Psychonauts’ psychedelics: A systematic, multilingual, web-crawling exercise

Catalani and colleagues frame psychedelics as a class of substances—principally tryptamines, lysergamides and a subset of phenethylamines—that act as agonists or partial agonists at the 5-HT2A serotonin receptor and produce characteristic alterations of perception and consciousness. The authors note that the early 2000s saw a proliferation of synthetic analogues within these chemical families, many of which have been described as Novel Psychoactive Substances (NPS). These synthetic derivatives frequently lack formal pharmacological characterisation, and several have been implicated in acute intoxications and fatalities. Parallel to chemical diversification, an online community of ‘‘e-psychonauts’’ has emerged who experiment with and disseminate information about psychedelic NPS via forums and other web platforms, potentially shaping demand and the online visibility of new compounds. The study set out to exploit this online activity by using an automated web crawler (NPS finder®) to identify and categorise psychedelic molecules discussed on psychonaut websites and other surface‑web NPS resources, and to compare the crawler’s output with entries in the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) and UNODC databases. The investigators restricted their operational definition of psychedelics to tryptamines, lysergamides and phenethylamines with psychedelic activity, and aimed to assess whether web‑based surveillance could reveal substances not yet recorded in official databases. The work is presented as part of a broader programme comparing online signals with real‑world detection of NPS.

The core method was a customised web‑crawling system (NPS finder®) developed by an IT company (Damicom) to run continuously and scan a predefined list of URLs for mentions of NPS. The scanned resources were primarily psychonaut websites and forums, supplemented by other online NPS resources. The crawler extracted structured information for each hit, including chemical and street names, chemical formulae, three‑dimensional images and anecdotal reports of psychoactive effects; when available the International Chemical Identifier Key (InChIKey) was added to minimise duplicate entries. Extracted records were stored in a restricted, password‑protected database hosted on secure servers. A multidisciplinary control panel of five professionals with backgrounds in medicine, chemistry and psychiatry manually reviewed all data that the crawler collected from 26 November 2017 to 24 February 2020. The reviewers performed additional checks against web sources, vendor pages and standard databases to assign unique chemical names and structures where possible. The researchers searched Google, Google Scholar, PubMed and PubChem to classify each identified molecule as a phenethylamine, tryptamine or lysergamide; natural and plant‑derived substances were excluded. Phenethylamines were sub‑classified where applicable (e.g. NBOMes, FLYs/dragonFLYs), and molecules lacking pharmacological data were labelled as putative psychedelics. For cross‑validation, NPS finder® entries were compared to the EMCDDA European Database on New Drugs (EDND) and the UNODC Early Warning Advisory (EWA), using InChIKey matching. Access to the restricted EMCDDA/UNODC resources for comparison was provided to one author. The study did not report inferential statistical modelling; comparisons were reported as counts and proportions drawn from the three data sources. The crawler’s activity was limited to the surface web and covered multiple languages (predominantly English, with Spanish, German, Russian, Italian, Dutch, French, Swedish and Turkish also analysed).

Over the >2‑year crawling period the NPS finder® database accumulated 4,368 unique NPS entries. Phenethylamines were the largest class detected on psychonaut sites, totalling 1,263 entries (28.9% of the crawler’s NPS), which included 31 NBOMes and 16 FLYs/dragonFLYs. Of these phenethylamines, 141 were described in the literature as having confirmed psychedelic activity, 109 as predominantly stimulant/entactogenic, and 966 had no available pharmacology/effects information and were therefore classified as putative psychedelics. Tryptamine and lysergamide entries comprised smaller shares of the crawler output, with 65 tryptamines and 16 lysergamides identified. By contrast, the combined EMCDDA and UNODC databases reported substantially fewer substances in these classes: across the two official databases the extractable counts included 27 NBOMes, 3 FLYs, 50 psychedelics, 58 stimulant/entactogenic phenethylamines, and 32 molecules with no reported effects; reported tryptamines totalled 48 and lysergamides 11. When the authors compared NPS finder® with EMCDDA and UNODC by InChIKey, the crawler detected an order of magnitude more psychedelic phenethylamines: 1,263 phenethylamine entries in NPS finder® versus 123 in the EMCDDA and 156 in the UNODC. For tryptamines and lysergamides the crawler still reported higher numbers but the gap was smaller (65 tryptamines in NPS finder® vs 50 in UNODC and 44 in EMCDDA; 16 lysergamides vs 11 UNODC and 7 EMCDDA). The extracted text reports that across the three classes 976 molecules were considered putative psychedelics due to lack of psychoactive data, of which 946 were previously unknown to both EMCDDA and UNODC. Elsewhere the authors state that the total number of either novel or putative psychedelics previously unmentioned by both UNODC and EMCDDA amounted to 994; the text contains both figures. Overall, the crawler identified many substances that were not present in the two official databases, indicating a substantially larger online repertoire of candidate psychedelic NPS than was captured by existing surveillance systems.

Catalani and colleagues interpret their findings as evidence that systematic, multilingual web crawling can reveal a large number of psychedelic molecules that are not yet recorded in official monitoring databases. The authors highlight a pronounced discrepancy for phenethylamines in particular, suggesting that psychonaut communities favour entheogenic/psychedelic phenethylamines and that this preference—combined with fast, anonymous information sharing online—may drive the higher counts captured by the crawler compared with the event‑driven, reactive reporting that populates EMCDDA and UNODC registries. They argue that web‑based surveillance can therefore complement traditional early‑warning systems by providing a more dynamic, near‑real‑time picture of substances being discussed or marketed online. The discussion also stresses public‑health concerns arising from the large number of uncharacterised molecules. The authors point to known examples of severe toxicity and fatalities associated with some novel psychedelics (for example NBOMes, Bromo‑dragonFLY, DOB and certain synthetic tryptamines), and note the lack of reliable pharmacology, ADME (absorption, distribution, metabolism, excretion) data, drug–drug interaction information and clinical toxicology for most of the newly identified compounds. Such gaps, they contend, create challenges for clinicians and emergency services and increase the need for prioritised toxicological characterisation and harm‑reduction strategies. Several limitations are acknowledged. The crawler’s findings do not prove that identified molecules are present in or prevalent within the real‑world drug market; surface‑web detection cannot establish consumption rates or geographic circulation. Mislabelling and inaccurate chemical information within online discussions are possible, although the authors attempted to mitigate this through manual curation and cross‑checking against vendor sites and chemical databases. The present work was restricted to the surface web and to a subset of languages, with the authors noting plans to expand to deep web/darknet sources and additional languages such as Chinese, Japanese and Arabic. Finally, the team reports planned follow‑up work using computational docking and other in silico methods, and suggests targeted in vitro and in vivo studies, to better predict which of the unidentified molecules warrant priority for toxicological and pharmacological characterisation. The authors propose that NPS finder® could serve as a supportive tool for early‑warning systems, surveillance, pharmacovigilance and policy responses, and could help produce prioritised fact sheets and reports to guide research and public‑health action.

The authors conclude that NPS finder® can identify a large number of new and previously unrecorded NPS, including many psychedelics, and therefore has potential utility for monitoring current and emerging drug trends. They suggest the tool could support European and UN early‑warning activities by detecting molecules shortly after they appear online, informing law‑making, surveillance, pharmacovigilance, law enforcement and the drafting of treatment, harm‑reduction and educational resources. The paper recommends follow‑up in silico, in vitro and in vivo investigations to generate predictive data on acute and long‑term effects, and proposes producing ad hoc reports for newly identified compounds considered most threatening to public health. The authors also note the tool’s potential value in tracking shifts in online drug markets during events such as the COVID‑19 pandemic, when online drug purchasing and use patterns may change.