This would run transparent to the user. At the end of the day, he'll have pure text files that can be searched. Implementing a google-like algorithm modified to work on flat files would be great. It'd be basically like the Remembrance Agent but more advanced.
I don't know how I'd feel about using such a system, but I'm reasonably sure that we're on the way there indirectly. Heck, most of us voluntarily not only record our lives, but also post it publicly. What's wrong with recording all our life (right from that websearch on the tablet, to what you're ordering at the restaurant whose name you forgot but you visited it on the day right after your best friend got married) if it's going to make it life easier. As such, I've noticed personally that I have stopped storing information in my head unless I use it regularly, but instead I store pointers to the location of the information, like whether I did a search {if yes, then what were the keywords, and what was the result's position}, or did I look it up on a forum or maybe even a real book.
/thinking-out-loud
All of this is possible now, and the lifelogging community has been doing it for years and years. "Why" you do it is still an unanswered question. The usefulness hasn't been proven out.
Over the past year, I've come to feel that perhaps recording every word isn't as important as recording the context. All the words you said are there, but the context is lost. In context, that was a really funny joke, but the words, by themselves, were rude, or cruel, or sexist, or racist. It doesn't record everyone around you laughing, or how good you felt, or that it was a really good day as far as your depression management was concerned, or that that was the last time you saw X before his accident, and when you do go back and reminisce, those are the important things, not the precise words you used.
As of yet, lifelogging hasn't saved the meaning.
If only there was a way to make a person relive the same "memory" or "experience", like a Pensieve[1]
What does the research tell us about how we feel? How is that different across genders? Across personality types? Across cultures?
How can we represent that? What sort of accuracy matters? Should it be relative or absolute? Does it matter? What will we be using it for?
How can we record it? How much can we trust self-analysis, like diaries or mood questions? How much can we trust biometrics? Is that dependent per-person or can we generalize?
What sort of biometrics would we need? How long would we have to wear them? Where? Power? Fashion? What about on the beach, can they survive sand and salt water and 110 degree F weather? What about military use, can they survive high-pressure sand and persistent sweat and the very different bodily reactions that someone in a firefight goes through compared to a soccer mom?
What is the goal of an emotional recollection? To understand it yourself? To remind yourself? To reminisce? To convey your emotions to another? To allow another to find something similar in their own recorded memories so they can empathize better?
Don't stop at fiction or wait for someone else to make a web service for it. This is something that can be designed and built today. This is something that could have been designed and built a decade ago.
Would life-logging help us find this essay or would the essay still be tricky to re-discover?
(Frustratingly, Wikipedia "Lists of X" is in fact "Lists of X where each item on the list must have a Wikipedia article". That's probably great in some respects, but it makes for a lousy comprehensive list.)
At least, it would only be fifteen years of personal saved items to go through, instead of the entire internet.
I'm still not sure this is a compelling application for a wide audience, but for several of the early everyday users of wearable computers (including myself) it had a lot of appeal.
"Implicit query" is another term for it. On the desktop side, one of the most realized solutions was the Dashboard prototype: http://nat.org/dashboard/
I've long argued that designing multi-modal interfaces is where applications need to go. We're seeing this now with desktop, web and mobile apps all running against the same APIs, but what it really means is you're more likely to be able to also design a UI that's really an audio-only interface, whispering to you continuously via a Bluetooth headset, or a wrist-mounted UI, or... etc.
It also means you can design and build something useful for desktop users, and then mobile users, and then scaled even further down for wearable users. Real-time continuous Google searching as a desktop sidebar, or rendered on an iPad or iPod Touch display linked to desktop input as a second screen, would be a start.
Was it Steven Johnson describing how he uses DevonThink (http://www.nytimes.com/2005/01/30/books/review/30JOHNSON.htm...)?
I notice there's no X or Z in DEVONthink, but there is a V.
I also notice the author of the custom software he used before that was Pinboard's Maciej Ceglowski. Fascinating!
I think that recording everything is almost useless, without a good filter (approaching AI levels). The real power, I think, comes in recall and referencing. Just as an example, I've been reading "The Stars my Destination" and two words came up that I didn't recognize (epileptoid and asthenic - funny, FF doesn't recognize them either!), and it would have been handy to just look them up quickly.
As to "forgetting", people already abuse current mnemonic devices (such as photos) to "hold on" too long, and using people's words against them is a problem with or without perfect recall.
For trade secrets, etc, you do what any high security facility does: recording/transmitting devices left at the door, outside, and simply accept the crippling that comes from lack of instant access to information.
Also, what's wrong with living inside Emacs? ;P
Edit to followup: Works of fiction I've found interesting in this vein include "The Final Cut" (even with the poor execution) and "Strange Days"; anyone have any recommendations along these lines?
Yes, it is! But one thing I've learned in managing online communities is that you don't give people the ability to do something you don't want them to do. I don't feel having perfect electronic recall should be a feature without more context sensitivity.
> For trade secrets, etc, you do what any high security facility does
Except for the Steve Mann argument: it's a prosthesis. When you take the hardware away from him, you're disabling him. He becomes disoriented and cannot function as well. It's like taking away a wheelchair, or crutches, or a hearing aid. And when more and more of your potential employees rely on these devices 18 hours a day, more and more will be negatively impacted. No, I think it's something that has to be fixed in the legal system or fixed in the entire design of a wearable system.
> Works of fiction I've found interesting in this vein
The novel "Snow Crash" has wearable computing used only by a subset of wired individuals derisively called "gargoyles." Everyone else still uses workstations.
The "Old Man's War" series of books has soldiers with embedded computers, including a subset of soldiers who have them embedded since birth.
The novel "Permanence" also has them, but used for pervasive IP rights enforcement.
The novel "Signal to Noise" has an elite intellectual class with implants that work with pervasive wireless sensors and fully immersive workstations.
The film "Stranger Than Fiction" has AR-style overlays at least during the opening sequences.
The short film "Sight" touches on AR and behavorial control a bit: http://vimeo.com/46304267
This kottke.org piece covers a few other shorts: http://kottke.org/12/04/the-real-google-glasses
I can't say you're wrong, but I can't say I'm convinced either...
http://en.wikipedia.org/wiki/Black_Mirror_(TV_series)#3._.22...
I think the most useful aspect of large digital archives will be when natural language processing and image recognition get good enough to automatically distil the archive into actual information (rather than just data), no manual reviewing ever.
This would be along the lines of being able to ask "What did Bob say about X?" or "Find things that correlate with weight loss" and get meaningful answers (with summaries and so on, as appropriate).
(However, this seems mighty similar to saying "sufficiently smart compiler" http://c2.com/cgi/wiki?SufficientlySmartCompiler.)
The thing about it is that I don't think the stories the software generates have to be precisely true. I don't even think they have to be internally consistent over the long term. Look at the data, generate a "good enough" story that helps push you toward your goal, and revise it later. Personal historical revisionism. In the long term, you won't remember the details, all you'll know is that your mood is generally better and you're two inches slimmer in your waist, so, what does it matter how it happened? The computer is looking out for you, that will be all that matters.
in:conversations, timeframe: 201x-11, location:office, present:Kalle Anka.
Fast forward 90 seconds and I'm proving he ACTUALLY told me to do exactly that.
Or maybe it's a personal fight, and you're throwing his words back in his face, and now he realizes he can't trust you with his feelings because you always take him so literally. He's sorry he's not a poet, that he's only human and has trouble expressing his emotions sometimes, but this just isn't going to work between us.
The only person who's ever looked bad when I've used someone's words against them is me.
Also, even though people may say the wrong things when they are emotional, a repeated pattern of them expressing those sentiments can reveal their real disposition. In this way, a record of what is said, word for word, can be useful to uncover someone's real thoughts. I already do this with my brain, but it is easier when it's in searchable, dated text.
This is the common wearable computing utility argument, but in practice, it doesn't seem to pan out.
Gordon Bell digitized much of his life, and everything for the past ten or twenty years. Phone calls, emails, a photo every sixty seconds, more when his heart rate increased. He hardly ever went back to it. Revisiting it was so rare as to be a notable event in and of itself. Rather, he found people with whom he had conversations would go to him and use him as a reference library.
Bradley Rhodes' Remembrance Agent was an Emacs thing which actively indexed and cross-referenced anything you were typing with things you had written before. He's working on Glass now, afaik: http://www.remem.org/ It isn't a generally useful solution because it requires that you live inside of Emacs. Today you'd need something that could index across multiple devices, multiple independent cloud storage systems, multiple independent accounts, etc.; or something that was locked into a single ecosystem and you lived entirely there, but then you'd develop "blind spots" for things that occurred elsewhere, like how you stop hanging out with certain friends because the only way they communicate with you is via Facebook but you turned off all of Facebook's notifications.
There was an essay a long time ago, a writer had been filing every link and every note into a pre-Evernote piece of notetaking/hyperlinking/PIM software, something with an X or a Z in the name, but I can't recall it or the piece. The essay was about an article he was writing during which the software brought up a saved article and a note he had written, and forgotten about, in a creepily timely and seemingly prescient moment. He wondered at what point the software needs to get credit for providing the research and the associations.
"Forgetting" is a key part of human existence that most wearable applications tend to ignore. "You were last here with [your ex]" says Foursquare. "You haven't talked with [your ex] in a while, make her day by leaving a message on her wall" says Facebook. Using someone's precise words against them can be emotionally cruel. Legally, you're expected to forget all of the specific details of things like trade secrets and company processes when you leave a job; how is that to be reconciled with your perfect, digital memory? None of these things are being actively explored.
We already hit the "second generation" of applications as described by the author; it's the third generation that interests me.