The computer mouse sat in a lab for almost twenty years before anyone bought one.
Douglas Engelbart built the first one in the 1960s at Stanford Research Institute, a wooden shell with two metal wheels. It worked. It was clever. And for most of the next two decades it stayed a research curiosity, something a few specialists knew about and nobody else could use.
That twenty-year delay had nothing to do with the engineering, which was finished early. Somebody still had to make the thing understandable to people who would never care how it functioned. Design is the usual name for that work, which is an unlucky name, because in most budgets the design line sits below the build and gets trimmed first when the timeline slips. Filed near the launch party.
Raw Capability Doesn’t Win On Its Own
A technology can be faster and cheaper and more advanced than everything around it and still lose to something worse that a stranger can figure out in a minute.
The mouse again. Engelbart had a working device; what he lacked was a system in which the device made sense to anyone outside a lab. Xerox built that system. On April 27, 1981, the company announced the 8010 Star, a $16,000 workstation with on-screen icons, folders, overlapping windows, and a mouse for pointing at all of it. You could see your work now instead of memorizing the commands that summoned it. The Star sold badly, largely on price, but it answered the question of whether point-and-click computing worked for people who weren’t specialists.
Worth noting what that labor consisted of. Engelbart’s group had tested the mouse against a light pen, a joystick, and a knee-operated control, and the mouse won on accuracy and error rate; they had also worked out that cursor movement should run at roughly twice the movement of the hand. When several of them moved to Xerox PARC in the early seventies, the researchers there read the test report and decided not to repeat the experiments. Any system they built would use a mouse. That is empirical work about human behavior, done years before anything shipped, and it looks nothing like styling.
Apple pushed the same idea further. On the Star, clicking an icon got you a pop-up menu that sat between your intention and the machine’s response. Apple’s software team went to direct manipulation instead: pull a window’s corner to resize it, grab it to move it, nothing in between. That version shipped on the 1984 Macintosh, got copied into Windows, and reached hundreds of millions of desks. The pointing device itself barely changed across those eighteen years. Icons, windows, menus, and the rules governing how they answered a click absorbed almost all the labor.
The Touchscreen Waited Forty Years
Eric Johnson, an engineer at the Royal Radar Establishment in England, described a capacitive touch panel in 1965 and held a patent by 1969. A graduate student at the University of Toronto built a working multi-touch system in 1982, and researchers at Bell Labs built a capacitive version independently that same year. Pinch to zoom and two-finger rotate were both running in university labs decades before anyone owned a phone that used them.
IBM shipped a real touchscreen phone in 1994. The Simon had been shown at COMDEX in late 1992 and reached buyers in August 1994, priced around $899, with a 4.5-inch monochrome screen you operated by pressing hard with a stylus. It could send faxes and email. It sold so few units that most people today have never heard of it. Everything technically necessary for a mass-market touch device existed. What existed alongside it was a set of interaction choices nobody wanted to make.
The iPhone landed in 2007 on top of all that prior art, and Apple invented none of the underlying pieces. The decision it made was narrower and more consequential. Your finger would be the only instrument, no stylus and no learning curve worth mentioning, and every interaction in the device had to survive that constraint or get cut.
The sequence repeats across most of the technologies now treated as inevitable. A breakthrough gets demonstrated twenty or forty years ahead of any market for it, sits in papers and prototypes, and the company that eventually owns the category tends to be whoever worked out how to put the thing into a person’s hand rather than whoever discovered it.
The Web Is The Clearest Case, And A Test For Any Studio
On the web, the capability and the translation were built by different people a few years apart, so the handoff is easy to watch.
Tim Berners-Lee wrote the first browser at CERN in 1990. Protocols, server, links, all of it running. It ran on a NeXT machine, which cost thousands of dollars and sold perhaps fifty thousand units in total across the company’s lifetime, so for the first couple of years the web was a working system available to almost nobody.
Then Mosaic. Marc Andreessen and Eric Bina released version 1.0 on April 22, 1993, out of the NCSA at the University of Illinois; Windows and Macintosh versions followed later that year, and it was free for most uses. Its notable trick was showing images inline with text where earlier browsers had opened them in separate windows. By December 1993 the NCSA counted more than 5,000 downloads a month. Nothing underneath had improved. A page had simply become something an ordinary person could look at and click, and the boom that followed is generally traced to that.
Which makes this layer a fair way to judge whoever builds your site. Making a page look expensive is not difficult and not rare. The work that separates the best web design agencies from vendors who decorate is taking a complicated product, service, or argument and getting a stranger to understand it inside the first ten seconds of scrolling. Ask a studio to explain how they did that on a specific past project. The answer sorts them quickly.
What This Means If You’re Building Something Now
The practical question is when the translation work starts. If it starts after engineering declares itself finished, the schedule has already classified it as finish work, whatever anyone says in the kickoff.
Put the people responsible for how a feature gets understood in the room while the feature is being specified, back when their options are still wider than color and copy. Your customer will never meet the architecture or the pipeline that took the whole quarter. They meet one surface, and on the record of the last sixty years that surface tends to decide how many of them stay.
Engelbart’s mouse worked in the mid-sixties. Johnson had his touch patent by 1969, and CERN had a running web in 1990, on a machine almost nobody owned. Each of them sat there, finished, until somebody did the translating. Whatever sits on your roadmap is on the same clock, and the only open question is whether that work gets done on purpose or gets done by whoever ships the version people can finally use.
