Slashdot Mirror
How Much Is Oracle To Blame For Healthcare IT Woes?
Nerval's Lobster writes "The state of Oregon blames Oracle for the failures of its online health exchange. The health-insurance site still doesn't fully work as intended, with many customers forced to download and fill out paper applications rather than sign up online; Oracle has reportedly informed the state that it will sort out the bulk of technical issues by December 16, a day after those paper applications are due. 'It is the most maddening and frustrating position to be in, absolutely,' Liz Baxter, chairwoman of the board for the online exchange, told NPR. 'We have spent a lot of money to get something done—to get it done well—to serve the people in our state, and it is maddening that we can't seem to get over this last hump.' Oregon state officials insist that, despite payments of $43 million, Oracle missed multiple deadlines in the months leading up to the health exchange's bungled launch." (Read more, below.)
"This isn't the first time Oracle's name has circulated in conjunction with the Affordable Care Act's digital drama. In November, USA Today published a piece suggesting that 'communication breakdowns' with Oracle Identity Manager had led to 'bottlenecks' in the registration process for Healthcare.gov, the federal online health exchange, which in turn prevented some users from signing up for healthcare. But a single contractor doesn't lie at the root of the federal Healthcare.gov's spectacular debacle: despite months of preparations, large sections of the site remained unfinished on launch day, and the completed parts crashed as soon as users began entering the site. According to multiple sources, the Medicare agency tasked with overseeing the project failed to adequately test, much less integrate, the site's complex elements ahead of launch day. Even if it didn't hold that much responsibility for the federal Website's issues, though, Oracle could find itself the target of much more blame in the Oregon case, where it was reportedly the sole contractor and overseer."
This after Oracle came out explaining how Open Source is not only dangerous but a cancer to development. I'm so glad Oracle has shown with out a shadow of a doubt that Open Source software leads to broken systems, I would hate to not know this, good work Oracle, from now on I'll always pick the closed source guys ...
It isn't healthcare.gov, it's CoverOregon.com, Oregon's own bungled system that only somebody who wants their identity stolen would fill out the "Download this 19 page PDF, fill it out, and mail it to us" "working website".
Though, you may be right- considering what is NOT working at CoverOregon, seems to be the part that links to Healthcare.gov
SJW: a person who perceives an injustice, and while correcting it, commits a greater injustice.
Outsourcing and privatization of coding is a disaster waiting to happen for any company or government of sufficient size. When you lack the wherewithal in your own organization to make the project you're planning, you also lack the wherewithal to judge how much time/money/manpower it would take someone else.
That in-and-of-itself is a problem, but it also, as you noted, injects a middle-man whose biggest incentive is to keep on earning money past the deadline for the project, not finishing it. When you hire your own coders, their biggest concerns tend to be keeping a manageable workload for themselves and keeping their jobs. Humans are (usually) much more reliable than corporations.
The so-called cost savings of outsourcing projects are a lie too, but that's another rant.
> But we are too quick to blame Oracle and the developer of healthcare.gov for problems that come down to what is simply, a bad and incomplete spec that is impossible to build a good system against.
No. All specs are incomplete or bad.
The Waterfall model that everybody seems to still love,in which you assume a spec is complete before you begin work, was discredited in the very paper that named it. Fifty years of waterfall model system develop has borne that out time and time again.
Part of delivering a working figuring out where the specs are flawed, and changing them so that the delivered system works for the users. otherwise it only works for the contracting officers and the lawyers who handle the ensuing lawsuits.
My Heart Is A Flower
It could have been worse.
Imagine what it would be like if it were running SAP.
Any insufficiently advanced magic is indistinguishable from technology.
My team has been talking about healthcare.gov and all the related woes for a while. Pretty much we're all in agreement that we should thank the baby jeebus every day it's not our project haha. Seriously though, for something this complex, if the team grows to over about 15 people it's doomed. And that's just YOUR side, I have a lot of experience interfacing to insurance providers' systems. Half the time the provider you're trying to connect to is broken and doesn't work per their API docs at a basic level let alone have proper capacity let alone have any sense of normal connectivity. I can't even imagine trying to talk to something as huge as the IRS. I bet it's 6 months before you can get a simple spelling fix on an API method pushed out to production.
I came to the datacenter drunk with a fake ID, don't you want to be just like me?
I keep hearing this, but it is hard to take it that seriously when other states' own exchanges are doing fine (so far).
If the specs really were that bad then I would expect the majority to be a disaster but that doesn't seem to be the case (again, so far).
I would agree with you but from TFA:
Oregon had an ambitious goal: to create a place where anyone, from Medicaid recipients and small-business owners to people in the individual market, could go to shop for insurance. "In hindsight — which is always wonderful — we made decisions that made our system much more complicated to build," Baxter says.
Initially, Oracle promised it could get the job done.
Yeah, it could have been a nightmare of a spec, but if Oracle promised it could be done, then I have a hard time cutting them any slack.
And you can bet that Oracle has lawyers ensuring they never actually meet the legal definition of fraud, and that the contracts have enough wiggle room to cover their asses.
But, I can also tell you that it's entirely common for companies contracting for this kind of thing to start off with the full knowledge that they've not asked for enough money to cover everything and get you a working system -- instead they rely on having to do changes and enhancements on a time and materials basis. And then they make a small fortune in quibbling over every little change.
I've seen several of these kinds of things where the contractors essentially knew there was no way to deliver the system on-time and on-budget. They just seem to build in the fact that once the client realizes it, the sunk cost is high enough they get to have a gravy train for some time to come.
It's not fraud, per se, but it's carefully managing the terms of your engagement with the knowledge the customer will end having to pay more and not really have much of a choice.
Sadly, it almost seems to be standard practice in the industry.
Lost at C:>. Found at C.
Allow me to rephrase:
After a project that I was biased against [1] failed, my bias was confirmed and I knew that my preferred solution would be much better.
I have a half-baked plan already, so surely the real thing can't be much more difficult.
Now I see evidence that contradicts my bias, so I'm completely surprised.
[1]: Why is a project I know little about using a database I know little about?
One of MarkLogic's strong points is that it uses that "example schema from a private insurance firm" as its starting point, keeping records arranged in the proper hierarchies for use in the healthcare industry. Yes, you could reproduce the constraints using another database, but why go to the extra work? Oh, right, there's that consistency point... but a quick search shows that MarkLogic is claiming ACID support.
So for a project in the healthcare sector is using a healthcare-oriented database. This doesn't seem to be a bad idea. The questionable part is how there are fewer MarkLogic experts than Oracle gurus, but that's not really a showstopper.
You do not have a moral or legal right to do absolutely anything you want.
>Grab example schema from a private insurance firm, adapt them to this task, and go from there.
You clearly have never worked with code from an insurance company. It's code that goes back to Rome, with layers of crap built on top of layers of crap. The code comments have remarks from developers begging for the sweet release of death.
---- The above post was generated by the Turing Institute. Maybe.
I just find it comical that this is one more in a long string of IT projects taken on by the State of Oregon to be completely botched together, launched to endless faults and problems, then fixed over a period of months if not scrapped altogether. And they have the balls to blame someone else.
To anyone that's lived in Oregon for any period of time over the last 10 years, this is business as usual.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
I think Oregon tried to be much more ambitious than other state exchanges,
That would be a fine argument if what failed was ambitious stuff. What wasn't working from day 1 on Cover Oregon was the ability for individuals to find out what any plan would actually cost and then to actually sign up for a plan. Those are two very basic features of any e-commerce site. Could you imagine anyone trying to claim that Amazon was being "more ambitious than other sites" because they wanted to tell you how much an item costs and then let you actually buy it? I don't know about you, but when I see a website that says "We have the following products, call for pricing and ordering..." I go somewhere else because I know these people aren't serious about their web presence or sales.
Yes, Oregon has some different requirements overall because of the existing state healthcare programs, but that should not stop someone from being able to get a price and say "I'll buy it".
And if Oracle was open and above board, it would have walked away from the contract very early, as soon as it was evident that the spec was incomplete and could not be implemented. That's what any reputable small business owner would do when faced with a similar problem. As soon as you realize you can't do the job, you quit. And start your legal guy on maximizing the smaller amount that is fairly due to you for the work that has been completed. There would be clauses in the contract to cover that.
Oracle is at fault. Or rather, persons in power at Oracle are at fault.
Will
Rumour within my organisation is that Oracle themselves have admitted to our architects that they don't know how their own Identity Management suite really works. They advised us to hire a systems integrator that had worked with all the pieces prior to Oracle's acquiring them.
> Oracle is so maddingly fragile (through their stored procedures
An entirely avoidable artificial and self-inflicted problem.
People pay for commercial databases because the biggest free one still gets delivered with something as simple as foreign keys turned off.
A commercial RDBMS might not be able to handle whatever Facebook is doing but supporting the ACA website should be no problem with 10 year old copies of Oracle, DB2 or even SQL Server.
A Pirate and a Puritan look the same on a balance sheet.
That's true, but if your developer can't even make the schema, they were just going to fail anyway. At least this way you learn they're in over their heads before all the money is spent.
This. I'm a huge data model bigot. I believe that even if your code is dodgy, if your data model is good, you can overcome the dodgy code. If your data model is a clusterfuck, though, all the hero programming in the world isn't going to make it work right.
Well there's the problem right there... they only paid 43 Million dollars. I think that's enough to buy one license of Oracle DB... for maybe a week or so...
http://www.beanleafpress.com