The Quiet Galaxy

The equation · piece 7 of 7 · L

How long a civilisation stays detectable, in years

Once a civilisation becomes detectable, how many years it stays that way before it dies out, falls silent or changes into something we would not recognise. Multiplied by the rate at which civilisations appear, this gives how many are detectable at any one moment, which is why Drake's 1961 meeting concluded that N is roughly equal to L. Our own technological civilisation has been detectable for about a century. The published figures run from a few hundred years, read off the lifetimes of historical empires, to billions of years for a civilisation that spreads beyond its home planet; the gap between them is the single widest in the equation after the chance of life starting.

The range this site uses 100 to 10 billion years reviewed
set on 16 October 2026

When the number on the home page is computed, a value for this piece is picked at random from this range each time, and every tenfold step between the two ends is equally likely.

Why that range

The 2018 paper's span of 100 to ten billion years, kept as it is. The floor is Westby and Conselice's century, which is how long our own civilisation has been detectable so far; Shermer's three to four hundred years sits just above it. Garrett's 'mere decades' (2025) is shorter, but it is a model of what might happen to a machine-based civilisation and not a measurement, and the one measured case has already lasted about ninety years, so the floor was left at 100 (ruled by John 2026-10-17 on the outside review, item 18). The ceiling is the age of the galaxy, past which the equation's assumption of a steady rate of appearance breaks down, and it covers Grinspoon's billion years and Drake's hundred million. Evenly spread across its powers of ten because nothing favours any power of ten between the two: the argument is between 'civilisations burn out' and 'civilisations become permanent', and both are published.

Rests on: Sandberg and others, 2018.

The published estimates

7 estimates on the record, oldest first. Each says who made it, when, how, and where to find it. A range is set from these by hand, with the reason above, never computed from them: the reason is the arguable part.

WhenWhoEstimateHow it was reached
1961 Frank Drake and the Green Bank meetingGreen Bank meeting of November 1961, as recorded in Drake and Sobel, Is Anyone Out There? (Delacorte, 1992), pages 55 to 62 (drake-1961) 1000 to 100 million years Educated guess, from a thousand years for a civilisation that destroys itself to a hundred million for one that does not; the meeting's spread of N, 20 to 50 million, came almost entirely from this factor.
May 1993 J. Richard GottGott, 'Implications of the Copernican principle for our future prospects', Nature 363, 315–319 (1993) 5100 to 7.8 million years Reasoned that we are unlikely to be observing our species at an unusual moment of its history: given about 200,000 years so far, the human species should last between 5,100 and 7.8 million more years with 95% confidence. A statement about our species, not about its detectable phase, recorded because it is the best-known attempt to put a number on how long we have.
August 2002 Michael ShermerShermer, 'Why ET hasn't called', Scientific American 287(2), 21 (2002) 304 to 420 years Averaged the recorded lifetimes of sixty historical civilisations on Earth, 420 years, and of the twenty-eight since the fall of Rome, 304 years (the column gives 420.6 and 304.5), and took that as the likely lifetime of a technological civilisation.
2004 David GrinspoonGrinspoon, Lonely Planets: The Natural Philosophy of Alien Life (HarperCollins, 2004) 1 billion years Argued that a civilisation which survives its first dangerous centuries may become effectively immortal, so that the factor should be replaced by the share of civilisations that become immortal times the age of the galaxy; a billion years is the figure the argument reaches.
June 2018 Anders Sandberg, Eric Drexler and Toby OrdSandberg, Drexler and Ord, 'Dissolving the Fermi Paradox', arXiv:1806.02404 (2018), Table 1 (sandberg-2018) 100 to 10 billion years A span evenly spread across its powers of ten, from the published estimates (which they read as 50 years to ten billion), cut off at the age of the galaxy because the equation assumes civilisations appear at a steady rate.
June 2020 Tom Westby and Christopher ConseliceWestby and Conselice, 'The Astrobiological Copernican Weak and Strong Limits for Intelligent Life', Astrophysical Journal 896, 58 (2020) (westby-2020) 100 years Assumed: the average lifetime of a communicating civilisation is taken to be 100 years, 'based on our own at present', and their count of 36 scales with it.
September 2025 Michael GarrettGarrett, 'Blink and you'll miss it: How Technological Acceleration Shrinks SETI's Narrow Detection Window', arXiv:2509.23632 (2025), accepted by Acta Astronautica (garrett-2025) · in the log 10 to 100 years Argued from a simple model that a civilisation whose technology accelerates fast, as a machine-based one might, passes through the phase in which we could detect it in 'mere decades' (the paper's abstract); the paper gives no number in its abstract, and 10 to 100 years is how 'decades' is read here. Added on 2026-10-17 (outside review, item 19); it lies below the range's floor of 100, which item 18 left where it is.

What would move this

The discovery that would narrow the range

A published argument with a number on how long technological civilisations last (from history, from risk studies of our own prospects, or from the physics of long-lived machines), or a detection whose age bounds it.

When a paper or a survey of that kind is published, it is logged in the story so far with a note saying whether it moved this range. A move is written as a dated revision below, with the old number, the new one and the reason.

The last reading

Once a week the records are read against the news, each against the test above. Open a line to see what that week held. Every reading ever made of this record is on the reading page.

  • 2026-10-16 Logged, no change By hand

    Four published arguments with a number or a reinterpretation, none in the log: Rahvar and Rouhani (2026-02-24, accepted in MNRAS Letters) turn the silence into an upper bound of about 5,000 years if life and intelligence are common; Blanco, Haqq-Misra and Profitiliotis (2026-04-15) simulate collapse and recovery and find civilisations active for 38% to 100% of their span, so the piece is better read as time spent detectable; Garrett (2025-09-28) argues the detectable phase may last mere decades; Bertolami (2025-10-08) proposes replacing the piece with an energy-and-search term. The 5,000-year bound is conditional on the other pieces being large, so narrowing the range on it would count the silence twice; the range of 100 to 10 billion years stays.

    Rahvar, 2026Blanco, 2026Garrett, 2025Bertolami, 2025 source 1source 2source 3source 4

How the range has moved

DateWhatFromToWhy
17 October 2026 range.why The floor is the shortest published figure, Westby and Conselice's century The floor is Westby and Conselice's century; Garrett's 'mere decades' is shorter but is a model and not a measurement, so the floor stays Ruled by John 2026-10-17 on the outside review (item 18, declined): the old wording called the century the shortest published figure, which stopped being true once Garrett's paper (2025) was added as an estimate. The range itself is unchanged.Because of Blink and you'll miss it: How Technological Acceleration Shrinks SETI's Narrow Detection Window.
17 October 2026 estimate six published estimates seven: Garrett (2025), 10 to 100 years, added Ruled by John 2026-10-17 on the outside review (item 19). Rahvar and Rouhani's 5,000-year bound (item 20) was declined as an estimate and stays in the log. The range is unchanged.Because of Blink and you'll miss it: How Technological Acceleration Shrinks SETI's Narrow Detection Window.
17 October 2026 range.status first-pass reviewed The outside review of the seven ranges ran on 2026-10-17 and John ruled on every item the same day (docs/reviews/2026-10-17-rulings.md).

The explanations that lean on this piece

An explanation for the silence is, in the end, a claim about one or more of the seven pieces. 9 explanations say this piece is smaller than it looks; 1 says it is larger.

The papers that bear on it

Books and films that take it up

Notes

Westby and Conselice (2020, westby-2020) make this the deciding factor explicitly: taking our own century as the average lifetime of a communicating civilisation, and life and intelligence as following on any suitable planet, they get at least 36 (between 4 and 211) communicating civilisations in the galaxy, the nearest at most about 17,000 light-years away, which is to say far beyond the reach of any present search. Their weaker assumption (intelligence at any age after 5 billion years; the paper's 'Weak Copernican' case) gives 928 civilisations, the nearest at most about 3,300 light-years away. They say a search is expected to succeed if civilisations last on average about 3,060 years under the strong condition and about 1,030 years under the weak one (read from the arXiv version of the paper, 2026-10-16). Checked 2026-10-16 against the primary sources: Shermer's two figures are in his column (reprinted on his own site); Grinspoon's could not be checked (see the open item above). The 'they are gone' family of explanations (the Great Filter ahead of us, Hanson 1998, hanson-1998; self-destruction) are arguments that this factor is short; the founding pessimist papers (Hart 1975, hart-1975; Tipler 1980, tipler-1980) argue instead that no civilisation has lasted long enough to spread here, which bounds this factor and the previous one together. The 2018 paper cuts its span off at ten billion years because a civilisation older than the galaxy cannot have appeared at a steady rate. Written 2026-10-16. Outside review, 2026-10-17, ruled by John the same day: lowering the floor from 100 to 10 years (item 18) was declined, narrowly: the only measured case, ours, has been detectable for about ninety years and counting, and the figures below a century are arguments and not measurements. Garrett (2025, garrett-2025) argues the detectable phase may last 'mere decades'; it is now in the estimate list (item 19) with the span of 10 to 100 years as how 'decades' reads, the paper giving no number of its own. To revisit if a paper gives a figure below 100 with a method behind it. Rahvar and Rouhani (2026, rahvar-2026) stay in the log and linked from this factor but are not an estimate (item 20): their 5,000 years is an upper bound derived from the silence if the other six factors are generous, and using it would count the silence twice.

Claims still to be checked

These were written from memory and have not yet been checked against a source. A later pass clears each one.

  • David Grinspoon's 'potentially billions of years' is quoted from the Wikipedia article, which cites Lonely Planets (2004); check the book, and whether a billion is his figure or the article's rounding. Tried 2026-10-16 and still open: the Internet Archive copy is not readable (lending only) and Google Books refused the query. Second-hand only: a Drake-equation paper (arXiv:1705.07816) says Grinspoon replaces the factor by the share of civilisations that become immortal times how long that has been going on, which 'may approach the age of the universe'; so the structure in the method line holds, but a figure of one billion is not confirmed, and the book's first publication was 2003 (the record says 2004, the paperback).

Think the range should move? Argue it, with a source. Spotted an error on this page? Say so. Or read how disagreement gets folded in. The record itself is data/factors/lifetime.toml in the public repository.