How Much SETI Has Been Done? Finding Needles in the *n*-Dimensional Cosmic Haystack

The paper addresses the common misconception, often invoked in discussions of the Fermi Paradox, that extensive searches for extraterrestrial radio transmissions have already been conducted and found nothing conclusive. The authors argue that existing searches cover only a minute portion of the relevant parameter space, so null results to date provide weak constraints on the existence of technosignatures. They set the context by reviewing articulations of the Fermi Paradox (e.g., Hart’s “Fact A”), noting that even if interstellar travel is feasible on galactic timescales, the absence of obvious artifacts or settlements in the Solar System does not imply a thorough search has been completed. They emphasize that quantifying search completeness via a well-defined, multidimensional “Cosmic Haystack” is necessary to place null results into context, much as early upper limits in other fields initially constrained only extreme regimes. The introduction motivates a formalism to parameterize and integrate over the many dimensions relevant to radio technosignature searches, enabling quantitative statements about search fractions and upper limits.

Prior visualizations of the SETI search space include Wolfe et al. (1981), who conceptualized a multidimensional “Cosmic Haystack” by compressing several parameters (e.g., polarization, frequency, modulation) and expressing spatial coverage via number of targets or beams. Papagiannis (1985) contrasted targeted versus sky-survey approaches by frequency coverage, sky fraction, and sensitivity. Tarter (2007) and Tarter et al. (2010) popularized a nine-dimensional haystack (three spatial, time, two polarization, central frequency, sensitivity, modulation) and argued that search completeness remains extremely low, akin to sampling a drinking glass from Earth’s oceans. The literature on survey figures of merit (FoMs) includes Drake et al. (1984), who proposed BW × Ω × φ_min^{-3/2} for narrowband searches, and Dreher & Cullers (1997), recast by Tarter (2001), incorporating EIRP, number of stars, polarization sensitivity, and frequency span. Enriquez et al. (2017a) proposed a Survey Speed Figure of Merit proportional to BW × φ_min^{-2}, tailored to Breakthrough Listen’s targeted strategy, and introduced a Continuous Wave Transmitter FoM involving N_stars and EIRP_min. Additional context includes early and ongoing debates about the Fermi Paradox and “Great/Eerie Silence” (Brin 1983; Davies 2011), and examples of initial broad but weak constraints on technosignatures (e.g., galaxy-scale waste heat limits; Griffith et al. 2015).

The authors formalize an eight-dimensional haystack for radio communication SETI and compute both total haystack volume and the fraction searched by selected surveys via analytic integration under simplifying assumptions. Dimensions: (1) Sensitivity to transmitted/received power, parameterized as S = 4π d^2 / EIRP with units of m^2 W^{-1}, chosen for linear improvement with integration time for unresolved signals; (2) Transmission central frequency; (3–5) Space: distance and direction (treated as a physical volume centered on the Solar System; sky coverage parameterized by solid angle); (6) Transmission bandwidth (BW_t); (7) Time/repetition rate (sensitivity as a step function up to total on-target integration time; haystack upper boundary at one-year repetition period); (8) Polarization (η_pol from 0 to 1 encapsulating polarization sensitivity). Modulation is treated as a selection of signal types to which the pipeline is sensitive (assumed here: constant-strength signals across a finite bandwidth with modest Doppler drift). Boundaries for the example haystack are: EIRP_min ≈ 10^13 W (Arecibo planetary radar scale), distance 0–10 kpc, full sky (0–4π sr), central frequency 10 MHz–115 GHz, transmission bandwidth 0–20 MHz, repetition period up to 1 year, and η_pol up to 1. Sensitivity versus bandwidth is modeled with three regimes: flat for spectrally unresolved signals (BW_t < channel width), falling as sqrt(number of channels) when BW_t spans multiple channels, and falling inversely once BW_t exceeds instrument bandwidth. The telescope beam is approximated as a uniform disk at the diffraction limit (no astrophysical background), and instrumental sensitivity is assumed uniform across bandpasses. The authors derive the multi-dimensional integral for haystack volume analytically by integrating sensitivity over central frequency, bandwidth, distance (with a parabolic EIRP boundary producing a critical distance d_crit where survey sensitivity equals haystack limit), and solid angle, then multiplying by repetition-time and polarization factors. The total 8D haystack volume for the chosen boundaries is given (Equation 11): V_haystack = 4π d_max^2_max BW_{t,max} (ν_max − ν_min) T_max / (5 EIRP_min), with units m^5 Hz^2 s W^{-1}. They provide a Mathematica derivation and a Python script to compute volumes and search fractions, including more general boundary cases.

• Under the chosen boundaries, the total 8D haystack volume is 6.4 × 10^116 m^5 Hz^2 s W^{-1}.
• For the Breakthrough Listen L-band campaign (Enriquez et al. 2017a), assuming 17 Jy single-channel sensitivity in 2.7 Hz channels, 800 MHz instrumental bandwidth (1.1–1.9 GHz), ~3.2 × 10^-3 sr effective sky coverage, 300 s total time on target, and η_pol = 1, the searched volume is ~2.4 × 10^98 m^5 Hz^2 s W^{-1}, corresponding to a haystack fraction of approximately 3.8 × 10^-19.
• Aggregating selected historical and recent surveys (including ATA and MWA surveys), the total searched fraction is ~6.0 × 10^-18 of the defined haystack.
• Interpreted via an ocean-volume analogy, the cumulative search corresponds to sampling ~8000 liters out of Earth’s oceans (≈1.335 × 10^21 liters), comparable to a large hot tub or small swimming pool, reinforcing that search completeness remains extremely low.
• The results are qualitatively consistent with Tarter et al. (2010), despite different haystack definitions, and show that existing null results cannot rule out even dense networks of strong transmitters within significant portions of parameter space.

The formalism demonstrates that despite decades of radio SETI, only a tiny fraction of the plausible parameter space for technosignatures has been explored. This quantitatively counters claims that SETI’s null results sharpen the Fermi Paradox or imply the absence of obvious radio beacons. The analysis highlights how survey design choices—bandwidth, sky coverage, sensitivity, and time sampling—map into haystack completeness, offering a common parameter space for comparing diverse surveys. The findings underscore that targeted high-sensitivity efforts and wide-field surveys are complementary: wide-field arrays (e.g., MWA) can sweep large solid angles quickly, dominating haystack volume in some regimes, whereas targeted programs improve limits for nearby stars and lower EIRP transmitters. While the haystack as defined includes vast swaths of space with low prior probability for transmitters, completeness within more astrophysically motivated subsets (e.g., continuous narrowband emissions from nearby stars) is substantially higher. Overall, the results argue for continued, diversified searches and more systematic reporting of completeness to frame null results appropriately.

The paper introduces a rigorous, eight-dimensional haystack framework to quantify SETI search completeness and provides analytic expressions, along with a Mathematica notebook and Python code, to compute haystack volumes and searched fractions. Applying this to representative radio surveys shows that the fraction of the defined haystack searched to date is extremely small, consistent with previous qualitative arguments that our exploration is far from exhaustive. The authors recommend that future SETI programs routinely compute and report haystack completeness or upper limits within a shared parameter space, ideally informed by signal injection and recovery to model sensitivity variations across dimensions. They emphasize the value of surveys with large bandwidths, wide fields of view, long dwell times, repeat visits, and robust sensitivity to accelerate progress, and point to the promise of “all-sky, all the time” architectures for rapidly increasing search completeness.

• The haystack definition and boundaries are somewhat arbitrary; results depend sensitively on choices such as d_max, EIRP_min, frequency and bandwidth ranges.
• Instrumental sensitivity is approximated as uniform across bandpasses and beams are simplified as uniform disks without astrophysical backgrounds; real systems have frequency-, direction-, and RFI-dependent sensitivity.
• Detection algorithms are assumed to be uniformly sensitive to both narrowband and (idealized) broadband top-hat signals; actual pipelines may have reduced or unknown sensitivity to broadband or atypically modulated signals.
• Repetition-rate sensitivity is modeled as a step function based solely on total on-target time; potential gains from stacking and complex repetition schemes are not fully treated.
• Non-orthogonality among dimensions (e.g., bandwidth-time coupling for short pulses, EIRP-distance coupling) is simplified for analytic tractability.
• The analysis aggregates heterogeneous surveys with differing strategies and thresholds under common assumptions, potentially introducing systematic mismatches.
• RFI contamination and practical frequency gaps are not fully accounted for, likely making some estimates optimistic.