Product Releases

🛡️ Lifeline and Immigration Status

The Justice Department ruled in May that the 1996 welfare law limiting benefits for immigrants applies to Lifeline, the federal phone and internet discount. We had been showing Lifeline to people whose status the law excludes.

We corrected it across all eight states:

• Non-citizens and green-card holders in their first five years no longer see Lifeline. Both the qualified-status requirement and the five-year bar apply to it.
• Refugees and asylees keep it — they’re exempt from the five-year bar.
• “Other lawfully present, with permission to work” keeps it, with a warning. That single option covers two groups the law treats oppositely: parolees and VAWA self-petitioners can qualify, while TPS holders, DACA recipients and student or work-visa holders cannot. Rather than drop the program for everyone in that group, we now say plainly that work permission isn’t enough on its own and point people to the Lifeline Support Center.
• Added the Lifeline Support Center’s phone, email and website to the five states whose listing gave people no one to call.

Doing this, we found last update’s green-card fix hadn’t fully taken hold on Lifeline: the restriction was removed in several states, but an age-based duplicate of the same rule was left behind it, so the program kept showing anyway. Both are gone.

The FCC hasn’t finalized its matching rule, so people in these statuses can still lawfully enroll today. We’ve moved ahead of the rule. Showing someone fewer programs than they might get is the safer mistake, but it is still one, and we’ll revisit when the FCC acts.

🗺️  State Updates

North Carolina — Two new options on the urgent needs step: “Help finding shelter” and “Free or lower cost health care.” The closest existing option covered help with rent, mortgage or utilities, which doesn’t serve someone with no housing at all, and there was no route at all to free care — a gap that matters more now that many North Carolinians are expected to lose Medicaid under the new federal rules. Both route into NC211, filtered to the right category.

North Carolina — Child care savings were overstated for families with a school-age child. North Carolina uses two income ceilings: 200% of the poverty level for children under 6, and 133% for children 6 to 12. We were adding every child’s cost of care into the household total without checking each child against their own ceiling. A Durham family of three earning $4,500 a month saw about $1,097 a month in savings when the real figure is $717 — their seven-year-old shouldn’t have counted.

Kansas and Missouri — The public charge link on the disclaimer page was invisible. The sentence ended “…please visit the .” with nothing to click. The address was right the whole time; the label was never filled in.

Washington and Massachusetts — Those same public charge links had been pointing at dead pages. They were fixed a few weeks ago, but only in the live site, not in the files everything gets rebuilt from — so the next routine update would have quietly restored both dead links. Now fixed for good.

✅ Accuracy and Content

Application times were wrong for WIC and school meals. We timed the real applications: WIC in Kansas and Missouri takes about one to five minutes, against the 30 to 60 minutes some states were showing. School meals in Missouri is about five minutes, not 15 to 30. Someone deciding whether they have time to apply today was being told to set aside an hour for something that takes two minutes. Corrected across states, along with making the apply button read consistently where the link goes to an interest form rather than a full application.

Nurturing Futures income wasn’t lowering SNAP estimates. Boulder County found this while modeling families’ before-and-after scenarios: entering the payment under Government Benefits and then raising the amount left SNAP unchanged, while the same increase typed into the general income field correctly reduced it. Fixed. We also answered their team’s outstanding questions on running estimates for mixed-status households.

🔧  Reliability and Maintenance

We removed a piece of shared logic that could silently change who qualifies. Thirteen programs asked one place “is this household eligible for Medicaid?” before deciding their own answer — and if Medicaid hadn’t been worked out yet, the answer came back “no,” indistinguishable from a real no. Whether that produced a wrong result depended on the order programs happened to run in, and nothing checked that order. Each program now states what it depends on, and we test the sequence. This surfaced a live bug in Illinois, where the marketplace program could run early, read “not eligible” for two state programs that hadn’t been evaluated yet, and pass a check it should have failed.

Fewer ways for the dashboards to disagree with themselves. About 50 metrics were written out twice — once for a single state, once across all states — with near-identical logic kept in step by hand. Each is now defined once. Nothing looks different; there’s just no longer a way for the two versions to drift apart.

Groundwork. We cleaned up the last loose ends from reorganizing the program code, and finished the tracking behind the Colorado heat pump results page so we can see where people drop off between reading about heat pumps and looking for a contractor. No addresses or personal details are recorded.