Make your key.
No real names. Choose a handle and a passphrase you'll remember — it's how you get back in. New here? This creates your key. Returning? The same one opens the door.
What got cheaper today, and what didn't.
Each day's science and industry news, scored against eleven basic needs, setbacks included.
The dates come from fitted cost curves, published with their miss rate.
-
Scored
daily -
Sources
graded -
Dates from
curves -
Misses
published
Sign in
Use the email you signed up with.
New here?
- The day's stories, scored across all 11 needs
- A source and an evidence grade on every story
- Dates with 80% windows, fitted from cost curves
- The engine's hit rate, misses included
Your email signs you in. We don't sell it. How the numbers are made →
the door is sealed
You passed. This is the last thing between you and the inside — a day's wait. Not to keep you out, but to make sure you meant it; anyone can click, few come back a day later. When the seal breaks, the inside opens: what's pulling the future forward, what's holding it back, and the others who see it too.
THE WRONG NOUN
The fear is one sentence long: you will become obsolete.
Wrong noun. What is becoming obsolete is the bill. The rent of being alive. The forty hours traded for the right to keep the lights on. A job was never the goal. It was the toll.
THE NUMBER THAT WANTS TO BE ZERO
In 1830, one hour of light cost you about three hours of work. Today it costs roughly seven hundredths of one second. Intelligence is the next thing to fall.
light · per hour, in work-time 1830 → 3 hours today → 0.07 sec solar · price per watt 1976 → $96.50 2024 → $0.26 (−99.7%) genome · cost to read one out-paced Moore's Law since 2008 ──────────────────────────────────────────────────────────────────── most things never fall like this. intelligence is one of the few that does.
verified: Nordhaus / NBER · Our World in Data · NHGRI
THEY LAUGHED AT THE TELESCOPE
Galileo saw moons circling Jupiter and proof the Earth was not the center of everything. For saying it he was made to kneel and take it back, and kept under guard until he died. He was right. It took the institution more than three centuries to admit it.
Seeing the direction is not the same as knowing the date. So we don't sell prophecy. We point at the curve and tell you the truth: the direction is certain, the timing is not. Bet on the direction. Then help pull it closer.
THE WAIT IS NOT A WALL
Yes — the powerful touch it first. A government can even switch a model off overnight; that has already happened. But "first" is not "only." Every wave — electricity, the car, the internet — began as a toy for the few and ended as air for everyone, and the wait kept shrinking: ~40 years to reach most homes in 1900, ~17 by 2000.
The wait is not a wall. It is a speed. Access is not handed down — it is pulled forward, by the people who refuse to stay quiet. Push harder, and the wait gets shorter for everyone. That is the whole reason this exists.
we are not afraid of the future. we are the part of it that isn't.
The daily pull
The current path to a free world
Eleven needs, scored against the day’s news. Each clock runs to the middle of its fitted window, and only measured data moves it.
- Prediction calibration
- —target 80% 80% bands · back-tested
- Engine confidence
- — High when the bands land within 8 points of 80%
- Signals on record
- — Curated stories, every day since launch
What this pace buys us.
the milestones ahead. where a measured curve exists, the cost-curve engine fits the window, and only a refit moves it. tap a milestone for its engine rows and this week's evidence.
how these dates move — the rules
a date moves when a new observation moves a fit, and for no other reason. each milestone's engine rows show the fitted 80% crossing window of its tracked gates (curves.html) and the latest measured move, dated, from the public prediction ledger: an append-only row for every changed window, so consecutive rows ARE the movement. headlines accumulate as evidence under a milestone; they move its date only by becoming a measured observation that changes a fit. the % and the un-gated year windows remain our editorial read, labeled. ● tracks a measured curve · ◐ emerging, real but unproven at scale · ○ a speculative window, not a schedule.
most friction is a filing, a proposal or a warning: something that might happen and can be undone. the biggest forward stories already happened. a single net number would hide that difference, so the measured side gets its own ledger below.
What reached people.
every countable figure a source stated, summed by unit and split by stage: proposed, approved, installed. households, people and patients stay in separate totals. the headline counts only what is installed.
Where it's growing. Where it needs hands.
all 11 needs, ranked by this week's momentum. tap a tile to filter the feed below to that need.
why a site like this — the straight-line habit
Many Minded plots the price of energy, health, knowledge, food and seven other needs as each one falls, and how fast. Solar is down more than 99% since 1976. A genome read once cost a hundred million dollars.
Ask someone where a price will be in ten years and they'll extend the last ten in a straight line. For anything on a learning curve, that guess has usually come in too high.
The feed live
The day's stories, newest first. Each one is weighed by its evidence; setbacks count against the index.
THE PROGRESS LEADERS.
ranked by verified, evidence-capped impact on the record — attribution, not endorsement; friction counted.
the record begins 2025-01 (401 verified landmarks to 2026-06, daily pulls since) — scores are credit ON the record, not a lifetime valuation · pull points = signed, evidence-capped impact of every signal credited to an actor (no day-conversion — measured "days" live on the horizon's prediction ledger) · the band is a relative read (share of the window's #1), not a rating ·
Every basic need, free.
how close each need is to free and universal — the mean of its sub-needs. a sub-need is measured where a global source exists (the share of people it reaches, and how far its price has fallen on a 100× log scale); the rest are editorial estimates, tagged as such.
how we measure progress →—
the full per-need record — evidence, the road to free, sources
What used to be expensive.
What became abundant.
What falls next.
every claim below links its source, and the strongest argument against it.
counted from its own first year.
each curve starts at 1×.
sources: Epoch AI · Our World in Data · BloombergNEF · NHGRI · ITU · projections are fitted fans, not promises — every curve, with falsifiers →
once machines and scale reach a problem, cost stops being the hard part and the blocker moves: technology → infrastructure → access → policy.
JOBS AT RISK.
the occupations most exposed to AI, next to what employment data shows. exposure is what AI could do; displacement is what happened. we keep them apart.
ⓘ exposure is a model's ceiling; displacement is measured employment data. every figure carries its basis and what would prove it wrong.
the full record — the three margins, the exposure track record, vendor reversals, sources
What's coming — and what it needs now.
the technologies that could remake a basic need — each labeled by how real it is, with the one bottleneck it has to clear next. we separate what a lab did from what a founder promised, and where the physics says no, we lead with why.
● measured — real today, the work left is scale not science · ◐ emerging — proven in the lab, a translation gap remains · ○ speculative — needs a breakthrough that has not happened · ⊘ physics-limited — the physics argues against it. every number is timestamped; some move monthly.
What we're building.
nothing here yet — check back soon.
The open line.
signals from nodes who found the door — transmit a thought, or run a command (try: help).