. The initial application did involve distributed servers in various cities (when I connected to it I got a Seattle based server), but very likely the software had the nice pyramidal shape so beloved by totalitarians, dictators and those in commercial IT who are very threatened by the power the average user has on their desktop.
Something to consider is that the ‘scope’ of the project was never clearly definied; therefore, CGI likely is in the clear.
I’m not in software engineering, but I am involved in Project Managment for large engineering mining projects.
The bottom line is… without a clear scope of what the contractor is responsible for; one CANNOT set benchmarks or milestones because the responsibility of the contractor has never been legally defined.
THIS is what I believe really happened with this contract. The government issued a Purchase Order for CGI to build a website for Obamacare; and they got exactly that. IT is not too much of a stretch to believe that Sebeliuos(sp?) has absolutly NO TRAINING or EXPIRIENCE in these matters(except for wasting taxpayer money), and threw a isht-ton of tax money at this project based SOLELY on HOPE.
In other words, she “hoped” the website would work.
Ask yourself this: if one doesn’t know how such a website is to function, then how can one expect to the contractor to build for them that which they have no vision?
Answer: they can’t! The only thing they can do is build ‘something’, bill for it, then claim they produced exactly what they were contracted to do. This is what I suspect is the true case.
In summery, I suspect that CGI burned 10s of thousands of hours(perhaps 100s) aimlessly building a website that they knew would never fall under legal scrutiny. In other words: free money!
BTW, South Park did an episode this season mocking Obamacare. very funny
I would have more sympathy for CGI if they hadn’t broken the cardinal rule of contracts. Don’t enter into one where the other side has the power to abrogate it at any time.
It is clear that requirements were withheld for political reasons til long after they should have been locked down, for one thing.
It is also clear that the scope of the project is in constant shift.
This was a no win for anybody who got involved. I am just glad it wasn’t me.
Everything old is new again…………..Accenture = Arthur Andersen = Enron Scandal.
The problem with widespread distribution of processing is that your attack surface becomes extremely broad, and multimaster replication is Hard.
I am familiar with engineering projects going off the rails – I learnt from my previous bosses 😉
However, the sheer incompetence of the Obamacare Act and its implementation, in my view, can only be explained by malevolence. The O-Team deliberately aimed to screw up the current health care insurers and delivery systems (AKA doctors and hospitals) so they can move in and socialize them.
Moving to Accenture is just a way to launder millions of Dollars . . .
Remember, 90% of any development project gets done in the first 50% of the time. The remaining 50% of the time is necessary to do the other 90%. (No, there are no typos in those two sentences, and, yes, I develop software for my living).
right on Daniel. and when those programmers (Scientist, Artists) are directed by Social Workers the outcome can only be Chaos. Engineers are trained to solve problems in that a re-produceable and in a very logical way. The Obama Administration is sadly lacking in this area as all of them are Progressives that have no training in logistics.
Here endeth the lesson. Go thou, and sin no more.
Typical Government IT Project. Much as I have no love for CGI I’m willing to bet the fault lies more with the government than CGI. Like most government IT projects the scope was not clearly defined and the requirements almost certainly changed daily until what the client wanted was nothing like what the RFP asked for. I’ve seen entire multimillion dollar gov IT projects ground to a halt because a button or tab is not the same color as it was in the previous system and some petty bureaucrat halts the project because well the button should be blue because it’s always been blue.