How lifelogging will become easy and automatic
- 19 November, 2016 23:00
Lifelogging is dying, right?
Gordon Bell, who served as leader of the lifelogging movement since the 1990s, told me earlier this year that he had given it up.
A couple of weeks after I dropped the news about Bell, the former editor-in-chief of Wired, Chris Anderson, tweeted that he had quit lifelogging as well.
What is lifelogging? Definitions vary. But generally, lifelogging is the automated collection of personal documents, data (including biometric data), photos and videos that can be used later for total recall or preserved for posterity.
Some lifeloggers (including Bell) use cameras, either clipped on clothing or worn around the neck on a lanyard, which take photos constantly.
In recent years, consumer lifelogging cameras have emerged. But the public is rejecting them.
One promising early consumer lifelogging camera, the Autographer from OMG Life, launched in 2012, then died off two years ago.
In 2012, a company called Narrative launched a tiny lifelogging camera called the Narrative Clip (initially called the Memoto camera). The company announced in late September that it would shut down. (A few days later, it emerged that a group of former Narrative employees were trying to save the camera and the cloud service where pictures were stored. )
These cameras are failing. Lifelogging cameras are awkwardly intrusive, inconvenient and tend to take lousy pictures compared to a smartphone. Most of the pictures taken by automatic cameras show boring or blurry scenes, or pictures of people that are unflattering.
Meanwhile, it's clear that glasses are a better place for lifelogging cameras.
Snap Spectacles are getting a lot of attention in the press, in part because the press is being played by a series of publicity stunts and contrived scarcity. These include temporary, randomly placed Spectacles vending machines and a re-shoot of the famous Robert-Scoble-in-the-shower photo, this time with Spectacles. (Snap is the new name for the company formerly known as Snapchat.)
To the best of my knowledge, none of the few Spectacles users deliberately does lifelogging, and Snap isn't promoting them as lifelogging cameras. Still, Snap's splashy marketing will help normalize camera glasses.
Why would anyone want to lifelog? That question is becoming increasingly irrelevant. You're already lifelogging. We all are.
Does anybody believe personal data isn't being collected?
Lifelogging has two parts:
- 1. Collect all your data
- 2. Instantly recall specific memories from that data on command
Is there any question that all our data is being collected? The collective data harvested and stored by Google, Facebook, Amazon and other companies retain everything you do, everywhere you go, everyone you know and everything you buy.
Let's start with what Google collects. Google Photos auto-collects every photo I place on my laptop (such as all the pictures from my DSLR camera) and every photo I take with my iPhone and iPad. Each of these photos is tagged with location, time and date. Google then applies artificial intelligence to them to recognize individual people and specific animals and objects. For example, I can search for "pizza in October, 2013," and Google Photos will show me every pizza I photographed that month. Google Photos, which are arranged by default in reverse-chronological order, is a rudimentary lifelog all by itself.
Consider the data exposed or listed on the Google Dashboard, Google's full-disclosure site for user data. This page reminds me that Google has all my relationships in Contacts, all my documents in Drive, all my emails for the past 10 years, all my Google+ activity, my location history, my package-tracking history, my music, my Play Store activity, everything in my About me profile, my search history, my tasks, my phone calls and voicemails and my YouTube videos. From the Dashboard, I can even listen to the recordings of my own voice interacting with Google Assistant, Google Home or Google Voice Search with a single click.
All that's just Google. I realized how much data is out there on social and other media when I turned my blog on Elgan.com into a public lifelog. I used IFTTT and other tools to funnel my social activity to the blog, including my two blogs, two email newsletters, my podcast, plus all my posts on three Twitter feeds, Google+, Facebook, Medium, Instagram, YouTube, Swarm and others.
(A service I told you about earlier this year, called Digi Me, will do all this for you as well.)
I probably create more public content than average. In general, however, most people already create enough content and data for an amazing lifelog.
So the first part of lifelogging is already happening. The question is: How do you do the second part -- instantly recall specific memories from that data on command?
My blog-based public lifelog is OK as a stream, but it's lousy for instantly conjuring up facts and memories on command.
If you look at the neglected category of lifelogging smartphone apps, which includes Fabric, Rove, Quantified, Instant, Logsit, Loca, Swarm and others -- it's really focused on the problem of accessing the same kind of data that's automatically collected elsewhere.
These apps in their present form are doomed, because much better ways to call up lifelogging information are being aggressively developed.
Just add search, A.I. and virtual assistants
What Gordon Bell, Chris Anderson and other lifeloggers realized is that collecting all your data creates a problem. You end up with terabytes of data with no fast way to extract the facts you're looking for.
But this problem is being solved in a major way.
Just as the collection of personal data for lifelogging "just happens" through normal activity, like using the Internet and sharing selfies and food photos, so will the tools for instant retrieval of "just happen."
As we speak, two broad technology revolutions will solve the lifelogging data access problem: artificial intelligence (A.I.) and virtual assistants.
This idea already works in a rudimentary form. I told you how Google Photos calls up the pizza I ate in October 2013. It turns out that by telling the Google Assistant in Google Allo, "show me my photos of pizza from October 2013," it will show me those pictures. The pictures are tagged with a location, too: Pizzeria Bianco in Phoenix, Arizona. My wife is in the pictures. Suddenly I'm remembering the entire day.
When the A.I. gets better, and all my data is accessible through Google Assistant and other virtual assistants, the fidelity of lifelogging retrieval will get ever sharper.
Also: A.I. will improve data collection as well. The next generation of lifelogging cameras is using A.I. to solve the problem of snapping and storing thousands of irrelevant photos.
A camera called the QindredCam from the startup Acumulus9 claims to use something like A.I. to choose when the camera takes photos. The camera is still under development, but the general idea is sound. It probably looks for changing scenery, faces, adequate ambient light and location to pick the perfect moments for taking snapshots.
Sony teased something similar in February. Sony's concept is called the Xperia Eye, and it uses face identification (not face recognition) and other factors to auto-snap photos with its 180-degree fish-eye lens camera.
The most ambitious A.I. lifelogging camera might be the Asteria camera by a San Francisco startup of the same name. The Asteria is an early concept for creating a lifelogging camera-like device that auto-captures photos, sounds, motion, location and even temperature, then uses A.I. to decide what to capture and retain.
Lifelogging used to be a difficult, complex and time-consuming project that didn't work very well -- a discrete research activity or hobby designed to result in total recall or computer-enhanced photographic memory. That brand of lifelogging is dying because it's antiquated or obsolete.
In its place, the natural evolution of personal data collection and picture taking, cloud storage, artificial intelligence and virtual assistants will automatically give us a better form of lifelogging than even the most visionary lifeloggers ever dreamed of.
Unless we intervene (for privacy or other reasons), the data will simply be collected, and in an increasingly A.I.-mediated fashion. Then, A.I.-based virtual assistants will simply retrieve for us whatever specific information about our past that we ask for.
Lifelogging as an independently driven project has no future. But the payoff of natural lifelogging -- the ability to remember anything instantly -- will simply happen for all of us who want it, thanks to A.I.