'

Ndiyo: Sharing PCs to bridge the digital divide

Rather than aiming for one laptop per child, start-up Ndiyo thinks the developing world would benefit from sharing PCs more effectively

MIT Media Lab guru Nicholas Negroponte has been grabbing the headlines recently with his One Laptop Per Child project. The aim is to create a cheap — about $100 — and robust laptop for use in the developing world. Negroponte is adamant that ownership of the device is key to helping children engage with technology.

Past attempts to give children in developing countries access to PCs have failed as children did not see the computers as their own, and as a result did not engage with them as expected. "People say we just gave 100,000 PCs to schools, and they are still sitting in their boxes. The problem is that you gave them to the wrong people — the kids don't think they are theirs, and see them as government property, or they are locked up after school," Negroponte told the Red Hat user conference earlier this year.

The scheme has not been without its detractors, most noticeably the Indian Government, which has decided not to pre-order any of the One Laptop Per Child (OLPC) machines claiming the project is too high risk and the money would be better spent on other learning tools. But despite concerns about the strategy, and whether it will come in on price — the laptop is hovering around the $120 mark at present — the overall aim of bridging the so-called digital divide is sound.

However, the OLPC project is not the only organisation to have focused on the lack of computing devices and infrastructure in the developing world. AMD announced last week that it had decided to pull the plug on its Personal Internet Communicator (PIC). The PIC was introduced in 2004 as part of AMD's 50x15 project, through which the company has pledged to help bring internet access to half the world's population by 2015. But sales of the product never made an impact on AMD's bottom line, and the chipmaker has stopped manufacturing it. Microsoft has also proffered its own solution to bringing cheap computing to the masses via mobile phones, while Intel has its own Community PC initiative.

But not all attempts to bring cheap tech to the developing world are based around new, proprietary technology. Charities such as the UK's Computer Aid currently take old PCs from enterprises, refurbish them and donate them to NGOs (non-governmental organisations) in Africa and South America. The organisation's chief executive Tony Roberts claims AMD and the OLPC project are risky strategies, as they are based around new bespoke technology. "They are looking to introduce a non-standard, untested platform... which they will only sell to governments," he said in a recent interview. "The decision to buy will be made by politicians who are elected every five years, and politicians generally don't take the decision to risk their political future on non-standard technology."

An alternative to both the refurbished PCs and the OLPC approach has been developed by two UK academics. Ndiyo, the Swahili word for "yes", is a project that aims to allow multiple users access to the same PC. Rather than trying to push more bespoke devices on countries with meagre IT budgets, Ndiyo allows one PC to be shared by five to 10 individuals by turning it into a mini-server networked to a series of thin clients.

The brain-child of Quentin Stafford-Fraser, a former research scientist at AT&T Laboratories Cambridge, Ndiyo is based around the untapped ability of the Linux operating system (Ubuntu) to support numerous simultaneous users. Together with his partner, technical author and Open University professor John Naughton, Stafford-Fraser decided that the traditional idea of one machine per user was a model that just didn't make economic or functional sense for the developing world. Instead, in the Ndiyo model, a Linux PC becomes a server to a series of "ultra-thin-clients" — called Nivos — which allow an extra display, keyboard and mouse to be connected to the computer via a standard network cable.

ZDNet UK caught up with Stafford-Fraser and Naughton recently to find out how their technology works and why it makes more sense than the strategies being developed by heavyweights such as Intel and OLPC.

Ideally, Ndiyo users would be able to simply install your version of Ubuntu onto their existing hardware and then attach the Nivo thin clients to turn their old box into a thin-client server — is that the case?
SF: With recycled Pentium IIs you might find there are performance issues but our aim is to make minimal changes to Linux. We could have had a substantial distribution of our own. People who understand Linux should understand this. Eventually the machine should look like it did when you started but you can add a number of users to it.

JN: It also depends on what level of user you are. There is a case for us thinking about semi-packaged solutions. One of the big issues in the developing world is uneven distribution of Linux expertise. Our Ubuntu distribution will have all the drivers you need. One of the big advantages of our approach is that,  if you have a set of CRTs, Nivo will work perfectly well with those and any keyboard and mouse.

One of the projects that has interested us a lot is the Skolelinux project in Norway — where they have a very large number of schools running Linux networks. One of their goals was that a non-technical teacher could get 20 workstations up and running in 40 minutes from a standing start, and there is part of me that would like us to do that with Ndiyo. They are doing for it real and it was pretty lavishly funded by the Norwegian Government.

But getting machines up and running is one thing — what about supporting them?
SF: One of the earlier concepts we had at Ndiyo is that they will work in situations where you have limited support on the ground. The terminals are appliances that take pixels that are sent to them and put them on the screen. What that means is that if you have a 10-user system, you are only administering one PC. By making terminals that are...

...like light bulbs — you can unplug and replace with another if you want to — then we reduce the hardware admin to a minimum, and when you have that and Linux you can do the software admin more easily. We may not have the expertise on the ground but it may be easier to administrate an Ndiyo network remotely. Essentially you're connecting to one PC and managing that one machine.

So where have you got to in terms of demonstrating that your technology can work in the field?
SF: We have had one trial running for two and half years. There is a small NGO in Cambridge called Aidworld that had the first Ndiyo systems. They are non-profit developers and have been thrashing it for a couple of years — doing Java development – not doing heavily intensive graphic stuff but making computers work hard. One of the guys went to Ethiopia and he visited various higher education institutions and came back and said to me, "Everywhere I went I wished I could give them an Ndiyo system". In countries where the price of single PC is high, the number of people who have seen PCs connected together is minimal, so getting network-management expertise is very hard. And if you find someone who can do it, they will get swiped when Cisco opens an office there. The nice thing about Ndiyo is that you don't need to understand networks and net management to use it.

Your system is based on Ubuntu Linux rather than Microsoft Windows. Have you encountered any resistance to the fact that users won't be able to run Microsoft Office and other applications, which are perceived as being tools of economic progress?
SF: That can be an issue. I don't think that is an issue for governments but more for individuals. A friend from India was telling me that call-centre people make most of their money in the evenings selling lessons in Word and PowerPoint to children in the village. There is a slight challenge we have had with some Linux distributions — Ubuntu in particular — in that it doesn't look like Windows; the fact there is no Start menu in the bottom left-hand corner, for example But it's easy to reconfigure so that it does look like that –— we don't want immediate barriers to people coming in and learning.

JN: What we decided was we are in the business of selling a meme [a unit of cultural information transferable from one mind to another] which is: networked computing doesn't have to be done the way we do it at the moment. One of the ways we could do that is that we could do a classroom in a box — could get a five-screen classroom in there. One of the most innovative uses of Linux in schools is Orwell high school in Suffolk. They have 400 thin clients running — all of them on open source, including Open Office instead of MS office. The thing that surprised me was that they had relatively little difficulty with that. Whereas most people predicted that trying to do non-MS stuff in a high school would end up with the kids moaning. But the school claims that most educational material is web-specific — so all need is a good browser and good web connectivity.

The cost and availability of the technology is important but so is the infrastructure to support — especially power infrastructure. Presumably your thin client approach is a more efficient way for multiple users to access applications?
SF: You can pull out the power plug from the Nivo terminal and it keeps running. You only need constant power to the server because all you have locally on the terminal is pixels. What that means is that if you are somewhere with an unreliable power supply you can use battery back-up or a generator to run your sever. You can use what supply there is to run the terminals — but if that proves to be flaky, you can reboot them much more freely than you can a normal PC — so it is much more robust.

It is also much lower power — around three watts on current machine and five watts if you have a mouse and keyboard plugged in — compared to 60 watts for a typical PC. That can be significant. In some parts of developing world electricity is much cheaper than here, and in some parts more expensive. If you go on UK figures then you can save a fair amount in hardware costs by buying Ndiyo type kit, but in a year or two you can save the same amount in electricity costs. That is what people don't think about — if you're running a 20-seat internet café, you can spend a thousand or a couple of thousand dollars on electricity.

Nicholas Negroponte has said that manufacturing of the OLPC laptop will begin when they have a key number of governments signed up and committed to take orders of the device. Are you working to the same model or do you have a more ground-up approach?
SF: Scale means you can do so much more, so much more quickly. Our stuff can be affordable in small numbers. The design from the word go is that this stuff should be very cheap so that it is nearly insignificant compared to the costs of the keyboard and mouse. At the moment we just have trial units, and we don't...

...have any of those because we are about to make a new batch shortly of a new design, but we hope to have more of those at the end of the year and then next year to start producing them in large numbers. As with everything, the cost is highly dependant on the numbers — if we made these in the millions they would cost next to nothing, but in the hundreds they cost more. But we can do them at a reasonable price now.

What is your model for financing this project — I see from your website that you have a connection to a for-profit company called DisplayLink (formerly Newnham research)?
SF: The underlying tech for Ndiyo is from DisplayLink — this company that we spun off three years ago. It is developing chip sets and is going into full-scale chip production. As the scale of that builds up then we can offer this at a lower costs. They are mostly using the tech for other things — not for thin clients, digital signage — adding multiple screens to one PC without using extra graphic cards.

We started DisplayLink because we wanted a philanthropic humanitarian effort which we thought could make a huge difference to millions of people but it's difficult to fund that on a conventional basis. But we realised there are applications of the same technology that would be more applicable to standard venture funding. What we hope is that the tech DisplayLink develops can come back and help Ndiyo. You can already buy some of the technology — Kensington makes a USB docking station for your laptop; you can plug in a keyboard, screen, mouse and VGA-out which uses Nivo technology inside.

So do you still have a direct relationship with DisplayLink?
SF: There is no direct relationship. I am no longer on the board but I have a very amicable unofficial one — we are helping to port some of their code to Linux.

So how many people are working on Ndiyo at the moment?
SF: It depends when you ask. It's me full time, John part time and some occasional help. Right now it is one-and-a-half people, but we are looking to expand that — when we move into manufacturing then we will have to boost that.

How close to manufacturing are you?
SF: The prototypes we have had so far have only been distributed on a small basis, as it's not suitable for manufacture on any kind of scale. But we have prototypes in the pipeline. So what I am hoping is that very soon we will have trial units we can get out to people who might be interested in larger quantities and we can get from that a feel of how many we should be manufacturing. So my hope is [to have] trial units by the end of this year and start manufacturing stuff early next year in larger quantities.

How big do you think the market is for the device?
SF: It's an interesting challenge — it is very hard to predict the size of any market that is specifically Linux based. We see evidence [of] huge numbers of Linux users out there and huge numbers who would use it if this kind of system was available. Of course they don't appear on any sales figures so it's hard to quantify the size of the opportunity here, so we need to do some exploratory research to work that out.

The OLPC project has been seized upon as an option for bringing the Linux desktop to schools in the developed world, too — do you have similar plans for Ndiyo?
SF: This is far from being something only for the developing world. We have always tried to say that we don't want to give the developing world something noticeably inferior to what we have — we want them to have good computing, [but] also a way to afford it. There are lots of good reasons to use it here, [for example] green motivation — it's much more environmentally friendly to manufacture and run.

And there are other reasons, such as flexibility. You can put Ndiyo in the kitchen — it is much better than a PC, it's quiet, doesn't generate lots of heat, and there are no moving parts. It is much more suited to these harsh environments — public kiosks and public settings for instance. In a way, it's more exciting to do in the developing world. Our hope is not that this will replace a windows PC on a network. This is a solution people who have nothing will be able to get. [It's] easy for us to say how does this compare with traditional PCs, when traditional PCs just aren't an option sometimes.

JN: We are very enthusiastic supporters for Microsoft's drive against piracy. The more they get a grip on pirating of Windows and Office then there are corresponding surges in other ways of doing it. We can see that in demand for training in Linux in the Philippines, which is steadily growing.