The Bandwidth Explosion
In the ’90s, CDs were expected to replace floppy disks. In some cases, such as the distribution of software, they did. In others, such as machine to machine copying, they did not. In the absence of good support for file systems such as UDF of CDs, it was very time-consuming to use them for copying small files. Their physical size did not help, either.
In spite of this, floppy disks are now pretty much dead. What killed them? The network.
The fastest floppy drives had a maximum speed of about 500Kb/s. Because a floppy disk could be in only one machine at once, this can be effectively halved because the writing and reading have to be done sequentially.
Home networks are several orders of magnitude faster than that, and even cheap Internet connections are a similar speed, if not greater. For small files, it became a lot easier to simply send them via email than to pass disks around.
In the mobile arena, this is starting to be repeated. Some years ago, I got my first phone capable of data. This was a second-hand Ericsson T68, which supported GPRS and Bluetooth, allowing me to connect to the Internet with my laptop while mobile.
GPRS was not a great system; it was a hack designed to be easy to deploy on existing GSM networks. It gave a similar amount of bandwidth in real-world use as a MODEM and latency that often topped a complete second, compared with tens or at most hundreds of milliseconds for a wired connection.
Later, GPRS gave way to UMTS. This brought latency down to usable levels, and increased the speed so that 50KB/s downloads were fairly common. While not as fast as a wired network, this makes sending files to or from a mobile user as fast as reading or writing a floppy disk.
Current plans for 4G infrastructure expect to deliver 100Mb/s of bandwidth to mobile users. Assuming that latency is kept in check, this makes it feasible to use a remote file server with a mobile device.
A year or so ago, the first rumors of the iPhone started to surface. At first, I was surprised. The U.S. mobile market is a great place to be for network operators, but far from ideal for device manufacturers.
In the absence of any real information, and assuming that Apple wouldn’t want to enter such a horrible market, I started speculating about what kind of device it could be.
The deployment of municipal WiFi networks in a number of U.S. cities seemed to present an interesting possibility. What, I reasoned, if the iPhone was not a phone at all?
Apple already has a voice and video conferencing application bundled with its desktop machines. What if the iPhone were a WiFi device running iChat? Users near an access point could call other iPhone or iChat users for free, and Apple could make more money by operating a gateway for connecting to normal telephones.
The limited availability of WiFi coverage might be a problem. I imagined this would be tackled in two ways. The first would be by using mesh networking between iPhones to extend the range of existing access points. The second would be to bundle the device with a wireless access point which would allow other iPhone users to piggyback onto existing wired networks when they were near the home of an iPhone owner.
One thing I failed to take into account was the awesome power of Steve Jobs’ patented Reality Distortion Field, which allowed Apple to persuade AT&T to pay for the privilege of allowing its customers to buy the iPhone.
This, combined with the lack of maturity of the WiFi infrastructure, made the final iPhone a much less interesting device.