Civilization Timelines

A periodisation scheme cuts the human record into named stages and then compares across them. It makes four separable claims and almost every dispute in this subject is about which one is being asserted. That the boundary is real — something changed. That the stages are ordered and the order is general — everyone passes through them in sequence. That the stages are comparable across cases — “Bronze Age China” and “Bronze Age Denmark” name the same kind of thing. And that the scheme has explanatory or predictive purchase — knowing the stage tells you something else. A scheme can satisfy the first and third and be useless; a scheme can be useful and fail the second entirely. The good schemes and the bad ones fail at different points, and holding the four apart is the only way to read the evidence without arguing past it.

The framing under test is that civilizational development follows stages that can be periodised and compared. Comparison works. Periodisation does not, and the field's own best programme is what demonstrates it. The most serious attempt ever made to test this framing returned the strongest possible result for the second half of the claim and then, using the same data, took the first half apart three times over. What survives is a measurable axis of social complexity with no natural joints on it.

Established Begin with the one clean success, because its structure explains every subsequent failure. Christian Jürgensen Thomsen took charge of a Danish prehistoric collection in 1816 and organised it by material — stone, bronze, iron — publishing the scheme in 1836. His predecessor Rasmus Nyerup had written that the objects were merely “endlessly old”, with only “conjectures and plausible hypotheses” about their age. Thomsen's method was empirical association: he examined “what other objects they were found together with.” He was demonstrating the arranged collection to visitors by 1827 and the scheme was broadly accepted by the mid-1830s. Established His own hedge is the important part. He wrote that the theory “needed to be studied more thoroughly before it could become more than just a presumption” — published as a working hypothesis requiring stratigraphic validation, and it got it. Frontier Three properties made it work: it was a claim about a specific region, it was grounded in a physical checkable proxy, and it was offered as a presumption rather than a law. Every periodisation scheme that has gone badly since has been global, proxied by something soft, and asserted. The source here is a Danish cultural institution recounting Danish archaeology, an interested party, used only for dates and for Thomsen's own words.

Established Now the framing's best case, and it is genuinely strong. Turchin, Currie, Whitehouse, François, Feeney, Mullins, Hoyer, Collins, Grohmann, Savage and forty further co-authors analysed 414 societies across 30 world regions over the last 10,000 years on 51 variables spanning social scale, economy, governance and information systems. A single principal component captures around three-quarters of the observed variation, different characteristics of social complexity are “highly predictable across different world regions”, and the authors conclude that “key aspects of social organization are functionally related and do indeed coevolve in predictable ways.” Established If one dimension explains 75% of variation across 414 societies, 30 regions and ten millennia, then societies are comparable along a single axis and locating a case on it is a defensible operation. Frontier It was contested immediately, in the same journal, by a historian and two statisticians. Only the bibliographic record of that letter was obtainable for this brief and no specific statistical criticism is attributed to it here — the disagreement is recorded, its content is not.

Established Then the same programme dissolved a seventy-year-old period. The Axial Age posits a mid-first-millennium-BCE pivot across China, Greece, India, Iran and Israel-Palestine: a transformation in the relationship between rulers and ruled and the emergence of moralising religion. Mullins, Hoyer, Collins, Currie, Feeney, François, Savage, Whitehouse and Turchin tested the proposals systematically against global comparative historical evidence, and the concluding chapter of the resulting volume gives the verdict in a sentence: “So, was there an Axial Age? We suggest the answer is 'sort of,' but it was not so much an age as a stage in the evolution of social complexity.” A coherent turn towards axiality “did not occur everywhere, and in the regions where it did emerge, the patterns and rates of emergence differ.” Established A named period held for seventy years dissolved into a position on a complexity axis when tested by researchers sympathetic to the idea of testing it. That is the cleanest demonstration available that a periodisation boundary can be falsified.

Established And then the programme's most prominent single result was retracted. Whitehouse, François, Savage, Currie, Feeney, Cioni, Purcell, Ross, Larson, Baines, ter Haar, Covey and Turchin published Complex societies precede moralizing gods throughout world history in Nature on 20 March 2019. It was retracted on 7 July 2021. The reanalysis that forced it, by a fifteen-author group from outside the programme, found that 61% of the data points on moralizing gods in the databank are missing values, mostly from populations under a million; that the original analysis re-coded those missing points as confirmed absences; and that confining the analysis to extant data, or applying standard imputation, reverses the finding — moralizing gods precede increases in social complexity. Their conclusion is that “the reported 'megasociety threshold' for the emergence of moralizing gods is an artefact of the decision to re-code all missing data as known absences.” Established The retraction note's own words are that the authors should have labelled moralizing gods as “'absent' or 'inferred absent' rather than 'unknown'… rather than converting 'NAs' to zeros during the phase of analysis”; they state their refined analyses still support their original conclusions but that the differences were “substantial enough to warrant a Retraction.” Frontier And it was not unanimous: “The authors John Baines, Alan Covey and Kevin Feeney do not agree with this Retraction.” That is a live minority position held by named people, and it is carried here rather than smoothed over.

Frontier The correct reading is not that the databank is discredited. The retraction demonstrates that it is auditable: outside critics obtained the data, found the missingness, reran the analysis and reversed the result. That is the process working, and it is more than most comparative historical projects can support. Established What it establishes is narrower and far more useful: in comparative historical databanks the treatment of absence is not a technical detail, it is the finding. A periodisation that says X appears at threshold T is a claim about where X was absent, and in this record most of the absences are unknowns. Frontier There is a live consequence for anyone reading the Axial Age work: its concluding volume restates the megasociety threshold, which no longer stands. The Axial Age assessment is a separate paper in a different journal and was not itself retracted, but it leans on a retracted result and anyone citing it must say so.

Established The most-tested stage theory in the record splits, and declaring the tie is the honest terminal position. Rostow's five stages, with a take-off into self-sustaining growth as the pivot, were comprehensively criticised within a year. Itagaki's 1961 collection carries the objections: Berrill asking “is there a genuine discontinuity rather than a simple acceleration of growth?”; Cairncross observing that “nearly all the economies one looks at historically were in continuous flux” and, decisively, that “there are no definitions of the successive stages that admit of their identification by reference to verifiable criteria”, allowing only traditional and modern industrial society as verifiable patterns; and Baran, that the theory “reduces economic growth to a single pattern by classifying any and every country only in respect to its position on their stepladder” and “tells us nothing except that there are stages.” Frontier The modern quantitative revisit reaches a partly opposite conclusion. Costa, Kehoe and Raveendranathan cover 223 countries grouped into 135 countries and 8 country groups over 1960 to 2010, with some series back to 1820, and find that take-off is measurable, real, and has become far easier than Rostow thought: “more than 50 percent of the world's population lived in countries in the Malthusian trap in 1960… By 2010, less than 2 percent.” Progression is not inevitable — countries can fail to progress — and catching the leader has become harder. Established A periodisation can be operationally unusable and substantively right about the one boundary that matters, and this one is both.

Frontier On the single most consequential boundary in modern economic history, serious scholars using overlapping data disagree by five hundred years. Historical national accounting places the divergence between Europe and Asia in the late medieval period, “already well under way during the early modern period, contrary to the recent revisionist views”, with Britain and Holland catching Italy and Spain from 1348 and forging ahead from 1500, against Mediterranean levels of roughly $1,100 to $1,500 per capita in 1990 dollars. The California School position dates the split to around 1800. Frontier That position is stated here as its opponent characterises it — its canonical statement was not obtained for this brief — and the column reporting the earlier dating is a research column by the author about his own work, not the underlying paper. Established Both sides use the same measurement infrastructure, a database covering 169 countries through 2022, and its release page carries no explicit uncertainty statement for the historical estimates. A dataset that dates the most consequential boundary in modern economic periodisation ships without published confidence intervals on its pre-modern figures. Frontier A five-century disagreement is the clearest available evidence that periodisation boundaries are argued into the record rather than read off it.

Established And the best-documented periodisation dispute in existence was settled by a vote, which both sides are entitled to read differently. The Anthropocene Working Group proposed a new geological epoch with its reference point at Crawford Lake, Ontario, dated to about 1950 and marked by artificial radionuclides from nuclear testing. On 4 March 2024 the Subcommission on Quaternary Stratigraphy voted 12 to 4 against, with 3 abstaining and 3 not voting; the international union validated the vote later that month. The arguments made against, as reported: that an epoch-level division is too large-scale next to the Oligocene or Miocene, and that human environmental impact spans “40,000 years or more” making a mid-twentieth-century start arbitrary; that “eighty years is a drop in the ocean in geological terms”; and that “the Holocene clearly hasn't ended, as its fauna and flora are still here.” Frontier The process was disputed by a named participant, who identified “breaches in the decisionmaking process” — including that the working group could not revise its proposal after discussion stopped, excluding recent findings — and “several statutory violations that call the validity of the vote into question.” Established The underlying empirical claim is separate and it stands. Steffen and colleagues tracked 24 indicators — 12 socio-economic and 12 Earth-system — over 1750 to 2010 and report that “only beyond the mid-20th century is there clear evidence for fundamental shifts in the state and functioning of the Earth System that are beyond the range of variability of the Holocene and driven by human activities.” The disaggregation also shows that in 2010 the OECD countries accounted for 74% of global GDP and 18% of global population. Established Both sides are right about different claims. Something measurably changed around 1950; a formal epoch boundary was rejected as the wrong instrument for recording it. The word survives the vote; the epoch does not.

Frontier Are there stages at all, or is the record one continuous accelerating curve? Kremer models population growth as proportional to population size — bigger populations generate more innovations, which support larger populations — producing super-exponential growth from a single feedback loop. He tests it on 37 population measurements from one million BCE to 1990, finding a robust relationship in which larger global population grows faster, and then on a design that is genuinely clever: five separated landmasses — the Old World, the Americas, Australia, Tasmania and Flinders Island — from 10,000 BCE to 1500 CE, where the model predicts that larger land areas support higher population densities and technology levels. Confirmed. Frontier If one continuous mechanism generates the whole one-million-year record, the stages are artefacts of where an author puts tick marks on a smooth accelerating curve. This is the strongest argument against the reality and generality of any periodisation boundary, and it should be weighed before any stage theory is described rather than after.

Speculative Against that, stages with doubling times attached. Hanson fits two million years of world product as a sequence of exponentially growing modes: hunting with a doubling time of 224,000 years from about 2,000,000 BC, farming at 909 years from about 4,860 BC, and industry at 6.3 years, with acceleration ratios of roughly 247 times from hunting to farming and 144 times from farming to industry. A projected fourth mode would double in “either .05, 1, or 4 years” on one model and in 9 to 16 days on the preferred one, with the transition “within roughly the next century”. It is an unpublished working paper and is flagged as one. Established Hanson's own caveats are more useful than his forecast and one of them is fatal to the method. He writes that “it is hard to see in much detail how the world economy could possibly grow that fast”, that timing estimates “seem less reliable than growth rate estimates”, and — the load-bearing admission — that the model would have mispredicted the industrial revolution's timing if built using only pre-1900 data. Handwave A stage model that cannot retrodict the most recent stage boundary from data preceding it has demonstrated no ability to predict the next one, and this is the author's own finding about his own model.

Handwave The other well-known fit reaches a similar date and that is not corroboration. Roodman fits gross world product from 10,000 BCE to the present, finds super-exponential growth in which “the bigger it has gotten, the faster it has doubled, on average” with a power law outperforming an exponential, and reports a trajectory approaching infinite output near 2047 — while acknowledging that the model is “inherently unstable”. The report itself was not obtained for this brief and is cited via a third-party forum summary of it. Frontier The deviation is the interesting part and it points the other way: the last thirty years sit at roughly the 25th percentile of the model's own distribution, and the last seventy years “are fit quite well by an exponential trend” rather than a power law. An attempt to include finite resources “does not go well.” Handwave Two hyperbolic fits to overlapping product series will place their asymptotes in similar places regardless of the model wrapped around them. A shared 2047 is a property of the functional family, not two independent pieces of evidence.

Frontier And the counter-case belongs beside them, because it is fitted to real data too and predicts the opposite boundary. Bloom, Jones, Van Reenen and Webb find research effort rising substantially while research productivity declines sharply, with Moore's law as the worked example: more than eighteen times as many researchers are required today for the same doubling of chip density as in the early 1970s. If the input requirement for fixed output rises exponentially, the accelerating-modes story and the harder-ideas story make opposite predictions about the next transition. Frontier Declare the tie. Both are measured; neither has been tested against the other on a horizon anyone has lived through. The extrapolation methods themselves belong to Technology Forecasting; what this brief owns is the claim that history divides into modes with characteristic doubling times.

Speculative Can a comparative databank date a transition at all? The retraction says the direction of at least one high-profile answer was decided by a coding convention rather than by evidence. Frontier The general form of the problem has no proposed solution: a periodisation is a claim about absence, comparative historical records are dominated by unknowns, and no standard exists for distinguishing inferred absence from missing data across a databank. Whether such a standard can be built, or whether the honest output is always a range of results under stated conventions, is open and nobody has posed it as a research programme.

Speculative And the synthesis question this brief poses as its own, flagged as such. If a five-century disagreement about the Great Divergence, a seventy-year-old Axial Age dissolving on test, and four professionally built technology trees agreeing on 30% of their nodes are all instances of the same thing, then periodisation is a research instrument rather than a discovery — boundaries are argued into the record, not read off it. Speculative No located author argues this as a thesis; it is assembled here and should be read as the brief's own construction. Technology Trees of Civilization reaches a structurally identical conclusion about technological rather than historical sequence.

Established The bottleneck in this subject is not measurement capacity. It is the absence-versus-unknown distinction, and it is present in every comparative historical record ever assembled. The retracted moralizing-gods result is the demonstration: 61% of the relevant data points were missing, recoding them as absences produced the threshold, and handling them properly reverses the direction. Established A periodisation says X begins at time T. That is not a statement about where X is present; it is a statement about where X is absent — and in every comparative historical record most absences are unknowns. The scheme is therefore a hypothesis about missing data wearing the costume of a hypothesis about history.

Frontier The second bottleneck is that the measurement infrastructure ships without uncertainty. The standard historical national accounts database covers 169 countries through 2022 and its release page carries no explicit uncertainty statement for the historical estimates. Speculative A five-century argument about the Great Divergence is being conducted on figures whose error bars are not published, and the argument cannot be resolved by more precision because no one has stated the current precision.

Frontier The third is that the observable set is one. Every claim in this brief about the sequence a civilisation passes through is fitted to a single case with internal variation. The cosmic-scale schemes make this explicit — a classification with no observed instances above the base type cannot be validated by comparison, because there is nothing to compare to. Speculative The comparative historical databanks partially escape this by treating 414 societies as cases rather than one civilisation as a case, which is exactly why the single-principal-component result matters, and exactly why the coding conventions inside it matter as much as the statistics.

Established Nothing on this map produces a result this brief waits on, and the honest record is a short list of things that do not exist. No brief here supplies a comparative historical databank, a missingness convention, or an independently originated biosignature. Frontier What the subject waits on is one scientific result and one institutional artefact, both recorded as typed constraints in section 13.

Frontier The scientific one is specific enough that an astrobiologist would recognise it immediately: an independently originated biosignature. The Great Filter argument's whole content is a claim about where along the path the improbability sits, and its evidential rule is unusual in being stated in advance — discovering independently evolved life elsewhere is bad news, because it implies the early steps are easy and the barrier lies ahead. Frontier A confirmed second origin would move the posterior in one direction and a confirmed absence across many surveyed worlds would move it in the other; nothing else on offer does either. Nobody is currently producing that result, and no brief on this map is going to.

Established The institutional one is a comparative historical databank with published missingness conventions, audited independently, versioned so that a reanalysis can be reproduced. The one large databank in existence turned out to be auditable in practice — outside critics got the data and reversed a Nature result — and that is a genuine achievement. Frontier What it lacks is a stated convention fixed before analysis, which is the thing that would have made the retraction unnecessary. Nothing about supplying it is a research problem.

Established The cheapest valuable experiment is a pre-registered periodisation test, and it has been run roughly once. The Axial Age assessment is the template: take a named period, state in advance what the comparative data would have to show for it to survive, and publish the result either way. Frontier The list of named periods that have never had this done to them is long — the Neolithic Revolution as a single event, the Iron Age as a threshold rather than a material, the military revolution, the Great Acceleration as a social rather than an Earth-system boundary. Each is testable against the same databank with the same method.

Established Second, and it follows directly from the retraction: publish the sensitivity, not the result. Re-run the moralizing-gods analysis under several explicitly stated missingness conventions — extant data only, standard imputation, inferred absence where a coder is confident, unknown otherwise — and report the full set of outcomes. Frontier The critics did exactly this once and it reversed a headline finding. Making it standard practice for every threshold claim in comparative history is a methodological change with no technical obstacle and one obvious cost: many results will stop being results.

Frontier Third: publish uncertainty on the historical national accounts. The five-century Great Divergence disagreement is conducted on point estimates. Bounding the pre-modern figures would not settle the argument, but it would establish whether the two datings are statistically distinguishable at all, which nobody currently knows.

Speculative Fourth: score a cliodynamic forecast independently. There is a standing quantitative prediction about instability in the 2020s and a claimed hit, and both the scoring and the database belong to the forecaster. An independent scoring against a database the forecaster does not maintain would convert a self-report into evidence. Frontier The comparison to draw is with the foresight literature: there the forecasting-tournament data exist and were never scored, and Public Policy Foresight owns that finding; here the scoring exists and was done by the person being scored.

Speculative Fifth, and it is the only experiment in this brief that could settle a cosmic-scale claim: look for a second origin. Biosignature surveys of the outer-system ocean worlds and of exoplanet atmospheres are the only instruments that bear on the Great Filter's location, and the argument's own evidential rule is already written down. Interstellar Civilization Models owns the search programme and its status; this brief owns only what a result would mean for where the present sits on a timeline.

Established The engineering object in this subject is a coding protocol. A comparative historical databank has to convert heterogeneous scholarship into variables with defined ranges, record who coded what and with what confidence, distinguish inferred absence from unknown, and version itself so that a reanalysis five years later runs against the same rows. Established The one large instance of this proved auditable, which is the property that matters most and is rarer than it sounds: outside critics obtained the data, quantified the missingness at 61%, reran the analysis and reversed a published result. Frontier What it did not have was a missingness convention fixed before analysis, and that single omission is what turned an auditable databank into a retracted paper.

Frontier The second requirement is a decision procedure that can be appealed. The Anthropocene vote is the only formal periodisation adjudication with published counts, a validating body above it, and a named participant alleging statutory violations. That is a functioning institution being visibly stress-tested, and it is more than any other periodisation dispute has. Speculative Whether a chronostratigraphic committee is the right instrument for a boundary defined by socio-economic indicators is a live question that the vote did not answer and was not asked to.

Frontier The third is a stated scope. Thomsen's scheme worked because it was regional, physically anchored and presented as a presumption. The engineering lesson is that a periodisation with a declared region, a declared proxy and a declared epistemic status is checkable, and one without them is not — and almost every scheme in general circulation lacks all three.

Four boundaries are tight enough to state. Interstellar Civilization Models owns the Fermi question, the Drake equation, colonisation-wave models and technosignature search status as astrobiology; this brief takes the Kardashev scale, the Great Filter and the dark forest only as periodisation claims — statements about where along a timeline a civilisation sits or fails. Technology Forecasting owns extrapolation methods; this brief owns long-run growth models as stage theories. Technology Trees of Civilization owns dependency structure and reaches the same shape of conclusion from technological rather than historical sequence. And Civilizational Planning and Public Policy Foresight own long-horizon planning institutions and the evaluation gap around them; neither is re-derived here.

Within this map, also: Long-Term Institutions, which owns the horizon over which any of this would have to stay comparable; Existential Risk Governance, which inherits the Great Filter's practical form; Multi-Planetary Civilization and Dyson Swarms, which are the engineering content of the upper Kardashev types; Interstellar Archaeology, where the detectability framing does real observational work; Civilization Resilience Planning, where secular-cycle claims meet preparedness; and Institutional Design, for the machinery any of these periods would have to run through.

Outside it: archaeology and stratigraphy, which supply the one clean success; cultural evolution and cliodynamics, which supply the databank and the disputes; economic history and historical national accounting, which supply the divergence argument; Earth-system science, which supplies the indicator record; and the philosophy of probability, which is where the anthropic arguments actually live.

Established This subject has an unusually visible institutional layer and it is worth reading as evidence. A stratigraphic subcommission voted 12 to 4 with published counts and a validating union above it; a journal retracted a paper 27 months after publication and printed the dissent of three of its own authors; a fifteen-author group outside a research programme obtained its data and reversed its headline result in the same journal. Frontier Every one of those is a mechanism working. Comparative history is often described as unfalsifiable; the record here shows a field with functioning correction machinery, which is a stronger position than most of the schemes it evaluates.

Frontier Interested parties are dense and they run in several directions. The databank's authors are the programme's principals reporting on their own instrument. The Anthropocene indicator record was assembled by proponents of the periodisation. The vote's counts reach this brief through a professional trade publication whose membership has a stake in Anthropocene framing. A museum tells the story of its own national archaeology. The growth-modes working paper is by an advocate of the fourth-mode thesis. Established Two sources here run against their authors' interest and are weighted more heavily for it: the retraction note, which is a research team dismantling its own Nature paper, and Hanson's admission that his own model would have mispredicted the industrial revolution from pre-1900 data.

Frontier And one case is the weakest evidential position this site recognises. A standing quantitative prediction of political instability in the 2020s has been scored as fulfilled — more than a hundred events in the latest five-year period — by the forecaster, on a database he maintains, published on his own newsletter. Speculative Report it as a claimed hit, name the scorer, and note that no independent scoring was located. His own caveats are on the record: that no scientific theory is absolutely correct and especially not in a young field, and that the instability observed is dominated by individual-instigated violence rather than by the collective action the framework anticipated.

Established Single-axis rankings of societies have a history and it is not a neutral one. A scheme that orders civilisations along one dimension of complexity is formally the same object as nineteenth-century unilinear evolutionism, and the difference is entirely in the evidential standard and in what is claimed. Frontier The measured result is defensible precisely because it is narrow: 51 variables about social scale, economy, governance and information systems covary, and one component captures most of that covariation. It is not a ranking of worth, it does not imply a direction anyone ought to travel, and it says nothing about which societies were good to live in. The distance between what the statistic supports and what a stage theory is usually used to argue is the whole ethical content of this subject.

Established The Anthropocene case shows the same gap in the other direction. The indicator record and the epoch proposal are separable, and a great deal of public argument treats the vote as a verdict on the first. Frontier Reporting “geologists rejected the Anthropocene” misdescribes a rejection of a specific epoch boundary as a rejection of a measured Earth-system change, and the disaggregated indicators — 74% of global GDP in countries holding 18% of global population in 2010 — carry a distributional claim that the naming dispute obscures.

Speculative And the cosmic-scale schemes carry policy freight that their evidence cannot bear. The dark forest argument, if believed, is an argument for silence — against deliberate transmission, and in favour of treating detectability as a hazard. Handwave It has no data, no instances and no proposed test, and it reaches most readers through a novel. A hypothesis in that condition should not be doing work in a transmission policy, and this brief says so while covering it in full rather than omitting it.

Established The terminal position is that the comparative axis is real and the tick marks on it are not. Human societies vary along a measurable dimension of complexity; that variation is orderable, cross-regionally predictable, and worth studying. What does not survive contact with data is the cut — the claim that the axis has natural joints, that those joints fall in the same places across civilisations, and that naming them tells you anything you did not already know from the position itself.

Established Every periodisation scheme that has been formally tested has either dissolved into a continuum, failed on its identification criteria, or been voted down. The Axial Age dissolved into a stage on a complexity axis. Rostow's stages failed Cairncross's verifiability test while the take-off claim underneath them passed a 223-country quantitative test. The Anthropocene has 24 indicators breaking unambiguously around 1950 and a 12-to-4 vote against making that a boundary, with a procedural challenge still standing. The Great Divergence is dated 1348 to 1500 by one method and around 1800 by another. Kardashev's scheme has no observed instance above the base type. And a stage model fitted to two million years of world product cannot retrodict the industrial revolution from pre-1900 data, on its author's own account. Frontier The three-age system survives, and it survives because Thomsen restricted it to one region, anchored it to stratigraphy, and published it as a presumption.

Established The mechanism is general and it is the most useful thing in this brief. A periodisation is a claim about absence. Comparative historical records are dominated by unknowns. So the direction of a threshold result is set by a coding convention, and the retraction is the case where that was demonstrated rather than argued: 61% missingness, recoded as absence, produced a finding that reverses when the missingness is handled properly. Frontier Anyone proposing a new civilizational boundary should be asked first what their convention for absence is, and second whether the boundary survives the alternatives.

Speculative The practical consequence for long-horizon thinking is that stage language is a poor planning instrument and a good communication one, and the two get confused. Saying a civilisation is “entering an information age” or “approaching a filter” conveys something; it does not license an inference about what comes next, because in every case where the inference has been tested it failed. Frontier What can be carried forward is the position on the axis, which is measurable, and the direction of change, which is observable. The name of the stage is a summary, and treating it as a premise is the error this brief is about.

These horizons carry the cosmic-scale and growth-mode material deliberately, at weak flags, because that is where the hypothesis space lives and the flags are what make covering it safe. Nothing below is a forecast:

• 10 yr: Frontier The testable things are institutional. Whether comparative historical databanks adopt published missingness conventions, whether historical national accounts ship uncertainty intervals, and whether anyone independently scores a cliodynamic forecast are all decidable within this window. Speculative The Anthropocene procedural challenge either produces a revised proposal or does not; the working group's inability to revise after discussion closed is the stated obstacle. Frontier Biosignature results from outer-system ocean-world missions and from exoplanet atmospheric spectroscopy are the only observations in this brief that bear on a cosmic-scale periodisation claim, and their status belongs to Interstellar Civilization Models.
• 25 yr: Speculative Growth modes. Two independent-looking fits place a hyperbolic asymptote near 2047, and the agreement is a property of the functional family rather than corroboration between models. The observable test arrives regardless: on the accelerating-modes account, a fourth mode with a doubling time measured in days or single-digit years would have to become visible; on the harder-ideas account, research productivity keeps falling and nothing of the sort appears. Frontier Both are fitted to real data and this brief declares the tie rather than choosing. Speculative Note the deviation already on the record: the last seventy years fit an exponential better than a power law, and the last thirty sit at about the 25th percentile of the hyperbolic model's own distribution.
• 50 yr: Speculative Kardashev. The scale orders civilisations by energy budget — planetary, stellar, galactic — and its fiftieth-anniversary review calls it often oversimplified, notes that many refinements and alternatives have been proposed, and still judges it “perhaps the worthiest legacy of the SETI 'founding fathers'”, because the emphasis on detectability obtains new significance under observational search programmes. Speculative A widely quoted continuous interpolation places humanity at about 0.7; that figure reaches this brief through a graduate-student explainer site and is flagged accordingly, and the commonly quoted energy figures for the three types are deliberately not printed here because the 1964 original was not obtained. Frontier Its honest status is that it is an observational search taxonomy wearing a developmental-stage costume, with zero observed instances above the base type, and that is a legitimate function which is not periodisation.
• 100 / 250+ yr: Speculative The Great Filter. The argument is that the absence of visible expansionist civilisations implies an enormous improbability barrier somewhere along a nine-step path from the right star system through prokaryotes, eukaryotes, sex, multicellularity and tool use to a colonisation explosion. It is a web essay, not peer-reviewed, and it is flagged that way. Its author stays agnostic on the location while being explicit about the stakes: “to support optimism regarding our future, we must find especially improbable past evolutionary steps.” Frontier The one serious quantitative attack argues the filter is behind us: a Bayesian model of evolutionary transition waiting times with an observation-selection correction, over abiogenesis, eukaryogenesis, sex and intelligence against an Earth habitable window of roughly 4.3 to 5.8 billion years, puts over 90% of posterior probability on rates giving less than a 1% chance of the final transition in a two-transition model and 99% in three- and four-transition models, with one three-transition case putting over 90% of posterior on rates implying about one star per galaxy. Frontier A separate line argues there is no paradox to dissolve, because Drake-style equations “implicitly assume certainty regarding highly uncertain parameters” and recasting them with realistic distributions yields a substantial ex ante probability of being alone — no probability figure is printed here because the abstract gives none and the body was not obtained. Handwave The dark forest holds that every civilisation passes through a stage in which it is detectable before it is defensible, and that the stage is lethal. That is a periodisation with a mortality rate attached, and it has no data, no instances and no proposed test; no peer-reviewed formulation was located and it circulates through a novel. Take it seriously enough to state that nothing has been proposed that would falsify it. Speculative The simulation argument belongs here for one reason: its first disjunct is a claim about the present. At least one of — “the human species is very likely to go extinct before reaching a 'posthuman' stage”, or “any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history”, or “we are almost certainly living in a computer simulation” — is true. Each is a strong statement about where the present sits on a civilisational timeline, and the trilemma is derived without a single observation. Its author declines to say which disjunct holds and cautions that “the implications are not all that radical.” Handwave Nothing at this horizon is a forecast, and the flags are the point.

• Depends on Nothing on this map. No brief here produces a result this one waits on: the material is historical, statistical and observational, and what is missing is a coding convention and an astronomical observation. No typed depends-on edge is claimed.
• Requires (not on this map) One scientific result and one institutional artefact, neither produced by any brief on this map. The scientific one is specific and an astrobiologist would recognise it: a biosignature from a second, independent origin of life. The Great Filter argument's entire content is a claim about where along the path the improbability sits, and unusually its evidential rule is written down in advance — independently evolved microbes elsewhere would be bad news, because easy early steps place the barrier ahead of us. A confirmed second origin moves the posterior one way; a confirmed absence across many surveyed worlds moves it the other; nothing else on offer moves it at all. The Bayesian work on evolutionary transition times gets to “intelligent life is exceptionally rare” from Earth's timeline alone, and it cannot be checked without a second case. Nobody is producing that result and no brief here will. The institutional one is cheaper and its absence has already cost a retracted paper. Comparative historical databanks need missingness conventions published before analysis, a coded distinction between inferred absence and unknown, independent audit rights and versioning that lets a reanalysis run against the same rows. The one large databank in existence proved auditable — outside critics obtained the data, found 61% missingness on the key variable, reran the analysis and reversed the direction of a Nature result — but it had no convention fixed in advance, and that single omission is what turned an auditable databank into a retraction with three authors formally dissenting. Nothing about supplying it is a research problem; it is a decision a databank's maintainers and its funders could take.
• Enables In principle a settled account of where the present sits on a civilizational timeline would bear on any brief that reasons from long-run trajectory. No typed enabling edge is claimed, because this brief's own finding is that no such settled account exists and that the schemes offering one fail on test.
• Adjacent Archaeology and stratigraphy; cultural evolution and cliodynamics; economic history and historical national accounting; Earth-system science; astrobiology and search programmes; the philosophy of probability, where the anthropic arguments live; and within this map Interstellar Civilization Models, Technology Trees of Civilization and Civilizational Planning.

Established “The Fermi paradox is Fermi's.” He published nothing on it. On the one occasion he is known to have raised it he asked “Where is everybody?”, apparently suggesting interstellar travel may not be feasible — “but not suggesting that intelligent extraterrestrial life does not exist or suggesting its absence is paradoxical.” Established The argument in circulation is Michael Hart's, published in 1975, and the historiographical correction is blunt about the cost: the misattribution “misrepresents Fermi's views, misappropriates his authority, deprives the actual authors of credit.”

Established “Geologists rejected the Anthropocene.” They rejected an epoch proposal with a specific reference point and a 1950 start date, by 12 votes to 4 with three abstentions and three not voting, and a named participant alleges statutory violations that call the vote's validity into question. Established The Earth-system evidence is separate and stands: 24 indicators over 1750 to 2010, with fundamental shifts beyond Holocene variability only after mid-century. The word survives the vote; the epoch does not. Reporting it either as a rejection of the science or as a mere technicality gets it wrong.

Established “Moralizing gods appear above about a million people.” That result was retracted on 7 July 2021, 27 months after publication. 61% of the relevant data points were missing, the analysis recoded them as confirmed absences, and handling them properly reverses the direction — moralizing gods precede increases in complexity. Frontier And the retraction does not settle it either. Three named authors formally declined to sign it, and the note itself records that the authors' refined analyses still support their conclusions while the differences were substantial enough to warrant retraction. Anyone citing the Axial Age volume should know that its concluding chapter restates the retracted threshold.

Handwave “Hanson and Roodman independently converge on 2047.” They are hyperbolic fits to overlapping gross-world-product series. A fit from that functional family will place its asymptote in a similar place regardless of the model wrapped around it, so agreement is arithmetic rather than corroboration. Speculative And the deviations point away from the date: the last seventy years fit an exponential better than a power law, the last thirty sit near the 25th percentile of the model's own distribution, and the modes model's author reports it would have mispredicted the industrial revolution from pre-1900 data.

Speculative “Kardashev types describe a sequence civilisations pass through.” There are no observed instances above the base type, so the classification cannot be validated by comparison — there is nothing to compare to. Frontier What justifies its continued use is not that it describes an observed sequence but that it orders things by detectability, which tells a telescope what to look for. That is a legitimate and useful function, and it is not periodisation. Established The commonly quoted energy figures for the three types are not printed in this brief, because the 1964 original was not obtained and a number remembered is not a number sourced.

Handwave “The dark forest is a hypothesis in the scientific sense.” It reaches most readers through a novel. No peer-reviewed formulation was located, there are no instances, no data, and no proposed test. Speculative Its content as a periodisation claim is precise and worth stating — every civilisation passes through a stage in which it is detectable before it is defensible, and that stage is lethal — and the honest treatment is to say exactly what would falsify it and then to note that nobody has proposed anything that would.

Speculative “A cliodynamic forecast of 2020s instability was verified.” The forecaster scored his own forecast on a database he maintains and published the score on his own newsletter, reporting more than a hundred events in the latest five-year period against a prediction made in 2017. Frontier No independent scoring was located. Under this site's interested-party rule that is the weakest evidential position available for a positive claim, and the correct report is: a claimed hit, named scorer, unverified. His own note that the observed instability is dominated by individual-instigated violence rather than collective action is a discrepancy with the framework, and he records it himself.

Handwave “Agricultural, industrial, information is the operative periodisation of the modern era.” Its two canonical statements were not obtained for this brief and no systematic empirical test of the triad was located at all. Handwave The step where the argument works by assertion is the third term: agricultural and industrial name changes in what most people do for a living, both measurable in labour statistics, while “information” names a change in what the most valuable activity is — a different kind of claim — and the boundary date varies by thirty years between authors. The criterion changes at the third stage and the scheme does not announce it.

Speculative “The simulation argument says we are probably in a simulation.” It says at least one of three disjuncts holds and explicitly declines to say which, and its author's own caution is that “the implications are not all that radical.” Frontier The reason it belongs in this brief rather than in a philosophy one is that the first disjunct is a periodisation claim about the present: it asserts that almost nobody makes the transition to a posthuman stage. Any of the three is a strong statement about where the present sits on a civilisational timeline, derived without a single observation. It is the purest example available of a periodisation produced entirely by argument.

Frontier And several numbers in circulation are deliberately not printed here, so that their absence does not read as oversight. The energy budgets attached to the three Kardashev types are omitted because the 1964 original was not obtained. The widely quoted probability from the paper arguing there is no Fermi paradox is omitted because its abstract gives none and its body was not read. No statistical criticism is attributed to the letter contesting the single-dimension result, because only its bibliographic record was obtained. Established A number whose source was not read is not a number, and a brief that prints one from memory has stopped being checkable.