Skip to content

Writing Feed

The Immortality Trap

Teddy Warner's GitHub profile picture Teddy Warner| Dec 2025 | 12–15 mins

This morning a friend sent me an intresting read: The Decline of Deviance an aggregate of data showing a massive collapse in risk-taking behavior over the past few decade. Teen alcohol use down 50%. Sexual activity down 40%. Fighting down 35%. Geographic mobility collapsed from 20% of Americans moving annually to 10%. Even our aesthetics became safer, British Museums analyzed photos of their artifacts over time and found a steady shift toward black, gray, and white, away from bold colors. Seemingly, we’ve entered an “epidemic of the mundane”.8

Changes in the color of objects over time Changes in the color of objects over time dark

Changes in the color of objects over time

At the time of writing this piece, I’m 20 years old. According to the Social Security Administration, my average life expectancy is 81.6, meaning I have roughly 61 years left on this planet.1

Every time I get in a car to go somewhere, I run about a 1 in 8.3 million chance of death.2 Every time I sit down to eat, about a 1 in 147 million chance of choking.3 Every walk I take, roughly 1 in 5.5 million.4 Every time I take up a simple convience, I wager the remaining 61 years of my life for said convience (i.e. I’m essentially wagering 61 years at 1:8.3M odds every time I drive).

Let’s conceive of a world of pragmatic ‘immortality’, one where the average lifespan is 1000 years. Our wager of years of life in exchange for convience still occurs, except in this world I’d be wagering 980 years for those same conveniences. The stakes increase 16 fold while the benefits remain near identical!

To state this precisely: the average drive is about 10 miles, that’s currently about 4 minutes of expected life loss per trip. With a 1000-year lifespan, that same trip costs 62 minutes of expected life. The wager is 15.8x larger.5

So what happens if we achieve our sought after ‘immortality’? Say we cure aging, dampen disease, and successfully extend the human lifespan to ~1000 years. The transhumanist dream.

I conjecture our lives would be no better than they are now.

But Teddy! Surely, by the time we have 1000-year lifespans, we’ll have much safer technology. And you would be right to claim this, the fatal risk associated with any given acction will continus to decline!

Between 1900 and 2024, life expectancy jumped 68%, from 47 to 79 years.6 During the same period, transportation got dramatically safer and faster. Traffic deaths dropped from ~18 per 100 million miles in the 1920s to 1.20 today, a 15x improvement.7 Average speeds increased from 8 mph (stagecoach) to 45 mph (cars), roughly 5.6x faster.

Life expectancy vs transportation speed 1900-2024 Life expectancy vs transportation speed 1900-2024 dark

Life expectancy vs transportation speed 1900-2024

For most of the last century, technological safety and speed massively outpaced our lifespan. Risk per trip decreased 110x while stakes only increased 1.3x. Subjectively, travel got safer despite having more years of life at risk.

Yet, unfortunately, this trend tapered off towards the end of the 20th century. Since 1970, transportation speeds have stagnated or declined. The Boeing 747 (1970) cruised at ~600 mph. The Boeing 787 (2024)? only ~560 mph. We had supersonic passenger flight with the Concorde at 1,350 mph - we abandoned it in 2003. US highway speeds have been stuck at 55-75 mph for fifty years. Meanwhile, mean life expectancy kept climbing, with about a 12% increase (from 70.8 to 79.2 years).

I find the Concorde’s retirement to reveal something profound: people with longer lives value safety over speed. They’d rather fly 7 hours at 560 mph than 3.5 hours at 1,350 mph if the crash rate is marginally higher. With 1000-year lifespans and those same risk ratios, would anyone fly at all?

Or look at playgrounds - I grew up playing in this massive wooden playground in the Chicago suburbs, with a somewhat ever-present risk of injury.12 Today, it is illegal to build a sandbox in a playground in San Francisco.13 As our life expectancy rises, it makes more sense to optimize toys and playgrounds for preventing lawsuits rather than developing children who know how to manage risk.

Our modest (~12%) lifespan increase since the end of the last century has already caused massive behavioral changes. The value of statistical life has jumped from $2.5M in 1982 to $12.5M in 2023 (inflation-adjusted).9 That’s a 5x increase in what society would ‘pay’ to prevent one death, while GDP per capita only grew 2.5x (so clearly wealth alone isn’t our driving motivator).

Value of statistical life 1982-2023 Value of statistical life 1982-2023 dark

Value of statistical life 1982-2023

If we try extrapolating this out over even the next 500 years, we can assume a near-complete eridication of risk. Not necessary because people are conscious of this, but rather because the expected value of any convience shifts dramatically. If people’s base state is one of higher comfort, they have substantially more time to do things, and doing any rapid convienent thing is wasting this additional time, of course, society becomes sluggish!

This derisking problem compounds even further when we consider that risk aversion grows with age. How much longer will a parent “baby” their child in a society of 1000-year lifespans? The age of those setting the behavioral norms will continously shift higher and hider, and society will become more and more risk adverse. In a society with 1000 year lifespans, the median person (presumably the cohort of people setting behavioral precidant) would be 500 years old. If risk adversion continues growing with age at the rate observed in current populations (roughly 1.5% per year of age), a 500 year old would be about 8.5x more risk-averse than the average 20 year old today.10 92% of the population would be more risk-averse than today’s 80-year-olds. Can you imagine what a pathologically conservative culture we would regress into? That all “young” people would inherit and thus compound?

Risk aversion vs age projected over 500 years Risk aversion vs age projected over 500 years dark

Risk aversion vs age projected over 500 years

This growing risk aversion with age likely already incorporates diminishing marginal utility of additional years - when you have fewer years left, each one is more precious. This only intensifies with radical life extension, making the compounding risk aversion even more severe.

Consider what drives progress in human society: coordinated overinvestment in ‘uncertain’ outcomes. Our tolerance for risk is everything if our fight against stagnation. Transhumanists want radical life extension, technological acceleration, and human flourishing (which in principle I find lovely), yet this dream contains a fatal contradiction: pursuing the first destroys the second, which destroys the third. More years of existence doesn’t mean more accomplished. It might mean less.

I believe the alternative is quite obvious: rather than drastically extending our lifespan, extend cognative peak within roughly normal lifespans. Extending cognitive peak to age 80 within a 100-year lifespan produces about 52 productive years of output, or 0.52 output per year of life. Extending lifespan to 1000 years with cognitive peak to age 300 produces 152 productive years, sounds better, but that’s only 0.15 output per year. The former is 3.5x more efficient at generating human value.

Cognitive extension vs radical life extension Cognitive extension vs radical life extension dark

Cognitive extension vs radical life extension

There’s something quite poetic about this. The obsession with living longer, our quintessential measure of medical progress, may actually hinder us in the long run. Every time I walk to my office, I’m making a bet. With 61 years left, I’ll take that bet. With 980, I’m not so sure. And when enough people stop taking enough bets, civilization stagnates.

The goal shouldn’t be more years, it should be more productive years. Quality over quantity if you will.

All my projections extend wayyyy beyond biological plausibility - we have emperical data to age ~85 and extrapolating to 500-1000 years require assumptions that may not hold. Nonetheless, the societal implications remain dire.

  1. Social Security Administration, Period Life Table, 2011. Life expectancy for 20-year-old males: 81.6 years. https://www.ssa.gov/oact/TR/2011/lr5a4.html ↩

  2. Driving risk calculation: NHTSA reported 1.20 deaths per 100 million vehicle miles traveled in 2024 (https://crashstats.nhtsa.dot.gov/). Average trip length is approximately 10 miles (DOE data). Per-trip risk: \( \frac{1.20}{100,000,000} \times 10 = 1.2 \times 10^{-7} = \) 1 in 8.3 million. ↩

  3. Choking risk calculation: Approximately 5,000 choking deaths per year in the US (CDC/Wikipedia, 2024). Of these, roughly 50% (~2,500) are food-related, with the remainder caused by non-food objects such as coins, toys, and other small items. US population consumes roughly 1,095 meals per person per year (3 meals/day × 365 days). Total annual meals: 335M people × 1,095 = ~367 billion meals. Per-meal risk from food-related choking: \( \frac{2,500}{367,000,000,000} = 6.8 \times 10^{-9} = \) 1 in 147 million. ↩

  4. Walking risk calculation: GHSA reported approximately 7,148 pedestrian deaths in 2024. Americans take roughly 39 billion walking trips annually (NHTSA). Per-walk risk: \( \frac{7,148}{39,000,000,000} = 1.8 \times 10^{-7} = \) 1 in 5.5 million. ↩

  5. Expected loss formula: \( E[\mathrm{Loss}] = \text{Years}_{\text{at\_stake}} \times P(\text{death} \mid \text{activity}) \times \text{Activity}_{\text{distance}} \). For a 10-mile car trip at age 20 with 61 years remaining: \( 61 \times \left(\frac{1.2}{100,000,000}\right) \times 10 = 7.32\times10^{-6} \) years = 3.9 minutes. For 1000-year lifespan with 980 years remaining: \( 980 \times \left(\frac{1.2}{100,000,000}\right) \times 10 = 1.176\times10^{-4} \) years = 61.9 minutes. Ratio: \( \frac{61.9}{3.9} = 15.8\times \) larger wager. ↩

  6. Life expectancy data: 1900 value (47.3 years at birth) from CDC NCHS Historical Data, “Death Rates and Life Expectancy at Birth,” available at https://www.cdc.gov/nchs/data-visualization/mortality-trends/index.htm and https://blogs.cdc.gov/nchs/2020/11/20/7035/ 2024 value (79.25 years at birth) from Macrotrends, “U.S. Life Expectancy 1950-2025,” https://www.macrotrends.net/global-metrics/countries/usa/united-states/life-expectancy Increase: 31.95 years = 68% growth. ↩

  7. Traffic fatality rates: 1923 value (18.65 deaths per 100 million vehicle miles traveled) from National Safety Council, “Historical Car Crash Deaths and Rates,” Injury Facts, https://injuryfacts.nsc.org/motor-vehicle/historical-fatality-trends/deaths-and-rates/ 2024 value (1.20 deaths per 100 million VMT) from NHTSA, “Early Estimates of Motor Vehicle Traffic Fatalities in 2024,” DOT HS 813 710, April 2025, https://www.nhtsa.gov/press-releases/nhtsa-estimates-39345-traffic-fatalities-2024 and https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813710 Improvement factor: 18.65/1.20 = 15.5× safer per mile traveled. ↩

  8. Mastroianni, Adam. “The Decline of Deviance.” Experimental History, 2024. https://www.experimental-history.com/p/the-decline-of-deviance Documents comprehensive behavioral changes from 1990-2024 including: teen alcohol use (-50%), sexual activity (-40%), fighting (-35%), drug use (-20-50%), teenage pregnancy (-67%), geographic mobility (20% → 10% moving annually). ↩

  9. Value of Statistical Life data from U.S. Department of Transportation. 1982: $2.5M (inflation-adjusted to 2023 dollars). 2023: $12.5M. Increase: 5x. Source: https://www.transportation.gov/office-policy/transportation-policy/revised-departmental-guidance-on-valuation-of-a-statistical-life-in-economic-analysis ↩

  10. Age-based risk aversion from Jianakoplos, N. A., & Bernasek, A. (2006). “Financial risk taking by age and birth cohort.” Southern Economic Journal, 72(4), 981-1001. Found risky asset ratio declined from 0.65 (age 30s) to 0.40 (age 70s), representing a 62.5% increase in risk aversion over 40 years. Dohmen et al. (2011) confirm risk aversion rises with age using panel data from Netherlands and Germany (https://cepr.org/voxeu/columns/identifying-effect-age-willingness-take-risks). The linear growth model \( R(\text{age}) = R_0 (1 + 0.015 \times \text{age}) \) is parameterized to match these empirical trends showing roughly 62.5% increase in risk aversion over 40 years, or approximately 1.56% per year. At age 500, this yields \( R(500) = R_0(1 + 0.015 \times 500) = 8.5 \times R_0 \). ↩

  11. Color & Shape: Using Computer Vision to Explore the Science Museum Group Collection, Science Museum Group, 2020. Changes in the color of objects over time. https://lab.sciencemuseum.org.uk/color-shape-using-computer-vision-to-explore-the-science-museum-c4b4f1cbd72c ↩

  12. Indian Boundary Park wooden playground constructed in 1989 with volunteer help from neighborhood residents. Source: Rogers Park/West Ridge Historical Society, https://rpwrhs.org/2014/08/04/indian-boundary-park/ ↩

  13. San Francisco banned sandboxes from city playgrounds due to bacterial contamination concerns. https://www.sfchronicle.com/bayarea/article/Citing-safety-hygiene-SF-parks-considers-11437854.php?psid=bLEge ↩

Comments

Copyright © 2025 Teddy Warner