Could one of the inventors of the webcam transform the economics of computing for the poor? In 1991, when he was a doctoral student at Cambridge, Quentin StaffordFraser hooked up a camera to monitor the department’s coffeepot so his fellow workers would know when it was fresh.
This was years before websites could host images or browsers display them, so he and a fellow student had to write all the code required to transmit and display a succession of changing images. Two years later, when the web’s protocol had caught up with the idea of images, other students rewrote the project and the modern webcam was born.
You couldn’t say the webcam has transformed the world, but now Stafford-Fraser has another idea for pushing pictures over networks, which very well might do. Instead of making PCs cheaper, he wants to make them easier to share so that one computer can work for half a dozen people.
There are about one billion people online in the world today, but even the cheapest PC costs more than a year’s income for about half the world’s population. This is ludicrous — there is more than enough computing power around to give affordable portions to maybe another one billion.
Stafford-Fraser says: “I do some consulting for a friend with a small business and he buys a new computer for his office every 18 months or so. While Moore’s Law holds true, each one is twice as powerful as its predecessor; in fact, each new computer will be as powerful as all the ones the business has previously owned put together, so it could do all the work that all of them are doing.”
Calculations like that led him to found Ndiyo.org, a company named after the Swahili word for “yes”. The name itself illustrates the condition he is trying to cure.
“When we were looking for a name, we needed something that would make a domain name, and almost all the short domain names in English have been taken. But Swahili is an important trading language, spoken by 15million people, yet hardly any domains have been registered using Swahili names.”
All that was needed, he thought, was a simple way to hang additional screens and keyboards off the one computer and an operating system that would make this possible.
The operating system was not a problem. Even Windows could manage this, technically, though the licensing costs would destroy any savings. But any form of Unix (the company is using the Ubuntu distribution of Linux) is a multi-user system out of the box — reliable, capable and with no licensing costs at all.
The real problem was the hardware. Connecting extra screens would normally need a new, expensive video card for every screen as well as long, expensive video cables.
Lots of companies had tried, and largely failed, to make successful thin clients — computers without hard disks that would get their operating systems and their programs from a server. But even the smallest of these was bulky and expensive: they all looked like small computers.
Stafford-Fraser wanted to build a widget, something so small and elegant it could be made in millions as a single chip that would cost not much more than a video cable.
In this light, this was much like the problem with which he made his name. After he got his doctorate and had made the department coffeepot a star, Stafford-Fraser became one of the original developers at AT&T of VNC, a free and extremely useful protocol that lets you see and operate on the screen of a distant computer: descendants of this are now built into to Apple Macs and Windows PCs to allow remote troubleshooting.
VNC works — very well — but you do notice the difference between using a VNC connection and sitting in front of the original screen. For Ndiyo, the idea was that there should be no discernible difference. This has been achieved: there are demonstrations showing full screen DVDs playing down an Ndiyo link, which would be impossible with VNC.
The solution, the Nivo box, is about the size of a cigarette packet and most of this size is actually the five ports around the sides: an Ethernet port for the incoming signal, two PS/2 sockets for the keyboard and mice, a VGA socket for the cable to the monitor and one for the power.
In the centre of the box sits a custom chip. This was needed for performance, but custom chips are expensive to design and the need to fund their development drove the second innovation of the company.
Although Ndiyo is an NGO built around free software, the funding has to come from somewhere. Stafford-Fraser set up a company, now called Displaylink, to design and develop the custom chips it uses. This raised more than £20million in venture capital, because there are all sorts of commercial applications for a system that will ship pixels fast down an ordinary network cable.
As an example Stafford-Fraser points to the railway station in Cambridge, which for months had three display screens high on a wall in the main hall: one said “arrivals”, one “departures” and one “Press Control-Alt-Delete to log in”. There must have been a complete PC behind the display, waiting for a technician to climb up on a ladder, plug in the keyboard and press the right keys. With a Ndiyo system, there would be nothing but a screen up there and all of the display screens would be managed from one central server.
Even in the developed world, this saves a lot of electricity. Stafford-Fraser reckons that a Nivo box draws about 3W, a 20th or 30th of the power of a full PC. Nor will it ever need upgrading. All these facts provide commercial opportunities and he is eager to weave the profit-making world into the non-profit-making one.
Though Stafford-Fraser is in favour of the much more famous $100 One Laptop Per Child (OLPC) project, he points out that his company is really aiming for a different market. The $100 laptop is clearly aimed at children and will be of less use when they grow up. That is where Ndiyo systems would be useful.
Many of the uses he envisages for Ndiyo systems in the developing world would themselves be profitable: cybercafés, for example, where one cellphone connection could support six terminals.
Stafford-Fraser points out that flat screen displays are cheap, getting cheaper and pretty much ubiquitous. On the wall of the Ndiyo offices is one that has been framed in wood and is used simply for a display of webcams around the world. He sees cheap, centrally managed terminals like this ending up almost everywhere in modern houses and it is clearly more sensible to run them from distant servers than for each to have its own computer.
Just last week the company took delivery of the first flat screens from Samsung that have its Nivo chips built in: beside the normal video inputs there is an Ethernet socket allowing the screen to be plugged directly into a Ndiyo network and start working at once.
At the moment, the project is being run by Stafford-Fraser, his friend John Naughton — a professor at the United Kingdom’s Open University, who is also a journalist — and a programmer. A fresh company, Camvine, has been set up to fund and expand further development of the software that streams the signal down cables. It is still all shifting pixels, just as the coffeepot webcam did; but this time around it might just really matter. — Â