2003-05-28 – Bandwidth of a lifetime – Updated

From \”Strange but true\” on the back of Computer Weekly, 27th May:

\”Boffins at the US Defense Advanced Research Projects Agency is (sic) planning to create a comprehensive searchable database of indvidual human lives encompassing every communication, encounter, transaction and emotion generated by a lifetime of social interaction.
Called Lifelog, the system would use medical and other sensors to record every scrap of data that a subject might \”talk or write about, or be observed to do or say or confront, or be resonably assumed to intend\”.

This seems to be a very similar idea to the one below, which I had about 18 months ago, which would record everything a person saw, heard, and so on, through their entire life. The idea of heart rate monitors, etc. would be an interesting addition!
Not sure how the database would work, though.. You would need an army of psycologists and data entry clerks to tabulate everything and put it into a computer database, and what would it tell you? What happens when they watch Watership Down, or the Teletubbies? You will have fun asking a 3 year old, too!

Originally published: 2001-11-20

Just a few years ago, the sheer volume of media involved in filming a person’s every waking moment, noise, eye movement, even the smells they experience, would have been improbably huge.

Now, due to the cavernous harddisk systems available, and the emergence of the div-x 🙂 video compression codec, we could quite easily manage hundreds of thousands of hours. So, how much exactly?

I have The Matrix on disk at home. It is a div-x 🙂 file, and clocks in at 598Mb for the whole film, some 136 minutes of high quality, fullscreen video, with slightly dodgy stereo sound. This translates to about 4.4Mb per minute. Adding in the sense of smell, though not currently technically possible, would only involve a single coded number, every few seconds, since smells change far slower than any other sensory input. The bandwidth cost would be negligible. The sense of touch would be tricky to capture, too, but again, the total bandwidth requied would probably be less than an audio feed. Although there is far more information, it changes far more slowly. (Audio frequencies are changing thousands of times per second, the skin stops detecting vibrations above about 100Hz, depending on location. Fingertips are more sensitve than the back.) Taste would also be a low-bandwidth feed, and, though again not currently possible, it would be on the order of the other half of the olfactory mechanism. Indeed, it is closely related.
I doubt I will hear much complaint if I increase the bandwidth per minute to 5Mb per minute, which is, I feel, on the generous side.

Now, unless we intend to try recording dreams, we will only need about 18 hours a day of recording. Few people need less than 6 hours of sleep per day, and most need 8, so again, this is on the high side.
18 hours converts to 5400 Mb per day. For a week, you would need 37800 Mb, or about 40Gb. This is fast becoming a trivially small amount of data!

However, we can go further. If a person listens to a song twice, that can be recorded once, and only the differences need be recorded again. If a person watches a film, you can take the film from elsewhere, as with The Matrix. Millions of people have watched it, and, again, massive savings could be made by simply recording it once, then simply recording the differences.
However, given the low costs of storage media, it is unlikely that, unless this could be automated, it would be worth doing. For £100 you can buy a 40Gb drive. For large corporations and the government, 127,764,000Mb of data isn’t that expensive any more.. 128Terabytes would hold the first 65 years of your life, in totality, for a cost, at today’s prices, of £338,000, and that is simply priced at £100 for 40Gb. In another 18 months, when prices have halved and the capacity has quadrupled, you could start using far cheaper drives and storage. Use of cheaper solutions, such as tape streamers or CD-R, pulls the price down even further!

Next week: How to build the data ports. 😉

Leave a Comment