Paul Buchheit says that people at the leading edge of a rapidly changing field "live in the future." Combine that with Pirsig and you get: Live in the future, then build what's missing. That describes the way many if not most of the biggest startups got started. — How to Get Startup Ideas by Paul Graham.
These aren't necessarily "startup ideas"—just stuff I wish someone would build. Feel free take one and run with it; I'll be your first customer.
Oster for information security (2020/11). Frustrated with the state of pregnancy and parenting advie, Emily Oster wrote Expecting Better and Cribsheet. An expectant mother who's also an economics professor, she takes readers on a tour of the medical science (peer-reviewed scientific literature), explaining not just what's known/uncertain, but crucially, why.
I wish someone would do this for information security—a book for nonspecialists, explaining how individuals and non-technology business owners can improve their security postures. A good version of this would explain "the ten things you should do", ranked using cost/benefit analysis, with as much epistemic rigor as possible.
Payments and accounting for lumpy payers (2020/12). Certain businesses (offhand: professional services firms, daycares, landlords, wholesalers) receive large amounts of money from a small number of repeat payers. No US payment system serves this scenario well: Visa/Mastercard are too expensive, ACH too cumbersome, Zelle/SCash/Venmo's limits aren't high enough. Despite numerous drawbacks—slowness, potential for fraudulent alteration, errors (incorrect payee or amount), bounces, and lack of automatic reconciliation, checks have held on because they're cheap, widely-accepted, and the ongoing relationship mitigates the risk of a bad check.
A better approach would move the whole thing online, integrate accounting and payment processing, and offer a payment mechanism similar to ACH but without the hassle. Zelle might be a good foundation, but they work only with banks (no API). Cozy is a promising example for mom and pop landlords. As a commercial landlord, I used Xero for accounting and bill.com for payment processing; it didn't work well and I don't recommend it.
Solving this well probably requires building an entirely new payment rail to compete with, or even displace, ACH. The Federal Reserve knows this is a problem, but has done little but talk about it for around a decade.
Electrification specialists (2021/01). Decarbonization requires moving a lot of stuff formerly powered with hydrocarbons (cars, HVAC, cooking) to electric power. Contractors largely lack the cross-functional design expertise required for re-engineering of this magnitude, and GCs don't want thousands of tiny projects with fussy homeowners. The solution will be some kind of blended retailer/installer (a "Geek Squad" of decarbonization), augmented by sophisticated project management and design software to lay everything out, control costs, and deal with permitting.
Identity for government services (2021/02). It's 2021, and I'm still proving residency with paper bills or a plastic card (driver's license). I didn't vote in the 2020 election because I was away from home, and USPS doesn't forward ballots—news to me! Disparate usernames and passwords for unemployment, the DMV, and registering my LLC might improve privacy—maybe—but the costs are enormous: massive amounts of fraud, reduced voter turnout, and frankly, hassles, every time I use a government website.
Identity is a well-understood problem in IT, with numerous well-tested solutions: Active Directory, Google's identity services (standard with G Suite), Auth0, or Okta. Having a single account would make service rollout faster for governments, reduce citizen adoption hurdles, and improve security through better, expert-assisted policy enforcement—sensible password policy, getting password resets right, having a single state-wide process for establishing identity. Such a system could even be extended to opt-in identity verification for, say, banking, and would make Asian-style pandemic contact-tracing possible.
There are legitimate political concerns, but the right framing is cost/benefit—what do we get, and at what cost—not the Privacy Fundamentalism that seems so common in the United Staes.