Hacker News new | past | comments | ask | show | jobs | submit login
What I learned in two years of moving government forms online (2018) (medium.com/jgee)
262 points by williamsmj on March 19, 2021 | hide | past | favorite | 133 comments



Well-written article, useful insights, total misunderstanding of the way things like this should be done.

There's an old saying: don't pave cow paths. That's exactly what he was doing. He was putting a different front end on an existing process. You end up with a total nightmare.

Instead, this should be done department-by-department, app-by-app. Design the process from the bottom up. For example: a form to request some document. Start with the documents: how are they stored? Should they be stored differently? Are they in a database? Should we really even be storing documents, or should they just be database entries? For that matter, should the department that manages them even exist? (Here in Chicago, the answer was no. They merged the County Recorder of Deeds office with the County Clerk.)

You should never just move things online. The web is different, and enables different business processes. Change the process first. Forms come later.


You're not technically wrong, but what you are suggesting simply doesn't work in government. Each department is a kingdom with varying amounts of political clout. Only the smallest and least-powerful groups will jump on board with a mandate that induces invasive and drastic changes to their workflows. The kingdoms with more political clout will laugh in your face and slow your progress down until your project simply runs out of clock time. The result is that you still have all the same legacy workflows and also this new workflow for the small number of groups you had more power over.

This guy managed to achieve his mandate AND managed to entice groups to investigate updating their workflows after the fact. I agree it's not the way things SHOULD work, but he managed to find a way to make the most amount of progress and along the way some kingdoms were enticed to cycle back around and improve their workflows too.


I haven't done a lot of work in government, but I have done a lot of work in orgs with distributed power structures, and this is 100% correct. Starting from the ground up means constructing a working group for every single workflow in which everyone who is even theoretically a stakeholder (and many people who aren't) is going to demand to be in the room and be heard. Once you get past 4-5 people, good luck holding a meeting more than once every few months, and if a single powerful stakeholder doesn't like what you are doing or perceives that you are stripping them of power, they will gum up the process until it fails. I've had 4-5 value adding initiatives crushed in this manner at past jobs.

IMO, the optimal strategy in such an environment is to get at least one powerful champion on your side who can give you political cover, and then to quietly construct a MVP while soliciting feedback from high performing rank and file in the departments of interest who you know (ie no other power brokers besides your champion(s)). If you can deploy a value adding MVP it becomes difficult for antagonists to intervene and kill the project without appearing petty, and then you can really get to work building the product you want.


In DC their in house 'innovation team' tried to overcome some of these issues and make them more useful to the people using the forms by having an annual formapalooza. Results were mixed but it is an interesting approach to digitization and human centered design.

https://thelabprojects.dc.gov/formapalooza


Great point. With the Federal Govt that often breaks down to "agencies that use GSA as a service provider" and "agencies that roll their own."


I sort of imagine some departments opting for a webform-to-fax gateway :)


Except what he did resulted in progress.

What you are proposing most of the time will not because it would require an immense, truly Herculean, amount of political will and backing.

Redesigning processes clashes with the #1 rule of big systems: processes are ruled by people whose jobs depend on them. And they WILL fight tooth and nail to keep those jobs. The bigger the change, the more threatened they will feel, the more they will sabotage you.

And you know what the #2 rule of big systems is: no matter what "the boss" may say, if the officials are not on board it's just not going to happen.

If you have not, I suggest watching "Yes minister" :).


People don't like to be told they have to change the way they do things because... the computer.

Years ago, I was at a very small company that used Exchange. We had some outages and it was just a lot of administrative overhead in a world in which Gmail was by then an option.

But one person in particular fought the change strongly because they were used to a business/sales process that used hierarchical folders to keep documents organized and Gmail, at the time, only had single level labels. The CEO mandated the change anyway but the other person never accepted that Gmail was suitable for our needs.


Sure. But also, lets admit that software developers are not qualified to redesign processes of institutions they dont understand, so fighting them is pretty often rational from institutional point of view.


If the processes of an institution can't be explained to a real, qualified software engineer, the issue runs much deeper than just resisting to change.


The author was given the mandate to "move forms online", so they chose the quickest, most agile option and did just that - moved forms online.

The author later brings up the questions you mentioned - what is a form, should these forms exist, etc, but instead of spending a long time trying to design an entire new system _when that wasn't even asked of them_ they decided to deliver results. Once they gained trust and got buy-in by the city, they could think bigger.


> Instead, this should be done department-by-department, app-by-app. Design the process from the bottom up."

Sounds great. If you can do this in less than a week for each form (the author managed 100 forms in 2 years) then you're perfecting processes _and_ moving them online faster than the author.

But just in case you missed it, let me quote a couple of bits of the article:

"Getting city workers to accept online submissions rather than traditional paper ones is the bulk of this work. On average, it took me about 30 minutes to make a digital form and five weeks to meet with, earn the trust of, and get buy-in from the employees who would use it. Even if they were excited, the nitty gritty details took a lot of back and forth."

"Some departments had sort-of insane business processes for submissions. If I tried to change those, I would spend a whole year on a single department. By focusing on the priority, moving forms online and making it easier for the customer, I could make consistent progress rather than be consistently blocked."

My guess is that you're going to spend decades perfecting processes, only to find the world has changed (or the politicians in control of the City of Boston have changed the law that governs the processes!) by the time you're ready to begin digitize forms. You're letting perfect be the enemy of good.

> You should never just move things online. The web is different, and enables different business processes. Change the process first. Forms come later.

Sometimes "just moving things online" is a big improvement over the status quo. But the author agrees with you in at least one case. See the paragraph about what they are doing about death certificate requests.


Yeah, I agree. A lot of what I deal with at my job is similar. The people doing the work just don't want people coming in telling them new ways to do things. It doesn't help that vendors are constantly promising the sky and leaving them with broken things. But once you start showing them small incremental improvements they start to discover the next step etc. Then they see how machines can help them and make suggestions.

In my field there's a very well-known example of a MRI manufacturer who has made very "smart" systems that end-users simply refuse to use. Technologists know how to operate a scanner, when a machine starts trying to help them it's very difficult for them to (1) understand what the machine is doing and how to work with it and (2) trust the machine, primarily because physicians yell at people, not machines. There are maybe 10-20% of the technologists that "get it" and want to tinker and learn how to use the new features. The rest still miss the way scanners worked 20 years ago--because they understood and trust them.

That's the thing with paper forms--people can understand what's happening, discover the process and figure out how to succeed and not to be yelled at. When it disappears into a machine it might as well be a blind and deaf new hire who keeps messing with your desk. It's a major issue of trust.

So the parent's "no it needs to be this giant reinvention when strangers who don't do the work come in and tell you that you need to do everything differently" attitude maybe works with management.

I honestly think the big problem here is that we don't seem to have systems that people can play with to build their own ideas anymore. Maybe sharepoint but my god I can't even figure out how to do anything on that. It really boggles my mind that we have things like iPads and iPhones but nothing like Hypercard.


This approach is likely to produce better experiences if it could be seen through to the end, but also likely to go nowhere inside a large organization.

Change is hard. Radical change, all at once inside a big organization is near impossible.

In this case, it’s a great first step to get all this info out of paper and wooden inboxes and into a more accessible form.

Doing so still leaves open the potential for reinventing high-priority forms and workflows. It also makes it easier, because now the team can easily ask questions like, “how many people use this?” And “how much time do we spend on this?” Etc.

In short, while you might be right wrt. producing innovation, sometimes the first step is to just put it on the internet.


Change needs to come from the top. The UK did it with Gov.UK, and their success inspired many other governments around the world to follow them.


And how long did it take?

I think Gov.UK was (and is) great, but it's entirely possible that doing a lift and shift of the paper based forms might have gotten more done in a shorter time.


This is 100% correct. I run a govtech startup (similar to Seamless docs, the company mentioned in the article), and this was one of the key insights we've picked up over the last couple years.

When your process happens on paper, there's nothing forcing you to ensure it makes logical sense. You can draw whatever you want on a piece of paper, so a lot of the paper forms that people call "permits" are just messy reflections of the thought process of the person who originally conceived the process. Naturally, products that start by directly replicating a paper form are accommodating and perpetuating the lack of planning.

We've kind of taken a backwards approach where we don't accommodate direct translation of forms and instead provide tools that encourage people to think through the steps of their permitting processes to get them to make sense on a digital platform. More often than not, they discover glaring inefficiencies in the process which when fixed make everything work more smoothly even independent of the move to a digital platform.

Shameless plug: https://citygro.ws


I will disagree. In the distant past, we had a bunch of forms to move online and our platform was SharePoint. I know, I know, but before you blame SharePoint, we didn't get to the point of SharePoint being the issue.

Instead, we got endless bikeshedding about "the process" and during that, many unrealistic expectations were conjured into existence as each process was re-imagined into kind of ultimate, abstract form of near infinite configurability. Try to imagine sitting at a table as someone says, with complete seriousness, that people ought to be able to submit a multiple one terabyte (not mega or giga, tera) files with a particular webform. Not any other transfer protocol, either. And this is back around 2008 or so. It should just ... expand as necessary.

Nor was this the only time I encountered this kind of behavior. I have seen functionaries whose visions of process suddenly bloat like that "elephant's toothpaste" when presented with technological options. I would do the forms first, given the chance.


That saying is strange because cow paths usually represent the most direct path somewhere that the designer of the space failed to imagine.

Is this picture a bad thing? https://i.redd.it/uj5ehfdxn5o31.jpg


That saying is also oddly appropriate here because the joke for anyone who lived in boston is that the city is built on paved-over cow paths.

(yes yes fun-police, it's actually a myth https://www.bostonmagazine.com/news/2018/03/06/boston-street... but the truth of it being a bunch of ad-hoc cities that coalesced into one entity isn't too far off from the truth. My favorite bit of Boston artwork is road silhouettes of all the Boston-area squares: http://andywoodruff.com/blog/boston-squared/ )


It is a strange saying. My guess is this is something that's lost in translation—a leaky metaphore—and the people using it don't think about real cow paths.

Real world space is (generally) 2 dimensional, and these cow paths are often very predictable and sensible. They're usually the result of architects thinking that sometimes people actually walk or bike, instead of always driving cars.

In the world of bureaucracy & information tech, the space can be very convoluted, and shortcuts aren't really the same kind of thing (queue XKCD comic on spacebar heating).


Cows tend to pass where other cows passed. Herd behaviour. Something like a real world effort optimisation.

No use trying to optimise in a week using one brain something that found a local minima (or maxima) over years and multiple people.

Normally the change (and effort) to achieve a better minima (or maxima) has to made on a variable you just can't see or imagine.

Something on the line of Chesterton’s Fence.


Those aren't cow paths, those are people paths. Cows are not known for their tendency to walk in straight lines. Edit: notice they still had to put up fencing to keep people from cutting off even that minimal switchback.


This might be a dialect issue. Where I'm from, "cow path" is the normal way of saying what also called a "desire path" by others. According to the wikipedia, I'm not completely out to lunch: https://en.wikipedia.org/wiki/Desire_path


I hadn't heard "cow path" before, but a quick google suggested it was a synonym for "desire path" so I went with that.


> Design the process from the bottom up

Bottom up IT system reimplementations fail much more often than incremental ones.

Bottom up reimplementations of operating human systems that also involve big bang IT implementations are even worse, by comparison to incremental evolutions, mixing incremental automation with incremental process improvement.

And it usually only takes one high-profile failure of a big-bang implementation to derail a broad reimplementation process, whereas the occasional incremental setback on an incremental improvement process is rarely politically significant to the overall process.

So, no, I disagree with your recommendation that this should instead have been done with a more waterfall-style approach.


The argument in the article for gradual change sounded reasonable to me. Just moving a process online is at least a first step.

> Some departments had sort-of insane business processes for submissions. If I tried to change those, I would spend a whole year on a single department. By focusing on the priority, moving forms online and making it easier for the customer, I could make consistent progress rather than be consistently blocked.

Re: should they just be database entries? he did do this or make other changes when the department agreed.


Except to redesign the processes you need buy in and co-operation from multiple different people and departments and good enough project management and communication with all of them to ensure that the thing you replace the forms with will actually do the same job.

Whereas moving existing forms online in the same structure can be done with minimal management agreement and provides a quick win.

Especially when you're talking about 344 forms. Redesigning all the back office processes for 344 forms would be a gargantuan task.


> Instead, this should be done department-by-department, app-by-app. Design the process from the bottom up.

It is a technical way. "Do not mind people, just make tech to work". This way have its benefits, like it tends to make result to have better design, but what about people? They work and they think that their work is important and valued. But "hey, some engineer came and started to make it his way without any reverences to us". It is humiliating, and such engineer could face a resistance.

The author made a genius' move which I might expect from a psychologist or any other social oriented person, not from tech-savvy engineer: he started to make life of a people better. He asked their opinion on how they think it would be made better. He showed respect. He showed people that he sees them as people, not as gears in a clockwork mechanism. They reacted as people generally tend to react to this: they welcomed him. Probably it helped that it was government bureaucracy which likely treat people as gears in a clockwork, so his way of treating people contrasted nicely to the way they used to.


I have expertise in the space and while I would have agreed with you 10 years ago, I disagree today.

There are billion dollar programs that can be replaced by a Google Form. You’re always going to lose if you are too accommodating. Every one of these dopey workflows has a unit of FTEs on payroll in the backend!

Get High level sponsorship, hard deadline and most things will be doable within reasonable constraints. The stuff that isn’t doable is the hard work, which you will never discover until it’s forced.


Great idea! I wonder why nobody has ever thought about that before! You must be the first! A modern day Prometheus!

The highlighted part about the fact that it takes 30 minutes to make a change, and FIVE WEEKS to get sign-off is spot on. I fucking hate bureaucracy. But you know what happens if you don't take the FIVE WEEKS to get sign-off? Nothing.


> For that matter, should the department that manages them even exist?

Yeah, nothing better then complete reorganization by someone who has no idea about the work being done, needs and pressures.

What will happen here is that they will quickly realize you dont know what you are doing but are threat, so whoever will have option will easily stall you.


Wasn’t one of the the points of the article the complete opposite of this comment? You can improve the process without having to overhaul the whole thing. They don’t need to flip their tech stack for forms.


It seems more like a classic case of refactoring except that the original infrastructure is paper-based, in which case the ideal digital redesign would be one that mimics the paper-based one precisely so that "regression testing" can be performed and components can be swapped in piecewise. Once enough components are digitized, the system as a whole can begin to be molded into more effective forms. This would have to be the plan at the outset, since it would require preserving enough flexibility.


> Change the process first. Forms come later.

In government that can easily mean the forms coming never.


That's just as true for the corporate world. Check out the business processes of, say, a global telecom that's been around since the era of party lines. SAP SE makes buckets of money from big corporations that just want computerized file cabinets.


> There's an old saying: don't pave cow paths

Not said by any dairy farmer. If you believe this, you never shovelled cowshit. Believe me, dairy cow paths and pens are paved, for a reason.

Pave cow paths, but start at the dairy and work backwards.


A lot of government agencies lack the budgeting required to go through a process of "as-is" and "to-be". This process is both lengthy and costly and is usually reserved for the main portion of that agencies workload.


Workflows and form-related processes are a little like desire paths, true. Over time people establish ways of working that define the business logic of a company. Some of it is codified on paper, some in the culture and interactions, and it all tends to slowly change over time.

There is a lot of value in the existing processes. That's what the business is built on. Proposing to drastically change how things work while promising that the result will be at least as good or better -- that's a tough sell for a lot of reasons. It also seems incredibly risky, projects like that are no easy feat.

I think a gradual, incremental approach would have a higher chance of adoption and eventual success. Small asks that produce tangible improvements, building towards a systemic changeover.


Did you try to do something like that - moving a government office into digital world? What is your personal experience with this?


> ...the way things like this should be done.

Do you have case studies for successful process reengineering efforts?


Your misunderstanding of how things work in government is at least as egregious as the author's misunderstanding of this idealized way to redesign processes.


Do you see any middle ground between the 'rebuild everything from the ground up' and 'rework the user interface' type approaches?


Didn't he specifically address this, in the section where he brought up "salami slicing"?


I'd love to read more about this view. Do you know any good books about this subject (government digitalisation)?


The iceberg image at the top of the article fails to show how much of an actual iceberg is beneath the surface.


What I've seen continually in government is the dreaded form-builder Frankenstein.

1. Gov thinks it needs some simple form building 2. Chooses a vendor form building framework, some mashup of ASP.NET usually with workflow tools 3. Oh, we'd like documents stored, and an integration with another system, and a custom reporting tool, and a payment function, and a appointment booking. 4. Vendor doesn't want to lose business so builds custom code into the "simple" form/workflow tool.

All of sudden the simple form tool is customized beyond belief, falls behind on updates and becomes impossible to change. Worst still, nobody wants to invest in it because it just-about-works.


I work in medical and there the top guys often make decisions that are really hard to implement in software. If they understood only a little of the consequences of their decisions have they would probably think a little more about the rules they are creating.

I see the same in product development. Marketing comes up with something that has words like "most of the time", "sometime" or "somehow" in the description. Really hard to impossible to implement. When I ask them what they really want to achieve we usually end up with something simpler and straightforward that fulfills the needs, is well understood and easy to implement.

I think in government with most of the work being outsourcend this feedback loop doesn't exist. A committee comes up with something and the contractor implements it without questions asked. The more complex the work, the more money for the contractor.


I work with authentication protocols and access control. Product people want magic.


I am fine with product people expressing their wants. But we shouldn’t accept their requests blindly but put them through a reality check. The customer is not always right or at least the customer doesn’t explain his real problem properly.


That's true, but power dynamics don't always align with wishes. If the professional recommendations of security people weren't so routinely ignored, you'd see a lot less stupid breaches. A great many disasters^1 could be prevented if experts weren't ignored.

[1] https://calteches.library.caltech.edu/3570/1/Feynman.pdf


Bah. People have to ignore security professionals all the time, because all too often, security professionals don't care if any software gets shipped.


That's unfortunately true too. I often break IT policies because I have to do my job and when I ask they don't have an answer. Security people definitely can't just say "No" but have to help finding a solution.


Everyone wants magic, even technologists.

Unfortunately, magic is hard work.


Nail on the head.


ASP.net with windows workflow foundation hasn't existed for like 10 years or something like that.


VBA hasn't been updated since 2010... yet it's still going strong.

Things like this create a niche and will live longer than cockroaches after a nuclear war.


VBA is still the main way of writing macros. There are plenty of jobs requiring VBA.

Try and find a job for asp.net webforms and windows workflow foundation.

So kinda pointless you posting that really. One is still 'current', but the other you'll be writing asp.net core with modern MVC.


first result in google is a handful of jobs... not many but definitely they exist.

https://www.indeed.com/q-What-Is-Windows-Workflow-Foundation...

And many more for webforms

https://www.indeed.com/q-Webforms-jobs.html

Niche things live on... as my post states.


FTA: That sort of works, but it also creates a weird dynamic because the tech teams don’t have the training nor the explicit mandate to mess around with other people’s jobs.

Here in Belgium, I did that kind of work for 14 years. The fact is that you have to have in your IT team people knowledgeable with business. When I say knowledgeable, I mean, they understand the business at least as well as the business people themselves. When you talk to business with that kind of team, it's much easier.

Now, the problem I had is not the form, it's the business process behind. Changing these is super hard because inertia is there. Also you have to take into account years and years of little shortcuts, exceptions, etc. that somehow now are part of the process. It's really tough to bring everything together.


Reminds me of this interview with Elon Musk, where he talks about Conway's Law and how they mitigate it at SpaceX: https://youtu.be/cIQ36Kt7UVg?t=206

It's very hard to design an efficient system when the people writing the requirements and the people implementing them don't really understand or talk to each other.


> Get rid of paper forms... moving those forms online ... made government services much more accessible

Not everybody uses computers. Not everybody has internet access.

Shouldn't paper forms still be available as a last resort?

I'm okay with "online first" but not "online only".


> Shouldn't paper forms still be available as a last resort?

No, not really. The edge cases are easily resolved by helping out those who cannot use the standard method. A friend, a random stranger or a clerk would gladly fill an online form when someone is having troubles.

Upkeep of a parallel infrastructure is hard and expensive.

If it’s not THAT rare it instantly becomes a service that is provided by a shop close to the place where the papers were supposed to be filled.

I’ve seen that in Bulgaria where clerks would fill online or paper forms for you in exchange of a small fee and in Turkey there’s a bookstore close to official buildings that will help you with all kind of form. There used to be people with typewriters hassling around public offices that would fill forms for you too.

Nowadays in Turkey the online government services are very good(you can do everything online, including everything from health services to subscription cancellations), so these seized to exist and if there’s some who cannot use the online stuff, a local shop or the clerks in a public office would help them out.


I am only okay with online only if all online forms strictly stick to open standards and protocols and require no data collection besides what is strictly necessary for the forms.

When I have to agree to a "Terms of Service" on a third party website to get my drive license and these Terms have a mandatory arbitration clause and leave this private entity with the ability to ban me for any reason, I will understandably be upset.

All government forms must be accessibly to all. If internet is required to use government forms, internet access must not be cut off to anyone without due process of law. The same goes to electricity and the ability to use electronic devices.

If the main computing devices available are Google Chromebooks, Apple iPads, and Microsoft Tablets, the user's ability to log in can be restricted at the drop of a pin by a faceless corporate entity. While it may still be possible to access the internet on some of these devices without agreeing to the TOS, this can change. If the government certificate expires, then the device will need to be updated. Updates almost always require a TOS agreement.

No private corporations should be able to ban people from using government forms without due process of law.

So yes, I am okay with online only, but only in a world where all software can be used by all users for any purpose without threat of revocation from extra-judicial private overlords.



Interesting edge case that can be solved by booklet or a person with half an hour time explaining how these things work or simply help them out instead of upkeep of a paper processing infrastructure through institutions.


My parents, who have decidedly not been in prison for the last 30 years, can barely fill out most digital forms (especially ones on the Web) on their own. They both have professional degrees and have used computers since the 1980s.

Assuming that someone who has likely never touched a computer before can operate at or above that level seems like a fast track towards ensuring that they remain marginalized and even further isolated from social services. That, in turn, has long term expenses that, I'd expect, dwarf the costs of a few hardcopies and a data-entry clerk.


I am not unfamiliar with this situation but I still think it is an edge case that the time will take care of.

In my case, family fills up the online forms for the elderly. I know that they are not stupid, my father is an Engeneer with masters degree, used to program with punch cards and he still uses Autocad but he is having trouble with online forms. When needed, he sends me a link and I fill up forms for him occasionally. I have other relatives too who are simply not comfortable with computers and we will help them out when needed.

When that's not possible, a clerk will help them out. People tend to be nice to old people, even if they are grumpy.


"edge case"? Take a look at how many Americans are under the control of the carceral system. Dismissing the reality of how we treat prisoners in the US smacks of elitism.


Government can never be anything-only. About 20% of adults in the United States are considered functionally illiterate and 10% are almost entirely illiterate. There will always need to be the option to speak to a human who can assist you.

https://nces.ed.gov/datapoints/2019179.asp


Boston has a number of ways to access online services, including free public wi-fi, libraries, and even a mobile City Hall that brings services directly to different neighborhoods: https://www.nationalcivicleague.org/promising-practices/bost...

More accessible than requiring someone to go to a City Hall basement!


Paper forms are available ... where? In my county, the only place to get a paper form is by printing it out. A lot of people use the library to get access to a computer ... to print the form.

I’m not saying “no” to paper, I’m just talking about the reality that having a computer in at least one part of the process is mandatory.


Kiosks at city hall and libraries help with this, but I think the simplest way to support this is to have each online form be printable (or have a form to request a paper version) and then an address to mail it with a message like “online forms usually processed in 1-2 days, mailing forms depends on postal service and may take 4-6 weeks”

And just do data entry on the few paper forms that make it past this and don’t think about it.

This is contingent on being able to have an alternate renderer that spits out pdf based on the form definition that should be dead simple.

I work in an org with lots of forms. People were weird that they wouldn’t accept a screen that said “Prepend approved on 3/18 at 1055” and would do weird things like print out the screen and ask for signatures or say funny stuff like “the law requires a paper signature.” So the pdf killer equivalent added a “view form” button that just displayed the old form, filled in from the same record with “Digitally signed by prepend on 3/18 at 1055” in the signature box. Seems stupid but worked wonders. Some people would print that out and save it, but that doesn’t slow anyone down but the printer.


100% spot on here.

There are a significant number of people who use their phone as the primary or even sole method of internet access. And that's not just low income people but also young adults. When you start counting homeless people who primarily use the library (if open), you have another problem.

I've worked in various "digital divide" issues for almost 10 years. Online first is okay but we can't wipe out the offline approaches without leaving large swathes of people without any feasible options.

Innovate but make sure you're considering your constituents' needs.


It is a significant reduction in scope and cost if you:

1. Make it online only

2. Let people use computers for free at city hall and libraries

3. Have folks around who can help if they get stuck


I don't think anything about this article implies that there's no non-tech way to utilize these services after this change. Most governments have laws banning that anyway.

A quick browse of boston.gov shows at least all forms I could immediately find have an in-person alternative listed on the website.


Moving the forms online makes it easier even if the person using the system is still the office clerk.

Sure, they'll have to fill the form themselves, but it is already online. So it is one (main) workflow regardless of who does it.


Nice writeup.

One likely factor for Josh Gee's notable successes is his youth. Because he's outside of the power structures and politics.

I started "adult" (office) work at 15yo. I got a lot done. Everyone loved the geeky computer whiz who could fix stuff. I got bonuses, perks, praise. This track record set me up for long-term failure.

By the time I turned 25yo, I hit a wall. I was now an adult. With my own power. So I was now perceived as a threat. Sadly, I was completely oblivious to the changed dynamic. And I didn't have a mentor or any one else to clue me in.

Since then, I've encouraged promising young talent to color outside the lines (eg be bold, don't wait for permission) while they still can.

Just from his writing style, I think Josh Gee will avoid my mistakes.


> I put in my earbuds, fired up a podcast, and started going line by line through the list. That took about 2 months. I would open the PDF, see if it was a form, and log it.

You can see this guy is not a developer because he didn't spend 2 months learning machine learning and computer vision while trying to automatize the form classification task only to give up after 6 months and do it by hand anyway.

And clearly government because he didn't spend 10 dollars in mechanical turks to do it in 24 hours.


Form Yes/No certainly but mechanical turks can't handle this part:

"I also reviewed contracts and documentation and interviewed employees in the various city departments to ask if they had other forms."


PDFs for forms are truly terrible and barely work[1] in anything except Adobe Reader. Online forms are not without their problems, especially from the maintenance point of view, but significantly more accessible to the normal people, especially for the mobile phones, if designed properly.

[1] https://gitlab.freedesktop.org/poppler/poppler/-/issues?labe...


I wished I had known that when I filled out a form in document preview on my Mac a few hours ago, including adding a signature.

The process was basically perfect and about as simple as can be.


I've successfully been using Preview to annotate, sign, fill in annoying PDFs for years. Works great


I've filled out lots of PDF forms in Firefox and Chrome without issue.


Because they license Adobe's engine to handle the PDFs


Firefox uses PDF.js, Chrome's reader is based on Foxit.


This is the first step of digitization: Putting electricity on the paper. But it's only the beginning. For the long-tail it may be good enough. But for lots of government forms there are so much more that can be done.

Like, we discovered some absurd amount of forms sent to us (like 90% or so) had errors that actually would make the applicant ineligible for whatever the form was for. Stuff like "if you answered X on Q14.2c, you have to answer Q32.5". All this is something good UX and logic can help make a breeze. Instead of a huuuge form to cover all edge cases, it only expands when veering from the happy-path.

Also, next step of digitization is actually getting the data as structured data. So that it's automatically put into whatever software the govemp uses to handle the form. Without that, majority of the amount of time spent from a caseworker can be to just transcribe the form into application before it's actually being worked on. Next step after that is of course to automatically approve/deny stuff. If not for everything, there are lot of simple cases that often don't need much scrutiny.

A big step is to actually get the gov to change the process/law. Putting electricity to paper is a small win over the incredible wins with just fixing the process. Like one form I worked on was so complicated (maternity/paternity leave) because the law had basically hundreds of small edge cases accrued over the years. Either to fix a loop-hole, or to help some group of people a bit extra. But that just ends up in regulation not even the experts approving these forms understand. It's a hard battle and can take years. Like, sometimes a new rule the politicians invent will cost far more to implement and execute for the small case, than just making it apply to everyone and grant them some extra money...


All very valuable and extremely on point for the person/team who will be taking over the process after the OP.

I've had my own very small share of experience delving into existing workflows and teasing out the early/simple cases and only getting into further processing once the basic conditions are met. It can be an interesting challenge to get those who do the current processing to see where efficiencies can be teased out.

The level of legal/higher management mandates for significant process changes is something I've never had to deal with, but I've seen the edges of just how crazy the complications can get. (I once did an implementation of Form 1040 handling via spreadsheet... I was surprised to find that proper math REQUIRED a circular reference and iterating to correct numbers.)


Great article. It felt true to my experience working with govt employees, and I’m impressed that this person had the tenacity to achieve the success that they did.

Unfortunately I also think this is the kind of stuff its really easy to write a cheap headline about.

“BOONDOGGLE: BOSTON SPENDS $150,000 AND 2 YEARS MAKING PAPER FORM INTO WEBSITE”

There, that’s 100,000 shares on a headline that no one will really care to read the details of.


The amount of GAFAM dependencies here is impressive. So Google has all the dead certificate request made by Boston resident ? And more ? I don't like this one bit. And the article didn't even mention if they try to use something else, so I guess they didn't even try.

From my point of view, paper is better than any Gafam


For those similarly confused as I was, Gafam seems to be the acronym for Google, Apple, Facebook, Amazon and Microsoft.

I'm sure I would agree to Google and Facebook being bad in this situation. I'm mostly convinced about Amazon, too, but why Apple and Microsoft?


Impressive credit to the author for writing an interesting article about the least interesting topic imaginable: government bureaucracy.


But it's not about government bureaucracy - it's about making both the lives of your coworkers and your fellow citizens just a bit better. That can actually remarkably fulfilling.

I'm working with a group of doctors practice in London on their digital plans at the moment, and I love it.


This is not about government bureaucracy. It's actually about optimisation of a complex system in a context of conflicting loss functions and hyper dimensional variable spaces.

He managed to find a better local minima (at least for the variables he was considering in the first place). The other uncountable variables will probably readjust to a unpredictable future state.


> It's actually about optimisation of a complex system in a context of conflicting loss functions and hyper dimensional variable spaces

So, what you're saying is that it's an interpretation of a social problem created by someone who's spent far too long doing function approximation? (As someone in this position, i entirely empathise with your interpretation).


> We’ve got a lot of users and a lot of submissions flowing through SeamlessDocs. They are a great tool, but our contract is structured in such a way that makes continued expansion tough.

I feel like an Open Source version of SeamlessDocs would be incredibly useful to tackle these issues not just in the US.

Are there people here who would be interested in building anything like this? Let me know.


I'd love a product that is just "drag a pdf into a window and we'll turn it into a form". It looks like seamlessdocs does that, and then lets you edit the resulting form. But they're focused on governments, and I still have to fill out stupid forms for my company on paper (usually health care or tax documents or paycheck stuff) because they send THOSE on to whichever company we've outsourced that work to.


The focus on not attempting to improve the forms or workflow alongside the porting of the docs online is key. Cuts down on potential errors, increases buy-in from stake holders and doesn’t turn this into some insurmountable task of being a consultant for several dozen city departments.


It's scary that they're now locked into SeamlessDocs which has stopped publishing its pricing.

2016: https://web.archive.org/web/20160303073934/https://www.seaml...

Today: https://www.seamlessdocs.com/pricing


Perfectly normal for this sort of product to not have fixed tiers, they've just pivoted off trying to sell to self-serve customers, to switching to a different pricing structure.

You can't make money like that if you have to negotiate contracts with government agencies.

Not sure how that's scary.

Imagine sending in a salesperson, and occasionally an developer, to 5 meetings, each lasting 2 hours, over the space of 6 months. Now calculate how long it would take to pay for that back at $55 p/m.

And you'll note that the page already had an "enterprise" tier, for which you had to enquire.


This is what makes paper forms so hard to crack. Paper and PDF is a predictable price that will be about the same in 10 years, 20 years, whatever.

Not knowing what SeamlessDocs charges (or will even be around) is hard to plan around. They do call out a new contract that I think is meant to help with this.


Sometimes the best form is no form at all. The Washington state license/registration process was incredible, just show up with the documents and they ask some questions and type the relevant stuff in for you. Compare to the ridiculous waste of time and paper in NY where you have to arrive with about half a dead tree of redundant information that they just retype into a computer anyway.

As a more general point, living in multiple states shows you how much bloated government is a complete waste of time when other states just don't do it and get along fine - other examples are car inspections and even income or sales tax.

Did the city decide to simply eliminate any of the functions the forms you found did instead of moving them online?


> A submission emailed to them. A PDF of the submission dropped into a Google Drive folder. A line added to a Google Sheet spreadsheet with the submission data.

This all sound great for the city, but it makes me concerned about privacy. At least with the form somebody had to be in physical access to steal the information. Now they just need access to one cog’s Google drive.


I also noticed the GitHub issues are full of city officials' names and email addresses, the perfect base to set up a social engineering attack. I really don't understand why all those GitHub issues need to be publicly accessible like this, all it takes is one team member to inadvertently post something sensitive and you're stuffed.


City official names and emails are usually public records and are in Boston. So it’s no more of a risk than current levels.

There is a risk of posting the wrong info, but that should be acceptable given the unlikely probability of that happening and the harm of when it does. Posting a citizens form is really bad, but posting meeting minutes probably isn’t too bad.

I think it’s important to consider the risk vs the benefit. How much is having better collaboration and avoiding a JIRA license worth?


Why?

It takes less than 5 seconds to pull out a form, scrawl a name and check a box. It can be done anywhere and with no tools other than the form and a pen.

Just finding a computer will take longer for people who don’t sit in front of one all day, then navigate an app, a browser, etc, etc.

You’ll find most people just end up clicking “print” when they do find your beautiful web form and then write on the paper copy.


At my company we use Github Issues for all task management. Like this author discovered, it is underrated.

SaaS products like Asana and Trello have prohibitive pricing models for what we consider pretty minor integrations. Jira is too complicated. I have seen some open source projects recently.


> Dogs (can this be a drop down as well from 0-9)

Ooh boy, I wonder how long the drop down for cats would be.


Fantastically written article. Fantastic article. It's always great to have both.


The entire landscape of this problem changed this year -- The latest version of Word now can open PDFs and edit the content, then save back to PDF. PDFs are no longer roadblocks, they are now a document type that just works.


only if you can afford to buy word, surely?


True, but most government entities I work with use Word as their primary authoring tool, so that isn't a major barrier.


I think the people filling these forms out are generally citizens of Boston, not city employees.


Very interesting!

side note but... _why_ do so many articles like to repeat things they just said in bigger text? It's usually somewhat spaced apart, but in this instance the sentence is repeated (but BIGGER) right after its written.

> On average, it took me about 30 minutes to make a digital form and five weeks to meet with, earn the trust of, and get buy-in from the employees who would use it.

> """ <same thing but BIGGER> """

I noticed this sentence is highlighted because it is popular; did Medium decide to auto-bigify a pull-quote, or do authors decide to do this?


For administrative fees maybe there could be a digital version of the tax stamp?


I was in tears with the amount of manual labor on easily scripted tasks the author spent. Months of work that could have been scripted in a few weeks.

You can't script the human element out, but you can save a whole bunch of time clicking through pdfs looking for forms and filing github issues.


I bet the author learned a huge amount of valuable context and domain knowledge by reviewing so many forms "by hand".


How would you go about programmatically determining if a PDF is a form or not?


Our product automates pdf forms in a specific area (HR).

From experience, the presence of form fields in the PDF will tell you that it is a form, but the converse is not always true.

Plenty of people flatten their pdfs to remove the form fields, or maybe they were never there in the first place.

I believe it's sometimes driven by misguided attempts to stop people from "hacking" (aka automating) them.

Or pure incompetence.

Some other things I would have like to see in the article would be some details of how they handled signatures. We do electronic signatures but it is certainly an area where accessibility comes into play.


Mechanical Turk?

HTTP get each pdf, look for tillable fields? Flag those for review.

Even manually, 2 months to review 2000 links is a long time. I don’t think the author meant he literally sat there for 40 days, but likely did other stuff too.


Puppeteer should work. I believe PDF form controls are turned in HTML form elements, which can be queried by Puppeteer.


+ check for actual form fields

+ check rectangles as % of form

+ flag burned pdfs for review


In my experience it's e-Redtape :)


It’s not even my final form


It sounds like you spent 16,000 hours to save residents 10,000 hours.


"They also made government services much more accessible."

I would say the value of making the forms accessible was worth it. Also with the flexibility to make changes to the forms in the future will save even more time.


Plus there's value in just making things better! Not everything should be viewed through the eyes of capitalism.


This is funny, but each year residents should be saving more and more.


"All of our submissions were handled via Google Drive." stopped reading there


Clarification: How can anybody be serious about such an approach that is clearly violating data protection requirements?


How is this different from forms being scanned and saved in a network folder at a file server?

Is CIFS a better data protection protocol?


That would almost as bad and is absolutely out of the question in today's regulatory landscape as it does not follow security best practices such as per record encryption or segregation of data.


The network folder & file server at least doesn't have an advertising business model incentivized to snoop on people.




Consider applying for YC's Summer 2025 batch! Applications are open till May 13

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: