Simon Greenwold

Simon Greenwold is a former researcher in the Aesthetics + Computation Group at the MIT Media Lab, best known for his 2003 master's thesis Spatial Computing. The thesis defined spatial computing as "human interaction with a machine in which the machine retains and manipulates referents to real objects and spaces", and argued that this was "an essential component for making our machines fuller partners in our work and play".

The thesis presented six experimental systems built at the Media Lab, among them a handheld, tracked video window for placing virtual objects in a real room, which Greenwold described as an exploration of augmented reality. After Apple introduced Apple Vision Pro in 2023 as "Apple's first spatial computer", technology writers and the Media Lab itself pointed back to Greenwold's thesis as the source of the term's modern definition.

Reviewed 15 October 2026. Checked the thesis PDF (definition, front matter, Installation and EyeBox chapters, closing chapter), the ACG project pages and resume, the DSpace record, Crossref metadata for the ACADIA 2000 paper, Processing credits, and the Hackl, TechTarget, GPS World (Esri) and arXiv citations of the term. About review dates.

Education

Greenwold's resume on his Media Lab page states that he studied at Yale University from 1991 to 1995 and graduated with a B.S. in English and Applied Math, with a concentration in computer science. The title page of his thesis also lists the Yale degree, dated June 1995. According to the resume, he spent 1999 to 2001 in the Master of Architecture program at the MIT School of Architecture and left after two years to join the Media Lab, where he studied from 2001 to 2003. Before graduate school he worked as a software developer and GUI designer at Bloomberg L.P. (1995-1997) and was a high school teacher at Friends Seminary (1997-1999).

While in the architecture program he co-wrote, with Peter Testa and Una-May O'Reilly, "AGENCY GP: Genetic Programming for Architectural Design", presented at the ACADIA 2000 conference of the Association for Computer Aided Design in Architecture in Washington, D.C.

MIT Media Lab

At the Media Lab, Greenwold was a researcher in John Maeda's Aesthetics + Computation Group (ACG). His project page describes his work as exploring "spatial contexts for computational constructs". Maeda supervised his thesis, and the readers were Joseph A. Paradiso, then co-director of the lab's Things That Think consortium, and Chris Csikszentmihályi.

Spatial Computing thesis

Greenwold submitted Spatial Computing to the Program in Media Arts and Sciences on 16 May 2003, in partial fulfillment of the requirements for the degree of Master of Science in Media Arts and Sciences; the title page is dated June 2003. The thesis is organized in three parts: a framework, a set of six experiments, and an evaluation.

In the definition chapter, Greenwold added that the real objects and spaces should ideally "have prior significance to the user". He separated spatial computing from 3D modeling and digital design because it "requires the forms and spaces it deals with to pre-exist and have real-world valence", and wrote that a screen showing a virtual space was not enough: the virtual space "must be meaningfully related to an actual place". His definition did not require a 3D display at all; he gave an auto-flushing toilet that senses a user moving away as a trivial case that still qualifies.

The introduction places the work next to mixed reality research, citing the "Extent of World Knowledge" axis of Paul Milgram's taxonomy, and says that spatial computing borrows "from cultural criticism, theories of HCI, and systems of mixed reality, but belongs to none of them exclusively". The closing chapter discusses the technologies the author expected to matter (thinner and more flexible displays including OLED, 3D displays, more reliable and affordable tracking, 3D scanning and 3D printing) and sketches a hypothetical office network, "ZapNet", in which laser pointer devices carry data between screens and physical objects.

Experiments

The six projects described in the thesis, with the dates given on Greenwold's ACG project index:

Project	Date on project index	Description
Installation	16 October 2001	A movable flat-panel "viewing window" with a small camera on its back showing a live video feed of the room, plus a stylus; both were tracked in position and orientation so that sculpted virtual forms stayed fixed in the real space.
Internaut	17 January 2002	Web site structures turned into 3D maps navigable in a modified Quake II engine.
Stomping Ground	28 April 2002	Visuals for a sensor carpet installed permanently at the MIT Museum, made with Paradiso's Responsive Environments group and Kai-yuh Hsiao.
Hotpants/LittleVision	2 April 2002 (Hotpants); 21 May 2002 (Tiny Stories workshop)	A palm-size 10 by 14 red LED display built for the Nylon microcontroller teaching platform, later used to play back tiny movies.
Pointable Computing	12 July 2002	Two handheld devices that exchanged data over pulse-modulated visible laser beams, so that a user selects a device by pointing at it.
EyeBox	15 December 2002	A low-cost 3D scanner built into a mini-fridge, reconstructing objects from silhouettes captured by webcams.

The thesis frames Installation in augmented reality terms, calling it "an exploration in what is traditionally called 'augmented reality'". Users held the plexiglass-framed screen in two hands and used the stylus to squirt out "blobby" forms, which then stayed in place in the room and could be seen from any angle as the screen moved. Position and orientation tracking came from an Ascension Technology Flock of Birds magnetic tracker, with one sensor on the screen and one on the stylus; the software was written in C++ with OpenGL. A networked client let users "throw" an object to other computers in the room, including a printer, which printed it.

In discussing precedents, Greenwold compared Installation with marker-based systems such as ARToolKit and with overlay systems by Jun Rekimoto and Steven Feiner, and traced handheld tracked screens to George Fitzmaurice's 1993 "eye in hand" concept. Because Installation gave more weight to the virtual forms than to the real room, he wrote: "Perhaps Installation is less augmented reality than augmented virtuality". He also reported that users tolerated a good deal of misregistration between the virtual and real views "as long as it is of the right kind".

EyeBox was the last of the six. In the version described in the thesis, three webcams, two fluorescent lights, a microwave turntable and a flat panel display were built into a mini-fridge; an object up to nine inches on a side was rotated once, and a visual hull reconstruction was computed from 24 silhouette images and refined up to 512 x 512 x 512 voxels. An earlier prototype documented on the project page was an 18-inch foam core cube with six webcams. Greenwold presented the device as much as an interaction idea as a scanner: the user opens the computer and puts an object inside.

Other projects

Greenwold wrote the interactive demonstrations for Active Statics, a set of eight Java applets for teaching graphic statics to architects; the site lists him as author of the demos and Edward Allen as author of the accompanying manual, with copyright 2003 to both. His resume lists a bronze award for Installation in I.D. Magazine's 2002 Interactive Media Design Review, and a showing of his "Weaver" software in the "Super-Ficial" exhibit at the New Museum of Contemporary Art in New York in winter 2003.

Later work

His resume, whose latest entries date from fall 2003, lists a lecturer post at the Yale School of Architecture (dated "Fall-Spring 2003"), where he taught two graduate classes on computational design, and a software engineer post at Metaserver Inc. from fall 2003. The Processing project's alumni list credits him with "Lighting and Camera" (winter 2005) and its special thanks go to Sami Arola "for writing the base of the original P3D" and to Greenwold "for incorporating camera and lights". No reliable source located for this article documents his current position.

Influence of the term

Greenwold's thesis was not the first use of the phrase. A 1997 World Scientific volume edited by Terry Caelli, Peng Lam and Horst Bunke was titled Spatial Computing: Issues in Vision, Multimedia and Visualization Technologies. Writing on Esri's blog in 2024, in an article later republished by GPS World, Greg Milner said that "the term spatial computing dates to the 1980s" and that "its modern definition entered the lexicon in 2003" with Greenwold's description.

The 2023 launch of Apple Vision Pro renewed attention to the thesis. In a Harvard Business Review article republished by the MIT Media Lab, Cathy Hackl wrote that "in 2003, researcher Simon Greenwold, then at the MIT Media Lab" had defined spatial computing, quoting his definition. TechTarget's definition of spatial computing states that the term "was coined by Simon Greenwold", quoting the "fuller partners" passage from his abstract. In academic work, a 2025 arXiv paper by Yibo Wang, Yuhan Luo, Janghee Cho and Junnan Yu, "What is 'Spatial' about Spatial Computing?", quotes the same definition and states that Greenwold "integrated MR with the concept of spatial computing".

See also

• Spatial computing
• Mixed reality
• Reality-virtuality continuum
• Ubiquitous computing