Methodology — how we score Elon Musk's predictions

The source rule

Every prediction on this site is tied to something Elon Musk actually said himself, on the record. We accept exactly four kinds of source:

His X/Twitter posts — linked by permalink, text verified.
Video of him speaking — interviews, keynotes, launch events, earnings calls on YouTube.
His own writing — Tesla Master Plans, the Hyperloop Alpha paper, company blog posts he authored.
Earnings-call transcripts — his words from the calls, via published transcripts.

Secondhand paraphrases, company PR, and "sources say" reporting don't qualify as the basis for an entry. News articles appear only as supporting evidence for outcomes (did the thing ship?), never as the source of the promise itself. Predictions we can't tie to his own words don't get published — they sit in quarantine until we find the tape.

How verdicts work

Pending — deadline hasn't arrived. Overdue — deadline passed, no delivery, no formal miss declared.
Delivered — happened by the stated deadline. Delivered · Late — happened, but after it.
Partial — happened in a materially weaker form than promised.
Missed — deadline passed and the thing demonstrably did not happen.
Disputed — the claim's truth is contested; we show the evidence both ways.
Cancelled — the goal itself was abandoned.

A goalpost is counted each time Musk publicly restated the same promise with a new deadline. Every prediction page shows the full progression — original promise, every slip, and where it stands today.

How we choose the deadline

This is where a tracker can quietly cheat, so here is the rule — applied the same way every time:

A specific date wins. "By 2020," "end of next year," "in two years" — we use exactly what he said, dated from when he said it.
A range uses the generous end. When he hedged with a window — "10 to 20 years," "two or three years" — we take the latest end, so a "late" verdict can never be an artifact of us picking the optimistic number.
Every restatement is kept. When the same promise gets a fresh date year after year (the Roadster has six), all of them sit on the prediction's timeline. Headline lateness is measured from his first dated promise — the one people actually planned around — and the full slip history is shown so you can score it from whichever date you think is fair.

We report lateness two ways: the raw gap in days, and the Elon-Time multiplier (how many times the originally-promised duration it actually took) — because "two years late" means something very different on a six-month promise than on a ten-year one.

And a deliberate choice that cuts the other way: a prediction is only graded against itself. A late or even on-time delivery is still logged as late or on-time no matter how impressive the achievement — Falcon Heavy and Crew Dragon are world-changing and years behind their promised dates, and both facts live on the same page. This is a scoreboard for what he said versus what happened, not a verdict on whether the thing was worth doing.

How updates happen

A robot scans every day: new statements from Musk, plus news that resolves pending predictions. Goalpost moved? Stance changed? Promise quietly shelved? It lands on the prediction's timeline and in the RSS feed, with the date it was last checked shown on every page. Corrections welcome — submit a prediction or a fix, or ping @howardyhuang.

Community rules

Members can comment on any prediction. The bar is simple — argue the call, bring receipts, keep it human:

Lawful free speech. Disagree hard, but nothing illegal: no threats, no incitement, no harassment campaigns, no doxxing.
Civil, not sanitized. Strong opinions are welcome; a wall of slurs or pure abuse is not. Make a point, not just noise.
No spam or link-drops. Don't flood the same take, and new accounts can't post links until they've been around 24 hours.
Stay on the prediction. Comments live under the specific claim — keep them about that claim.

Most comments post instantly. A filter holds anything that trips the lawful/spam lines for a human to review, and any member can Report a comment — enough reports auto-hide it pending review. Repeat offenders get banned. It's the X model: open by default, moderated by exception.

The images

Illustrations are AI-generated in a consistent retro-futurist style — fitting, for a site about futures that haven't arrived yet. They're illustrations, not photographs of real events.

Referral disclosure

Tesla and Starlink links are personal referral links. Tesla referrals currently include 3 months of Full Self-Driving for the buyer. Starlink referrals give the friend one month free (applied to their second month) and pay us $100 per activated referral (cash to bank or service credit). They cost you nothing extra and keep the lights on. They have zero effect on verdicts — as the 49 misses on the books suggest.

Who runs this

Built and maintained by Howard Huang. Independent fan project; not affiliated with or endorsed by Tesla, SpaceX, Neuralink, The Boring Company, X, xAI, or Elon Musk. It's a scoreboard, not a hit piece — when he delivers, the stamp says so.

Common questions

What is Elon Time?

Elon Time is the gap between Elon Musk's promised deadlines and reality. Across 215 tracked predictions, when a deadline slips the median prediction takes ×2.1 the promised time, and late deliveries land an average of 553 days behind schedule.

How many predictions has Elon Musk missed?

Of 215 predictions tracked here, 49 were missed or cancelled and 71 were delivered — many of them late. 94 are still open, and the remainder sit in the partial or disputed buckets.

Where does the data come from?

Only from Elon Musk's own words: his X posts, video interviews and keynotes, things he wrote himself (the Tesla Master Plans, Hyperloop Alpha), and earnings-call transcripts. Every entry links the original tweet, video or transcript. News articles are used only as evidence of an outcome, never as the source of a prediction.

Who decides the verdicts?

The rules above decide them, and they're applied the same way to every entry. A prediction is graded against itself — against what he actually said, not against whether the achievement was impressive. Every verdict shows its receipt, and corrections are welcome.