Hacker News Reader

Comment by cjpearson | Hacker News Reader

That price is going to scare away most voters. $1000 per citizen for something that few people care about.

You could argue that the cost estimation is a worst-case scenario. It assumes that 90% of the affected software will be made compliant, but it's also possible that they just use proprietary exception for most applications. It would cost much less in that case, but it would do so by abandoning the purpose of the bill.

The bill died in a unanimous vote last time around, and it's not going to anywhere close to success this time either. However some other parts of the 2022 bill (considering libre solutions for new applications, limiting NDAs and non-competes, allow audits of software used in criminal cases) are more feasible and might stand a chance if they're also split into new bills.

View on HN

> $1000 per citizen

How did they reach that figure? $1.4 billion dollars seems like far too much.

Edit: Interesting. Their methodology talks about rewriting all the state's public-facing server side applications to be open source. (Hence the cost.)

I thought the bill was just requiring that the public be able to use libre browsers and apps on their own devices to access state services, not that the state's server code had to be open source.

You can read their methodology in the bill text [0]. In short, there are 271 applications they found that must be brought into compliance and estimated the timeline and staffing costs based on similar recently completed projects.

[0] https://gencourt.state.nh.us/lsr_search/billText.aspx?id=188...

Suggestion: Pass a bill requiring all new or upgraded systems to be non-proprietary. Then it has no current-year budgetary cost.

The problem with doing it this way is you're accounting for the bill as the cost of upgrading all government systems [to non-proprietary ones], which is on the same order as the cost of upgrading all government systems in general -- something that happens eventually but on its own budget.

So require upgrades to be to non-proprietary systems. Then you just wait. In maybe ten or fifteen years pass a bill requiring any remaining proprietary systems to be replaced, which by then will be fewer of them and by then will be the ones that most need to be replaced.

I don't think the bill requires the server side to be open source. I think it is sufficient as long as the client side can use libre software to interact with the government. (But this is not explicitly stated in any of the links I followed.)

Another comment mentions fillable PDFs not working in non-Adobe software. This seems like the most likely culprit to me. And in general, if someone has to go back over the entire surface area of an application and re-test everything to find all the quirks like this, I can start to understand how that huge cost estimate might make sense.

Yeah, these estimates blew me away. https://gencourt.state.nh.us/lsr_search/billText.aspx?id=188...

I guess there are all kinds of aspects of the government I'm not even aware of, but still... Everything I see is always browser-based with maybe some PDFs. What even is "NH FIRST (statewide ERP)"? Why does the general public interact with an ERP at all? I'm assuming this bill means that individuals should not be compelled to use proprietary software, but if it includes businesses also then the price tag makes more sense.

> You could argue that the cost estimation is a worst-case scenario.

It probably should be. What might improve the bill is trying to get a similar bill passed in a few other state legislatures and working with them as a group to lower costs.

The bill pegs the cost at nearly $300M/year for 5 years, with the funding coming from the state's "General Fund". According to the State of NH, that fund supports about $1.7B in total spending. There's no way a state with such a limited fund and desire for taxes is going to spend such a large percentage of the budget on that effort.

As someone who I respect once told me:

Freedom isn't free.

Reply on news.ycombinator.com