[kthxbye123]

The (very reasonable) comment this reply was originally threaded with is now dead, so I’ll post it here:

Some level of fraud in a social welfare system is absolutely acceptable, if tightening controls to eliminate the fraud would mean that otherwise eligible people didn’t receive benefits they qualified for, or even if it meant that accessing those benefits became much more onerous. I don’t doubt that $36bn is too high, but this kind of analysis is almost never present in these articles and politically in this country the spectre of fraud is usually used as an anti-welfare cudgel

[wpietri]

Definitely. I can trivially design an unemployment system that has zero fraud: we just never give out any money at all. We have to balance fraudulent payments against deserving recipients who don't get the money.

I used to work with the people who made GetCalFresh: https://www.getcalfresh.org/

The basic problem they addressed was that a lot of people who qualified for food stamps didn't get them because it was too hard to apply. You either had to fill out an intimidating set of paper forms or a quirky and intimidating set of web versions of paper forms. GetCalFresh basically just applied the lean startup playbook: make something very simple, drive people to it with ads, and, as Paul Graham suggests, they did things that didn't scale. Initially, they filled out the paperwork by hand. Then they kept adding automation for filling out and faxing in PDFs. They talked with users, removing barriers and solving problems. Rolling out county by county, they've made a huge difference. Average online application time dropped from 45 minutes to 8. Last I checked, they were helping tens of thousands of people per month.

I think it's fantastic that all of these people previously going hungry are now getting fed. But for years they weren't, and there are still plenty of deserving recipients who aren't. A lot of what GCF is doing is taking information the government already has and giving it back to them. Why is that necessary at all?

I get why unemployment is still a mess. But given the power of networked computers, eventually I'd like to see all safety-net programs get rid of the need to apply at all. For a lot of pandemic layoffs, it would have been possible for state governments to contact people who were probably eligible, have them confirm a few details, and start sending them the money. And as a bonus, a proactive system like that would be much harder to scam.

[ghaff]

There is certainly a contingent on HN who belong to the camp of "we need tamper-proof national ID secured by technobabble." I'd just point out that it took me two tries to get a new RealID license because, even as a homeowner and passport holder, I didn't quite bring all the required forms of identification and proof of residency the first time.

Personally, I'm just fine with a system that has some cracks and limitations.

[Sebguer]

Agreed. There should absolutely be anti-fraud mechanisms, but those are not the same as what people tend to start talking about when they start talking about welfare fraud.

You can protect against fraud of this scale without adding more rigorous means testing, or mechanisms that really just hurt the people who need the most help.

[newacct583]

That's exactly right. The total in the headline (I didn't read through to do a serious accounting) comes to about a 10% overhead vs. CARES act funding. That's bad. States should work to make the process more efficient. But efforts at stopping fraud like this by increasing bureaucratic friction can easily end up cutting out 10% of entitled recipients who just don't have the right paperwork.

In fact, most state unemployment systems reliably and measurably shortchange their citizens by being needlessly restrictive. They need to be easier to qualify for, not harder.

[dang]

Edit: I misread this originally—but please don't copy-paste comments on HN. It strictly lowers the signal/noise ratio.

---

I can't tell what comment you were referring to, but please don't copy-paste [dead] comments on HN, or any comments. If there's a [dead] comment that shouldn't be dead, you can vouch for it as described at https://news.ycombinator.com/newsfaq.html, or email us at [e-mail].

[SilasX]

Parent meant that the comment s/he replied to is dead, not that the replying comment (that s/he made and is reposting) is dead.