From Weiser to Matter: a short history of ubiquitous computing
How the idea of computing woven into everyday surroundings travelled from a 1990s research vision to the Matter interoperability standard of 2022.
Most people meet computing without noticing it now. A thermostat learns a routine, a doorbell recognises a face, a speaker on the kitchen bench answers a question about the weather. None of this looks like a computer, and that is rather the point. The idea that computing should dissolve into the surroundings, rather than sit on a desk demanding attention, is older than the smartphone and older than the phrase “smart home”. It has a traceable history, running from a research lab in the late 1980s to a technical standard published in 2022.
It is also a history worth reading with a sceptical eye. The vision of environments that quietly anticipate what we need has been declared imminent many times, and the gap between the promise and the room you are actually sitting in has often been wide. What follows is the through-line — the ideas, the names given to them, and the moments when the technology caught up enough to matter.
The calm-technology vision
The most influential single source is a researcher named Mark Weiser, who worked at Xerox’s Palo Alto Research Center (PARC) around 1988. Weiser used the term “ubiquitous computing” to describe a future in which computers were so numerous and so well woven into the world that they became effectively invisible. His much-quoted essay “The Computer for the 21st Century”, published in Scientific American in 1991, opens with the line that “the most profound technologies are those that disappear” — the way writing or electric lighting disappeared into the background of ordinary life.
Weiser sketched devices at three scales: inch-sized “tabs”, foot-sized “pads”, and yard-sized “boards”, scattered through a building and aware of one another. He later framed the goal as calm technology — machines that inform without shouting, moving from the centre of attention to its edge and back only when needed. Set against the personal computer of the day, which insisted on your full focus, this was a genuinely different way of thinking about the relationship between people and machines.
The important thing to hold on to is that this was a vision, not a product. The hardware of the early 1990s could not deliver cheap, plentiful, networked devices. But the framing stuck, and almost everything that came later is a variation on it.
From ubiquitous to ambient
By the late 1990s the same idea was being reframed in Europe, often under the label “ambient intelligence” (AmI). Where Weiser’s version leaned towards the engineering — many small computers, quietly co-operating — the European framing put more weight on the human experience: environments that are sensitive to the people in them, that recognise context, and that respond in ways that feel supportive rather than intrusive.
Ambient intelligence became a popular organising theme for research and technology policy around 1999 to 2001, complete with scenario documents imagining a day in a responsive home or office. The emphasis on context, personalisation and unobtrusiveness carried Weiser’s ideas forward and gave them a more explicitly human-centred vocabulary. If you want the fuller picture of how that strand developed, the ambient intelligence topic hub goes into more detail than there is room for here.
A name for the plumbing
While researchers were describing responsive rooms, a separate phrase was coined for the network underneath them. In 1999 a technologist named Kevin Ashton used “the Internet of Things” as the title of a presentation about radio-frequency identification (RFID) tags and supply chains. The point he was making was practical: if everyday objects could report their own identity and location, computers could track the physical world without a person typing anything in. Ashton went on to help set up a research centre at the Massachusetts Institute of Technology focused on exactly that.
The phrase “Internet of Things” (IoT) outgrew its origins in logistics and became the general term for physical devices that sense, connect and act. It is the plumbing that ambient intelligence had always assumed. For a closer look at what that plumbing actually consists of — sensors, networks, gateways and the rest — the guide on how IoT works is a good next step.
The sensor and smartphone era
Ideas do not become real until the parts get cheap, and that is what happened over the following decade. Two things mattered most. First, micro-electromechanical systems (MEMS) sensors — accelerometers, gyroscopes, microphones, pressure sensors — fell in price to a point where putting them in almost anything became reasonable. Second, from around 2007 the modern smartphone bundled a cluster of those sensors, a camera, satellite positioning and a permanent network connection into a device that hundreds of millions of people chose to carry.
That combination did more for ubiquitous sensing than any research prototype. It normalised the notion of a pocket computer aware of location, movement and sound, and it created the app-and-cloud pattern that later “smart” devices copied wholesale: a small networked gadget, a paired phone app, and processing done on a distant server. The trade-offs of that pattern — the pull between local and remote processing, and the latency and privacy questions it raises — are examined in the guide on edge versus cloud.
Here is the rough shape of how the vision and the reality lined up over time.
| Era | Roughly | Guiding idea | What it added |
|---|---|---|---|
| Ubiquitous computing | late 1980s–early 1990s | Computers woven invisibly into surroundings | The founding vision and vocabulary |
| Ambient intelligence | late 1990s–2000s | Context-aware, human-centred environments | A focus on experience and responsiveness |
| Internet of Things | from 1999 | Everyday objects that sense and connect | A name for the underlying network |
| Sensor & smartphone era | from around 2007 | Cheap sensing carried by everyone | Scale, and the app-and-cloud pattern |
| Smart speakers | 2011–2016 onward | Voice as a shared interface | A hands-free way into the home |
| Matter | 2022 | Devices that interoperate across brands | A shared language between products |
Talking to the room
The next visible shift was interface, not infrastructure. Voice assistants brought speech recognition to ordinary devices: one arrived on smartphones in 2011, and the first mainstream smart speaker for the home followed in 2014, with competing products from other large technology companies appearing by 2016. For the first time a computer in the living room could be addressed out loud, by anyone, without a screen.
Smart speakers made the ambient-intelligence idea tangible for a mass audience — a room you could simply talk to. They also made its limits obvious. Assistants mishear, misunderstand context, and depend on always-listening microphones that many people are uneasy about. The calm, anticipatory environment of the research papers turned out, in practice, to involve a lot of repeating yourself and a fair amount of thinking about who might be listening.
Making devices agree
For years the biggest practical obstacle was not sensing or voice but fragmentation. Devices spoke different wireless languages — Zigbee, Z-Wave, Wi-Fi, Bluetooth and various proprietary schemes — and often would not talk to one another without a specific hub or a particular brand’s app. A light from one maker and a sensor from another could sit a metre apart and remain strangers. This is the everyday reality that the tidy vision tended to skip over.
Matter, an interoperability standard released as version 1.0 in 2022, was an attempt to fix that. Developed by a large group of technology companies, it defines a common application layer so that certified devices from different makers can be set up and controlled together, over standard internet protocols, running on top of Wi-Fi, Ethernet and the low-power Thread mesh network designed for small battery-powered devices. It is deliberately unglamorous: not a new vision, but an agreement about how existing devices should behave so they stop being strangers. Whether it delivers on that is still being worked out in practice, and the guide on Matter and Thread covers where it helps and where it does not.
In short: the dream of computing that fades into the background was articulated at Xerox PARC around 1990, renamed and humanised as "ambient intelligence" in Europe, given a network to run on by the Internet of Things, made cheap by smartphone-era sensors, given a voice by smart speakers, and finally handed a common language by the Matter standard in 2022 — with each step delivering less, and messier, than its billing.
The gap between the vision and the room
Read across this history and a pattern repeats: a compelling idea, an announcement that the future has arrived, and then a slower, more awkward reality. Homes filled with gadgets that need their own apps. Devices that stop working when a company shuts down a server. Automations that are fiddly to set up and quietly break when a network name changes. The “calm” that Weiser described is hard to find in a house that pings you whenever a battery runs low.
That is not an argument against the technology, but it is a reason to treat grand claims carefully. The useful parts of this history are concrete and modest: a sensor that saves energy, a reminder that helps someone live independently for longer, a standard that lets two products co-operate. The sweeping promises have a poor track record; the specific, well-scoped improvements have a much better one.
Where this leaves us
The through-line from Weiser to Matter is really a slow move from vision to plumbing. The early decades produced the ideas and the vocabulary; the recent ones have been about the unglamorous work of making devices cheap, connected and — at last — able to co-operate. None of it has produced the effortless, invisible environment of the original papers, and it may be that no single standard or product ever will.
The practical takeaway is to judge each new “smart” thing on its own terms. Does it solve a real problem, does it keep working if a company loses interest, and does it respect the people in the room? If you want to place any particular device or claim in this longer story, the overview of what intelligent environments are is the natural companion to this one — a map of the field this history helped to build.