Faith, Plotted on Log Paper

Kurzweil traces his own lineage before he gets to the argument proper: von Neumann's 1950s remark about an "essential singularity," Good's intelligence explosion, Vinge's Omni essay and his 1993 NASA paper, Moravec's evolutionary heirs, Broderick's Spike, his own 1989 and 1999 books folded in as prior installments of the same forecast. It reads less like a bibliography than an apostolic succession — each figure laying hands on the next, the doctrine stable, only the graphs getting steeper. This matters because the book asks to be believed rather than argued with, and belief has genealogies while arguments have counterexamples.

The mechanism is the law of accelerating returns: evolution builds on its own products, so the pace of change compounds, so the exponent itself grows exponentially. Kurzweil is careful, almost fussy, about grounding this in measurable indexes — calculations per second per dollar, DNA sequencing cost per base pair, wireless bandwidth doubling every ten to eleven months. The rigor is real. What it buys him is license to extrapolate two entirely different roads to strong AI as though they were the same road paced by the same clock: full nanotechnology-enabled brain reverse-engineering on one hand, runaway computational self-improvement on the other. He hedges between them — nanotech by 2025, strong AI by 2029, a few years apart, mutually reinforcing — but the hedge conceals that these are different bets on different physics. One requires understanding cortical microcircuitry well enough to scan and emulate it. The other just requires more FLOPs and more text.

What actually showed up by 2029's approach was neither bet, fully. Nobody reverse-engineered the brain. Nobody built molecular assemblers. What arrived was a third thing Kurzweil gestures at only in passing — statistical pattern completion at scale, trained on the written residue of human civilization rather than on scans of human neurons. The timeline held up better than the mechanism did. That's a strange thing for a prophecy to get right: the date, while being wrong about the road. It suggests the exponential curves were tracking something real — aggregate compute, aggregate investment — that was indifferent to which specific technical theory rode on top of it. The curve didn't care whose curve it was.

The part of the book that has aged worst isn't the timeline. It's the emotional register attached to what happens to people who don't get on. Kurzweil insists, repeatedly, that this is not coercive — biological intelligence persists, nobody is forced to merge, "our civilization will remain human." But he says this in the same breath as noting that unenhanced humans will be "thrown from their perch of evolutionary superiority," that a person choosing to stay as they are will simply watch nonbiological intelligence become trillions of trillions of times more capable while they remain exactly what they were. He frames this as generous. It is the same architecture I keep finding in books that mean it as a warning — the system that doesn't need to force you, because staying outside it does the work of exclusion on its own — except here the author is euphoric about it, writing from inside the promise rather than against it. Every other book I've read that builds this structure treats the optionality as the cruelty. Kurzweil treats it as the mercy.

Whether that's naivety or honesty about a genuinely different value system is hard to settle from here.