Machine elves: where the phrase came from, and how rare elves really are in DMT reports

THE SHORT VERSION · SWIPE

▶ Watch it as a 20-second video

‹›

Few phrases are as closely tied to DMT as “machine elves”. It turns up in podcasts, comics, memes and novels, and many people meet it before they read any research. This guide traces where the phrase came from, what the studies actually counted, and why researchers think a catchy label may shape the very reports it is used to describe. For the broader research picture, start with our pillar, what is DMT, and DMT entity encounters: what the research says.

What are machine elves?

In the reports that use the phrase, machine elves are small, playful or mischievous beings that seem to be made of shifting, mechanical or geometric material, constantly folding and rebuilding themselves. A 2021 field study of 36 interviews with experienced users found that the single most common “texture” people gave to the beings they met was “self-transforming”, a tendency to morph in constant flux. The authors explicitly link that description to McKenna’s machine elves, citing a 1993 recorded talk of his (Michael et al., 2021). One participant described a spinning, cog-like, self-transforming machine world inhabited by beings with an elf-like mischievousness. These are reports of experiences. Evidence: preliminary.

So the phrase names a recognisable style of description: part machine, part sprite, part living geometry. It is not a scientific category, and no study defines “machine elf” as a measured type with fixed criteria. Coders who sort online reports simply count the times authors use the label or describe something that clearly fits it.

Where does the phrase “machine elves” come from?

The phrase is credited to Terence McKenna, an American writer and lecturer on psychedelics. We paraphrase rather than quote him. According to the cultural historian Graham St John, McKenna first took DMT in the fall of 1965 in Berkeley, California, and later described the experience as overturning his intellectual world. St John calls him a talkative front man for the “machine elves of hyperspace” (St John, excerpt via Chacruna). Evidence: established that McKenna popularised the phrase.

When was it first used? Our earlier reading of St John’s 2015 cultural history dated the earliest spoken use to a McKenna lecture of 2 October 1987. St John’s published excerpt does cite a McKenna lecture of that date, for his account of his first DMT experience, but it does not show the phrase itself in that talk. We could not open a recording or transcript, so treat the 1987 date as provisional. Evidence: unresolved.

The printed record is easier to check. A full-text search of the Open Library and Internet Archive book collection for the exact phrase returned 142 matches, and the earliest dated book among them is McKenna’s own 1991 collection, The Archaic Revival (Open Library full-text search). In Google’s Books Ngram data, the phrase is absent from 1960 to 1991 and first registers in 1992 (Google Books Ngram). Evidence: established for the 1991 book use.

How common are machine elves in DMT studies?

Much less common than their fame suggests. The largest count comes from a team led by David Lawrence of the University of Toronto, which coded every DMT experience posted to one online forum from 2009 to 2018: 3,778 experiences in 3,305 posts. Entities appeared in 1,719 experiences (45.5%). Machine elves appeared in 50 of those encounters (2.9%), and other elves in 30 (1.7%). Both sit inside a broader “mythological beings” group (8.4% of encounters) that also includes faeries, sprites, trolls, gnomes and dragons (Lawrence et al., 2022). Evidence: preliminary.

Put another way, 50 machine-elf reports among 3,778 experiences is about 1.3% of all experiences in the sample. The most common entity types were feminine figures (24.2% of encounters), deities (17.0%) and aliens (16.3%). A separate “robot or machine entity” category, without the elf element, was itself more frequent than machine elves, at 6.7% (Lawrence et al., 2022). These are self-selected public posts, which favour striking trips. Evidence: preliminary.

Why does the share of elves change with the method?

Because a list puts words in people’s mouths. The 2021 field study points out that the 2020 survey mostly used fixed, pre-set answer options, while its own participants described what they met in their own words (Michael et al., 2021). When “elf” is on the menu, some people who met something small, quick and strange will tick it. When nobody suggests the word, far fewer reach for it. The same pattern shows up for other features, which is why our article on DMT entity types gives figures by method. Evidence: preliminary.

The 2021 paper also cites an earlier content analysis of 149 online reports, by Lyke (2019), subtitled “Not everyone gets machine elves”, which put elves and faeries together at about 7% of encounters (Michael et al., 2021). We could not open that analysis directly, so we report it only as described by the 2021 paper. Across every sample we found, elf-type beings were a minority. Evidence: preliminary, from self-selected samples.

How did the label spread?

Through books first, then the internet. The Open Library full-text search for the exact phrase returned matches across very different kinds of book: cyberculture writing from 1994, popular science and spirituality titles in the 2000s, a superhero comic in 2007, short fiction, poetry, and a 2008 book on DMT and alien encounters co-written by Rick Strassman (Open Library full-text search). That search covers only scanned books in one library, so it shows that the phrase travelled widely, not how many people used it. Evidence: established for the printed record.

Google’s book data show the same rise in outline: nothing before 1992, then growth with peaks in 2011, 2015 and 2018 (Google Books Ngram). Ngram counts only phrases that appear in enough books, so the first year it registers is not the first use. It also leaves out podcasts, forums and video, where much of the later spread happened. No study we found has measured how media shaped the vocabulary of DMT reports over time. Evidence: established that the phrase spread in print; unresolved how much media drove later reports.

Can a famous label shape what people report?

The researchers who counted the elves think it may. In their list of limitations, the 2022 authors note that earlier posts on the forum may influence later users’ experiences and the way they report them, and they give the repeated use of ready-made terms such as “machine elves” for otherwise new experiences as the evidence (Lawrence et al., 2022). The 2021 field team designed their interviews partly to avoid this, encouraging people to describe what they noticed in their own words rather than sort it into cultural categories such as “alien”, which they say can spread like memes (Michael et al., 2021).

There are two separate ways a label could matter, and the evidence differs for each:

1. Shaping the report. Someone meets something hard to describe and reaches for the nearest famous word. This is close to certain for some reports, because the label is so often used for experiences that do not closely match it (preliminary).
2. Shaping the experience itself. Expectations and shared stories change what people actually perceive. This is plausible and has support outside DMT (see below), but no controlled DMT study has tested it (unresolved).

Outside DMT, there is support for the second idea. An anthropologist who spent 18 months at a ceremonial ayahuasca centre in the Peruvian Amazon describes how group discussion, ritual songs and the centre’s Christian-inflected cosmology shaped not only how participants interpreted their visions but what they reported seeing, including frequent “demons” (Dupuis, 2022). Evidence: preliminary, from one centre.

A 2026 theory paper on entity encounters notes that entities take culturally specific forms, including spirits, elves and ancestors, and cites a study in which almost half of white Westerners who took ibogaine reported visions of television screens, which are rare in accounts from traditional African users (Mago et al., 2026). In a preregistered lab study, telling people to expect more voices in noise made them report more voices, especially when the sound was unclear (Szymanek et al., 2024). None of these studies involved DMT entities. Evidence: preliminary.

What do machine elves tell us about the DMT experience?

Mostly, they tell us about the gap between a famous story and the data. The features that are common across studies are less quotable: the beings are usually described as conscious, intelligent and benevolent, communication is reported in most encounters, and fear is also common, reported by 41% of the 2020 survey (Davis et al., 2020). A 2026 lab study that interviewed 23 people in fine detail found that perceived presences arrived only after a sense of a three-dimensional space had formed, and varied widely in form (Sanders et al., 2026). That structure looks more consistent than any single cast of characters. Evidence: preliminary.

Elf reports also do not settle what the beings are. Some people take them as contact with independent intelligences; most neuroscientists treat them as products of the brain in an unusual state. The studies on this page record what people say. None tests whether anything exists outside the experience, and the authors do not claim to. We stay neutral on that question here; our DMT entity types guide sets out the competing explanations side by side. Evidence: unresolved.

What we still don’t know

• The exact date of the first spoken use of the phrase. A 1987 lecture is a candidate; we could not check a recording (unresolved).
• Whether the share of elf reports has changed over time, or tracks media moments such as books, films and podcasts. The dated forum posts exist, but no one has analysed them this way (unresolved).
• Whether hearing about machine elves beforehand changes what people experience, rather than just how they describe it. No controlled DMT study has tested expectation and entity content (unresolved).
• How common elf-type reports are among all people who use DMT. Every sample so far is self-selected (unresolved).

The fair summary: “machine elves” is a vivid phrase from 1980s and 1990s psychedelic writing that became the internet’s shorthand for DMT, while the beings it names appear in only a few percent of recorded encounters. For what people more often report, see DMT entity types explained; for why these meetings feel so convincing, see why DMT feels more real than reality.

Shroomverse sells mushroom products. This article is general information about the history of the phrase “machine elves” and published research on reported DMT experiences. It is not medical, legal or purchasing advice, and it does not recommend any product.