I'm a (non-US) federal public servant who worked for a small, and now defunct, agency with the mandate to embed 'user-centric' design principles in the creation of public services. At the time we were wound up, I concluded we had failed. Now I'm less sure. I still occasionally see instances where our work has influenced other public servants, so perhaps we've had some measure of longer-term success. You should carefully think about and define your metrics for 'success', both in the short and long term.
All that aside, the key reason for poor design in government is this: lack of incentive. And no, the ballot box is not an effective incentive; very few voters will punish a government because their tax return form was poorly designed. Private companies possess this incentive: if your product is hard to use, people won't buy it and you'll go out of business. Government doesn't have this incentive: if people find your products hard to use, they'll use them or they'll (eventually) go to jail, won't get their welfare payment, won't receive a driver's licence etc.
Outsourcing or privatising these kinds of services is not the answer: you're just substituting a public monopoly for a private one, which arguably leads to worse outcomes for citizens (and great ones for the private monopolist). This kind of action only makes sense if a competitive market will form after government gets out of the way. It's helpful to think of government services in this fashion: a service market that is dominated by a monopolist.
So that's the problem you are attempting to solve: how do you get a monopolist with no profit incentive to design better products and services?
Could a setup help with this unfortunate incentive situation, wherein independent frontend providers would get a small, fixed fee from the state for each case handled through their systems? They would invariably compete for users on the usability front. And if those fees (oh no, fixed! Where is that sacred free market?) turn out too big it would not be as catastrophic as one may think, because they would not just get filthy rich/reinvest in shady lobbying (like traditional government contractors, e.g. military technology), but would put much of the excess money into advertisement, keeping the bullshit-job flywheel spinning that our postindustrial societies are relying on to keep an acceptable number of people employed.
I see a few potential pitfalls, but it should be doable and cheap. The meta-system would have to be balanced between two failure modes. One would be to allow frontend providers to optimize their revenue by case-inflation, like making their users run to that virtual tax office more often than strictly necessary, e.g. by shoddy implementation of corner cases. The other would be discount providers specializing only on the easy cases, making it unnecessarily hard for more complete providers to make their living.
I've often wondered about similar things, but admittedly haven't done any formal analysis.
I suspect it's situation specific: there are probably situations where a model like this would work very well, and others where it would be a terrible idea. I'm also a little wary of government created 'pseudo-markets', as these often become bureaucratic nightmares that cost more to administer than if the government had just done the thing itself. 'Pseudo-markets' are kinda like machine-learning algorithms: they'll optimise towards your success metric, but sometimes in unexpected and undesirable ways (i.e. game the system).
Also, for certain services, you'd need to ensure equity of access (i.e. cover all use cases). Tax returns are a good example. There's some pretty esoteric stuff in tax returns that are only relevant to a handful of people (e.g. reporting franked dividends distributed from closely-held unit trusts via an interposed corporate entity, or whatever). It might never be profitable for a private market to cover this case, meaning the government would have to further 'pseudo-regulate' its 'pseudo-market', offer further subsidies, or cover this use-case itself.
Interestingly, on your second 'failure mode', you could flip it around and view it as a desirable outcome. Sticking with income tax reporting, most reporters have pretty simple tax affairs. They just report their annual income (which the government already knows), maybe claim a deduction or two, and that's it. But because the government must cover every possible use case, people are forced to wade through a 40 page form instead of a 1 page form. There might be more gain to society from doing this:
1) Government offers the fixed subsidy, but makes clear it is only guaranteed for, say, 2 years.
2) After 2 years, offers a lower (or no) fixed subsidy for the 'cherry-pickers' (who will still be profit positive)
3) Reallocate the savings as higher fixed subsidies for the remainder of the market
By iterating this process a few times, the market would naturally segment according to complexity, allowing the government to accurately 'price discriminate' on the basis of complexity. I dunno, I'm just spit-balling here, no clue if this is actually a good or bad idea. What are your thoughts?
Of course, all of this is only possible if the government publishes a 'tax return' API that's easy to use (i.e. does not create high implementation costs for private providers). Even though publishing an API sounds (and, frankly, is) simple, you'd be astounded at how often the government screws this kind of this up (often by contracting out to IBM, Fujitsu and their ilk). Or not publish one at all, even though the potential benefits are blindingly obvious...
I see two differences hidden between much overlap: expert/layman and embrace constant retuning/embrace creeping detuning. Since we are only taking about allowing investors to tap into possible efficiency gains in the execution of bureaucratic processes (and not into actual resource allocation, as it happens for example in highly regulated but not fully state-run healthcare systems or in renewable energy programmes), there is a natural upper limit for disoptimization. So in a way this problem is much easier than other, very similar regulation/gaming the system scenarios and a loose reins approach should be less risky.
Indeed. Much of what I see from the UK GDS is bang on; there are clearly some very bright people there. Take a look at their guidance to other government agencies on exposing APIs: https://www.gov.uk/service-manual/making-software/apis.html
It's concise, in plain English and very cleverly avoids using the terms REST or RESTful even though they're basically describing RESTful architectural constraints. It's clever because they avoid getting dragged into religious wars about API technicalities and provide flexibility for situations where fully REST-compliant APIs are not practical.
I wish more public servants (and politicians) understood web APIs and the economic value that would be unleashed if governments exposed more (both data and transactional). But instead, most are obsessed with building 'portals' and 'mobile apps', because those are 'innovative'!
Agreed - and echoes a similar case I read about when the folks behind the Ordnance Survey (government agency and makers of truly exceptional maps for outdoor enthusiasts in the UK) proudly showed how they made their data online for a licensing fee that generated about £10m a year.
The people they were presenting to? A group of European governments that endorse massive, free access to this sort of data - as a means of spurring innovation and economic growth.
Now contrast that with London Mayor Boris Johnson's movie to open up the data for their city bike sharing scheme... released on a Friday, by Monday there were two apps that were slicker and better-developed than the city was planning, plus were now two small businesses.
Yeah the whole 'charging for government data' thing drives me crazy!
In almost all cases, the marginal cost of producing the 'good' (data) is approximately 0. Furthermore, because infinite copies of data can be made, there is no reason for price rationing. And there's no argument for granting monopoly through artificial scarcity (like there might be with patents and copyright) as the government would have produced the data anyway (often as a byproduct of some other activity e.g. income tax reporting).
Government data is a public good, in the strict economic sense (non-excludable, non-rivalrous).
Setting prices above marginal cost (i.e. 0) causes significant dead-weight loss; it is absolutely inexplicable and inexcusable public policy. This isn't complex economic theory. This is economics 101; you'll find this information in any first-year uni economics textbook (in the 'market failure' chapter). I can only conclude that public servants who do this kind of thing either aren't familiar with extremely basic public economic theory (which is a bad sign), or are actually trying to reduce public welfare (making the public service a rather odd career choice).
Conceptually, charging for government data is equivalent to levying a super-narrowly based and highly inefficient sales tax. The most baffling thing is that it's conservative governments that generally approve these kinds of policies, even though the net effect is to increase the size of government at extremely high efficiency costs.
> For example, the Department of Defense and Department of Veterans Affairs spent $1.3 billion on a program to build an electronic health-records database and abruptly stopped the project in 2013 after it failed to progress as planned. More staggering? In the past 10 years, about 96 % of all government IT programs that cost over $10 million were deemed failures, meaning they didn't meet their budget, timeline, or user expectations.
If this startup (and ones like it) is successful in getting traction taking over some of these projects, there are going to be a lot of very unhappy IT companies that were riding the gravy train for decades.
> A lot of the work we did last year for retooling the Healthcare.gov application process was figuring out which questions were necessary to ask of everyone and which ones were only necessary for certain people. Instead of having one online form with dozens of entry fields on a single page, the new Healthcare.gov application process asks a few general questions—like income and household size—then directs you to more specific questions based on your replies.
Jesus. You'd like to think there'd be some minimum baseline of common sense required to design an application of this level of importance.
> The newfound optimism about the government's technical future is inspiring, but can a 10-person startup really make a difference?
Obviously. You'd almost have to be trying to screw it up as bad as the original team did, unless this story isn't an accurate portrayal.
> If this startup (and ones like it) is successful in getting traction taking over some of these projects, there are going to be a lot of very unhappy IT companies that were riding the gravy train for decades.
While $1.3 billion is a massive amount of money to spend on a project, the article summarizing that effort as being "an electronic health-records database" is incredibly trite. This characterization implies "some RDBMS someplace that holds the information." The scope of trying to bring all parties involved into a set of computer systems collaborating in workflows which would make Descartes[1] blush is immense.
Stating that the tens of thousands of people working on that effort are "riding the gravy train for decades" is naive at best.
Sorry but as someone with many decades of experience with various levels of self-interest and corruption, I don't buy your casual "it's a hard problem" excuse.
> In the past 10 years, about 96 % of all government IT programs that cost over $10 million were deemed failures
Would you have us believe these were all hard problems?
> > In the past 10 years, about 96 % of all government IT programs that cost over $10 million were deemed failures
> Would you have us believe these were all hard problems?
First, the statistic the article references appear to be from a Computerworld article[1] which uses "The Standish Group" analysis. This analysis indicates a 96% failure rate for all IT programs costing more than $10 million and is not limited to "government IT programs." In other words, the article failed to meet their user expectations of due diligence in representing a pivotal metric such as this.
Second, as the size of an effort increases in personnel, of course there will be inefficiencies. Waste, self-interest, and possibly corruption as well. But you specifically used the phrase "unhappy IT companies that were riding the gravy train", implying ulterior motives of those organizations and discounting the difficulty involved in a healthcare records project involved two huge organizations.
I can't see things improving any until they fix the bidding process for government contractors, and that would require congress to fix it. I can't see that happening. The Congress as a whole, left or right, has an army of corporate campaign contributors that benefit from keeping the system the way it is, a labyrinthine mess only certain, usually larger or specialized, players have the time and legal expertise to navigate.
It's another problem where the folks most able to fix things are unwilling because they benefit from keeping the status quo or have no incentive to take risks. Many, many politicians in DC have private investments among contractors.
But even if all of these contractors and politicians were removed, the inertia of public sector work is immensely bad still across the government. Lack of correctly aligned funding, poor morale, and a huge body of highly invested folks that just wanted a paycheck and don't want to do more than the minimum. The attitude is very rational when change is so impossible over decades that it'd drive you insane to try to change anything of consequence. This is why I quit public sector work entirely - I care too much to spend years, decades of my life pushing a stubborn mule that's gotten complacent and has been beaten and demoralized among a lot of its life. This doesn't mean that it's like this absolutely everywhere but it is the case for almost everyone without lots of connections / credentials to get you into the actually good projects with actually good people with actually good leadership. But why should I try so hard just to have an acceptably enjoyable job when I can try a lot less in private sector and get something pretty reasonable right away?
Hi! We don't have a jobs page up yet, so I'll leave this here in case anyone's interested in working with us (Nava)
Nava | Washington DC* | Experienced full-stack developers/devops/product manager/operations | On-site - Full Time
We're a small team of engineers from Silicon Valley that came out to DC last year to help fix Healthcare.gov. It turns out there’s a lot more to fix. And it’s surprising how much can be fixed by a small group of resourceful people with a Silicon Valley mindset, deep technical experience, and the willingness to work closely with dedicated civil servants in government.
Our revamped Healthcare.gov application has been used by millions, converts 35% better, and halves the completion time. The login system we rebuilt is about two orders of magnitude more reliable and two orders of magnitude less expensive; for example, it’s about $70M less per year to operate. We’re just getting started, and we’ve started Nava to help fix everything else. [0]
People die because the Veteran's Administration is months behind in processing claims. The Social Security Administration pays benefits to millions of deceased Americans. $80 billion is spent every year on federal IT contracting, and 96% of projects are deemed failures [1].
That’s not because there’s some conspiracy or because government is inherently incapable of doing it right. These are complicated legacy systems and processes, and there are very few people with modern tech industry experience who are aware of these problems and willing to help fix them. You can help change that.
Our team is 10 people (Stanford, Google, YC alums), and we plan to bring on a few people every month through 2015.
We’re looking for:
- experienced full-stack engineers
- experienced devops engineers
- a product manager with a technical background
- a hyper-resourceful operations person
We have a social mission (we just incorporated as a public benefit corporation (PBC) this week), but we pay market compensation (above market, for DC) and equity (above market).
If you'd like to build software and infrastructure that radically improves how our government serves people, we’d love to hear from you at jobs@navahq.com.
*Not in DC / able to relocate, but intrigued and in SF? Talk to us.
I think what you're doing is admirable and just wanted to wish you good luck! I've been fighting similar battles within government (non-US) and it can be both very frustrating and very rewarding. Frustrating because 'being right' is almost never the way to win the argument.
Although I'm terrible at this myself, I've observed that opposition to change is best overcome by persistence. The best change agent I've worked with used a particular (and highly effective) technique that I've since termed 'objection exhaustion'.
He'd hold a meeting of all the 'major stakeholders' (read: people who's stuff we were messing with, but who's co-operation we needed) once a week/fortnight to update them on our progress (read: grind them down into submission).
When someone raised an objection to our ideas, he'd listen carefully, take their concerns seriously and commit to examining the issue and bringing findings back to the next meeting. We'd then do that, and most of the time present evidence at the next meeting that showed the objection was baseless (in a nice, 'give them an out', kind of way). Tedious as it was, by doing this over and over we'd exhaust all their objections and they'd be forced to agree with us (or continue to object for no apparent reason and look like a mindless fool in front of their senior colleagues).
Depressingly, the true but hidden objection is usually "don't touch my stuff" or "don't reduce my budget". Objection exhaustion works great here: few public servants are shameless enough to admit, perhaps even to themselves, that these are the reasons for their opposition.
Fixing it too well may be incompatible with some personal liberties and privacy. A big problem is that the Government has no really solid way to authenticate you online. If it did, you'd need to provide far less info when signing up for something. Some of the Scandinavian countries work that way.
Should the US? Should Government web sites work on the policy that you should never have to tell the Government something it already knows?
I can only speak from an Australian perspective, but you'd be surprised how sensitive many public servants (outside of the 'intelligence community') are to public backlash against a national identity system. Admittedly they're concerned for the wrong reason, political backlash, rather than the right reason, protection of citizen privacy. But their reasoning is unimportant as long as it drives them towards a privacy-enhancing solution.
Standards like OAuth (and profiles like Open ID Connect) lend themselves well to federated auth systems where user-authorised attribute disclosure by mutually trusted third-parties, rather than authenticated identity, forms the basis of access to government online services. This could even have privacy benefits outside of citizen to government transactions; an online bottleshop wouldn't need to ask for a scan of your driver's licence (causing information leakage), it would only need to ask a trusted attribute holder if the customer is over 18 (or 21).
> Should the US? Should Government web sites work on the policy that you should never have to tell the Government something it already knows?
Should is a form of judgement, so I'll weigh mine in if you don't mind :-).
As I am certain you know, while it is technically possible to have government sites pull up everything known about the citizen given a modicum of uniquely identifiable information, the US culture would likely produce a tremendous "big brother is watching us" paranoid kickback. If this is warranted or not is moot.
For a smaller scale example, some years back the Home Shopping Network (HSN) made available caller-ID to CSR's for people calling to purchase product. They still have it now, of course, but at the time it was initially made available, CSR personnel would greet the caller "by name" when answering. Something akin to, "Hello Mr. Smith, how may I help you?"
This freaked out a lot of people. Especially the elderly. In any event, as those of us reading/posting in this forum know, the information is most certainly still there. Its presence is just not revealed to the HSN consumer.
Put that same type of convenience onto a US government web site and tech-trolls will light up the web with NSA conspiracy articles ad nauseam.
Right. Suppose the Obamacare web site had been really good. You go there, and it checks your IP address with your ISP, identifies your account, and greets you with "Hello. It looks like you're Mr Smith of 1234 Scott St, Raleigh, North Carolina. Is that correct?" The user answers yes, and gets back "You're qualified to sign up for Obamacare. We've checked your tax records, and your income qualifies you to get extra assistance. You don't have any other medical coverage right now. If you sign up, your premium of $42.00 a month will be deducted from your paycheck from Walmart Inc, along with your taxes. Do you want to sign up for medical coverage?
Yes, thanks for signing up. You now have medical coverage. Here's your certificate of coverage. Print this page and you're done.
But from the 24x7 headline-starved troglodytes looking to whip the populace into a frenzy? Loud and heard from every corner. And what would those which slavishly follow their pundits-of-choice do? Would they remember the ease with which they received health coverage or would they incorporate whatever mantra was beaten into them?
I spoke to one of them on the phone recently and was very impressed! If I didn't have a couple of other compelling things occupying me right now I'd seriously consider joining their team in DC pronto. They are preparing to introduce more government clients to speedy iterations in clean modular application development and I'm eager to see them continue to impress.
I think the startup has good intentions, but I think is focusing on the wrong end of things. With good reason, though, as I don't imagine there's much money to be made in the areas that really require attention.
My hypothesis is if we had competent (I mean most, not just some) elected officials in the first place, almost none of the chronic issues would exist to begin with. Voters need to be far more critical of who we put in office (all offices at national, state, and local levels).
But how do we educate ourselves on them? We need people working to make raw, unfiltered data readily available to voters BEFORE they vote so we end up with a far more productive / capable government to begin with.
I'd tolerate the terrible design if the core functionality just worked. Apparently, you're not allowed to make EDGAR filings on the SEC's website if it's a federal holiday. Go figure.
The team who was recruited had to work as typical government contractors but are now forming their own company which they want to operate differently from all the other companies that build stuff for the government. One of the team members told me a while ago they were interested in organizing into a public benefit corporation or something like that (I had never heard of such a thing).
Hi Stephen! Which one of us did you talk to? We did just incorporate as a public benefit corporation. It's basically a for-profit corporation that also has a social mission baked into the charter.
All that aside, the key reason for poor design in government is this: lack of incentive. And no, the ballot box is not an effective incentive; very few voters will punish a government because their tax return form was poorly designed. Private companies possess this incentive: if your product is hard to use, people won't buy it and you'll go out of business. Government doesn't have this incentive: if people find your products hard to use, they'll use them or they'll (eventually) go to jail, won't get their welfare payment, won't receive a driver's licence etc.
Outsourcing or privatising these kinds of services is not the answer: you're just substituting a public monopoly for a private one, which arguably leads to worse outcomes for citizens (and great ones for the private monopolist). This kind of action only makes sense if a competitive market will form after government gets out of the way. It's helpful to think of government services in this fashion: a service market that is dominated by a monopolist.
So that's the problem you are attempting to solve: how do you get a monopolist with no profit incentive to design better products and services?