A ranked canon of the greatest experiences ever made in virtual, augmented and mixed reality — from Wheatstone's mirror stereoscope to the 2025 Venice Grand Prize. Two hundred and ten works, scored on one instrument, so a barefoot walk on desert sand and a billion-download phone game can be argued about in the same column.
Ranked on a single seven-part instrument, published in full under The Record. Each of these reached the top by a visibly different route — one won on jury validation, one on install base, one on an interface nobody has improved on in thirty-one years. Where a work's score is low on a criterion, that is stated rather than smoothed over.
Everything the seven research beats surfaced and could verify, in one ranking. Entries marked MILESTONE are machines, tools, platforms or institutions rather than experiences — the Sword of Damocles is the most important object in this document and would be a dismal ten minutes. They are ranked for their inheritance, not their pleasure.
| # | Work | Year | Domain | Access | Score |
|---|
Almost every argument the industry had after 2012 had already been had. These are the five traceable descents: what each line solved first, and who is still living off it. Read left to right; the vertical position of a node carries no meaning.
Krueger → Kinect → your hands
Myron Krueger built responsive mixed reality in 1975 and refused the headset on principle. His finding still underwrites the field: with zero haptic feedback, participants physically pulled away when their video silhouettes touched another person's. Visual co-presence alone produces bodily reflex. Kinect shipped that idea to 35 million homes in 2010; its team went on to build HoloLens; hand tracking on a Quest is the same premise with better silicon.
Heilig → Osmose → Tree → teamLab
Every claim made about “multisensory storytelling” in 2026 was made concretely, in wood and 35mm film, in 1962. Sensorama's failure was not conceptual — Heilig simply could not raise capital, and the content pipeline cost more than the box. That is the same arithmetic that killed The VOID in 2020 and Galactic Starcruiser in 2023, and it is the open problem Positron's chair economics and teamLab's headset-free model are two competing answers to.
A ranking whose instrument is hidden is an assertion. This is the instrument, the award record it draws on, the myths the research corrected, and the list of works that no longer exist.
Seven criteria · 100 points · one instrument for all of it
The problem is that these objects do not share a unit. Pokémon GO has 709 million downloads and zero XR awards. Osmose has roughly 25,000 immersants across thirty-one years, zero awards, and a University of Toronto Press monograph. A single unweighted “greatness” number always collapses into whichever axis was measured most precisely — in practice, install base. The fix is not a common unit. It is to score every work on the same seven axes and let works win differently.
| Criterion | What it measures | Pts |
|---|---|---|
| Medium‑essentialism | Could this exist in another medium without losing its meaning? Is presence, embodiment, scale, gaze, breath or co-presence load-bearing — or is this a film with a headset attached? Highest weight, and it is the criterion that correctly penalises VR modes of flat games. | 20 |
| Peer & jury validation | Tiered by body. Academy / Emmy / Peabody / Venice / Cannes / Golden Nica score highest; DICE, Tribeca, IDFA, Annecy next; TGA, VR Awards, Raindance below that; outlet lists lowest. | 18 |
| Influence on practice | Did other makers copy it? Gravity gloves, hand-presence comedy, redirected walking, the empathy documentary, locative AR — all traceable to a single origin work. | 15 |
| Craft & execution | Graded against contemporaneous hardware, always. Videoplace on 1970s silicon and Alyx on a 2020 GPU are both maxima; comparing absolute fidelity across eras is meaningless. | 13 |
| Critical reception | Metascore where there are 15+ critics; otherwise substantive long-form criticism. A Leonardo paper, an MIT Press chapter or a museum catalogue essay counts the same as an aggregate score. | 12 |
| Reach & penetration | Log scale, banded. 10³/10⁴/10⁵/10⁶/10⁷/10⁸⁺ = 2/4/6/8/10/12. Linear reach makes Pokémon GO mathematically unbeatable and the whole exercise pointless. | 12 |
| Durability & access | Can a person experience this today? Applied as a modifier, never a gate. Osmose is nearly unrunnable and stays at number three. | 10 |
The disagreement that has to be named
These are two separate canons that share a medium and almost nothing else. The games press measures hours of replayable play, mechanical craft and platform performance; the festival circuit measures duration-limited authored work, subject matter and formal invention. One typical length is 10–20 hours, the other 8–45 minutes. One you own forever; the other you see once, in one city, if you got a ticket.
In ten years, Venice has never awarded a game and DICE has never awarded a festival work. Across roughly 150 award-winning works, exactly three appear in both canons: The Line / A Linha (the only work to win both a Venice Immersive prize and an Emmy), Spheres (Venice top prize, plus Cannes's own 2024 retrospective selection), and Carne y Arena (Academy, Cannes, VR Awards).
And neither canon has any mechanism for the two most-used things ever made in XR. Pokémon GO and VRChat are award-invisible. Raindance's Best VRChat World is the only prize in the entire record that touches platform-native social XR — and its winners could not be retrieved.
Top prize by year
Game of the Year · by year of work
Recognition from outside the XR bubble
Where the festival canon lost its institutions
Sundance New Frontier — the strand that incubated the entire 2015–2020 generation — is gone. Reduced to two labelled projects in 2024, zero mentions in 2025, absent from the 2026 programme. It never had a competitive prize; its power was curatorial, and a New Frontier premiere was worth a Cannes Official Selection slot.
Tribeca Storyscapes lapsed after 2023. The Proto Awards ran 2014–2016 and the domain is now a squatted gambling site. The VR Awards have been unreachable since 2022.
Against that: Cannes launched an Immersive Competition in 2024, and Venice is stronger than ever. Net effect — the festival canon is consolidating around Venice, Cannes, IDFA and Annecy. Any canon built now is, whether it admits it or not, weighted toward European curatorial taste. This one admits it.
Five things most XR histories get wrong
26 works · the medium's most urgent open problem
Flagged rather than smoothed over
Unretrievable: Raindance Immersive winners 2021–2026, including Best VRChat World — the single most important gap, since it is the only award touching platform-native social XR. SXSW XR Experience Competition winners before 2021. Tribeca Storyscapes 2013–2016 and 2018–2020. Emmy 2022. Ars Electronica Interactive Art 2017–2026. MoMA and Museum of the Moving Image XR acquisitions.
Unverifiable: two titles requested in the brief — Eve of the Hunt and Mother Bunker — appear in no reachable record: not the Venice Immersive competition lists 2017–2026, not the SXSW winners roster, not Cannes Immersive, not the XRMust database. If they are real, they are not in the festival record. They are excluded rather than guessed at.
Least reliable column: access status. Meta has repeatedly retired legacy Rift content, Oculus Story Studio and Penrose no longer exist, Magic Leap One was bricked on 31 December 2024, and installation works tour unpredictably. Re-check anything marked archival before citing it.