S.F. programmers build alternative to HealthCare.gov [video]

http://www.cbsnews.com/8301-18563_162-57611592/s.f-programmers-build-alternative-to-healthcare.gov/

by r721 • 13 years ago

103 82 13 years ago

cbsnews.com

aram• 13 years ago

Hats off to the guys - good job.

Here's the link to the site: http://www.thehealthsherpa.com/

willchilcutt• 13 years ago

I don't understand how this site works. A few days ago when it was first posted here, I tried my zip, and it returned

"Plan Results

Found 0 exchange plan."

Today I searched my zip again, this time even messing with the different options of insurance type, and still zero results. Is this saying there are no insurances available for people in my area?

joyeuse6701• 13 years ago

Apparently it only works for California zip codes:

frakkingcylons• 13 years ago

Not necessarily, I tried it with Arizona and Texas zip codes and those both returned valid results.

dmethvin• 13 years ago

You haven't provided enough information for anyone to give you advice about what might be the problem. Perhaps you live in a state that has its own exchange (e.g., Kentucky or Maryland). Or maybe there is no data for your area yet. Try Chicago (60609) for some sample data to use for an experiment. If that doesn't work, perhaps some error is occurring on the page; check the JavaScript console and report it to them.

_pius• 13 years ago

This looks great, but haven't the big, embarrassing problems been with scaling the back end? Even if this were a superior UI and flow, it wouldn't address the issues that made the news.

hga17• 13 years ago

Perhaps, and no.

E.g. the front end registration problem is due to a late requirements change in August that had the effect of preventing window shopping, plus we know there were changes ordered though the week before launch, and no integration testing until 1-2 weeks before launch, and that discovered it locked up if 200 simultaneous login attempts were made or if it had ~1,100 users on it.

So, yeah, those are scaling problems, but not ones we can necessairly blame the techies for, given that they didn't have enough time, and couldn't do integration testing (the government's CMS was the integrator and integration tester, now replaced by QSSI, the ones who did the "data hub"? I think it's called, which is supposed to be not so bad).

But also no, in that the fix-it czar's #1 declared priority is to fix the garbage 834 EDI transactions that are being sent to insurers. That includes incorrect data and incorrect transactions (e.g. multiple enrollments and cancellations for the same person). Based on that, the front end's problems have been a blessing in disguise, because the insurers are only getting a few hundred enrollees a week and that low volume allows the insurers to call each up and straighten out problems.

An estimated 16 million policies in the individual market, and an unknown number in Obamacare's interim high risk pool must sign up for new policies by December 15th or so or they'll suffer a lapse in coverage, which for many of the latter will be fatal. The only way to get subsidies is through the Federal system, i.e. the state exchanges use it for that (maybe with inaccuracies, but it's evidently working for the state exchanges that are working, but don't judge that for any given state exchange until you find out which are getting Obamacare and the much larger numbers reported to be signing up for Medicaid), and that makes sense to avoid fraud, since it results in direct payments from the Federal fisc.

dr_• 13 years ago

Valid point with the window shopping comment. A huge mistake on the part of the administration though. The site was apparently flooded with visitors and you have to wonder how many of those visitors were people who already had decent coverage, privately or through their employer, but were just curious to check it out because they had heard so much about it. I know I was one of those people.

jorgem2• 13 years ago

> An estimated 16 million policies in the individual market

What percentage of people with individual plans (who need to get new plans) would actually qualify for subsidies? I don't, so I have more options: (1) I can buy another individual plan from my insurer (2) I can shop other insurers or (3) I can buy from california exchange

In none of those scenarios do I actually care whether healthcare.gov is working. I suspect I am not alone.

Tloewald• 13 years ago

It doesn't need a back end at all. They've done exactly what I suggested in my blog but was too lazy to do — make a rough guess based on static data and hand you off to the vendor. It can be done client-side off static content. Using the IRS service to verify income would be something the real site could easily do, but it's not really necessary.

http://loewald.com/blog/?p=5349

danso• 13 years ago

From the convoluted articles I've read, not only was the main bottleneck in the identification process, but the issue seemed to be that the authentication was necessary to just show the plans, because the gov wanted to make sure you were 25 before showing you a range of prices...was that the case? If so, whatever happened to letting people just browse the entire table of prices....remember when people had to look things up and scan tables once in awhile, instead of living in a Minority Report world where every piece of info is handcrafted to the user?

I wonder how much of Healthcare.gov's problems could have been eliminated if the sign up was more modular...users were allowed to browse before signing up for anything. Then they could sign up via email and tentatively pick a plan. Then in 3 days, they go through the full authentication process...this is similar to how Amazon does things...you can sign up without a credit card, you can browse without signing up, and you don't go through the processing bottleneck till the very end.

hga17• 13 years ago

The contractors were ordered in August to change the system to require up front registration and submission of "correct" information on penalty of perjury, no window shopping allowed, nor checking out "what if" scenarios (e.g. using the Kaiser Foundation's calculator, someone was unable to find any scenario where being married wasn't more expensive, and of course subsidies are based on income, and that can be very flexible for a lot of people).

Since then window shopping has been allowed, but I don't know if the "under penalty of perjury" bit was dropped.

As to "what happened", I'll just point you at this Washington Post article and let you come to your own conclusions: http://www.washingtonpost.com/politics/challenges-have-dogge...

Note also the front end traffic jam has avoided sending the insurers more garbage 834 EDI enrollment and cancellation transactions than they can manually clean up by contacting enrollees. That's the #1 thing to correct according to the fix-it czar.

And I do agree with you, an Amazon system, which I'm sure is based on the BASE eventual consistency back ends, e.g. you don't get your confirmation email until it decides who "won" the last copy of an item, would have been a lot better. But then again the contractors were doing a lot less of that sort of top level design that you'd think (see the Post article).

analog312• 13 years ago

I just tried it, and it's great. Nice job, on many fronts: Recognizing a problem, coming up with a simple, uncluttered solution, and making it work.

In my view, a site like this is useful for a lot of purposes beyond actually signing up for health care. For instance, I've got employer sponsored care, but I was simply curious about what my costs would be through the exchanges. I also want to be an informed citizen, and "how much does it really cost" is an obvious question that anybody might be inclined to ask. In my view, knowing the magnitude of the numbers should help provide a reference point for political debate.

hga17• 13 years ago

Have you tried this: http://kff.org/interactive/subsidy-calculator/ ?

analog312• 13 years ago

Thanks for that! I had assumed that I wouldn't qualify for a subsidy, and indeed I don't.

As others have mentioned, these sites don't actually solve all of the problems, but it seems to me that a few free tools that just help people gather information would take some of the pressure off of the main system. And if enough of these tools were to crop up, then none of them would have to be 100% reliable.

Simply making people less nervous about signing up will make the official process work better.

meshko3• 13 years ago

Not even going to click to find out how stupid people are. Did they implement integration with data streams from hundreds of national insurance companies? Didn't think so.

theg2• 13 years ago

It is stunning just how much a lack of education and expertise went into this. This says nothing about HIPA, security, scaling, data source integration. It's staggering how much misinformation this PR stunt is going to cause.

misterjangles3• 13 years ago

Their site doesn't ask for any data except your zip code then forwards you on. Where does HIPA come into play?

otterley• 13 years ago

HIPAA is not implicated in the new health insurance marketplace. HIPAA protects personal medical information, and under the Affordable Care Act, you cannot be asked any personal health information on any insurance portals because insurers are required to cover any pre-existing conditions.

ddebernardy• 13 years ago

Now now now. They're only 20 but "have plenty of experience working at places like Twitter and Microsoft." Surely they \cough\ know what they're doing.

gruseom• 13 years ago

Your point may be valid, but you put it so maliciously that I winced when I read it. Smart but malicious comments are the worst thing about HN. They are like dioxins where stupidity is mere sewage. PG has written about implementing a comment filtering mechanism. I hope he does and that it will catch stuff like this.

I don't mean to pick on you; it's just a cumulative reaction, plus it happened to be the first thing I read this morning. I agree that back-end integration is likely orders of magnitude harder than UI for a system of this nature.

meshko3• 13 years ago

You are right. I was a little too annoyed.

_pmf_• 13 years ago

But ... but ... CSS3 transitions and WebGL surely are worth more than domain expertise!

misterjangles3• 13 years ago

What is that supposed to mean? Did you even look at the site?

minor_nitwit• 13 years ago

Of course not, the big government contracts did all that work, and they just sip it off from the top of healthcare.gov.

That said, they definitely did improve upon the government's front-end/implementation.

misterjangles3• 13 years ago

They dodged the hardest issues and politics by piggy-backing on the actual site to obtain the data. So you cannot really say that they did what the government contractors were unable to do.

But without a doubt they eliminated all of the cruft and whittled the site down to exactly what consumers want. I want the ACA to succeed and would love to see some variation of this "quick lookup" on the official site. The government would be wise to pay it some attention.

grey-area2• 13 years ago

To say they have completely replaced healthcare.gov would be stupid, but have the developers said that? I went and had a look at thehealthsherpa vs healthcare.gov. The people how made thehealthsherpa don't seem stupid to me - in fact they've identified the main pain point with the healthcare.gov website for the user, and tried to do something about it. The experience was as follows:

http://www.thehealthsherpa.com/

Enter your zip

Look at a list of plans with filters for my info.

Click one of the plans for info and get a phone no

I didn't have to fill in some long forms, I got straight to the info which counts without giving out any identifiable info.

https://www.healthcare.gov

Click on see plans now - this leads to a page of info with no plans information - great! Hit back.

Go back to home and click 'apply online'

Choose a state - read some info on the problems with this state

Click another apply online button

Click 'get started' to really get started this time!

Am confronted with a form asking for all my personal details, I stopped here.

Now of course that doesn't mean they have replaced healthcare.gov, or done in two weeks what all those developers working for the gov took $400m, but I'd say they've delivered something which is more valuable because it delivers what people actually want - a quick set of choices about healthcare for their specific area, and contact details for the same. It's a good summary of a complicated situation, presumably based on lots of the hard work done for healthcare.gov (assume they scraped the data).

You're not going to sign up for a plan without verifying costs anyway (by phone or on the providers website), so providing a quick summary with contact info is really all they need. I'm not convinced the apply online side of obamacare is actually helpful...

In the UK and quite a few other countries, this monthly insurance cost comes out of taxes and we don't choose an insurer, I think I prefer that single-payer solution.

kalleboo2• 13 years ago

The original healthcare.gov allowed for as they call it "window shopping", but a decree from the White House less than 2 months before launch demanded they change it to require a signup.

edit: fixed time period

pron• 13 years ago

They didn't build a healthcare.gov clone, but a useful site that can give people much of the information they need. It looks like a very good MVP, and it's absolutely clear that they didn't try to brag how easy it is to build healthcare.gov, but actually tried to help people until the gov site is fixed.

It is a beautiful and commendable project.

meshko3• 13 years ago

I don't understand how it can help if it is just a facade on top the healthcare.gov. Creating alternative sites like this can only create confusion for average users. I don't know their motivation, but the reporting definitely as an undertone of "look, all these professionals couldn't do it for millions of dollars, and 3 kids did it in a week for free".

butner• 13 years ago

Seriously -- this tool is equivalent to a cardboard cut-out of computer. It is embarrassing this is getting attention. They're dumping a publicly available spreadsheet that contains example age-group pricing... i.e. unless you're 27 or 50, it's giving incorrect information.

ollymorgs2• 13 years ago

The government contractors should of just built healthcare.gov as a data service and asked people like these guys to build an intuitive front-end for it. Then if people don't like the interface, they can do something about it :) This may also help to dissipate the traffic load and present the same data in more relevant ways to different types people. The idea of building apps as single, monolithic system is out-dated.

Even though I'm a Brit, some governments are making moves to modernise and open up their data for others to play with, over here we've got a great government-funded team working on exactly that (called the GDS). (http://digital.cabinetoffice.gov.uk/)

Frankly I think this is a massively expensive and missed opportunity for the US Government's digital service team.

drivingmissm• 13 years ago

Britain is far worse than the USA -- your health care system just flushed $18 billion down the loo on a massively failed IT system.

http://www.independent.co.uk/life-style/health-and-families/...

hga17• 13 years ago

A closer reading reveals that 11 billion (I assume that's thousand million) pounds was the projected cost at the time for completion of the system. "Only" 2.7 billion, or 4.3 billion USD, has been spent, although I don't know if there are contract terminations of serious size to be paid.

Still, it's an epic failure, said to be the greatest single one in government IT contracting.

(E.g. I'll bet the FAA's eternal quest to upgrade from their 1940's (sic) technique and '60s or so architecture has cost more over time, but not any single failed effort.)

ollymorgs2• 13 years ago

Yea we get stuff wrong too. That was a massive balls up, but well before the time of the GDS team.

Also note that the NHS system was about building a database for every NHS health record across the country; prescription notices, medial history, etc. etc.

I'm not sure it'll even be legal now. HealthCare.gov isn't going nearly as far as that.

hga17• 13 years ago

Well, the government contracts didn't have that level of control, political and bureaucratic types from the White House on down to HHS's CMS were running the show, delivering requirements very late in the game and making constant changes to them, including the "no window shopping" registration first in August, and from many reports change orders through the week before launch.

So you ought to say "The government should of just built healthcare.gov..." ... except that's very clearly not the style of this government.

E.g. one simple way to take traffic off most of the system would be to provide an interface to the subsidy calculation to entities beyond the state exchanges (who have to use it; when Verzion accidentally took the site offline, they couldn't do much), like insurers. It's said to be on the todo list, but at a very low priority.

semerda• 13 years ago

For an MVP this is a great start. Front-end code is clean, concise and the flow far better than the gov version. If the gov is looking for solutions to their problems these guys just showed what local talent can do vs outsourcing an important piece offshore and messing up.

robomartin3• 13 years ago

I must admit I have not devoted any time to understanding the real technical issues behind healthcare.gov. What little I know is from interpolating and extrapolating what I read here and elsewhere. One impression I got is that this $600+ million dolar website's authors tried to do it all. I don't get that at all.

So, here's what I would have done. This is completely sit-of-the-pants. I'm sure it has holes. Nothing that couldn't be sorted out by applying real thought to the problem beyond this quick and admittedly faulty definition:

My approach would have been to first develop a specification for an API all insurers would have to implement. They would be required to have a full implementation of this API in order to participate in the service. Now you have hundreds of teams working for you. This would cost almost not taxpayer money.

The next step would be to develop an automated test suite to test for API compliance. Open source it and release it. Setup a method through which this API could be used to certify compliance. Perhaps this would be an insurer's portal in healtcare.gov. Upon certification they would receive an expiring token identifying the plans submitted as being fully compliant with the API. More on this later.

Now everyone knows what they have to do to communicate their plans and how to make sure their API is fully compliant. No tokens. No participation.

Each plan has a token. The token is also the GUID for each plan.

Unit and regression testing are your friends. Insurers would do their own testing and they would be forced to continue to guarantee compliance by means of the tokens expiring at preset intervals (a year?).

Next the API would be released for anyone interested to implement health insurance browsing sites. Yup, let entrepreneurs battle it out for best-in-class discovery services. You would have hundreds of sites pop up with all kinds of really good solutions.

You would then publish another API anyone could use to actually communicate with government servers and "book" insurance.

Now you have a potential of tens of thousands of people working for you. And still virtually zero cost to taxpayers.

If that's too open, perhaps the plan browsing sites simply redirect you to a government site for the actual booking. They pass the token for the plan that was selected and a secure government site takes it from there.

Not done yet.

I would also hold a pre-release contest with cash prizes in the range of $1,000 to $100,000 for faults found during penetration testing of the government booking site. Release the code to the .gov site and let hackers mount massive attacks from every angle. Do this until top hackers agree the site is pretty damn hard. Give the top ten contributors cash as well as some kind of a presidential recognition at the White House. In other words, make it very public and very prestigious. Money and time well spent.

Once there, load test for something in the order of a million simultaneous users. The site must be built with elastic capacity as it will have to be able to manage insane traffic for a few months and then scale down to a more reasonable baseline.

Then open the doors.

However. This does not happen all in one shot.

Segment the population around data centers. Perhaps a couple of states on the east coast and a couple on the west. Announce it across all media. Schedule additional population segments across a period of a few months in order to average out a reasonable load on the system.

All independent browsing sites would have API access to the schedule. They could then save your selection and email/sms you when your turn comes up.

In other words, decouple the act of browsing and researching plans from actually booking it. And, at the same time, utilize a mechanism through which the rate of booking per unit time might be controlled. The browsing sites would simply not be able to forward to healthcare.gov unless it is your time slot. Now you have a situation where you simply can't have a hundred million people hitting healthcare.gov.

One could further segment booking by implementing time slots based on other criteria. For example, the last digit or two digits of your SSN.

Yet another approach would be to offer a limited number of booking time slots per hour. Browsing sites, through the API, would have access to this information. This means they could provide a virtual appointment system where you pick the time and date that is most convenient for you to actually book your plan. If a slot is available you are in. Otherwise you are offered alternatives. With this you could easily throttle the number of people who hit healtcare.gov per unit time. Again, healthcare.gov would NOT be a browsing site. You'd have to go through private sites for plan discovery.

Once booked, healthcare.gov would --via a pre-published API-- forward a completed and validated application to the insurance company.

I am sure further checks and balances would be required. And, as I said above, this game-plan surely has holes. It is seat-of-the-pants, I am coming up with it as I type. No way it is perfect.

However, there is one thing I know: The general idea here is to leverage the free market and private enterprise in order to delivery a massively better product. I realize this goes counter to the ideas the current administration must have about how a country ought to be run. However, if this healthcare disaster has done anything at all for all of us is to potentially prove, once and for all, that central planning and management by government is a horrible idea.

This is exactly how everything is done in government. We simply don't get to see it and experience it in daily life. With the ACA there was no way to hide it. This has been and continues to be a very unique opportunity for the American public to really see how bad government is about specifying, planning, managing and executing anything at all. If this makes people understand the fallacies in that school of thought this could very well have been the best $600 million we ever spent.

hga17• 13 years ago

You need to add to your seat of the pants concept how the government is going to control the payment of subsidies to prevent fraud. Those result in direct payments from Federal fisc to insurance companies....

Right now the Federal system is doing that for everybody, Healthcare.gov and all the state exchanges (single point of failure as Verizon's goof demonstrated). In theory insurance companies will get access, but it's said to be a very low priority.

Then there's, oh, the straight out fraud problem, i.e. you sign up, pay your first payment, then come January you or one of your fellow enrolees find out when you try to get services or prescriptions, the insurer you thought you signed up with has never heard of you. That pretty much requires starting at a blessed government site, I don't think your stopping at a government booking site would work since that could be spoofed, right? So could close the loop snail mail confirmations from insurers.

robomartin3• 13 years ago

That's why each plan has a GUID. This GUID would be authenticated when you go to healthcare.gov. They would not let you purchase a plan that does not authenticate. They are the ultimate authority and filter. They'd also handle details such as subsidies, etc.

In addition to that the transaction from healthcare.gov would, of course, ultimately reach the actual insurer who would have to further authenticate it, communicate with you and possibly also send some kind of an ACK to healthcare.gov.

The process could be full circle with many layers of confirmation and verification. This does not need to be overly complicated.

The idea is that the free market would handle the implementation of the massive first layer of a national system such as this one. Get government involved only where and when absolutely necessary.

xbeta• 13 years ago

As a Canadian living in the U.S. (SF bay area), I still don't understand why American make healthcare so complicated. At beginning, HMO vs PPO vs whatever makes me very confused. Why not just move to a public national health care that works for all?

wildgift6• 13 years ago

money * low information voters * paranoid messages = tea party

Also, around 80% of people are in private insurance plans through a plan at work. Some fraction of these people are happy with their HMO or PPO. These are the middle class and upper middle class or rich. They are also politically resistant.

The support for national health care or even single payer would come from the uninsured and people who have lower-end insurance (high deductible plans) and are forced into a kind of expensive PPO (insurance that makes you pay most of the costs of doctor visits).

Hilary Clinton's single payer plan was demolished by the insurance industry. So this time around, they kept insurance companies in the reforms. That negated some of the resistance from the well-insured.

meerita• 13 years ago

They bootstrapped a lot, that's nice :)

x00542• 13 years ago

This makes me so infuriated! This site is great, it's exactly what needed to be done, it works, and of course, it was NOT designed by our government, because it works, and that just makes way too much sense! I used the CA site before to look at available covarage. It's much more dificult to use, and it hides from me the more affordable "Catastrophic Plan" for which I do qualify, because I am under 30 years old. They do not want people to sign up for that plan, so they just hide it! This is outrageous beyond belief.

All this site now needs is the ability for the insurance companies to upload information about their plans directly to thehealthsherpa.com and we can forget about healthcare.gov all together. People signed up for years for insurance coverage by phone, directly with the companies, and they can continue to do so. The exchange should be just be an informational tool. This site does the trick.

I think thehealthsherpa.com should provide links, or better yet, add pages to the site directly, to explain the different plans and regulations for different states, so it can be more of an information hub. Great Job!

greg5green• 13 years ago

Why would you want to sign up for the "Catastrophic Plan"? From my calculations here in Ohio, you could have actual coverage for not much more?

x00542• 13 years ago

For me it comes out to $250/month for Bronze Plan, and only $170/month for Catastrophic Plan. I don't go to doctors unless I need to, and when I need to, I can afford paying cash. I can take a $6,350 bet on my health, but I still want to be covered above and beyond that.

And the question isn't about whether or not I should get catastrophic plan. I have the right to get it under the new law, and CA's Exchange site should have the duty to show me it as an option. Instead, they hide it, which is very underhanded.

wildgift6• 13 years ago

For other readers, here's a comparison of these plans. http://health.usnews.com/health-news/health-insurance/articl...

chucknelson• 13 years ago

A little skeptical of this, as others are. How was this tested? Just because it seems to "work" with responsiveness and items on the screen does not mean the info is correct.

Not saying healthcare.gov is any better, but there is just no mention of the information given and/or calculations done being correct.

hsitz• 13 years ago

I'm assuming the task these guys have done is tiny portion of the actual work needed for the website.

I've been wondering for a while, though, why the federal government can't just buy/beg a copy of software running one of the exchanges for one of the states doing things themselves. Then they could modify it to work for other states. Maybe they'd then have to replace existing federal exchange with 26 different instances (one for each state that's a member of the federal exchange), but I can't imagine that the technical difficulties of modifying one state exchange's software to run with another state's data would be as bad as fixing what seem to the problems with the existing federal exchange website.

hga17• 13 years ago

Probably too late for the very hard mid-December deadline for those losing their insurance, maybe too late for getting enough young healthy people enrolled before adverse selection forces insane 2015 premium filings from insurance companies, which are due by the end of March.

They also have only 4 well functioning states' systems to chose from as I understand it.

The centralized subsidy calculation/verification/etc. process, which the Federal system does for all the exchanges is another issue, although I've only heard that some time ago is was known to produce incorrect results under testing.

Also would require the Federal government to admit a much bigger screwup than they've been willing to so far, in a process entirely driven by politics (http://www.washingtonpost.com/politics/challenges-have-dogge...) until the last week of last month when CMS was fired and the fix-it czar was appointed.

If you've been in this game long enough you've seen projects and companies destroyed by a failure to take decisive action when it could have made a difference (or read, for example, Churchill's The Gathering Storm, or any shorter version of how WWII could have been so easily averted by preventing German rearmament). The odds are pretty high this is going to be another example of that.

wildgift6• 13 years ago

Or allow one state to operate another state's exchange.

The states were given federal money to build their state ACA programs... so the government already owns it.

Now, imagine if they had been forced to open source this stuff, at least between the contractors.

S.F. programmers build alternative to HealthCare.gov [video]

Related Stories