Date: 2026-08-06. Scope: architecture #5 (subscriptions, vouchers, and credits). This late pass does not change a scorecard cell; a revised row needs the required independent re-score.
Two independent randomized US studies find that making a local digital-news subscription free did not materially create use or civic outcomes:
- Hopkins & Gorton (2024) randomized 2,529 Pennsylvania residents to a free local-paper offer. Only 44 subscribed (1.7%), with little evidence of effects on knowledge, engagement, or attitudes.
- Trexler (2026) randomly offered 500 North Carolina registered voters two free months of a local paper plus newsletters. Activation was 3.8%, content access 2.4%, and there was no meaningful consumption or civic-outcome effect.
These tests are close proxies rather than refundable vouchers, and their settings are limited. Together, however, they refute a flat, low-salience free-subscription mechanism as an evidenced route to expanding local-news use. Price relief alone is not enough.
The separate claim that a flat voucher would disproportionately fund national incumbents rather than undercovered counties remains indeterminate. Neither experiment reports payment destination by outlet size, ownership, or news-desert geography. Canada's former credit was a small, non-refundable post-purchase credit (15% of up to C$500, closed after 2024), and no causal allocation, staffing, or coverage evaluation was located. CRA.
Design implication
If #5 remains on the board, test a geographically targeted refundable credit through randomized phase-in. Publish redemption by outlet size, ownership, and county; measure newsroom FTE and local-government coverage against matched non-pilot areas. That is the missing test for both payment allocation and coverage supply.