800 Megs of Data Per Person Last Year?

Ant writes "Growing net, computer and phone use is driving a huge rise in the amount of information people generate and use. US researchers estimate that every year 800MB of information is produced for every person on the planet. Their study found that information stored on paper, film, magnetic and optical disks has doubled since 1999. Paper is still proving popular though. The amount of information stored in books, journals and other documents has grown 43% in three years."

Comments

This article is so clearly a dupe...

[FrankoBoy]

Damn... Go read the first article from like frou days ago...

On the edge of the Singularity

[headkase]

With information growning exponentialy, one must wonder if we're on the edge of the Singularity as anticipated by Vernor Vinge.

What IS information

[EvilTwinSkippy]

According to Claude Shannon's groundbreaking work on the subject, Information is measure by its surprise quotient. 80 bits that you are expecting to see carry no information at all. 1 bit that is a surprise is a tremendous amount of information.

The game of scrabble is a good illustration. Common letters (a,e,i,o,u,s) have 1 point. Uncommon letters (z,x) are 10 points. All letters have a different point score based on their frequency of use in the english language. (At least for the english version of the game. I know the scores on the letters are different for the German version at least.)

All of the modern compression algorythems work on this principle. They detect the parts of your "signal" that contain the least information, and convert them to a smaller form. Of course you have to know a bit about your signal before you can be good at all at predicting what is common or not.

LZW compression, for instance, is great at compressing text. Images OTOH LZW is not so good at. At least color images (GIF actually uses LZW.)

For full color images we use JPEG, which breaks the image into 8x8 tiles and then compares the tiles to the output of the inverse cosine transform. So instead of storing the actual RGB information it actually stores the coefficients of the transform needed to reconstruct the tile, and the varience of the original from the ideal.

MPEG uses a JPEG-like compression for key frames, and then simply stores what pixes change in between frames. Some implementations also attempt to compensate for motion, which is starting to get beyond what I can explain in the space provided.

Suffice to say information is the level of surprise inside a signal. It doesn't really matter what form of signal it is.

Re:...and 3/4 of that are raw bitmap images

[EvilTwinSkippy]

Actually images with a large field of one color compress quite well. The compression algorythems sees "Oh blue again" and represents all the blue pixels as a single bit. Some just say "it's blue until I tell you otherwise" or "It's blue for the next 65242 pixels."

Images with complex shapes compress terribly. I was out at a botanical garden trying to photograph the ends of tree branches as they fork off into millions of buds. It looks crappy in JPEG form.