Dead End: Google Glass & The First Wearable Wave

Abstract

Google Glass was meant to take the computer out of the pocket and put it in the field of view. Instead it made the wearer conspicuous. Bars banned it, eight members of Congress wrote to Larry Page about it, a California driver was cited for wearing it, and in February 2014 Google published an etiquette guide telling owners not to be “creepy or rude (aka a ‘Glasshole’)”. The American Dialect Society had already put the word on its 2013 ballot. Google ended the consumer program in January 2015, relaunched Glass for factories and hospitals in 2017, and stopped selling that in March 2023. The technology mostly worked. The problem was that the people around the wearer could not tell what it was doing.

Project Glass

Google X began work around 2010 on what was announced on 4 April 2012 as Project Glass. The idea was old, a head-up display of the kind fighter pilots had used for decades. The form was new: a spectacle frame carrying a prism display, a camera, a microphone, Bluetooth, Wi-Fi, and a link to an Android phone.

Sergey Brin ran it as his own project and wore prototypes in public. At Google I/O in June 2012 he put skydivers in Glass and had them stream their jump onto the Moscone Center screen on the way down, then took $1,500 pre-orders from developers in the room.

The Explorer Edition shipped from 16 April 2013. It was a developer release, described as such, and treated by everyone else as a product launch. The display was a 640 × 360 prism sitting above the right eye, equivalent to a 25-inch screen at eight feet. There was a 5-megapixel camera with 720p video, bone-conduction audio, a touchpad along the temple, and voice control that started with “OK Glass.” Battery life covered a day of glancing at notifications and well under an hour of continuous recording. In direct sun it got hot.

The Glasshole Effect

The camera was always visible and only sometimes recording, and nobody outside the frame could tell which.

The 5 Point Cafe in Seattle banned Glass in March 2013, weeks before the first units shipped. Las Vegas casinos, cinemas, and strip clubs followed. A West Virginia legislator introduced a bill against driving in a head-mounted display. In May 2013 eight members of the Congressional Bi-Partisan Privacy Caucus sent Larry Page a letter with eight questions, including whether Glass could collect data on people who were not wearing it and whether Google would revise its privacy policy for facial recognition.

On 30 October 2013 a California driver, Cecilia Abadie, was cited near San Diego for wearing Glass at the wheel under a statute about screens visible to the driver. On 16 January 2014 a court commissioner threw the citation out: there was no proof the device had been switched on. The legal system had run into the same problem as the bystanders in the bar.

Google’s response, in February 2014, was an official etiquette page. Do ask permission before filming people. Don’t stare into the prism. Don’t wear it where phones are banned. “Don’t be creepy or rude (aka a ‘Glasshole’).” The company was now using the insult in its own documentation. When the American Dialect Society voted on the words of 2013, “glasshole” was on the ballot in the Most Likely to Succeed category, defined as “a person made oblivious by wearing Google Glass, a head-mounted computer.”

The privacy-by-design problem

Phones can record too, but a phone announces itself: it comes up, it points, the hand and the arm and the screen are all part of the signal. Glass removed those signals on purpose, because removing them was the entire product idea. Capture was supposed to be seamless.

Social norms about privacy in public rest on being able to see that you are being recorded. That mutual recognition is what makes consent possible without anyone saying a word. Glass did not make recording invisible, but it made it ambiguous, and ambiguity is enough: the bystander now has to ask, and asking is the cost.

The problem is unsolved. Apple’s Vision Pro (2024) animates a white pattern on its outward-facing EyeSight display while capturing, which reviewers immediately criticised for looking nothing like the red light people already understand. A computer worn on the face is a camera pointed at everyone else, and the indicator problem comes with it.

The wearable that found a market in this period sat on the wrist rather than the face, and even there the independent pioneer was squeezed out: see Dead End: Pebble.

Selling to the Public, Then Not

Google opened Glass to any US buyer for one day on 15 April 2014 at $1,500, sold out, then made it generally available in the US that May and in the UK in June at £1,000. Sales were never disclosed and were, by every account since, small.

On 15 January 2015 Google closed the Explorer program, stopped selling Glass on the 19th, and moved the team out of Google X into its own division under Ivy Ross, reporting to Tony Fadell, who had come to Google with the $3.2 billion Nest acquisition. Coverage treated it as a burial. Internally it was a change of customer.

The Enterprise Edition

Glass Enterprise Edition was announced on 18 July 2017 by project lead Jay Kothari, after a two-year quiet pilot. More than fifty companies had been using it, among them AGCO, DHL, Dignity Health, Sutter Health, Volkswagen, and Boeing. At AGCO, putting assembly checklists and manuals in the field of view instead of on a tablet across the room cut machine production time by 25% and inspection time by 30%.

The consumer objections mostly evaporated in this setting. A warehouse picker being told “aisle B7” hands-free has no bystander to disturb. A surgeon reading vitals without turning away is not surveilling anyone. A quality inspector photographing a part is doing what the job already required.

Enterprise Edition 2 followed in May 2019 with a faster processor and a USB-C battery. Google stopped selling it on 15 March 2023 and ended support that September. It died as a small business inside a large company, not as a rejected product.

Why Consumer Face Computers Keep Failing

Glass was the first case, not the only one.

Snap Spectacles (2016) were camera sunglasses sold from vending machines, and they showed how thin the demand was in numbers: Snap sold roughly 150,000 first-generation pairs after ordering around 800,000, and wrote off $39.9 million in unsold inventory and cancelled commitments in the third quarter of 2017. Three more generations followed through 2021 without changing the picture.

Amazon Echo Frames (2019) put Alexa in a frame with no display and stayed a niche.

Ray-Ban Stories (2021) and Ray-Ban Meta (2023) sold better than any of them by hiding the technology inside a frame people already wanted to wear, and by shipping a camera with no display at all.

Apple Vision Pro (2024) at $3,499 is a headset rather than a pair of glasses, priced where mass adoption is not the question being asked.

The obstacles have not moved much since 2013. An all-day display in a spectacle frame needs more energy than a battery that size holds. A bright, high-resolution, low-power microdisplay with waveguide optics that puts an image into the pupil remains expensive and optically compromised. Visible technology on the face still reads as visible technology on the face. And the applications people name (navigation, notifications, translation) are worth minutes a day, not hours, which is a poor return for something you must wear all day to have when you need it.

Dead End: Utility for the Wearer, Cost for Everyone Else

Glass worked. It was light enough, the display was legible, the voice control was serviceable, and in a warehouse it paid for itself. What it could not do was give the people around the wearer any reason to accept it.

Most consumer technology asks nothing of bystanders. A phone in a pocket, a watch on a wrist, headphones on a head: the cost falls on the owner. Glass moved part of the cost onto everyone in the room, in the form of not knowing whether they were being recorded, and offered them nothing in exchange. The bans, the letter from Congress, the traffic citation, and Google’s own etiquette guide were all the same objection stated by different institutions.

That is why the enterprise version survived and the consumer version did not. In a factory, the bystanders are colleagues who agreed to the arrangement as a condition of the job. In a bar, they did not.