Hacker News Reader

Hacker News

Comment by DannyBee | Hacker News Reader

Parent Full thread

DannyBee

Err, no.

You understand this isn't a dumb mobile app someone types something into with a simple server backend right?

It's the system that manages and handles the operational data collection and statistical analysis for the census.

It's not a CRUD app bob types something into at the end of the day. It manages the whole effort.

View on HN

And the cost likely includes wages/costs of people for things like training users on the system, setting up and running a help desk for census staff to call when they're completing the census, costs for managing and running procurement for the different components, etc.

You've told us what it isn't, why not tell us what it is. "manages the whole effort" is pretty meaningless, a time sheet application can make that claim.

Imagine building software to handle all the HR needs, IT needs, accounting needs, CRM needs, etc, for a 600,000 person company. Assume that this is also going to be the mother of legacy-tech-debt projects, too. Then, add on the feature of ingesting and actually processing and storing the raw census data, too. Then imagine that it's not a company, but a government agency doing that.

>Then imagine that it's not a company, but a government agency doing that.

I think that is the key takeaway here.

I think we all agree that's it's not trivial. What is far less convincing is your case for it costing $900 million. Dispatching people to addresses is not a novel problem, package carriers do it every day. There isn't some massive R&D effort needed here. The tech already exists, the task at hand is merely a matter of application.

I know several people who worked for the last census. The issue you run in to is that people don't only live at addresses. For instance, they attempt to count all homeless people. They attempt to count everyone in prison. They try to count every single person in the United States.

This is not as easy as you make it sound. For instance, how would you make sure you counted every transient effectively?

The way the census solves that problem is to go out and count all of them in a single night.

That's a level of accuracy mot people couldn't even begin to plan for. If you tasked FedEx with that problem, they'd probably find all the easy people and then hand off the rest to the USPS like they do for excessively rural customers.

Now you're back to harping on operational costs. It doesn't take a $900 million application to assign people areas of a city to canvas.

Tracking that you've performed the operations correctly is extremely complex, necessitating an extremely complex application.

What are you talking about? How is it extremely complex?

So far this is a simple "business" program (there's nothing that requires inventing new Computer Science things, or new hardware, or writing a kernel module, or designing for ultra large storage, or ultra low latency, or global consistency), the requirements should be listed by order of importance by the Census stats and ops team, field teams, administrators; then ask a few teams to come up with a high-level sketch; select 3 of the designs, let 3-5 teams prototype and then assign the remaining of the work to the team that was the most competent.

It's not a 900 000 000 dollars job. That means you can employ thousands of Google engineers for a year to build it. That's insanely expensive.

As someone pointed out, here's the doc and scope:

  - Research and Data Analytics
  - Strategic Planning, Program Development, and Integration
  - Communications Support for Decennial Census Operations and Other Programs
  - Field Recruitment Advertising and Communications
  - Traditional Advertising and Media Buying
  - Digital Advertising and Other Communications Technologies
  - Social Media
  - Public Relations
  - Communications Planning and Materials for the Partnership Program
  - Statistics in Schools Program
  - Website Development and Digital Engagement
  - Rapid Response Activities
  - Project and Financial Management
  - Stakeholders Relations
  - Communications Support for the 2020 Census Data Dissemination

ftp://ftp.census.gov/about/business-opportunities/2020-comm-final-rfp-1-21-16.pdf

I don't think your simple plan will ever come up with something that covers all the requirements in finite time.

How long do you think it takes "3-5 teams" to prototype something covering even a miniscule amount of the above?

This is not surprising though, SV always thinks it has some idea how to solve government scale problems, but most "problem solving" in SV is done by eliminating the customers you don't want to care about.

All I claimed was that the "product development" should not cost 900M USD.

But this is a communications campaign, a PR/PSA whatever, and recruitment and so on.

The actual IT systems for the census seem to be in-house developed by the Census Bureau. (Or at least the prototype was, and it's already in place.)

Reply on news.ycombinator.com