“I think there is a world market for maybe five computers.”
– Thomas Watson, chairman of IBM, 1943
The currently accepted version of Moore’s Law states that the number of transistors per square inch on an integrated circuit doubles every 18 months. Because this happens, computing power also doubles every 18 months, and that is why we are now able to own a smartphone or watch with more capability than almost any mainframe of 40 years ago.
The first smartphone, Simon, went on sale August 16, 1994, at the bargain price of $1100—$900 with a BellSouth contract. It had a touchscreen and 10 apps (called features) and weighed less than a pound!
As technology has evolved, so has its pricing. Does anybody remember the first flat-screen monitors? I remember walking into Best Buy and seeing the future, which was 17 inches measured diagonally and cost well over $1000. You can still buy a refurbished 17″ for $29.60. A new 24″ can be had for under $120.
Software and networking are similarly affected by Moore’s Law. A 10 megabyte download used to take most of a day. Now, for typical internet connections, download time is measured in seconds. Personal computers used to bog down if three applications were open at the same time; now we often have three dozen, all more resource-intensive. My first company’s mainframe had 32 kilobytes of RAM and 1.2 million instructions per second. My current work PC has 16 gigabytes (half a million times the size) and a 4-core 2.8 gigabyte CPU (more than 9,000 times the cycles) of that mainframe, which serviced an entire company.
What does this have to do with the language industry?
In the 1950s, machine translation (MT) developers claimed: “In five years, we’ll have done it!” Unfortunately, the bar for “it” kept moving. However, today’s language industry is finally approaching that goal—and the primary reason is based upon the increased computing power predicted by Moore’s Law.
Twenty years ago, it took a supercomputer or, at least, a very powerful mainframe to generate a reasonably useful MT engine, and it probably took a full day’s run. Today, thanks to improved programming, clever and complex algorithms, and faster hardware, we can generate the same engine on a capable PC in a few hours. Twenty years ago, it was ridiculous to imagine video conferencing. Ten years ago, it required expensive special equipment installed at each site. Today, we can do not only video conferencing but three-way video interpreting on our laptops, phones, and tablets.
What’s next? Virtual reality to transfer and translate one environment to another in real time? Implanted chips that will furnish content in any language desired? Already there are heads-up displays so that we can see the original text as well as the translation or interpretation? It’s all possible, due to the effects of Moore’s Law.
Bob May, Guest Blogger, CTO, Tembua:the Precision Language Solution