In 2015, Raspberry Pi became the bestselling British computer of all time.
Earlier this year, it passed the 12.5 million mark in sales, taking its place as the third highest selling general purpose computer ever built.
When the project got underway, though, its primary objective wasn’t to sell millions of units. The Raspberry Pi was conceived as an educational device. Its enormous popularity is proof of how well it executed upon that vision.
In just five year’s time, the hardware went from a promising idea to a globally recognized brand – and we’re only going to see the full effect of how it makes computing more accessible as the next generation of programmers mature and flourish.
To see how the project came to fruition, Digital Trends spoke to a few of the people who played key roles in its early history.
Eben Upton considers his trusty, second-hand BBC Micro one of the most important purchases he’s ever made. He used this system to learn the ropes of computing, programming in assembly language through his teenage years. When he finished his degree in physics and engineering at Cambridge University, it was this passion that prompted him to pursue further education in computer science.
Towards the end of his PhD course, Upton became increasingly interested in computer hardware. Looking back fondly on his own youth tinkering with his BBC Micro, he wondered what could be achieved if a similar computer was designed for the 21st century.
“There were kind of two strands for a long time; there was a strand of me trying to build cheap hardware, and then there was a strand of panicking a bit about admissions numbers at the university,” remembered Upton when he spoke to Digital Trends via Skype last month – his ‘panic’ was due to the fact that at that time, he was the director of studies for the computer science program at University of Cambridge’s St. John’s College, giving him chance to see how student admissions were failing to meet goals. “Raspberry Pi was those two strands coming together.”
It took several years for the device we know today to take shape. Starting in 2006, Upton worked on various ideas in his spare time, alongside his studies and later, his day job. He created several generations of “bodged together” hardware. They’re still in his possession, and fully functional, to this day.
One early iteration was based around an Atmel chip, boasted a whopping half a megabyte of RAM, and was only capable of driving a low-resolution display. “As far as it goes, it’s kind of fun. It doesn’t have a fully developed software stack on it, but you can imagine that it would be somewhat like a BBC Micro or an Amiga – the sort of software you could write for it would be the same kind of software that you would write on […] an Amiga.”
Upton managed to develop this version of the hardware to the point where it could display a proof-of-life animation on a monitor. Then, he was sidetracked by other responsibilities – primarily, a job he took with chipmaker Broadcom. His decision to join the company would end up playing a pivotal role in the success of Raspberry Pi, but first, it stopped development in its tracks.
“One of the first things I saw there were a series of chips that were quite cheap, and could do properly what I had been doing in an ad-hoc way; they could drive a display, they had SRAM, they had a processor,” explained Upton. “When you’ve been bashing together something by abusing the chips it was using – finding ways to do things that the chip designers had never imagined – when you’re confronted with the proper version, it’s sort of demoralizing, I guess. It makes you not want to keep hammering away, hacking at the stuff.”
Upton began to wonder if Broadcom’s chips could be used to realize his idea for an educational computer. And hardware wasn’t all the company would offer the project, as some of the workplace friendships that Eben struck up would form the backbone of the Raspberry Pi team.
“I started at Broadcom, I think the week before Eben started at Broadcom,” recalled Gordon Hollingworth, Raspberry Pi’s director of engineering, when he spoke to Digital Trends over the phone. “We were sat pretty close to each other from the beginning, and we got on really well.”
Hollingworth was a software engineering manager at Broadcom, while Upton developed architectural systems for the company’s chips. “It was always fun watching Eben grow and become a different person,” he said of their time at the company. A MBA he attained at Cambridge Judge Business School contributed to this growth.
It was during these studies that Upton decided to revisit the idea of an educational computer. He remembers it as a period of “quiet work” while juggling his responsibilities at Broadcom and his degree.
“He came up with a number of different, random things which either he’d built by hand, or by taking some of the hardware we had at Broadcom and altering it to try and make it into this theoretical system that could be used to teach kids programming,” said Hollingworth.
One such project used one of Broadcom’s development boards as a foundation for a Python shell. Upton ported the MicroPython libraries onto the experimental platform, allowing the device to boot straight into an environment where the user could write code using the language.
Because it’s pretty much completely open-source, it enables anyone to do pretty much anything.
“The issue really is that with something like that, it becomes very closed – all the software is very closed, it’s very limited,” said Hollingworth. Raspberry Pi as we know it is a fully fledged computer, but sticking with MicroPython would have made it more of a microcontroller, along the lines of an Arduino. “One of the more important things about Raspberry Pi is because it’s pretty much completely open-source, it enables anyone to do pretty much anything.”
This experiment didn’t run Linux, which meant that everything from its keyboard and SD card drivers, to its text editor, had to be written from scratch. It was never going to be able to be produced in large quantities. Still, it did contribute to a couple of the decisions that made Raspberry Pi what it is today.
It confirmed that using Broadcom chips offered up some major advantages, not least because Upton and Hollingworth were very familiar with the hardware. It also highlighted the need for a chip with an ARM core, which would allow the device to run Linux. The ‘Pi’ in Raspberry Pi is a holdover from when the computer was based around a Python environment, but Upton looks back on introducing Linux support as the watershed moment for the project.
“The decision to build a Linux computer, rather than some ‘special’ thing was an early one, and very valuable,” said Upton. By 2011, the hardware that would underpin Raspberry Pi was taking shape, at least conceptually – but there were lingering questions about what the audience for such a device would be. It was intended as a platform to help young people take their first steps into the world of programming, but a viral video was about to grow its audience exponentially.
Because the BBC Micro was an inspiration for the Raspberry Pi project, Upton initially hoped it could be an official successor. He even booked a meeting with the BBC’s technology correspondent, Rory Cellan-Jones, about a potential partnership.
The BBC ultimately decided not to put its name on the product. Instead, Cellan-Jones published a video of his meeting with David Braben — Raspberry Pi Foundation co-founder and legendary British game developer –to YouTube. The response was enormous.
“That was the point when it crossed over from something that we were working on among ourselves – maybe a few people in Cambridge knew, but it wasn’t widely known – to being kind of a mass phenomenon,” said Upton. “It became that mass phenomenon over the course of two days in 2011”
Liz Upton, the director of communications for Raspberry Pi — and Eben’s wife — described her reaction to this response as one of “absolute shock.” Then working as a journalist, she was brought on board as a volunteer, helping to manage the community that sprung up around the project.
“It was a kick up the arse,” said Eben. “It was an ‘oh shit’ moment. You’ve been thinking about this as an educational thing, and then all of a sudden, 600,000 people have looked at Rory’s blog in two days.”
We suddenly realized that we had promised tens and tens of thousands of people that we were going to make them a computer — for $25.
“Eben and I were watching as tens and tens of thousands of people watched the video, sort of congratulating ourselves,” said Liz. “Then we suddenly realized that we had promised tens and tens of thousands of people that we were going to make them a computer for $25.”
It would be no small feat. Even with an army of potential customers – the beta version of the Raspberry Pi operating system was downloaded 100,000 times before any hardware was made available – manufacturing the first devices wouldn’t be cheap.
“Eben and the other trustees of the Raspberry Pi Foundation put together some money to create the first 10,000 units,” remembered Hollingworth. Liz Upton recalls some long nights sat with her husband at their kitchen table, stuffing stickers into envelopes that would be sent out to supporters at £1 a pop. An initial run of 10,000 seemed more than sufficient. Upton even had half-joking concerns that he would end up with 9,000 unsold Raspberry Pis lingering in his shed.
Instead, a total of 100,000 pre-orders were submitted. It became clear that satisfying demand beyond the initial run would be a problem, so Upton devised a plan that would take the financial burden off the Raspberry Pi Foundation. “The decision to be a licensing company, to license our technology to RS and Farnell, was very important because it reduced those capital constraints for us,” explained Upton.
“Eben got in touch with the guys at RS and Farnell,” said Hollingworth. “These are big distributors and they have loads and loads of money, because that’s all they do: they buy products and they sell them on. The one thing they deal in quite heavily is basically capitalizing these products.”
Upton proposed that the Raspberry Pi Foundation would show RS and Farnell how to manufacture the Raspberry Pi, and set up production at the factory. The Raspberry Pi Foundation would then receive a license fee for every unit sold. Effectively, the manufacturers would be paying the set-up costs, but the Foundation could expect income that grew in line with the success of the product.
“The great thing about this is, we don’t need to raise money, we don’t need to raise capital,” said Hollingworth. “And as we are, at the time, just a charity, we can’t really raise capital. That was one of the big changes that made Raspberry Pi a real thing.”
With this manufacturing model in place, the Raspberry Pi Foundation was well-placed to serve the masses of individuals who had already expressed an interest in the project. “It felt like before Raspberry Pi, there were people out there that wanted Raspberry Pi,” said Upton. “Sometimes you launch a product and you don’t have to educate people about why they want it. You just have to tell them it exists.”
The Raspberry Pi Foundation could have thrived for several years selling the first iteration of the hardware, as single-board computers were still something of a novelty. Yet the continued success of the computer hinged on what would come next.
After a successful launch, Upton and the rest of the Raspberry Pi Foundation set about adjusting everything from the way the hardware was manufactured, to the design of the device itself.
While the licensing agreement that had been established was a major boon for the project, there were soon plans to improve the manufacturing situation further. The first wave of hardware was made in China, a decision that was made purely to attain the all-important $35 price point. However, the Foundation aspired to bring production closer to home.
A manufacturing plant in Pencoed, Wales sat almost unused. It had been constructed using funds from Sony and the Welsh government, and briefly served as a hub for the creation of LG television sets. By 2012, it only served a small number of contracts on top of its primary, niche function of building high-end cameras used by television broadcasters.
“Eben talked to those guys and said, ‘well, wouldn’t it be awesome if we made Raspberry Pis in the U.K.?’” said Hollingworth. “Because that would be the dream, right? To make them here rather than make them in China – not because it’s a terrible thing, making them in China, but because we love the idea of making things in the U.K. There was a time when engineering in the U.K. was something to be proud of, and we got left behind somewhat, in that respect.”
The Foundation also brought on James Adams, a colleague from Broadcom, to redesign the original model of the Raspberry Pi. The follow-up would become known as the Model B+. “There were some obvious issues with the Pi 1,” said Adams when he spoke to Digital Trends. “Nothing that was a particularly big problem, but it was obvious that there were things that needed fixing.”
Adams executed what amounted to a complete redesign, in his words a ‘reimagining.’ The form factor had to be similar, and certain design features like extra USB ports were mandated by the team, but he was given leeway to make decisions like adjusting the position of mounting holes, and reworking where the connectors are placed.
“The big challenge there is that I had to be true to the original design,” Adams reflected. “But also, this one, we were going to make hundreds of thousands before we’d ever launched it. One bug that wasn’t found could have cost the company. If they were returned, that wasn’t something we could ever support. It really was, ‘you have to design this thing right, first time.’”
The Sweeter the Berry
Redesigning the original Raspberry Pi board is one thing. Pushing the hardware forward with every new iteration is another. The device has such a broad range of applications that a new version that’s dramatically different could throw a spanner in the works for many owners.
“We’ve got that form factor now, and it works quite well, we don’t really want to change it – we’re more about evolution rather than revolution,” said Adams. Each generation is similar to the last, so you’re not going to have a problem getting hold of something that will fit in the Raspberry Pi-shaped hole in your product.”
It’s encouraging […] that we can rebuild the computer industry in a way that’s more representative of society.
The team told me about various implementations of the hardware over the course of our conversations, ranging from custom tech for factories, to cattle-counting devices. If the product’s design suddenly changed dramatically, many owners would be left in the lurch.
That isn’t to say that the Raspberry Pi has grown stale, or that it will never change. Many owners are pushing for new features and functionality, and have been since the very beginning.
“There’s a very simple calculation,” explained Upton. He gave the example of an analogue-to-digital converter, which is frequently raised as the biggest missing feature of the hardware. Arduino, the Pi’s biggest rival, does feature the component. It costs less than 20 cents, but the Raspberry Pi Foundation is reluctant to include it.
“It’s a $35 product that doesn’t have this 10-20 cent component, why not?” said Upton. “The answer is, we don’t believe that many of our customers want them. We have this rule of thumb: suppose you have a 20-cent device that you can add to the build materials, and 10 percent of your customers will use it. You think of it as being a $2 device.”
The Raspberry Pi is manufactured under very small margins, so even a seemingly trivial addition to the build needs to be scrutinized. “We have that golden figure of $35,” said Hollingworth. “The question really is, what can you fit into $35, and what’s useful? People have said, can’t you have SATA? And can’t you have gigabit Ethernet? Why can’t I have analog inputs? The answer is always, always, always, because you can’t afford it.”
That said, the Raspberry Pi Foundation is always using economies of scale to push down pricing in various areas as its products have become more popular. The result is extra funds that can be used to make each new version compelling.
“There are usually obvious, standout features that people want,” shared Adams. “It’s always nice to have more memory, it’s always nice to have a faster processor, it’s always nice to have better Wi-Fi. A lot of these things are obvious, you know what you’re going to add, you’re just waiting for the tech to enable it at the right price.”
His example was the addition of Wi-Fi between Pi 2 and Pi 3, something that added a huge amount of utility to the device. In a perfect world it would have been there from the outset, but the price of the technology had to be right before it was introduced.
“B+ to Pi 2 was, I think, nine months, and then there was a year until Pi 3, and now we’re having a bigger gap,” said Adams. “It’s largely driven by the technology, what we can do, and what we can get in at the price point. Which is kind of good, right? We don’t want to give out the expectation that we’re gonna have a Pi every year, that forces you into a cycle that you shouldn’t be in, in this kind of world.”
“This is something that Raspberry Pi are quite proud of doing,” Hollingworth agreed. “We’re not like Apple, where your new phone is going to be like three hundred quid more expensive than your old phone. We’re trying to hold our price as much as we possibly can. We think that’s a cool thing.”
Pricing has been an important consideration for the Raspberry Pi project since the very beginning. Eben paid £220 for his BBC Micro, which he described as being “very second-hand.” Adjusted for inflation, that’s around £869 ($1,147) — a ton of money for a young person to invest in a computer, and far out of the reach of most families.
“I was lucky that I came from a family where I had a savings account with a few hundred quid in it,” said Upton. “The barriers were substantial. That probably means that there were people of my generation who would have been amazing computer programmers, who never had the chance to discover that.”
Upton likens it to entire classrooms being given a recorder to see if they have any affinity for playing music. Some might go on to be a concert pianist, or a virtuoso guitar player. The majority might never play a clean note, and never pick up an instrument again. Yet the availability of a cheap and accessible option can find talent that’d otherwise remain undiscovered.
“It’s no coincidence that our cheapest machine is $5, right?” Upton continued. “It’s really important. It would have been easy for us to get cocky or complacent, once we’ve done a $35 computer. But you’ve got to remember that a lot of people don’t have $35 of discretionary money – but most people in the developed world do have $5 of discretionary money.”
The low price point doesn’t just lower the cost of entry; it lowers the cost of screwing up. People that might want to tinker with hardware, rather than concentrate on coding, can experiment without the fear of destroying expensive equipment. It’s difficult to fry a Raspberry Pi, but it’s not impossible. If it happens, the penalty is low, especially if the system in question is a $5 Zero model.
Next Year’s Crop
“It’s not realistic to say, ‘everyone should be a computer programmer,’” said Upton. “I think everyone should have a chance to find out whether they want to be a computer programmer, and that isn’t where we’ve been in the past.”
Even back in the ‘glory days’ of the 1980s computing boom, Eben remembers it being a primarily male pursuit. It wasn’t a matter of young women not being interested, or not having the necessary skills. The activity was simply pigeonholed as soon as it became mainstream. Liz experienced this firsthand.
“I went to an all-girls boarding school, and there really wasn’t much of a science education there at all,” she said. “They tended to diagnose you as being artistic rather than engineering-oriented.”
Liz was fascinated by computers, but she wasn’t given the opportunities to pursue this interest. Her computing lessons were all about typing, so she learned about coding from computer magazines at home. When she was at university, studying law, she spent free time indulging her enthusiasm for engineering – but she had to chase down this experience, rather than being exposed to it by her education.
“There are an awful lot of things that can exclude people from this stuff,” she said. For as long as computing has been a viable career, it’s been easier for men from a privileged background to get involved than anyone else. As of 2017, everyone uses the software and services that programmers build – so for the most effective product, it would follow that the industry itself should be as diverse as its user base.
Liz and Eben had a daughter earlier this year – Liz was still on maternity leave when we spoke over the phone. “We just had our first kid, so this has got more important, rather than less important, over time,” Eben told me.
Eben made it clear that becoming a father has reinforced his belief in the importance of giving young people access to computers. Today, 40 percent of the attendees of the Code Club events supported by the Raspberry Pi Foundation are female. That’s good news for everyone involved with the organization.
“I think there are some really encouraging signs – and as the father of a daughter, it’s become particularly, personally important to me over the last few months – that we can rebuild the computer industry in a way that’s much more representative of society,” said Upton. “That’s not just gender. That’s race, and class.”
Eben conceived the Raspberry Pi as a means of stimulating admissions to a computer science course at one university. Within five years, it’s given hundreds of thousands of people around the world a way to experiment with computers and programming. The parameters of the project have grown with its audience, and there’s no sign that will end.
“When I started this, if you told me that we would get back to where we were in the 1980s, I would have been over the moon,” Upton reflected. “I think in a lot of ways we’re past where we were in the 1980s, in terms of levels of participation. Now it becomes about, what can we do internationally? Can we do this in developing countries? Can we make sure that we get enough girls? Can we make sure that we go past the 1980s? It evolves. We get more ambitious as time goes on.”
Published at Tue, 07 Nov 2017 11:15:46 +0000