The folks at Development Seed are doing really awesome work between Prose and MapBox and it's great to see them get some recognition for it.
Also a great plug for Jekyll in general. While there can certainly be more worrisome factors than cost for a site, site performance should be one of the most important issues when launching. By completely getting rid of a backing database, page lookup times can be reduced dramatically. As Prose continues to mature, it's getting more and more flexible by the day. Just last week they implemented an extensible "tagging" feature for posts: https://github.com/prose/prose/issues/464.
In case you haven't noticed, I'm a fan. Kudos to dev seed and big props to the White House team for the willingness to move away from Drupal.
Agreed - this is great work and I'm really happy to see Government moving in this direction. Key point for Federal leadership here (in addition to the cost savings) are that a small team of developers and designers working together using open source tools can build great products that scale to meet the demands of government. Helps that the team seems very capable!
It's a risk for any CIO / CTO to kick off a project like this in government, and I applaud HHS for taking the step.
The folks at Development Seed are doing really awesome work
I am also a member of the Development Seed fan club, having met many of the team members. Though, I feel like they should update their PR statement. Every time they are mentioned in an article, the Drupal past comes out: The startup first made its mark in the DC tech scene consulting on Drupal I think we can all agree that mentioning one's history working with Drupal is not going to convey a feeling of cutting edge work by a, frankly, team of brilliant engineers like Dev Seed. I have done work with Drupal in the past, yet it does not appear on my resume.
Instead, they could be saying something like: Development Seed, makers of Mapbox and Tile Mill, which directly competes with Google Maps products, and in many ways is better, develops and promotes technologies and methodologies reminiscent of Bay Area start-ups.
These guys really do not get enough credit, even with a minor mention in the Atlantic article.
Their history with Drupal is a key part of their value story in DC, where Drupal is cutting edge--it is rapidly replacing expensive proprietary CMS solutions for many federal agencies and nonprofits.
Although it might be hard to believe here on HN, saying something is "reminiscent of Bay Area start-ups" means jack squat to the folks in DC who spend a lot of money on web projects.
Last week, the United States Department of Health and Human Services (HHS) relaunched Healthcare.gov with a new appearance and modern technology that is unusual in federal-government websites.
The article seems to be about a promotional/propaganda CMS for the government, am I missing something?
"This is such a lean site," said Jon Booth
"This is our ultimate dogfooding experience," said Eric Gundersen
"The work that they're doing is amazing," said Sivak
the site is just one component of the insurance exchanges. Others may not be ready by the October deadline.
It is definitely amazing how brilliant and successful every decision on a project turns out to be when you only interview the people who made those decisions!
I didn't just interview the people who made those decisions. I've talked to dozens of federal and state CIOs over the years, along with many contractors and Web developers. The consensus amongst insiders and industry observers is that this approach and outcome are genuinely novel and innovative in the government space, which is why I wrote the feature to begin with, starting months ago.
This would have been more evident if any of those people had been quoted in the story. It also would have made the story more compelling -- as it stands, it reads like a bunch of colleagues congratulating each other on what fine work they have done.
Fair point. And ouch, I think. Just so it's clear, I'm not their colleague -- and the contractors in question weren't obligated to praise CMS. I'd hoped that readers would find the perspective of the writer useful, in addition to the comments of the principals, whose interests are obvious. Thank you for the feedback.
"and the contractors in question weren't obligated to praise CMS"
If you are a government contractor and some reporter asks you about a government project, the response is going to be positive (or at least obfuscated) unless you would like to have no more government contracts.
Sure, propaganda can be educational. I read white papers, ads or other industrial propaganda just like anyone. My first word was promotion and I added propaganda afterward, maybe it added too much negativity, sorry.
On the other hand, when reading the article I was disappointed that the article wasn't about the actual exchanges to facilitate finding and purchasing insurance. It was about building a CMS to help the government communicate about [ vaporware (too negative) | something coming soon (rose colored glasses on)].
[Adding more:]
Since you seem to be identifying as the author, thank you for sharing this information about the site. It must have been very interesting to see the modern web dev meet gov.
Adding/editing content by github pull request isn't tremendously impressive. I'm sure the Centers for Medicare and Medicaid Services (CMS) will appreciate volunteer effort. However, pull requests aren't an avenue for contributions by people outside of a very limited population. It isn't a community/software moderated hive-mind like Wikipedia or StackExchange and I think that the difference goes back to the word propaganda.
Education takes many forms and propaganda is a description of education at its most one-sided and didactic. This website will have the information that the CMS thinks will (and wants to) use for many purposes including "educating the uninsured." As modern as the technology stack is, the human stack did not change. I'm sure there will be non-governmental initiated wikis and Q/A websites for people to cooperatively educate themselves about the real deal, so a fresh coat of paint on the status quo isn't so bad.
Thank you for the additional clarity on what you meant! Yes, I wrote this. I'd like to tackle a piece on the exchanges themselves too, at some point, but that would be another feature, not least because of the diversity of state approaches and the fact that the federal version isn't done yet.
tbf, there is a good social engineering function to propaganda. Propaganda is not necessarily false, it just can be. If you look at the roots of the word (which originates with the Jesuits) it has the same meaning as "to go viral" in modern Internet speak. To propagate, to organically spread and grow on its own.
My father, who worked with MLK and the SCLC referred to his work in the civil rights and union movements as being "a propagandist for the right reasons" -- a dry jest that should have made your brain trainwreck instantly if you are paying attention. He was an incredibly smart and funny -- and very ethical man. I inherited the family business, which is why I ended up as founding executive director of the Tor Project. You of course are welcome to judge that bit of work on its own basis. I didn't write the software, I just took it from a framing as a gray hat cypherpunk project to being the acclaimed champion of human rights, journalism, democracy rights activists, and apple pie in about six months flat along with seven figures of funding most of it from the US government without which we'd probably be dead meat from the NSA now (IMNSHO, and I don't speak for the current project's management -- only a volunteer these days).
Ultimately, my point is, I've worked with folks in government, and around Alex. And I've worked around these health care exchanges from RomneyCare here in MA.
The fact that they can relaunch, have a database like this work, even get this decision through, is pretty damn revolutionary. That it worked and they didn't get immediately hosed for it -- you have no idea how risky this was for them on an eGov basis, especially in "obamacare" where they have targets on their back for everything.
Give the dudes credit for a little excessive celebration. It's not propaganda -- and if it were...?
The use of Jekyll (or any static HTML generator) is a really huge, positive conceptual step. I mean, we can now make edits -- in Markdown -- to individual blog posts and submit pull requests.
Whether they actually honor those requests* is another thing, but this is a really cool step. Even the use of Markdown means that the site/content designers have to really work to obfuscate the content with unnecessary modules/widgets/etc.
Regardless of the merits of the administration's policies and initiatives, I hope this catches on. 95% of the government websites I've seen, from city council to federal, could be rendered in static HTML...and maintained much more easily.
The huge leap that has to be made for Federal Government CIOs is the psychological attachment they have made between software that works and huge price tags.
Sometime, long before I ever came along, CIOs in the Federal sector became convinced that the only way for software to work was for it to be big, sold by SAP/Microsoft/Oracle, and have a huge pricetag. Anything else is a huge risk in their world.
They just don't believe that something that takes so little money, man-hours, and time to implement can possibly satisfy their requirements and replace such a huge portfolio of bloated legacy systems.
I built a metadata management system in a few weeks as a "rapid prototype" for a branch of the US military. It satisfied all of their requirements, was fast, secure, and stable. It was storing all of the production data and was made to be production ready by simply flipping a switch. Then the CIO saw it.
"Wow, this is amazing. Can you give the code to Team X so they can build it in Sharepoint?" 8 months later, a stunted, awful version of my prototype was demoed. The thick-client interface was gone, replaced by the awful, non-responsive Sharepoint bullshit. The Postgres back end that cost nothing was replaced with an expensively licensed SQL Server DB. The Google-like "search as you type" capability that was in my "prototype" was gone. Infinite scrolling in the grids was gone. Did I mention that they force their users to use Internet Explorer 8 because of "security reasons"? Never in my life have I witnessed such a huge level of IT incompetences as that on display in the United States military. This is almost surely due to endless bloated budgets which allow wankers like this to continue to keep their jobs.
So, does this mean they learned some important lessons from the $18 million recover.gov site?
As for Percussion, is the Sivak person saying that it takes a month to make a simple navigation change? Is this really about the CMS or the red tape involved in getting approval to make the change?
There's a lot to be learned from Recovery.gov, not all negative. In this case, the HHS CTO was referring to a change in the CMS itself, not the website(s) it controls. Getting code updated in the normal .gov contractor/sub-contractor model would generally take more time, both in terms of requirements and implementation.
I am skeptical. Seems more in the implementation of the CMS. Not a fan of percussion, but I have seen drupal implementations that tied content to presentation also take an stunningly long amount of time for a navigation change. Bad design. Bad implementation.
Sorry, not feeling very patriotic about this. "by the people, for the people, released back to the people." Meanwhile those of the people who are physicians have had their fees sequestered by 2% http://www.medscape.com/viewarticle/781630
and are awash in extra paperwork for trying to prescribe something as simple as diabetic test strips as funding is tightened: http://www.pharmacist.com/cms-cuts-reimbursement-diabetes-te...
I'm very eager to see how they handle the actual exchange code.
Jekyll and static HTML make a lot of sense for a marketing/informational site (which this appears to be right now), but at some point, they're going to need a dynamic backend (right?). These exchanges will need to aggregate plan info from lots of different insurance carriers and display plans to people based on various bits of user-specific info. Perhaps they can generate static info pages for all of the plans and redirect people to these pages via JavaScript and then pass the actual enrollment off to the insurance companies.
If they have to support enrollment and dynamic content, it
will be interesting to see how they do so while adhering to their principles of simplicity.
If they don't, it will be an interesting example of how to embrace simple solutions where possible while preventing scope creep.
Either way, I'm glad to see this type of innovative development happening.
One "website" can have multiple backends: static files prepared with Jekyl for the CMS part of it, and a full backend for any dynamic bits. You can feed both in the same page with judicious use of AJAX.
Of course. And if they do implement such a JavaScript-dependant approach, it will be entirely different from any other government site. As I said, I look forward to seeing how they progress.
This reads like the worst marketing mashup this year:
"We knew there were performance and reliability benefits from building the stack on HTML."
I know there are performance and reliability benefits from talking in English.
"You're just talking about content. There just needs to be one server. We're going to have two, with one for backup. That's a deduction of 30 servers"
I'm very happy the government decided to skimp on reliability for a public healthcare market.
"You can put Wordpress into Amazon but then you'll have problems; it wasn't designed to be a cloud application."
Wait wait wait Wordpress isn't a cloud app? Fuck, someone go tell Hostgator that--apparently it can't be done!
"Notably, the HHS team has been using feedback to drive design choices from the beginning, applying the 'lean startup' model to development."
Wow, even the lean startup namedrop.
I'd be a lot happier if, instead of putting the Obama webdev team on the gravy train, they spent that same money--you know--actually fixing healthcare. As it is they seem to be just doing free advertising and lead generation on the public dime for the insurance companies.
To your substantive point, no one skimped on reliability here. The content was pushed into Akamai precisely so it would always be available, quickly, to end users. Reducing servers in this context isn't skimping -- it's saving money by using modern Web tech. Developing this site also wasn't an option, as your comment implies: the law required its development and deployment. If you'd like public funds directed elsewhere, call your Congressional representatives.
but that doesn't seem super innovative. others in government have used akamai as origin servers--TSA then DHS did this in 06 and got rid of a bunch of servers.
From what I know of the project, CMS has a GitHub enterprise account that they used internally to "iteratively create on github," this Repo is the one that moved over after release.
It's a big step to have government releasing ANYTHING on Github. And yes, it's the next logical step for them to start projects on GitHub. But they're not there yet -- its worth noting that neither Sunlight nor Code for America start their projects out in the open 100% of the time either. It's new ground, and government will, I think, move more towards developing out in the open as we progress. But for now, I'm pretty excited that they embraced technology from this decade.
I should note I'm the former Director of Sunlight Labs, and am an advisor to Code for America.
As I understand it, that's correct: built using Github Enterprise internaly and then published publicly. There was code in public, on the Development Seed account, with respect to coding for the project; anyone can go look at the commits there. Your other points speak for themselves.
Did GitHub go through FISMA certification for their public site? How are these govt agencies using the system to host code without getting the platform accredited?
41 comments
[ 5.8 ms ] story [ 94.1 ms ] thread(The article buries the link to the page with the link: https://www.healthcare.gov/developers )
Also a great plug for Jekyll in general. While there can certainly be more worrisome factors than cost for a site, site performance should be one of the most important issues when launching. By completely getting rid of a backing database, page lookup times can be reduced dramatically. As Prose continues to mature, it's getting more and more flexible by the day. Just last week they implemented an extensible "tagging" feature for posts: https://github.com/prose/prose/issues/464.
In case you haven't noticed, I'm a fan. Kudos to dev seed and big props to the White House team for the willingness to move away from Drupal.
It's a risk for any CIO / CTO to kick off a project like this in government, and I applaud HHS for taking the step.
I am also a member of the Development Seed fan club, having met many of the team members. Though, I feel like they should update their PR statement. Every time they are mentioned in an article, the Drupal past comes out: The startup first made its mark in the DC tech scene consulting on Drupal I think we can all agree that mentioning one's history working with Drupal is not going to convey a feeling of cutting edge work by a, frankly, team of brilliant engineers like Dev Seed. I have done work with Drupal in the past, yet it does not appear on my resume.
Instead, they could be saying something like: Development Seed, makers of Mapbox and Tile Mill, which directly competes with Google Maps products, and in many ways is better, develops and promotes technologies and methodologies reminiscent of Bay Area start-ups.
These guys really do not get enough credit, even with a minor mention in the Atlantic article.
Although it might be hard to believe here on HN, saying something is "reminiscent of Bay Area start-ups" means jack squat to the folks in DC who spend a lot of money on web projects.
http://digital.cabinetoffice.gov.uk/
https://www.gov.uk/
https://www.gov.uk/designprinciples
The article seems to be about a promotional/propaganda CMS for the government, am I missing something?
"This is such a lean site," said Jon Booth
"This is our ultimate dogfooding experience," said Eric Gundersen
"The work that they're doing is amazing," said Sivak
the site is just one component of the insurance exchanges. Others may not be ready by the October deadline.
If you are a government contractor and some reporter asks you about a government project, the response is going to be positive (or at least obfuscated) unless you would like to have no more government contracts.
On the other hand, when reading the article I was disappointed that the article wasn't about the actual exchanges to facilitate finding and purchasing insurance. It was about building a CMS to help the government communicate about [ vaporware (too negative) | something coming soon (rose colored glasses on)].
[Adding more:] Since you seem to be identifying as the author, thank you for sharing this information about the site. It must have been very interesting to see the modern web dev meet gov.
Adding/editing content by github pull request isn't tremendously impressive. I'm sure the Centers for Medicare and Medicaid Services (CMS) will appreciate volunteer effort. However, pull requests aren't an avenue for contributions by people outside of a very limited population. It isn't a community/software moderated hive-mind like Wikipedia or StackExchange and I think that the difference goes back to the word propaganda.
Education takes many forms and propaganda is a description of education at its most one-sided and didactic. This website will have the information that the CMS thinks will (and wants to) use for many purposes including "educating the uninsured." As modern as the technology stack is, the human stack did not change. I'm sure there will be non-governmental initiated wikis and Q/A websites for people to cooperatively educate themselves about the real deal, so a fresh coat of paint on the status quo isn't so bad.
My father, who worked with MLK and the SCLC referred to his work in the civil rights and union movements as being "a propagandist for the right reasons" -- a dry jest that should have made your brain trainwreck instantly if you are paying attention. He was an incredibly smart and funny -- and very ethical man. I inherited the family business, which is why I ended up as founding executive director of the Tor Project. You of course are welcome to judge that bit of work on its own basis. I didn't write the software, I just took it from a framing as a gray hat cypherpunk project to being the acclaimed champion of human rights, journalism, democracy rights activists, and apple pie in about six months flat along with seven figures of funding most of it from the US government without which we'd probably be dead meat from the NSA now (IMNSHO, and I don't speak for the current project's management -- only a volunteer these days).
Ultimately, my point is, I've worked with folks in government, and around Alex. And I've worked around these health care exchanges from RomneyCare here in MA.
The fact that they can relaunch, have a database like this work, even get this decision through, is pretty damn revolutionary. That it worked and they didn't get immediately hosed for it -- you have no idea how risky this was for them on an eGov basis, especially in "obamacare" where they have targets on their back for everything.
Give the dudes credit for a little excessive celebration. It's not propaganda -- and if it were...?
Triangulation sucks too.
Whether they actually honor those requests* is another thing, but this is a really cool step. Even the use of Markdown means that the site/content designers have to really work to obfuscate the content with unnecessary modules/widgets/etc.
Regardless of the merits of the administration's policies and initiatives, I hope this catches on. 95% of the government websites I've seen, from city council to federal, could be rendered in static HTML...and maintained much more easily.
* edit: Only one way to find out: https://github.com/dannguyen/healthcare.gov/commit/6438498ea...
Sometime, long before I ever came along, CIOs in the Federal sector became convinced that the only way for software to work was for it to be big, sold by SAP/Microsoft/Oracle, and have a huge pricetag. Anything else is a huge risk in their world.
They just don't believe that something that takes so little money, man-hours, and time to implement can possibly satisfy their requirements and replace such a huge portfolio of bloated legacy systems.
I built a metadata management system in a few weeks as a "rapid prototype" for a branch of the US military. It satisfied all of their requirements, was fast, secure, and stable. It was storing all of the production data and was made to be production ready by simply flipping a switch. Then the CIO saw it.
"Wow, this is amazing. Can you give the code to Team X so they can build it in Sharepoint?" 8 months later, a stunted, awful version of my prototype was demoed. The thick-client interface was gone, replaced by the awful, non-responsive Sharepoint bullshit. The Postgres back end that cost nothing was replaced with an expensively licensed SQL Server DB. The Google-like "search as you type" capability that was in my "prototype" was gone. Infinite scrolling in the grids was gone. Did I mention that they force their users to use Internet Explorer 8 because of "security reasons"? Never in my life have I witnessed such a huge level of IT incompetences as that on display in the United States military. This is almost surely due to endless bloated budgets which allow wankers like this to continue to keep their jobs.
As for Percussion, is the Sivak person saying that it takes a month to make a simple navigation change? Is this really about the CMS or the red tape involved in getting approval to make the change?
Jekyll and static HTML make a lot of sense for a marketing/informational site (which this appears to be right now), but at some point, they're going to need a dynamic backend (right?). These exchanges will need to aggregate plan info from lots of different insurance carriers and display plans to people based on various bits of user-specific info. Perhaps they can generate static info pages for all of the plans and redirect people to these pages via JavaScript and then pass the actual enrollment off to the insurance companies.
If they have to support enrollment and dynamic content, it will be interesting to see how they do so while adhering to their principles of simplicity.
If they don't, it will be an interesting example of how to embrace simple solutions where possible while preventing scope creep.
Either way, I'm glad to see this type of innovative development happening.
"We knew there were performance and reliability benefits from building the stack on HTML."
I know there are performance and reliability benefits from talking in English.
"You're just talking about content. There just needs to be one server. We're going to have two, with one for backup. That's a deduction of 30 servers"
I'm very happy the government decided to skimp on reliability for a public healthcare market.
"You can put Wordpress into Amazon but then you'll have problems; it wasn't designed to be a cloud application."
Wait wait wait Wordpress isn't a cloud app? Fuck, someone go tell Hostgator that--apparently it can't be done!
"Notably, the HHS team has been using feedback to drive design choices from the beginning, applying the 'lean startup' model to development."
Wow, even the lean startup namedrop.
I'd be a lot happier if, instead of putting the Obama webdev team on the gravy train, they spent that same money--you know--actually fixing healthcare. As it is they seem to be just doing free advertising and lead generation on the public dime for the insurance companies.
To your substantive point, no one skimped on reliability here. The content was pushed into Akamai precisely so it would always be available, quickly, to end users. Reducing servers in this context isn't skimping -- it's saving money by using modern Web tech. Developing this site also wasn't an option, as your comment implies: the law required its development and deployment. If you'd like public funds directed elsewhere, call your Congressional representatives.
https://github.com/CMSgov/healthcare.gov/commits/master
Still cool its open sourced though.
It's a big step to have government releasing ANYTHING on Github. And yes, it's the next logical step for them to start projects on GitHub. But they're not there yet -- its worth noting that neither Sunlight nor Code for America start their projects out in the open 100% of the time either. It's new ground, and government will, I think, move more towards developing out in the open as we progress. But for now, I'm pretty excited that they embraced technology from this decade.
I should note I'm the former Director of Sunlight Labs, and am an advisor to Code for America.