The Psychotic State Number 5 December 31, 1998 A MILLENNIAL RIFF ON mp3 The proximity of the end of the year has me in an unusual psychotic state. I've been thinking a bit about where our industry is going in the next century. I'm not the guy who has to think about all the Y2K stuff the company has to do (Tony Ward's in charge of that) so my delirious mind has gone in directions rather different from 2 digit or 4 digit numbers and whether there's a February 29th two years from now. (There is.) There's an interesting development that seems on its face to not be millennial in import, but I think that it's a significant portent. What I like about it is that involves technology at its heart, but it's really a creative revolution. What I'm talking about is the recent dust-up over mp3 and the Diamond Rio. For those of you who haven't been following it, here's a brief synopsis. As you may know, there's an organization called the Moving Pictures Experts Group, which works on video transmission and storage. The standards they produce are the ones that govern the internal format for all the files you see with the extension MPEG. They have come up with various file formats with different features, of which the simplest, MPEG 1, is a video with a stereo audio track. Recently they proposed an improvement to their audio storage algorithm, and they added it to the standard. It's officially denoted MPEG 1 layer 3. The number 1 is version of the standard; layer 3 refers to the fact that a layer 3 compliant MPEG player has to be able to play the layer 3, 2, and 1 files, while layer 2 compliance only requires a player to play layer 1 and 2 files. The right extension for such a file would be .mpeg1-3 or something; instead it's called .mp3 because of the persistent malign effects of DOS. Recently a company decided to manufacture a device called a Rio that allows people to download mp3 files into it and carry them around and play them - an mp3-man as it were. Since the device has no moving parts, it's quite small, and I'm sure it will ultimately be very cheap. Naturally, the recording industry companies that make their livings selling CDs were less than thrilled at how easy these devices made it for people to copy and trade around recordings, perhaps without buying them. In response to someone doing something they didn't like, the music companies did the All American thing - they sued. They tried to get an injunction to keep the Rio devices from being sold. Recently they lost, and the next great revolution is on. Because I'm a techie, and because it's cool, I thought I'd explain a bit about why mp3 is different from what you're used to. Let's begin with how CDs got the way they are. Every sound can be decomposed as the sum of waves of different frequencies. The human ear is sensitive to at best to the range of 20 cycles per second up to at most 20,000 or so. Middle C is an easy-to-remember 256 cycles per second, and each octave up in pitch represents a doubling of the frequency. Therefore the human ear can detect a range of about 10 octaves. Now here's how they designed CDs: to represent a sound digitally, you have to break the continuous shape of the wave into discrete points. To do this, you sample the wave at regular intervals. If you do this, you are going to lose some information - the shape of the wave between the points you sample is unknown to you. Suppose you want to record a wave of 20,000 cycles per second. To do this, you'll have to sample the wave at least twice per cycle. (Roughly: once to get the top of the wave, once to get the bottom.) Thus, to get all of human hearing, you have to sample at last 40,000 times per second. To solve a technical problem, the designers of the CD standard did a little better than that, so the sampling is at 44,100 times per second. A stereo recording is two channels, and each sample is 16 bits. (If you used 8 bit samples there would be a noticeable rounding error on each one, yielding a detectable hiss.) The designers of the CD wanted to record to Beethoven's 9th symphony on a single disk, and it's about 70 minutes long. In total there is (44,100 samples/second/channel) * (16 bits/sample) * (1/8 bytes/bit) * (2 channels) * (60 seconds/minute) * (70 minutes) = 740 Mbytes. Just about what a CD holds. The interesting thing about mp3 files is that they're way, way smaller than the calculation above indicates. There's no "smarts" in that way of doing things, it just brute-force tries to reproduce the wave shape as accurately possible. But while its true that the human ear is sensitive to waves up to 20,000 cycles per second, there are lots of things that you can do to the wave that will yield waves that are _mathematically_ rather different but that _sound_ the same to us. So what the clever folks at MPEG did was to realize that there's a lot of information on the CD about the wave that we don't need. Using the mp3 compression you can shrink the file by a factor of _twelve_ without _any_ noticeable degradation in quality. You can do way better if you're willing to suffer some audible distortion. For example, a sound can be shrunk by a factor of 100 or so if you're willing to have it sound roughly like an AM radio broadcast. You can see why the music industry freaked out. With compressions like these, you can easily e-mail entire albums to your friends. You can store lots of music on your PC's hard disk. And then you can download your favorites to the Rio and listen to them while you jog. To retaliate, the music industry has been forcing artists to remove the mp3 files of their music that some groups had put up on their web sites. It seems to me to be a rearguard action to delay the inevitable. But I think that the mp3 is even more significant than the possible demise of the music industry as we know it. In a way, the mp3 represents a big step toward what I think of as The No-Medium. You see, in our history we've always had to think about _things_ when we think about information. So, for example, when we think of music, we have those round physical disks on which they're recorded somewhere in the backs of our minds. It's manifest in quite a number of ways: *) The reason that most popular music is organized into pieces of about 3 minutes in length is that shellac 78 RPM records - the major medium on which music was distributed for nearly 60 years - holds just that much. When the 45 RPM single was designed, it was smaller in size and lighter in weight, but, not coincidentally, the same length of time. *) The word "album" originally referred to the binder in which the 78s were stored, which usually contained between 4 and 12 records. When the LP was introduced, it held an album's worth. *) When we write a book about everything someone has recorded it's called a discography - even if the recordings were on tapes. *) Even today we refer to Director (or MPEG) as multi-media, even though there is in fact no celluloid film or vinyl disk anywhere involved. It has just one medium, if we count the CD ROM on which it's generally distributed. It seems to me that the limitations of the physical medium on which the information is stored or reproduced functions in some ways as a paradigm, in the sense I discussed last month. You see, when we approach an scientific or financial or artistic problem - such as how to write a piece of music - we bring to it certain preconceptions. This is not a bad thing. We use the paradigm to make sense of our world, and to communicate. We can't renegotiate every aspect of the meaning of our work with its intended audience every time. So we westerners know the musical scales, we know the language in which the song is sung, and so on. In Mozart's day the audience expected the third movement of the symphony to be a Minuet and Trio - and they knew what one sounded like. When we hear a blues work we expect the pattern to repeat every twelve bars. Likewise, the composer of a popular song nowadays does not generally want to challenge the expectations of the listeners, who have been hearing three-minute pieces all their lives. (I once took a course on the history of Jazz. The professor bemoaned the fact that when the LP was invented, the three minute restriction was lifted, and some artists had boring, flabby solos in their music that they wouldn't have had time for on a 78.) We still live in a world of limitations of the medium. In our company we have sitebuilders who are expert in coaxing interesting things out of HTML. Our designers know how to create "for the web" - things that work on 256 color screens in 640x480 pixels. And the files I'm talking about still have those annoying three-letter extensions. The mp3 cracks open the door to the No-Medium world. The information in an mp3 file is totally abstracted from the way it's stored and how it's reproduced. It can be on someone's web site or on a CD or in the Rio's memory. The songs use so small an amount of the system's physical resources that we just don't care. Our culture will be free to reinvent the musical paradigm, without the choices that were made for us by technical constraints. Nothing much may change. Even if it doesn't, the No-Medium won't stop there. The same things are happening in other spheres - text and video for example. As these get abstracted from limitations we don't even think about, who knows what people will do? Turn Marshall McLuhan on his head: The No-Medium is the message.