Hacker News Reader

Comment by danso | Hacker News Reader

I was surprised at how cheap, relatively speaking, this system is:

> The state police used a vendor to build their program, at an estimated cost of $195,000, and the courts used in-house tech experts at an estimated cost of $50,000.

I don't have any experience in government IT, but even in corporate terms, this seems relatively cheap given the scope and lifetime of the system (though I'm assuming these stated costs don't include maintenance and ongoing operation) and the consequences when there's a system failure, e.g. lawsuits.

View on HN

Yes, that's extremely cheap.

In Chicago, there was a system being developed in partnership with University of Chicago to make city-owned database schemas open to the public through a PHP web app.

The system cost 300k to develop, was never finished, and was quietly put into forever-maintenance mode after a few years of use.

As a general rule, Government is pretty bad with tech projects and their associated costs.

Just look at Canada's Phoenix payment system. That thing did wild amounts of damage and cost billions.

In practice, I agree: this is pretty cheap given the market.

But now take a step back and think about it: the actual problem is probably pretty simple. If the databases are well made, it would just take a few requests for this operation to be done. Even with the tests, that's one or two programmers' job for two weeks, and maybe one week of a partial review of results by a legal specialist to validate the quality.

Imagine if the state had two in-house programmers up to this task available. $195,000 would cover their salary for a year.

> and the consequences when there's a system failure, e.g. lawsuits

Another saving grace for doing things in house: you take actions internally. You can correct very quickly, shuffle incompetent people, save all the litigation costs and the bad faiths over misinterpretation of a SoW.

Those two in-house programmers would never get the time to do this. They would be swallowed in various maintenance task on the other 500 custom systems the state owns.

(from experience in the private sector)

That's a difference I observed (in France) between the private and public sector. In several administration, public workers have a reputation of being lazy because they don't always have work to do. They are on demand, and work on projects as appropriate.

That's what a well-funded public task force looks like.

In such a case, a manager would look at the schedule, reserve the time necessary for this dev, take into account the other missions of the devs and get it done.

> the actual problem is probably pretty simple.

Yeah but writing the code is not what's expensive on this project.

That's my point. What is expensive is to interact with a private entity to do it, to manage risks and potential litigation. All these costs would be saved if the solution was developed in house.

It's all about liability transfer. Government is basically crippled by risk aversion. The first 165k were to sign a contract accepting all sorts of liability in case something bad happens. The other 30k was to write some SQL, test it (hopefully) and then run it in prod.

I doubt risk aversion is the reason. The govt is still on the hook for data errors regardless, also data errors are can be corrected with existing processes.

$195k would be enough to probably only cover one state-employee programmer for one year. Think about your salary, now double it. That's how much you cost your employer. I'm sure a state employee costs even more with how inefficient overhead is for government agencies.

In most of the world, public administrations tend to have lower overhead than similar private sectors. Administrations usually benefit more from the economy of scale and if you consider dividends as a cost for the end user, they don't have those.

Is a 100k salary the norm for people writing DB queries in the US? That sounds a lot.

I work in county government IT in Ohio, and the $50K number, per county court, seems like a reasonable estimate. I don't know anything about Pennaylvania, but in Ohio each court (of which there could be multiple in each county-- a court of Common Pleas, a Juvenile Court, and one or more Municipal courts) would each have their own disparate court management systems. In my county there are currently 3 different systems covering 3 courts. Each one would need its own integration.

A few years ago the Supreme Court of the State of Ohio undertook a multi-year data warehouse project to build a centralized reference database for all state courts. I assisted the vendor in our County and easily had a significant fraction of that $50K in my labor over a period of a couple of years.

Pennsylvania has a similar structure as far as the different types of courts, but they have had their Unified Judicial System [0] for quite some time now. It's actually quite usable (once you know where the links you need are) and really ties together all the different courts and provides a web portal [1] where most things one would need to do (except actual court appearances) can be done pretty quickly.

[0] http://www.pacourts.us/ [1] https://ujsportal.pacourts.us/

Police use grant funds to do everything and usually have a stable of good smaller consulting firms that get shit done. They probably had 2-3 contractor FTEs for 4-6 months and built off an existing system.

Government IT is good at building things like this. They suck at large projects and ops.

Wow. That's like practically free. Would love to see a case study on how this system was implemented.

Where you see efficiency, a silicon valley MBA sees opportunity.

Reply on news.ycombinator.com