Chris McKillop, the former head of engineering for Google’s Fuchsia project, has provided us with some information on the project’s origins and future plans for the operating system.
Chris McKillop, the director of engineering for Google’s Fuchsia project, left the company earlier this year, as we previously reported. He spent the first few years of his ten years at Google as a member of the larger Android team before contributing for seven years to the development of Fuchsia from its inception to its initial deployment and beyond.
I recently had the pleasure of speaking with McKillop about those early stages of the development of Fuchsias and the direction he sees the project taking. We made a few illuminating stops along the road to discuss the Nexus Q’s shortcomings, Flutter’s accomplishments, and Meta’s abandoned Fuchsia project.
You were a part of the Android TV and Nexus Q teams when you first started working at Google in 2012. When did your move to Fuchsia start, and how did it feel?
So, when I first started working for Google, I joined a project called Android@Home that was developing the Nexus Q. I sort of walked right into the action. Absolutely not from the beginning. I participated in both the shipping and unshipping processes.
You are undoubtedly aware of its legacy. The lesson here is to be careful about the cords you store in the box. For a product that was mostly about speakers, we included an HDMI cable instead of speaker wires, and we were demolished as a result.
Then, in 2014, Google acquired Nest, and as a result, most of what we were doing at Android@Home was rendered obsolete. Although it and Nest overlapped, Nest was unmistakably where “Google” was headed.
So I took what we were working on at Android@Home, combined it with the Live TV work that was coming out of the original Google TV from 2010, and created Android TV from that. That was created from zero to one.
The Google TV Nexus Q (in 2011) Then, in this group that Larry Page, the co-founder and CEO of Google at the time, and Sundar Pichai, the product director at the time, had created, an opportunity presented itself to essentially look at the future of everything that Google was doing so, unrestricted, go look at this stuff.
It may have been referred to as a senior engineering talent retention program by some people in the outside world. I have heard those words. That’s kind of how it was. However, I viewed it as a chance to enter and improve on a few things.
I started working on the Fuchsia pitch around six months after I joined Google in January 2012. The name was chosen based on that. The original version of the deck was presumably constructed in the fall after we de-launched Nexus Q.
Operating systems have a number of drawbacks, including the fact that it takes a very long time to develop them and an even longer period for them to succeed. You must begin tackling those issues simultaneously. There was no one working on that when I arrived at Google.
Given that the windows were closing, I found that to be incredibly shocking. You would be in a similar position to Apple in the 1990s prior to their acquisition of NeXT if you didn’t start working on it. Or you can be in a position comparable to those operating system manufacturers who were supplanted and are now either nonexistent or no longer provide operating systems.
We all started developing ideas. For Fuchsia, I actually created a pitch deck, but I first received no financing. However, I didn’t stop thinking about it or working on it, and then this chance to contribute to this next-generation computing field came up.
I made an effort to enthuse a few folks about “Fuchsia,” but I was unsuccessful. I believe this is primarily because I had only recently joined Google.
Fuchsia was never intended to be about creating a new kernel at that time. It was really a comment I made about how each team at Google had their own Linux kernel team, including the Chrome OS team, the Android team, and the Chrome OS team. There was also a Linux desktop version at Google called Goobuntu , and later gLinux , as well as a Linux kernel team in the data centers. They were all independent, which seems absurd and ineffective.
Because the architectures varied, outside developers were unable to truly work on and target all the platforms that Google was providing. You had to complete custom work.
It was absurd, I know, that the “companies Google was cooperating with” had different teams for their engagements with Chrome OS and Android. Fixing that in the long run was actually part of what we were discussing in the original pitch.
A Core OS team exists at Apple. Windows and the NT teams are both handled by Microsoft teams. I essentially wanted to follow a pattern from others who have been creating “operating systems” for a while, and I also wanted to incorporate some aspects of Apple’s approach that I hadn’t seen elsewhere. Prior to going to Apple, I had not seen it. I didn’t see it again. I had failed to reproduce it.
How do we acquire qualities like full stack ownership and genuine concern for the performance of each layer of the stack so that we can give users the best experiences, for example?
Speaking of Apple’s influence, you’ve frequently claimed that Pink and Purple initiatives within the company served as Fuchsia’s primary inspirations and gave the project its name. Taligent, which Pink split off from Apple to become, was ultimately canceled. Purple, on the other hand, was the initial iPhone OS, which revolutionized computing as we knew it. Could you explain how each of them impacted the design of Fuchsias?
Actually, my interest in operating systems began because of Pink . When Pink first became popular, I was a high school student who sort of followed the music industry like a fanboy. At the incorrect layer of the stack, they were working on some pretty fascinating projects involving developers, developer techniques, and object-oriented design. I wanted to consider it from top to bottom of the stack and draw inspiration from that.
At Apple, I then worked on Purple. No one outside of Apple has been able to duplicate a lot of the very outstanding work done in the areas of touch, touch latency, and graphics speed. Although I don’t believe it’s wonderful for the world that only Apple has that, it’s excellent for Apple that they have it.
My intention was that if you put some effort into it and were prepared to move up and down the stack the same way they did, you might reproduce what’s there.
The Linux system call interface is one of the issues that many people run across when they create stuff on top of Linux. You create things on top of that material that is underneath you without giving it a second thought. When utilizing the Linux kernel, most users don’t completely understand that stack.
Simply put, it’s really challenging to do because of the community’s culture and the size and complexity of that codebase. Part of what I wanted to do was that. What if we simplified it and only used the parts of Linux that we actually needed?
You may not be aware of this, but Apple did not create all of Darwin , which serves as the foundation for other operating systems, including iOS and macOS. In order to have such control, they took FreeBSD and placed it on top of Mach , creating an alternate universe of drivers. Could we then carry out a similar action? The queries I was posing were those.
The original logo for Fuchsia (left) and its revision for 2021 (right) Additionally, while creating something fresh like this, you must choose your area of focus. That also contributes to the Purple’s “influence”: being consumer-oriented not concentrating on data centers Avoid concentrating on strict embedded systems when working on real-time robot projects. nonetheless, concentrated on consumer goods and the issues that consumers and developers for consumers encounter.
Thus, that is where the name originated.
You claimed that it was difficult to generate buzz about Fuchsia on Google. What ultimately caused them to change their minds, and how did things finally get going?
This other project was looking at new ways of developing applications, new ways of developing UIs, and new ways of doing things when I moved over to it and left the Android team. They weren’t necessarily considering novel approaches to operating system development. That was something I brought up, and I also took over those other tasks.
Flutter , a new method of creating user interfaces, is one of the things that emerged from that. When I started, there were only Adam Barth, Eric Seidel, and Ian Hickson. I assisted them in building their team before pushing Eric “to expand and do it alone.” There, excellent work is “happening.”
Editor’s note: The Flutter team revealed that Eric Seidel is leaving Google to embark on a new adventure just before this article was published.
Then we launched Fuchsia, initially utilizing Flutter. Because we were developing this new operating system and you wanted to discuss these novel concepts, Flutter supports Fuchsia. You’ll never get to the point of being able to demonstrate anything if your first action is to declare, “We have to get all of Android running.”
It took a pretty long time to say, “I want to have all of Chrome running,” which was the second thing we did after Flutter. In Chromium, the commit history shows how long something took. These require enormous codebases and applications to execute.
Editor’s note: In early 2017 , work on getting Chrome to run on Fuchsia started in earnest, with an early prototype being made available in late 2018. The whole browser didn’t start there until March 2022.
Thankfully, Chrome at least operates on Windows, Mac, Android, and a variety of other platforms, and Windows has many Fuchsia-inspired themes. They have the proper cuts in the proper places so that we can enter and carry out our tasks. On Android, that’s less of a problem.
2018’s chromium-fuchsia
In 2022, Chrome on Fuchsia We needed to be able to swiftly demonstrate something. We want the opportunity to put these theories about how things might develop in the future to the test.
We actually came up with some really strong examples of how we could make a difference in areas that are important to people. That is when the project’s excitement really began to take off.
At that point, we began concentrating on where we could ship first. Because a lot of individuals start operating systems, the fact is. Operating systems are launched continuously. Numerous GitHub projects, IRC chat channels, and college students can all be found everywhere.
Actually, starting an operating system is not that difficult. On the other hand, only a few occasions have it been shipped to millions of individuals. I think it’s crucial that projects get done and are more than just academic research endeavors.
Finding that vessel for us to board was a positive thing. It doesn’t matter if a “device like the Nest Hub” is only using a small portion of Fuchsia’s capacity. We discovered so much. When reality sets in, you realize where you were residing in an idyllic environment.
Operating systems are launched frequently, as you mentioned. You could say that one of Fuchsia’s primary distinctions is that an All-Star squad of programmers and thinkers from WebOS, BeOS, iPhone OS, and the list goes on worked on it. What was it like to collaborate with these All Stars?
We actually had a PowerPoint with all the 1.0s that the original Fuchsia team had “previously” shipped at one point, but I don’t have a copy of it anymore because it’s a Google thing. We could hardly fit everything on a slide.
Wow.
It was incredible. We were aiming to convey the wonderful point that this team is capable of seeing an idea through to its initial launch. A truly remarkable collection of people.
Naturally, I had previously worked with some of them. With the operating systems’ “community,” you won’t have to go through numerous NeXT 0. Three degrees is all that is necessary to essentially cover the entire planet.
As an example, you can connect from me to Steve Sakoman, who was co-founder of NeXT 1 and my employer at Apple when I first started working there. I’ve now “connected” to everyone at Be thanks to that link. How long does it take for this to “be connected to everyone” now? It goes quickly, don’t you think? The community is quite close-knit.
At some point in the early 2000s, academia developed an obsession with microkernels, and everyone sort of plunged in. Then, as Android and the iPhone grew in popularity, operating system research started to slow down a bit. Therefore, the same individuals responsible for operating systems back then are also responsible for operating systems today. The gang doesn’t even expand significantly.
After a period of relative stagnation, I believe new, fascinating people are now entering the field. The process of “connecting” everyone doesn’t take long. So word spreads that a new operating system is being developed and a new event is taking place.
Because we were so dedicated to doing it in the open, many individuals who had previously worked on operating systems “got intrigued.” They were quite frustrated by the fact that their work was kept confidential within the organization where they were employed.
One member of the “Fuchsia” team has a history of leaving a company and then devoting time to putting his thoughts onto GitHub once he has done so. Because if he uploads the one thing that is in his thoughts to GitHub, it will belong to him forever.
That is why, for instance, LK exists. The most widely used operating system in the world that no one is aware of is undoubtedly LK. Every mobile phone that has been sold in the last ten years has it or one of its offspring.
And wasn’t LK a fundamental component of Fuchsia as well?
Yes, that’s how we launched the kernel. because it knew how to start things off and boot up on ARM and x86. Although it lacked a user space or an application model, it did have a good, simple codebase that you could sort of bolt into the center of before beginning to construct the shell around it.
That was a fantastic bootstrap, but later we went back in and removed it as well. In fact, Travis Geiselbrecht, who wrote LK, is a team member. There were numerous team members who contributed to LK in various capacities and at other “businesses.” LK was a really logical place for us to start as a result.
Additionally, we looked at incredibly tiny embedded devices in the early stages of that project. So when we concentrated on Fuchsia, we really stopped using LK for these other projects that we were working on.
LK is employed in numerous contexts. It is utilized by Qualcomm and is currently used in many embedded systems. So, yes, it’s not difficult to locate the locations where it’s being used.
Do you believe that a forthcoming generation of operating systems will usher in innovative new ways to use a computer? A few operating system projects you worked on include iPhone OS, webOS, QNX, and Fuchsia.
Maybe. Let’s see. How we get to the point where we aren’t always switching from static application to static application while we are doing things will be a big part of what I believe has to happen next, in my opinion.
They weren’t mistaken in their reasoning, if you look at the work that NeXT 2 and NeXT 3 did in the 1990s, right? It’s absurd that I have to completely rewrite this application just to add this one functionality.
They weren’t approaching it the appropriate way, in my opinion. They weren’t considering security, privacy, or the problems that might occur when combining live Internet code with other Internet code whose developers you don’t know and expecting those things to function together.
So I believe whomever can figure that out will bring about a significant change. You can see Apple is considering it if you look at how its widgets and surrounding applications interact together between the watch and the phone.
iOS 14’s Widgets Nearby Clipboard sharing in Android 13 The Android development team is clearly considering it. The Fuchsia crew was considering it at the time. People are considering this, based on current events throughout the world.
It will take some time, and it’s not yet apparent if there is a solution. It’s a challenging issue.
The basic security designs in silicon, in my opinion, are one of the things getting in the way. Take a look at the exploits that are being used, such as Spectre and Meltdown. Historically, it has been believed that this section is secure and that we don’t need to be concerned. Various hardware attributes can be used to guarantee the proper operation of these software entities and objects. That is no longer accurate. There will always be new iterations of that kind of exploit.
The next breakthrough, in my opinion, will be that “silicon” level, which will open up these higher tiers. As a result, it will simultaneously originate from the top and bottom, from the silicon design and app architecture. When those things properly lock into place, something will follow.
It will be something brand-new, and I have no idea how long it will take. These processes take time. The original RFC for the Internet was NeXT 4, I mentioned to someone the other day. When you start discussing that with other people, you begin to see the scope of these massive, standards-based initiatives, which are how technology succeeds.
0
Work on standards and interoperability if you want a world in which everyone can contribute, which is something I strongly believe in. Also challenging are standards and interoperability.
Additionally, “that ideal” has been the antithesis of technological advancement for the past 15 years. Right? Technology has always been about creating walled gardens of its respective worlds that they can then provide to customers.
You can’t just turn up and start adding features to Gmail. It just doesn’t work that way. I believe it is time for new schools of thought.
Where do you personally expect or envision Fuchsia being in ten years, having been there all along?
I believe there is a slim possibility that everything Fuchsia has accomplished will end up in the Linux kernel. They attempt! It’s excellent that they’re actually taking action right now. Because it actually is about the features we were aiming to add, it would be a fantastic result.
Although I don’t subscribe to the NIH’s “not invented here” or similar ideas, I do believe that people sometimes become complacent. Given that Linux had prevailed, it was simple to grow complacent. When new things appear and demonstrate the viability of new concepts, other people start to use those concepts. That, to me, is fantastic. There is always a chance of that.
1
However, I believe that in ten years everyone will be attempting to determine the “perfect” way to use fuchsia. Companies will need to decide how they’re going to adopt Fuchsia because I believe it will give them some significant competitive advantages. In around 10 years, I predict that “Fuchsia” will be there.
2
By that time, some people will already have “accepted” it. When the Fuchsia fork at Meta was canceled, I was disappointed. In fact, I believe that they split it too fast and that we didn’t collaborate enough throughout that time.
I firmly believe that Fuchsia’s design and its advantages over current operating systems are real, and I am confident in the team’s performance. I believe everyone will be attempting to determine how they may use “Fuchsia” as well, assuming they complete the tasks they are now working on.
One of the wonderful aspects of it being an open source project is that. Due of its presence, it will inevitably become something that people will notice and consider. Oh, that is also useful.
Yes, the “Fuchsia” squad is succeeding admirably. I continue to interact with individuals on the Discord server.
It’s amazing to consider that. Really, it is. I had exactly the same thoughts on Meta’s Fuchsia-based project. It seemed so thrilling, but naturally we didn’t find out about it until Meta had already canceled it.
Only a small portion of what Fuchsia is doing has been forked by “Meta.” Fuchsia once had a model that we liked to refer to as a layer cake. Zircon, garnet, and the strata above are where the stones were located. The issue was that the layer above Zircon, called Garnet, was already too opinionated for the Meta team, so they only forked the layers below Zircon.
The “team at Meta” had already made significant progress along this path by the time we had independently realized that these layers were overly opinionated and had redesigned “Fuchsia” to be much more modular. So even though I was trying to get them to think about how we could be more collaborative, it was difficult to envision how we would be able to work together on that.
The person who created the “Meta” fork now works at Google. He was one of the NT guys when he worked at Microsoft. That’s actually a contributing factor. Man, that seems a lot like what we were doing back in the day, he thought upon seeing “Fuchsia.” He was thus thrilled about it.
We sincerely appreciate Chris McKillop taking the time to provide these views.
FTC: We employ automatically earning affiliate connections. NeXT 5
NeXT 6
