Resource — idea leaderboard
Ten pitched topics, each already carrying a verified research spine and a before/after demo concept. Vote a topic up or down — no sign-in, no account, votes save in your browser. Click any row for the full spine, the sources, and what we still owe.
How this works
Nothing reaches this page on a hunch. Every entry below already has an angle we can defend, a claim nobody in the niche has made yet, primary sources we have read, and a demo we know how to build. What we do not have is a running order — that part is yours.
Ten pitched deep-dives
The starting order is our suggested release sequence — early entries chain off videos already made, because callbacks compound the sense of one connected body of knowledge. Move them.
The XP-badges-leaderboard stack does not add motivation — it converts it, and the exchange rate is terrible.
Fixed-length challenges are built on a number that never existed — and the real spread is 18 to 254 days.
Granted once in a dialog, re-litigated on every single send — and most "push lifts retention" stats are vendor marketing.
Every user sees it, almost no team designs it — and it is where three verified literatures collide.
Offboarding as a designed surface — plus the regulatory whiplash story nobody in design content has straight.
The labor illusion is real and reversible; the skeleton screen has no peer-reviewed support at all.
Service or stalking turns on one variable: whether the user knowingly handed you the data.
Your sort order is a nudge with a Gini coefficient — and popularity defaults can lock a market onto mediocre winners.
Everyone covers setting prices; nobody covers changing them — the moment that actually torches subscription businesses.
The methodology manifesto disguised as a topic — winner's curse, guardrails, and the most-tested UI on the internet losing to a regulator.
Not yet ranked
Four more with promising spines that have not yet earned a slot on the board — either because the causal evidence is thinner than the pitch needs, or because they overlap too heavily with a video already made. No voting on these; they move up only when the research does.
The sequel to the $300M button. Baymard's measured averages — 5.1 steps and 11.3 form fields in a typical checkout — plus address autocomplete and the error-recovery literature. The evidence is there; the question is whether it reads as a second forms video rather than a new one.
Sources: Baymard Institute on checkout form fields · Seckler, Heinz, Bargas-Avila, Opwis & Tuch (2014), CHI '14
Query-suggestion psychology, and the argument that "no results" is the deadliest empty state in software — the one screen where the user told you exactly what they wanted and you answered with nothing. Currently an angle in search of a literature.
Sources: none tiered yet — this one needs a research pass before it can be pitched properly.
WCAG contrast ratios are the one solidly citable number in the whole Playbook, and the aging-population arithmetic is genuinely compelling. But the conversion claim needs a causal spine it does not yet have, and we will not run a "do the right thing, it pays" video on correlations.
Sources: WCAG 2.x contrast success criteria (W3C recommendation, cited directly); no conversion study strong enough to headline.
Ovsiankina resumption — the 66.79% figure that survived when the Zeigarnik effect did not — combined with endowed progress, applied specifically to setup checklists. Well evidenced, but it risks being a subsection of the progress-bars video rather than its own.
Sources: Nunes & Drèze (2006), JCR 32(4) · Ovsiankina resumption work, carried from the progress-bars brief and still owed a direct read.
Applies to all ten
Which is why this page shows its working. Where a source exists and we have read it, it is linked. Where a claim is still carried by an author and a year, it says so in plain text. A pitch is not a citation, and the gap between them is the whole point of the channel.