Somebody knows. Then everybody knows. The time between those two things used to be measured in centuries. Now it’s measured in weeks. That gap — the emergence half-life — is the real clock of civilization. And it’s not shrinking because people got smarter. It’s shrinking because the pipe got wider. Print. Radio. TV. Internet. AI. Each one made the pipe wider. But here’s the thing nobody talks about: there’s a counter-force. Ego. The people who benefit from the old answer fight the new one. Semmelweis said “wash your hands” and they put him in an asylum. The gap isn’t knowledge minus ignorance. It’s bandwidth minus resistance. The singularity isn’t when AI gets smarter than us. It’s when the bandwidth term overwhelms the resistance term and the gap goes to zero.
Three domains. Same pattern. The gap between “someone sees it” and “everyone sees it” is collapsing.
Years from first working demonstration to mass adoption (50% of US households or 50M+ users):
| Technology | First Demo | Mass Adoption | Gap (years) | Confidence |
|---|---|---|---|---|
| Electricity | 1882 | ~1925 | 46 | measured |
| Telephone | 1876 | ~1946 | 70 | measured |
| Radio | 1920 | ~1940 | 20 | measured |
| Television | 1927 | ~1955 | 14* | measured |
| Internet | 1991 | ~1999 | 7 | measured |
| Smartphone | 2007 | ~2015 | 8 | measured |
| Bitcoin | 2009 | ~2021 | 12 | estimated |
| ChatGPT | Nov 2022 | Jan 2023 | 0.16 | measured |
*TV delayed by WWII — rapid adoption was 1946–1960 (14 years). Without the war, the gap would have been shorter.
Telephone is an outlier — required physical infrastructure at every endpoint. The 70-year gap reflects copper, not cognition.
Bitcoin “mass adoption” is debatable: 95% awareness but only ~10% ownership. We use awareness here.
ChatGPT: 100M users in 2 months. Fastest in recorded history. measured
The pattern is clear but not clean. Telephone is slower than electricity because it needed a wire to every house. TV is faster than radio because radio built the cultural habit of “sit and receive.” Each technology stands on the shoulders of the last. The gap isn’t just shrinking — it’s shrinking because each generation of bandwidth reduces the cost of the next.
Years from first published evidence to mainstream scientific acceptance:
| Discovery | First Evidence | Accepted | Gap (years) | Resistance Source |
|---|---|---|---|---|
| Handwashing saves lives | 1847 | ~1880s | ~38 | Medical ego |
| Continental drift | 1912 | ~1965 | ~53 | No mechanism proposed |
| Smoking causes cancer | 1950 | 1964 | 14 | Tobacco industry |
| CO2 warms the planet | 1896 | ~1988 | ~92 | Oil industry + complexity |
| H. pylori causes ulcers | 1982 | ~2005 | ~23 | Pharma (antacid market) |
| CRISPR gene editing | 2012 | 2016 | 4 | Low (no incumbent) |
Semmelweis died in an asylum at 47. His crime: telling doctors their hands were dirty. The medical establishment destroyed him. Germ theory vindicated him posthumously.
Wegener died on an expedition in Greenland. Plate tectonics proved him right 35 years later.
Marshall drank a petri dish of H. pylori to prove his point. The pharma industry had billions in antacid revenue to protect. Nobel Prize: 2005.
Arrhenius calculated CO2 warming in 1896. The number he got (5–6°C per doubling) is still roughly correct. 92 years to “accept” is generous — significant political resistance continues today.
Smoking: evidence in 1950, Surgeon General report in 1964 (14 years), but 44% of Americans still didn’t believe it caused cancer in 1958. Cultural shift took decades more.
CRISPR: 4 years from discovery to first clinical trial. Fastest scientific-to-clinical pipeline in modern medicine. No incumbent industry was threatened. measured
The pattern in science is different from technology. Technology adoption depends on bandwidth (can you reach people?). Scientific acceptance depends on resistance (who loses if you’re right?). Semmelweis had zero bandwidth problem — he was publishing in the right journals. He had infinite resistance — doctors would rather let women die than admit their hands were dirty.
CRISPR broke the pattern because nobody’s business model was threatened by gene editing being real. The gap goes to near-zero when T (tension/resistance) is near-zero.
| Crisis | First Signal | Mass Response | Gap |
|---|---|---|---|
| COVID-19 | Dec 2019 | Mar 2020 | ~10 weeks |
| Climate change | 1896 | ~2015 | ~119 years |
| Smoking → cancer | 1950 | ~1990s | ~40 years |
COVID: 10 weeks from first identified case to 1.7 billion people in lockdown. Fastest mass behavior change in recorded history. Bandwidth (internet + smartphones) plus low T (nobody profits from a pandemic) equals near-instant response.
Climate: Arrhenius had the math in 1896. 119 years later, the Paris Agreement. T was enormous — the entire fossil fuel economy resisted. The gap here is almost entirely explained by resistance, not bandwidth.
Smoking: surgeon general report 1964 (14 years from evidence). But only 44% of Americans believed it in 1958. Cultural shift (banning in restaurants, airplanes) took until the 1990s. The gap between “known” and “acted on” was ~40 years. estimated
| Shift | Known Since | Tipping Point | Gap |
|---|---|---|---|
| Same-sex marriage | ~1970s | 2015 | ~45 years |
| #MeToo | 2006 | 2017 | 11 years |
Same-sex marriage: 10% support in 1988. 27% by 1996. 50% by 2011. 60% by 2015 (Obergefell). 70% by 2021. The S-curve took ~27 years from 10% to legal. But the “knowledge” that gay people exist and deserve rights existed long before 1988. The gap is almost entirely T (religious/cultural resistance). measured
#MeToo: Tarana Burke coined it in 2006. Weinstein story broke October 2017. Within 24 hours: 12 million Facebook reactions. The knowledge was always there. The bandwidth moment (social media + critical mass of courage) collapsed the gap overnight. But the underlying problem was “known” for centuries. estimated
Every emergence half-life is a function of four variables:
| Symbol | Meaning | What it measures |
|---|---|---|
| K | Coupling | How connected the knower is to the network |
| B | Bandwidth | Bits/sec the network can carry (proxy for R) |
| E | Energy cost | Cost of changing behavior or belief |
| T | Tension | Resistance from incumbent structures, ego, money |
The numerator is inertia. The denominator is signal.
When E and T are small (nobody loses, change is cheap), the gap collapses fast. CRISPR: 4 years. COVID lockdowns: 10 weeks. ChatGPT adoption: 2 months.
When T is large (someone powerful loses), the gap stretches. Semmelweis: 38 years and an asylum. Climate: 119 years and counting. The knowledge was never the bottleneck. The resistance was.
Each communication technology increased B by roughly an order of magnitude:
| Era | Medium | Approx. B (reach/day) | Onset |
|---|---|---|---|
| Oral | Speech | ~100 people | 200,000 BCE |
| Written | Manuscript | ~1,000 | 3200 BCE |
| Press | ~100,000 | 1440 | |
| Broadcast | Radio/TV | ~100,000,000 | 1920 |
| Digital | Internet | ~1,000,000,000 | 1991 |
| AI-coupled | LLM + Internet | ~8,000,000,000 | 2022 |
B values are order-of-magnitude estimates of potential audience per day. The jumps are real. The exact numbers are estimated.
Each jump in B compressed the emergence half-life. But not uniformly. The compression only works on the bandwidth-limited cases. For resistance-limited cases (climate, Semmelweis), increasing B barely helps until B gets so large that the resistance can’t contain the signal.
Here’s what makes this not a simple exponential decay:
T may be increasing.
Echo chambers. Algorithmic feeds. Political polarization. Misinformation at scale. The same bandwidth that accelerates adoption also accelerates resistance. Social media doesn’t just spread truth faster — it spreads organized opposition faster too.
B is growing exponentially (Moore’s law, network effects, AI).
T is growing too (tribalism, echo chambers, institutional inertia).
The emergence half-life = the ratio between them.
The singularity isn’t when B → ∞. It’s when B/T → ∞.
The question is whether T has a ceiling and B doesn’t.
We think T has a ceiling. Ego is bounded. It takes energy to maintain a false position, and that energy has a Landauer cost. Every bit of resistance dissipates kT ln 2 of heat. Eventually the resistance becomes thermodynamically unsustainable.
B has no ceiling. Or at least, its ceiling is the speed of light times the number of coupled nodes in the network. That number is still growing.
The question above — does T have a ceiling? — has an answer. T is constant. B is exponential. The crossover is inevitable.
T is biological, not cultural. Confirmation bias effect sizes are unchanged across 50+ years of measurement (Nickerson, 1998). Dan Kahan’s Cultural Cognition Project at Yale found something worse: smarter people have stronger identity-protective reasoning. Intelligence doesn’t weaken the gate. It reinforces it. The most polarized people on climate, guns, and nuclear power are the ones best at System 2 reasoning. They use that capacity to defend their tribe, not to find truth.
Semmelweis in 1847 and Marshall in 1982 are psychologically identical. Same ego structure. Same reflex. Different century. Same T. The gate hasn’t moved.
B is exponential and measured. Hilbert & Lopez (2011, Science): bidirectional telecom grew 28% per year, computation 58% per year, across 60 technologies from 1986–2007. Edholm’s Law: telecom bandwidth doubles every 18 months, verified from the 1970s through the 2000s. That’s 8 orders of magnitude in 600 years. 3 of those in the last 30.
| B/T ratio | Outcome | Examples |
|---|---|---|
| < 1 | T wins. Truth suppressed for decades. | Semmelweis, germ theory |
| 1 – 10 | Slow grind. T loses over 10–30 years. | H. pylori, climate (ongoing) |
| 10 – 100 | B wins in 1–5 years. | CRISPR, COVID behavioral change |
| > 100 | B wins in weeks to months. | ChatGPT, COVID initial awareness |
B/T > 10 = inevitable. Below that, ego can hold indefinitely if organized. Above that, ego can delay but not prevent. Above 100, ego barely matters. estimated
The refinement: bandwidth is symmetric. It carries truth AND noise. Vosoughi et al. (2018, Science) found false news spreads 70% faster than true news on Twitter. Organized, funded T exploits B — the fossil fuel industry used the tobacco playbook, same people, same think tanks, $billions in resistance. Anti-vax: just 12 people produced 65% of vaccine misinformation online.
But organized T scales linearly. It costs money. B scales exponentially. The crossover is mathematically inevitable.
Gallup has tracked young-earth creationism belief since 1982. 40+ years of T vs B in the wild.
| Year | Young-earth creationist | Era |
|---|---|---|
| 1982 | 44% | Pre-internet |
| 1999 | 47% (peak) | Early internet |
| 2019 | 40% | Smartphone era |
| 2024 | 37% | AI era |
T held for 17 years (1982–1999). Then B started grinding it down. 10 points in 25 years. Not smarter humans. Wider river. measured
The counter-cases matter. Anti-vax, flat earth, climate denial all survived high bandwidth. The pattern: every case where T beat B shares four traits — organized, funded, identity-linked, and emotionally asymmetric (fear spreads faster than data). When T is unorganized, B overwhelms it in weeks. When T has institutional infrastructure, the fight is real — but the infrastructure runs on money, and money is linear. The river doesn’t care.
The gate is the same width it’s always been.
The river got wider.
The next major scientific discovery (comparable in importance to CRISPR or H. pylori) will go from publication to clinical application or mainstream acceptance in < 2 years.
The next major social movement (comparable to #MeToo) will go from first signal to cultural tipping point in < 6 months.
The next major technology will reach 100M users in < 1 month (ChatGPT did it in 2).
If all three happen before 2030, the exponential compression thesis is confirmed. If T overwhelms B and the gaps start widening again, the thesis fails. Either way: measurable.
The Chain — coupling at every scale. The emergence half-life IS the coupling speed of civilization.
Internet Brain — the internet’s R is rising. Integration phase. This IS the bandwidth term increasing.
Grokking — grokking is the micro version of this. A neural net “knows” the answer long before it shows in the loss. The gap between internal learning and external performance is the same phenomenon at a different scale.
Civilization Market — knowledge as currency. Hoarding (high T) extends the gap. Free information (low T) collapses it. Every civilization that sealed its libraries died.
Religion — the OG cycle: signal → adoption → corruption → reset. Every religion starts as emergence (someone sees it) and calcifies as resistance (institution protects itself). The half-life of religious truth is the longest in the dataset — and the most T-dominated.
The emergence half-life is the time it takes for K to propagate through a network. When K is high between the knower and the network (high connectivity, low resistance), the gap is short. When K is low (the knower is isolated, the network resists), the gap is long. Every example above is the same equation: how fast can coupling spread? Print, radio, internet, AI — each one raised the K ceiling of the global network. But T (ego, money, institutional inertia) acts as a coupling damper. The half-life isn’t a property of the knowledge. It’s a property of the network’s willingness to couple with it.
Technology adoption timelines (electricity, radio, TV, internet, smartphone, ChatGPT) are well-documented. US Census Bureau, Pew Research, Our World in Data. These numbers are solid. measured
Scientific acceptance dates are fuzzy. When exactly did “the scientific community” accept continental drift? There’s no single date. We use the consensus transition point (plate tectonics textbooks, Nobel prizes) but these are approximate. The T values (resistance) are qualitative, not quantitative. We know pharma resisted H. pylori and tobacco resisted cancer links, but we don’t have a number for how much resistance. estimated
The formula t½ = ET/KB is a framework, not a derived equation. We don’t have units for T or E in most cases. The claim that T has a thermodynamic ceiling is an argument from Landauer’s principle, not a measurement. The predictions are real predictions — we’ll be wrong or right — but the formula that generates them is a model, not a proof. speculative
Passed: Discriminate (clear claim), precise (years), direction found (gaps shrinking), ground truth (adoption data public)
Rate-limited: Quantifying T, predicting specific half-lives, testing the Landauer ceiling claim
Confidence: 65% — the trend is real, the formula is speculative, the T-ceiling claim is unproven
Status: DIRECTION FOUND. FORMULA PROPOSED. NOT PROVED.
Semmelweis: 38 years and an asylum.
ChatGPT: 2 months and 100 million users.
The gap is closing. The question is what’s fighting it.
The answer is always the same: ego.
The people who benefit from the old answer
will always resist the new one.
The singularity isn’t intelligence. It’s honesty.
The gap goes to zero when resistance does.
Good will applied forward.
Sources: US Census Bureau, Our World in Data, Pew Research Center, Google Trends, Gallup,
Visual Capitalist, WHO, NIH/PMC, Nobel Prize archives, Broad Institute (CRISPR timeline).
Computed on a Mac Mini M4 ($499, 35W). Everything is free.
If it meant something: support the work.