The RWA Market Quiz — Twenty Questions

Five dozen tokenization reports. Twenty questions.

A timed, single-page knowledge check built on the same source folder as the RWA Research Snapshot — the RWA & tokenization research published Apr–Jun 2026. Not the headlines: the figures that quietly contradict each other, the products nobody can name, and the details most people skim past.

Every question carries a sourced explanation. There's a clock, a streak multiplier, a 50/50 lifeline, and a shareable result card.

No build step, no dependencies, no framework, no backend. One file — index.html. Everything below runs client-side.

---

How it plays

Three toggles on the cover, all independent:

Setting	Options	Effect
Mode	Random 20 · Daily challenge	Daily seeds the RNG from the date — same 20 questions, same order, same shuffled options for everyone that day
Topics	Mixed · Market · Issuers	Mixed draws 10 from each category; a single topic draws all 20 from one
Pace	Chill 45s · Normal 28s · Blitz 15s	Per-question countdown. Running out reveals the answer and scores zero

Twenty questions per run, drawn from a bank of 68 — so no two random runs look alike.

Keyboard: 1–4 to answer, Enter or Space to advance.

Scoring

gained = round( (100 + speed) × multiplier )
  speed      = round(120 × timeRemaining / duration)   →  0–120, answer faster to earn more
  multiplier = 1 + 0.1 × min(streak − 1, 10)           →  1.0× to 2.0×, caps at an 11-answer streak

A perfect, instant, unbroken 20 tops out at 7,590 points. Wrong or timed-out answers score nothing and reset the streak to zero. The streak counter bursts confetti at 3, 5, 10 and 15+.

The 50/50 lifeline is once per run and multiple-choice only — it dims two wrong options and is disabled on the other question types.

Question types

Four formats, mixed into every run. The draw guarantees a minimum of the non-multiple-choice ones (~30% of each category's slice, floor of 2) so a run never comes out as twenty identical MCQs — see drawCat() at index.html:656.

Type	Count	How it works
Multiple choice	53	Four options, shuffled at render. Correct answer flashes green, wrong shakes
Higher / Lower (hl)	5	Two labelled sides, pick the bigger. Values stay hidden until you commit
Guess the number (slider)	5	Drag and lock in. Scored on relative error against a per-question tolerance (15–30%), and the reveal tells you how close you got
Order by size (order)	5	Tap four items largest-first. All-or-nothing — the reveal ranks and colours each slot

Personas

The end screen grades you on the 20:

Score	Persona
18–20	The Insider — could have written the report
15–17	The Allocator — would survive a due-diligence call
11–14	Directionally Correct — right shape, fuzzy decimals
6–10	The Skimmer — you've read the executive summaries
0–5	The Tourist

If your market and issuer scores diverge by 3 or more, the persona picks up a line naming which side is the gap.

Sharing

• Emoji grid — 🟩/🟥 in rows of five, Wordle-style
• Share on X — opens a tweet intent prefilled with score, persona, grid and points
• Copy result — same text to the clipboard, with an execCommand fallback for older browsers
• Download card — renders a 1200×630 PNG on a <canvas> (paper background, score, persona, result squares, R3 mark), waiting on document.fonts.ready so Fraunces and Inter land in the image rather than the fallbacks. See makeCard() at index.html:805

Best score and play count persist in localStorage (rwaquiz_best_pts, rwaquiz_played, rwaquiz_sound). Nothing leaves the browser — there is no analytics, no backend, no network call after the fonts load.

---

The question bank

68 questions, split exactly 34 market / 34 issuer, each with a sourced one-paragraph explanation shown immediately after you answer.

• Market parameters — the definitional fights: free-float vs. full settlement layer, the 2030 forecast spread ($400B → $14T), penetration at ~1 basis point, who published which number.
• Issuers & products — the fine print: BUIDL vs. USYC vs. Figure's HELOC token, ERC-3643 and why compliance tokens can't sit in Aave, Spiko's AMF-supervised fund suite, Apex/Tokeny, bStock trade sizes, Tether's Treasury book.
• Castle Labs (Jun 2026) — a later block on DeFi utilisation: only ~$3B of a ~$28B market is actually deployed, reinsurance is the highest-utilisation category, tokenized "commodities" is really just tokenized gold.

Sources cited per answer — Binance Research, BeInCrypto × rwa.xyz, Pantera, CoinGecko, Citi, BCG, Castle Labs and others.

---

Repo layout

index.html    the entire app — markup, design tokens, CSS, 68-question bank, game engine, PNG card renderer

That's it. There is no package.json, no config, no CI.

Run it locally

python3 -m http.server 8000

Then open http://localhost:8000. Opening index.html straight off the filesystem works too — only Google Fonts needs network, and the layout degrades to Georgia/system sans without it.

Deploy

Any static host serves it as-is: GitHub Pages, Netlify, Cloudflare, an S3 bucket. Push index.html to the root and you're done — no build command, no output directory.

Note that unlike the sibling snapshot repo, there's no wrangler.jsonc and no .gitignore here. If you want the Cloudflare deploy path, copy both across.

---

Editing the bank

Questions live in the BANK array at index.html:322, grouped by comment banner. Every entry needs c ('market' or 'issuer'), q (the prompt), e (the explanation) and s (the source line). The rest depends on type.

Multiple choice — no type field. a[0] is always the correct answer; options are shuffled at render, so write them in any order you like as long as the right one is first:

{c:'market', q:'Distributed (free-float) tokenized RWA value in May 2026:',
 a:['$31.4B','$12.8B','$58.9B','$94.2B'],
 e:'Distributed value hit $31.4B – up from ~$21.5B at the start of 2026…',
 s:'Binance Research · May 2026 (RWA.xyz data)'}

Higher / Lower — val is compared numerically and rendered with fmtB() (billions, rolling to $XT above 1000), so keep both sides in the same unit:

{c:'issuer', type:'hl', q:'Bigger tokenized product by value?',
 left:{label:'Circle USYC', val:2.96}, right:{label:'BlackRock BUIDL', val:2.42}, e:'…', s:'…'}

Slider — tol is fractional, not absolute: 0.15 means within 15% of answer. min/max/step set the track, dec the displayed precision, prefix defaults to $:

{c:'market', type:'slider', q:'Guess Citi’s 2030 tokenized-securities forecast:',
 answer:5.5, min:0, max:20, step:0.1, dec:1, suffix:'T', tol:0.2, e:'…', s:'…'}

Order — four items, scored strictly descending by val, all-or-nothing:

{c:'market', type:'order', q:'Order these 2030 forecasts, largest first:',
 items:[{label:'BCG', val:14000},{label:'Citi', val:5500},…], e:'…', s:'…'}

Two things to keep an eye on when you add questions. The 34/34 category split is what makes the Mixed draw land 10 and 10 — skew it badly and one topic starts repeating. And each category needs enough non-MC entries to satisfy the specials floor: single-topic runs ask for 6 of them out of a 20-question draw, and the current bank has 8 market / 7 issuer, so there isn't much headroom before the draw quietly falls back to filling with multiple choice.

Design tokens are CSS custom properties in :root at index.html:12 — paper #F2EBDD, ink #1A1816, accent #C44A36, green #4F7A54, highlight #F5E69A, with Fraunces for display and Inter for text. The same palette drives the canvas card, but hard-coded as hex literals in makeCard() — change a token and you have to change it there too.

Sound is off by default (WebAudio oscillators, no audio files): a rising triad on correct, a low saw on wrong, a tick in the last five seconds, a fanfare above 14/20. The preference persists across sessions.

Motion respects prefers-reduced-motion — confetti, point floaters and the score count-up all switch off.

---

Known limits

• Daily mode uses the local date, not UTC. todayKey() reads getFullYear/getMonth/getDate off a local Date, so "same 20 for everyone today" holds within a timezone but not across them — players either side of a date line get different sets at the same instant.
• The answer key is in the page source. a[0] is always correct, and the higher/lower and order values sit in the DOM at opacity:0 rather than being withheld. That's inherent to a no-backend quiz; anyone who opens devtools can win. Fine for a knowledge check, not for anything scored competitively.
• No Open Graph or Twitter card tags — despite the built-in share buttons. A shared link renders bare, and the nicely-rendered PNG card has to be attached by hand. Four <meta> tags in <head> would fix it.
• The run length is fixed at N_TOTAL = 20 (index.html:611) and referenced in copy throughout — changing it means touching the strings too.

---

Prepared by Mike Blank · figures as of 1 Aug 2026. Sources cited per answer.