Let's get this out of the way first: this is only for those who actually like the idea bringing desktop Linux to a wide audience.
Do you feel like nothing in your use or enjoyment of Linux depends on it going mainstream? That's fine. It just doesn't make you interesting on this topic.
For everyone else, a mere emotional sympathy for the outcome might be enough to carry you through, although I make no promises. I warn you fairly that this document is far too long, badly organised and sprawling in the observations and ideas it offers, many of which are made poorly and submitted for improvement.
I mostly mean it for those who have (or would like) a commercial or professional motivation to make this work. This is a very small group of people, but it's one that might do great things.
Let's jump right in.
“I would love to change the world, but they won't give me the source code.” - Anonymous
“Software is like sex: it's better when it's free.” - Linus Torvalds
Why don't more people use a Linux desktop?
The obvious assumption is it's either quite difficult or just a bit shit.
I mean, they're giving it away free and still hardly anyone wants it.. what does that tell you? You get what you pay for, right? Clearly it's so painful that only hardcore nerds, cheapskates and weird fanatics could put up with it. Otherwise more people would use it.
Look at that closely and you see a circular argument backed by no evidence. But few people are prompted to do so; they're busy with the rest of their lives. In the meantime, it's a view that makes intuitive sense, is self-reinforcing, and consistent with the ordinary expectation of unfamiliar tech. You even meet IT pros who think like this.
For the select few who actually use desktop Linux, this doesn't add up at all. They see daily that it runs so much faster, it's more stable and secure, it offers a simpler, less sprawling, more visually consistent graphical interface and it's basically just nicer and easier to use. When they're forced to use Windows 10, it's painful.
So how do they explain the market share? Usually with an “outstanding issues” narrative.
This sees the many advantages, but also the many problems: software support, hardware drivers, installation headaches, the miscellaneous little things that still send new users to the terminal, UEFI secure boot, frames per second on games, the list continues.
This is very persuasive; this is all very real and clearly it matters. These truly are the very best reasons not to use Linux.
I mean, let's say you're a web designer or a front end developer. Wouldn't something free and UNIX-y be ideal?
But when all your projects require Adobe products, Linux sucks. Even if you manage to run them in a compatibility layer, you'll live in fear of updates breaking everything. That's no way to maintain a livelihood.
Or what about passionate gamers? If particular games or graphics cards don't run as well.. or at all.. that's a big deal, right?
This narrative seems so convincing that it would be easy to miss that it makes no sense at all.
I mean, it's hardly an untested theory. Linux has been extremely good at identifying these issues, taking seriously that they matter and doing something about it. It's made immense progress on this front.
Has this been matched by similar progress in market share? No? So how could they be the big thing at play?
And what about that great mass of ordinary users with the simplest needs? Who aren't into the latest games, don't rely on bleeding edge hardware, don't need to run Mac or Windows-only applications. They just want to browse the web, write documents and emails, stream some music, video chat with colleagues and family.
These people experience relatively few of these issues, so you'd expect Linux to be relatively strong here. The opposite is true.
These conversations happen among technically focused people. They rarely zoom out to view the context around it: matters of reputation, media presence, social proof, first impressions, attitudes, perceptions and so on. But when you market share is this low, this context is the only part that 97%+ of humanity has any contact with. It must be of primary importance.
Saying this out loud isn't a magic spell to wish the tech issues away. Rather, it recognises that there is already excellent momentum, organisation and leadership on that front. The work to fix the social and behavioural context has no momentum at all.
It also recognises that only users have any real idea what these tech issues even are. When you talk to non-users about desktop Linux, their objections include little to nothing about the installer or proprietary graphics drivers or whether their wireless driver is in the kernel on the LTS distros. They know nothing about it.
The alternative is to ignore social context entirely and just improve the software. You might as well exhibit da Vinci in the darkest depth of an abandoned mine.
Should it be surprising that Linux has ignored this context for so long? It is, after all, a creation of technology professionals and hobbyists, many of whom have little to no background in these topics. That's not through any individual failing. They've just cultivated very different skills.
In traditional businesses, there's a marketing department or perhaps external advertising and publicity agencies to handle it. Open source has tended to provide very little basis for professional marketing. It's often not monetised with a sale, for a start. In any case, sharing your work without requiring payment or paperwork is already about as frictionless as it gets – between tech savvy types anyway.
But we now head in to 2020 with a desktop ecosystem that is maturing and commercialising. Across Europe and America, specialist hardware companies have shown you can sustain and grow a viable business selling Linux desktops and laptops. Some mass market outfits like Dell and Huawei have Linux offerings too. There is a wealth of independent media around the Linux desktop and much of it's quite good.
There are even a few desktop focused distributions that have figured out how to monetise a free operating system beyond donations.
Let's now take a good look at the psychological and reputational challenges in the way of desktop Linux, how they could be affordably and effectively tackled, and who has the most to gain from making it happen.
“If your perception is so much worse than your reality, what on earth are you doing trying to change the reality?” - Rory Sutherland, Vice Chair Ogilvy UK
When I write on this topic, I'm always curious to compare my ideas against published data on how the general public perceives desktop Linux. The trouble is, this doesn't seem to even exist. I think that says a lot on its own for how little thought the Linux world has given to psychological context.
Still, it should be possible to go part way to a useful picture by reasoning from casual conversations with non-users.
Let's start by noticing that few non-users have a direct line of sight to a Linux desktop. The market share statistic (between 1 and 3%, depending who you ask) goes some way to describing this, but vastly understates the case.
This market share is not just low, it's lumpy. It skews heavily to tech professionals and hobbyists. If you're outside these circles, it's not 1 in 100 or 1 in 50 machines you see it on, it's more like zero. Any opinion they have of desktop Linux – if they even have one – has little direct basis in the software itself or the experience of using it.
Their view is instead formed from inferences, observations and chinese whispers sewn together from media visibility, snatched glances, assumption and word of mouth. It's like a painting of a photograph of a photocopy. It's not assembled by a conscious process of reason, but by impressions gathered while they were busy with other things.
The most visible thing about Linux is where it's mentioned, who talks about it and how. These people are visibly enthusiastic about and adept with computers. They're comfortable with jargon, acronyms and technical concepts. There's a high chance they work or have worked in IT. This on its own positions Linux for a technical elite.
This group shows linguistic and cultural diversity, but very little gender diversity. Even compared to other technology communities, such as those around video gaming or Linux servers, the maleness of the Linux desktop is striking. Perhaps more women use it than we see, but perceptions rely on what's visible.
Linux also has a reputation for toxic elitism. For many years it was utterly routine for new users engaging with the community to find hostility, mockery and abuse. This still exists in pockets. Linux communities in the main though have realised how bad this is and changed. Still, it's left a big impression.
For most of its life, a Linux desktop really was only for the very technically capable. It could be incredibly difficult to install. Just obtaining installation media could be non-trivial. Configuring applications and hardware tended to involve lots of trial and error. This has been old news for years now, but it still lives in people's heads and in word of mouth.
Linux dominates many areas of computing beyond the desktop. These forms of Linux keep making new inroads into public awareness. This is a mixed blessing. There is real value in raw visibility and brand mentions. And yet, this almost always casts Linux as software for a technical elite to do difficult things.
Many people encounter Linux on servers at work or through an educational institution, while programming or setting up servers. Dealing with Linux servers means bash commands and text files, and perhaps the occasional (often ugly) web interface.
There is really nothing about this experience that would make you realise that the Linux desktop has become a polished and easy experience. Linux for Everyone's interview with Chris Titus gives a first hand account of a senior IT pro with no idea how great the Linux desktop was, despite interacting with Linux servers all the time.
Millions of people have first encountered Linux through the Raspberry Pi. This does offer a basic desktop environment, though there's not all that much you would do with it. Once you get properly stuck into projects, you're quickly up to your armpits in bash commands and text files again.
This is no shortcoming of the Pi: it aims to promote the study of computer science, not to promote Linux as an operating system for everyone. And yet, it's still out there showing millions that desktop Linux is very technical. It's quite possible to love the Pi to nerd out on and yet still think you wouldn't want desktop Linux as your daily driver.
It's now quite easy to use Linux command line tools on Windows through the Windows Subsystem for Linux. This strengthens the identity of Linux with the bash terminal and with development and systems administration work.
Linux and open source has reached mainstream media: see this CNBC feature and this article in The Atlantic. The emphasis here is again on infrastructure environments and on how code is written, maintained and accessed.
People are also in contact with Linux through embedded systems: in their Android phone, in their wireless router, even their electric cooking pot. We can discount this; Linux is in there, but it's not really visible in such terms.
“Good news everyone!” - Professor Farnsworth
This makes it sound like Linux is just plain bad at media, communication and marketing. Is this really true?
How about all those Linuxy content creators producing so many excellent blogs, videos and podcasts? You could never hope to keep up with all of that, even if it was all you did. Then there are the established businesses large and small who profitably sell Linux computers.
There are even a couple of desktop distributions who have figured out how to get end users to buy the damn thing. You can't totally suck at marketing if you're persuading customers to pay for free software.
But very little of this activity really grows the market share, because it's almost entirely aimed at people who already know they want the Linux desktop.
Even a billion dollar business with the global reach of Dell pushes a Linux range only to those looking for it.
This is a instance of the principal-agent problem. Whether your aim is to sell hardware, software or ad impressions, you'd be downright silly not to serve the audience most interested in what you do. That's just good targeting.
Yet when this is replicated across an ecosystem, we end up with desktop Linux's best advocates, marketers and communicators addressing their best and most persuasive pitches to existing users.
This is of course a very general picture; there are clear exceptions. Desktop Linux stories make it to the front page of Forbes. It reaches wider audiences of technology enthusiasts through Linus Tech Tips and PCMag features.
For the most part, though, this ecosystem leaves it to chance to reach the wider world. They, by necessity, are where you grow the market share.
“I use Arch btw” - Ancient Proverb
Why does anyone use Arch? It's challenging and tedious. Just running updates offers careless users scope for catastrophe. It does more or less the same stuff as any Linux.
There are even projects like EndeavourOS that offer something very close to a minimal Arch install, with a super easy install and a community who won't ritually humiliate you when you get stuck.
Keen Arch fans might point to the possibilities to customise, the leanness of an operating system built just from what you need, the chance to really get your head around Linux, the breadth and bleeding edge of the software library, the power you gain once you're over the learning curve, the pride and satisfaction of building it yourself.
These truly are excellent reasons. But what's actually wrong with the most superficial one? To tell people you use it?
That want of many Arch users to be conspicuous inspires much humour among Linux users. As a badge of Linux accomplishment, however, it seems quite good.
Arch communities are notorious for handling beginner questions with RTFM or even mockery and abuse. If you use any kind of helper tool to install it, you're told it's “not Arch anymore” and are instructed to go elsewhere. They quite emphatically don't want it to be easy.
The attitude is that if you can't work it out from the documentation, you don't deserve Arch. To many people, this just seems like asshole behaviour. Once you consider Arch's signalling value, there's a logic to it.
And yet, if you lurk in these spaces, you'll never see this strategy outlined. You might even open yourself up to ridicule for thinking they'd be interested in anything as effete as a “brand”. They've not planned this. This social intelligence operates at a more automatic level of instinct and intuition.
A human is a kind of monkey whose success rests much on social factors. Our survival, reproductive success, and the prospects of our offspring are all at play in our social performance. We've evolved to be quite good at sending and understanding social signals without even being conscious of it.
The signals we trust most are those with visible costs; these are much harder to fake. This instinctive sensitivity to costly signals means people can organise and cooperate in situations of limited information and limited trust. This theory of costly signalling goes a long way to explaining many behaviours that otherwise seem entirely against our interests:
Because Linux is reputed to require more effort than Windows or MacOS, using it signals technical ability and interest. Even among insider who realise how much simpler and nicer a Linux desktop can be, it still places you in an elite group to have figured it out for yourself.
As a way of advertising your nerd prowess as widely as possible, Linux might be the sweet spot in balancing fame with obscurity: BSD, Risc OS, Haiku and so on set you apart from the mainstream too, but how many have heard of them?
This signalling can have professional advantages; it also serves a more basic want for our peers to take us seriously. This sets up a variation of the principal-agent problem: the Linux desktop's highly technical reputation is a major barrier to wider use, but it's quite beneficial to many of those using it and indeed championing it.
As open source communities have matured and professionalised, it's much less fashionable to treat beginners with outright contempt. Still, there are all manner of subtle and not-so-subtle cues to present the software as a technical business for technical people.
Desktop Linux can't go mainstream by simply taking what exists and scaling it. It can't be about building a userbase that looks exactly as it does now, only twenty times the size. It will require a different brand personality. This will mean talking about the software in new ways, in new places, between very different people.
Let's look at Microsoft's enormous advertising campaign for Windows 95. This was the first time anyone really tried to position an operating system as relevant to everyone. It was also a huge success in doing so.
The most prominent element of this campaign was the music of the Rolling Stones. Apart from the lyrical association of “Start me up” to the Start button, it's hard to think what an old rock band from the vinyl era has to do with computers. Jennifer Aniston and Matthew Perry, stars of the decade's most popular sitcom Friends, also featured in the longer form video marketing. This seems similarly remote from a computing world.
And surely that's the whole point. This communicates instantly that Windows 95 is not just for professionals and hobbyists to notice, it's for everyone. It only enhances this signalling power that many of us believe the Rolling Stones and Friends actors are costly to work with.
(Note that Microsoft did nothing at the time to correct reports that the fee for the music was several times what they really paid; to be thought credible enough to print, this much inflated figure was likely leaked by someone inside the Windows 95 campaign.)
Would there be the same kind of budget to market desktop Linux? Not anytime soon. But a marketing effort should look for creative uses for the same kind of thinking.
Repositioning the Linux desktop as less technical by necessity means diluting the signalling value that much of the Linux community instinctively values. This is a problem for any marketing effort that relies too heavily on a really broad based, community driven, grassroots push. It's reasonable to expect a total lack of interest, or even a deep hostility to such a project from wide sections of the Linux world.
Linux users are quite a mix; not everyone's so protective of the technical brand. Costly signalling isn't the only driver of Linux behaviour. Reciprocity is also visible. It might be possible to mobilise volunteers who see promoting the Linux desktop as their way of giving back to open source.
Still, it will likely only be a few special people with an enthusiasm and feel for communication that could resonate with non-users. This will likely hold true even among those with the most ideological commitment to open source, who would love nothing more than to see this market share grow to dizzying heights.
Because desktop Linux needs to position as less technical, the most useful visibility will be in contexts that look quite remote from established Linux communities.
Signalling instincts will always be there. Sometimes they prompt unpleasantness. More often, they're healthy, functional and help us cooperate. If desktop Linux's technical brand is diluted, people can and will find signalling value elsewhere – perhaps specific distributions or BSD.
In the meantime, any effort to bring desktop Linux to the mainstream should be fluent with these issues to manage them usefully.
“Nobody gets fired for choosing IBM” - 1970s industry cliche
Social proof is a profound principle of human behaviour with massive implications for Linux. I urge anyone encountering it here for the first time to read further on the theory, and on the fascinating psychological experiments that delve into it.
The basic idea is that we're all highly influenced to do what we see others do. That's wired into our biology and happens below our awareness.
This influence is strongest when we feel uncertain. We are also more influenced by social proof from people we admire, or from those who seem similar to ourselves.
On face value, this seems like a mindless herd behaviour. But it must have helped us immensely as a primate species in the wild. When we're unsure what to do, we default to the things that visibly aren't causing death or injury to those who seem most like ourselves.
And how would we cope with the countless daily decisions of our modern world without this kind of mental shortcut? When you consider how much less humans have at stake in making perfect decisions than they do in avoiding catastrophic ones, social proof looks like a work of evolutionary genius.
Advertisers knew about social proof many decades before academics came to write about it. That's why it's so common to see popularity claims in advertising: “join over 334,000 people”, “over 99 billion burgers served”, “New York Times Bestseller” and so on. Advertisers also understand that this is malleable – that different perspectives and contexts can create a sense of popularity – and so look for creative ways to frame their offering.
But what if you're on the wrong side of social proof? You can be unpopular just for being unpopular. At an instinctive level, mass audiences decide desktop Linux is unsuited to them just because they don't see people like them using it.
The general invisibility of women on desktop Linux boosts this social proof penalty among the majority of adults who aren't men. This is a big barrier to market share.
Most conversations around how open source can better include women tend to focus on attitudes: whether women are welcome and taken seriously. Of course we should get this right for its own sake, but this doesn't address social proof. That hinges just on whether women see people who look like them using desktop Linux.
Desktop Linux does a lot better with age and cultural diversity. It's still important that any marketing push reflect this.
Social proof is stronger when we're uncertain, so it makes sense that desktop Linux does better among a technical elite. It's not that Linux in 2020 needs such expertise; this certainty negates social proof.
For everyone else, this social proof penalty is compounded by feeling dissimilar in technical ability to Linux users.
Social proof also feeds into the more learned and conscious behaviour of defensive decision making.
In the workplace, just as in nature, it's far less important to find perfect decisions for the whole organisation than it is to steer around those that could be terrible for you personally. This is a convincing reason to adhere to visible norms.
Most employees, most of the time, want to do the right thing by their team. But where there is ambiguity, the incentive is to choose what will bring themselves less trouble.
Recommending Windows is safe. That's not because it lowers your risk of catastrophes – the opposite is true. But when they happen, you won't be held accountable. Choosing Windows is so unremarkable, it won't even be questioned.
Recommending Linux means sticking your neck out. It will be suspected for any problems that arise. Managers in non-technology businesses often know little about computers, so it may backfire on you even for problems with nothing to do with the desktop operating system.
For anyone trying to sell support services for desktop Linux, this is utterly diabolical. You might totally persuade a decision maker that Linux is absolutely better for their use case, and they could still be entirely rational to choose Microsoft Windows.
Social proof and defensive decision making don't only come into play when users are selecting an operating system. Engineers, developers, and managers are humans too. These factors exist when they consider what level of hardware or software support to offer Linux.
It may suit some of these individuals to think that they're beings of pure commercial reason, but they're subject to the same unconscious biases and diverging incentives as the rest of us.
Neither social proof nor defensive decision making will disappear soon. They must be addressed head on. That means generating buzz and visibility for desktop Linux in a wider variety of places.
“Ah, your brother blows bubble gum!” - Bugs Bunny
Hypothetically, the Linux Foundation might take a lead here. It might devote a tiny part of its time and budget to a market research and publicity effort to really push desktop Linux. With an annual revenue over $80m, throwing just a fraction of a percent toward growing the desktop could be huge.
This would fit well with its formal mission of “building sustainable ecosystems around open source projects”. It kinda makes intuitive sense too.
But the Linux Foundation is reluctant to even acknowledge desktop Linux.
When you search the foundation's website for the word “desktop”, the results date from 2008 to 2013. A search of the foundation's Twitter shows mentions of desktop Linux as late as 2016. The foundation's Linux.com website does still publish desktop content occasionally, but it doesn't merit its own category. It's not seriously pushed.
Is it too conspiratorial to observe that in 2016, Microsoft joined the Linux Foundation as a Platinum Member? There's no solid evidence of a quid pro quo from the Foundation to not back desktop Linux.
But all it might actually take is for Microsoft to show up and be of real value to their work. Other decision makers would then decide on their own to set the priorities least likely to alienate Microsoft. This wouldn't ruffle too many feathers because these infrastructure concerns are what most interest the other large backers too. Developing the kernel and picking up the slack on other crucial open source projects is surely enough to keep busy.
Exactly how this has played out is ultimately a lot less important than just whether or not the Foundation backs the desktop. So far there's little reason to think they'll seriously take it on.
The Free Software Foundation has a brand and a following. Growing desktop Linux would be a big win toward their aim of promoting free software. Historically, this hasn't been the most PR and marketing savvy organisation: could this change with new leadership coming in?
One problem is growing desktop Linux will take a pragmatic attitude to proprietary drivers and firmware. They don't do that. There's real value to a strong voice for pure software freedom, so it's not even desirable that they would.
There's no telling how long the wait would be for an existing organisation to take this on.
“Yeah, well, I'm gonna go build my own theme park.” - Bender Bending Rodriguez
Many businesses and professionals could gain much from growing the market share of desktop Linux. What if they come together to do so?
They might fund and drive a small organisation with an exclusive focus on making desktop Linux go mainstream. It would finally have a marketing department.
This exclusive focus is essential. A Desktop Linux Alliance should never be sidetracked by other goals. It wouldn't have any.
It might be about moving large numbers of people from a proprietary OS to the GPL, but could take an expedient approach to closed source drivers and applications to make it happen.
This would be a non-profit organisation, but would nevertheless be commercial in outlook, to the benefit of the businesses and professionals involved in desktop Linux.
“The Linux philosophy is 'Laugh in the face of danger'. Oops. Wrong one. 'Do it yourself'. Yes, that's it.” - Linus Torvalds
The big thing about free software is that nobody needs any permission to try something new.
This is ultimately a strength. It gives free reign for new ideas while insuring against the bad ones. It's not that those involved are individually less capable of terrible decisions; but in a forkable universe, it's that much more difficult to enforce them.
It also means that growing market share for desktop Linux is a very different task than growing a proprietary operating system.
A proprietary software business has full command and control. A voluntary association of open source interests can lead, inspire, inform and communicate but ultimately has no authority to give orders.
Neither should it really need to. It just needs to focus on the best things it can do to support the ecosystem. Other participants, being free to do what they want, can then leverage this work in their own way.
So instead of planning the whole campaign down to the last detail, like the D-Day landing, it would be more like the SAS in North Africa: a small, fast and lightly equipped band of shameless opportunists relentlessly probing for places where short, sharp action can have an outsized effect on the battlefield of public perception. The regular soldiers still fight the war, with the odds shifted significantly in their favour.
One useful way to set priorities might be the AIDA model. This groups the stages the stages of consumer decision making as Attention, Interest, Desire and Action. AIDA doesn't capture everything that might happen, but it's been taught in universities and used by business for over a century because it's of great practical use.
It's possible to push through all four stages at once, but this kind of marketing tends to be idiosyncratic and rely much on time pressure: infomercials, sales letters, in-store promotions, door to door salespeople and so on.
Few non-users will switch to Linux in an instant. They'll progress gradually through AIDA, across multiple points of contact.
Attention and Interest are the easiest phases to work on when you are pushing a whole family of options. That's also where desktop Linux needs particular help.
That's about getting non-users to first just notice desktop Linux, and then to become passively curious about it, to learn what it has to offer, to start wondering if it might be relevant to them, and to be receptive to hear more. Desktop Linux has for the most part left these phases more or less to chance until now.
An alliance could also help with the Desire phase – to turn passive curiosity to an active want – but much of this has to rely on individual software, hardware and services vendors to push their specific offering. The Action phase – closing the installation or sale – would be entirely on the vendors.
An alliance effort could still support the Desire and Action phases with a standard of advice, research and analysis that would be cost prohibitive for individual players to develop in-house. To the greatest extent practical, this work should be published to everyone, in the spirit of open source.
“In praising Antony I have dispraised Caesar” - Cleopatra
The fragmented nature of Linux poses unique marketing challenges. Linus Torvalds has gone as far as saying it's “why the Linux desktop failed”. Is he right about this? Is it truly the biggest reason?
We actually have no idea. What it might take to verify this is some published data on how many non-users have either considered switching or tried switching but were too deterred by the confusing options. That would be some measure of the problem's size.
In my own casual efforts to ask non-users why they haven't switched to desktop Linux, fragmentation has never been mentioned. There are limits to the strength of the conclusion you'd draw from one guy asking many friends and family. Even if people don't claim this objection specifically, the quantity of jargon and the apparent complexity of the decisions involved surely contribute to their general impression that Linux is for a technical elite.
For those developing a sales pitch, it makes sense to present a very simple choice architecture. This is also a great recipe to alienate large sections of a Linux world that would otherwise be very supportive of an effort to build market share.
Fragmentation is largely a problem of perception and communication. Having an extremely long list of options is no hard technical barrier to switching.
But it does place in front of would-be users a set of decisions that feel extremely convoluted and carry a weight of unfamiliar concepts and language. In the face of this, the easiest choice to make is to not do anything.
It's only once you're using it that you realise how optional it is to get to grips with this stuff.
The common way to help new users through this is to jump right in and clarify as much as possible: what a distribution is, how they relate to each other, desktop environments, package managers, LTS branches, point releases, rolling releases and so on. You can take quizzes that ask many questions and then give you a list of answers, all neatly detailed with pros and cons. You can stream half hour videos from Youtube on this.
This is a great way to engage highly analytical personalities.
But with a wider audience, this could do more harm as good. It continually reinforces an underlying message that choosing the right distro is crucially important and you should be fluent with them to use Linux. This is a terrible pitch to most humans who will never enjoy distro hopping as a recreation.
It might be worth testing an opposite approach: to acknowledge that it's diabolically complex and then laugh it off as something that hardly anyone understands and doesn't really matter. This is in fact broadly true. The main distros built for broad audiences really are far more similar than they are different. They all run slightly varying versions of the same applications. They all offer the same benefits over Microsoft Windows.
That's especially true for “average joe” users who aren't developing software, setting up systemd services or using the package manager from the command line. Here, the consequential differences are the most superficial: default themes, the app store experience, running updates through a GUI and other things close to meaningless for many sophisticated users.
There is also room to avoid this question entirely and just promote “Linux”. This simplifies the communication immensely. The logic is that different distros share similar features and benefits compared to Windows 10.
Certain worthwhile activities force your hand though. It's hard to do an infographic without picking either a single distro or a shortlist of distros to highlight.
Windows 10 also comes in many flavours. Very few Windows users have any idea of what most of these varieties even are, let alone how they differ. This has not been a big problem for Microsoft because their positioning is such that few people need to even notice this.
Exactly how you market a fragmented ecosystem is a question for which there is no perfect answer, but a range of answers that weigh these competing considerations differently. Getting this right might mean experimenting with a range of approaches and comparing how they each perform in the wild.
There is danger here that the difficulty of this question could lead to a paralysis of indecision. To dither endlessly over this would be the worst decision of all.
“It's not the having, it's the getting.” - Elizabeth Taylor
Billions of people across all the world's countries have a personal computer in their household. This is too many people for a small team with a limited budget to reach all at once.
The challenge then is setting priorities.
One factor to weigh heavily is who can deliver a financial return. Delivering some real benefit to those backing the effort is the best basis to sustain it.
Desktop Linux's social proof challenges mean there's still value to new users who won't immediately spend money. That's especially true if they're in territories where market share is most valuable.
There's an obvious case to chase English language media first. English speaking countries form a full quarter of world GDP. English is also the world's most widely spoken non-native language. There may be a further case to repurpose content and promote it to other major language markets.
Within these territories and language markets there are different sectors, niches and use cases to target. This decision will inform which media channels are targeted, what messages are delivered and which benefits are highlighted.
There might be some scope to sidestep this decision by pursuing media coverage in the most mainstream channels. Large news outlets have recently given prominent coverage to open source, though so far not to Linux on the desktop.
There is also a strong argument to do the opposite: to target really narrow audiences that have strategic or where the use case for desktop Linux is particularly good. Concentrating efforts in small groups makes it much easier to build social proof within these niches.
Any decisions on who to target might be provisional - to be revisited based on new data and research. In the meantime, let's take a quick overview.
Gaming is a big thing in much discussion of desktop Linux's market share. This seems like an organic outcome of the fact that many who Linux also like games, rather than a strategic consideration.
Which doesn't mean this a bad thing to focus on.
PC gamers will by and large be more willing and able to deal with installation difficulties, driver issues, configuration problems and so on. They're already dealing with a lot of this stuff in Windows, where it's sometimes not as nice.
This would also be a relatively easy audience to reach. Many gamers engage with lots of gaming media, and are interested enough in computers to engage with other computing content.
Gamers are also hungry for the newest and fastest hardware and software, so winning market share here shapes an environment where hardware and software companies have to care more about Linux support. This helps smooth these issues for everyone else.
Gamers are still computer hobbyists. Gaming communities also tend to look very male. This is not an ideal niche to chase to address desktop Linux's social proof issues. An exclusive focus on this audience would fall well short of the work necessary to position desktop Linux as a truly mainstream operating system.
In spite of Linux's huge improvements as a gaming platform, the Steam hardware and software survey shows the market share is currently less than 1%. It's difficult to say how much these figures are skewed by dual boot systems using a Windows partition for gaming.
There's also the inconvenient fact that it's still not the absolute best platform if you are exclusively interested in gaming. For all the immense progress, gaming on Linux still means giving up performance on many games, while a few won't run on multiplayer or at all. The best way to convince gamers to try Linux might be within a broader campaign that emphasises other benefits to a wider group.
Many people who use Linux very regularly on servers, embedded systems or hobby projects have no idea how good the desktop Linux experience has become.
Is this low hanging fruit? They already use Linux, which gives them a big head start on navigating the choice of distribution, installing drivers, using the package manager and the terminal if they have to.
Targeting these Linux users shares many of the same advantages and disadvantages as targeting gaming communities. There are media channels to reach them through, they are curious about tech. But winning this crowd won't by itself build the social proof or develop the messaging needed to win over a truly mass audience.
Many people in this group can code and are interested in open source projects. There could be some good contributors to desktop Linux projects to be found here.
One way to magnify reach might be to promote Linux to independent media creators across all topics: history, philosophy, sports, business, politics, parenting, comedy, film, all sorts of things.
This is a relatively narrow audience to target. Tend to be curious people who crave novelty. Desktop Linux really does make particularly good sense for people who work with websites and digital media.
They're also of strategic value because they, almost by definition, have their own diverse audiences, the vast majority of which will never in a million years listen to a Linux podcast.
If they can be persuaded to use Linux, some of them would occasionally mention it in passing during conversations otherwise unrelated to computers. This is worthwhile kind of visibility to cultivate, as it helps establish desktop Linux as a new norm and a thing that non-technical people might use.
Content creators could be reached with little to no budget by a multi channel content marketing effort. This might involve in-depth features that explain the many advantages of Linux for their use, along with detailed instructions on how to go about it. This could then be promoted across relevant online communities.
This is content that a Linux community is placed to do well, because there are many Linuxy content creators already. Showcasing how they do what they do isn't just useful content for aspiring content creators across other subjects, it leverages the similarity effect in social proof.
This might be the most desirable place to gain market share because these businesses pay money for support services. Desktop focused distributions could offer support contracts directly. They might additionally or instead offer training and certification to others who provide support.
This is a place where real revenues can go to developing and marketing desktop distributions. A couple of desktop distributions seem to be already positioning themselves to take on this space.
This is also where the defensive decision making problem is at its most pronounced. Addressing this means doing more than persuading the decision maker that desktop Linux makes sense – it means reassuring them that it will go well with their bosses and coworkers who know nothing about computers.
To that end, basically any visibility for the Linux desktop outside of technology contexts will be helpful. Especially useful would be mentions in business publications read by managers of departments other than IT.
The very smallest businesses and sole traders have a very different decision making process. These people manage their own IT, which makes things much simpler. Instead of worrying what everyone else will think of their operating system, they only have to like it themselves.
This is a price sensitive group. They tend not to pay much or anything for IT support (even when it would would make sense to).
Linux can well serve this wish to avoid expenses. That's a little because they can avoid the licence cost, but mostly because of the user experience they can get on modest hardware.
These people are also very worried about software support. For some, this is well-founded – they really do require a particular application that can't be relied on to run in Linux. Others think they “need Microsoft Office”, even when their word processing needs are incredibly basic. Some will worry about this even when everything they use runs in a web browser.
These very small businesses are of strategic value because a portion of them do go on to become larger. Winning these small business users also creates excellent social proof for the larger businesses.
Making inroads with this audience will also prompt developers of business software to rethink their attitude to Linux support.
These people are on a budget and try to make their cheap laptops last. They also tend to have simple needs that Linux could easily cater to.
Another interesting angle would be to try to get in front of people who are shortly about to install an operating system.
One big advantage that Windows has is that you usually don't have to install it. People building their own computer don't have this luxury, and often discover that installing Windows 10 can be quite a headache.
For small businesses that refurbish second hand computers, installing a lightweight Linux distribution could be a good way to extract value from old machines that would be painfully slow with Windows 10.
“If it's a good idea, go ahead and do it. It is much easier to apologise than it is to get permission.” - Grace Hopper, inventor of the compiler
So what could a Desktop Linux Alliance do to grow the market share?
Here are a few ideas.
One of the most surreal features of two decades of “year of the Linux desktop” predictions is that it looks like nowhere along the line has anybody really had a good look at the people who'd have to be won over.
These conversations go ahead on the basis of casual observations of non-users, or worse, by users putting it all together in their own heads. When you try to track down what the market research says, none seem to be published.
Some Linux businesses may have unpublished data for internal use, but this is of limited use to an open source ecosystem where the work and the decision making is distributed across so many small business and community projects.
Entrepreneurs in this space who want to pitch products and services to new users have little to no coherent picture of what this audience actually sees, thinks or knows about desktop Linux.
We don't even have a good idea of the market share. The figures you see quoted tend to be collected from website traffic, which can skew in all sorts of ways. It's not a great measure, but at the moment it the moment it's the best we have.
There are more questions to ask. Who has heard of Linux? Who realises what it is? Who realises it can be used as a laptop or desktop operating system? Who has ever used it? What has used it in the past 12 months? How much do they know about it? How aware are they of the features and benefits?
Until we have this data, we don't even really know what the challenge we're facing really is. We could make reasonable guesses, but we don't actually know.
There's also scope to study audiences of specific interest, such as PC gamers, podcasters or small business people.
Such data would firstly be useful for the Alliance's own marketing and publicity efforts.
The data could also be published. That way it can inform business and community efforts to compete with proprietary operating systems.
Market research data can also be used to provide the media with interesting stories. In this way it can be leveraged for publicity across multiple channels.
Some things you can't learn about an audience just by asking.
People are generally pretty good at telling you what they like and what they're aware of. But what we do and what we'll say we do are often quite different things.
That's partly because we're not always totally honest about these things, but mostly because our real reasons aren't always visible to the conscious mind. This makes it difficult to distinguish the real drivers of our own behaviour from the rationalisations we construct afterwards.
That's why it's interesting to run tests and measure the response. Fortunately, the modern web makes this inexpensive and easy.
With a modest budget, you could use digital advertising to split test different messages and compare how they perform.
With no advertising budget, you could run split tests on pages that already receive inbound traffic.
The front facing pages of desktop Linux websites would be worth particular attention. The better these pages perform in terms of driving interest, purchases and installations, the faster the market share can grow. The more we can learn and share about how to make these pages perform best, the better.
There could be other marketing case studies beyond running split tests. This could be done by the Alliance on its own, or it could be done in cooperation with a participating business or organisation. By participating, they would receive help with improving their marketing, in return for opening their marketing channels up for testing and agreeing to have the findings published for the benefit of open source projects generally.
If you have half a billion dollars to spend, you can reach big audiences across multiple major world markets through advertising. If you have much less, you reach big new audiences by becoming the feature story on other people's media channels.
Open source has recently started appearing in mainstream media. Desktop Linux is also a good story that merits prominent coverage in top-tier publications. If it was pitched in the right way, it would surely secure some.
This is a great opportunity to visible to more people. It's also a chance for desktop Linux to appear in contexts less exclusive to technology professionals and hobbyists.
Publicity methods are also a direct way to craft social proof. There is a lot of scope here to leverage the multiple source effect by appearing in a range of media.
So any kind of story about people switching to Linux is worth pushing. That's especially true if those people are not technology experts. Even small gains should be publicised aggressively.
One way to do this could be a data driven approach, based on finding the little niches and micro-niches where the Linux desktop is either increasing in popularity, or is just more popular than average.
There is also scope here to appear in niche publications that cater to target audiences. This is powerful because it leverages the similarity effect.
Content marketing is publicity's close cousin. It has some crossover in that it's about publishing and promoting content, and the two approaches generally pair very well.
It differs in that it's more inbound in emphasis – it's about nurturing existing attention. It can work particularly well for targeting specific niches.
Content can be primarily text based, such as blog posts and well curated resources sections. Infographics and other visual content can also be a great way to quickly communicate features and benefits, and can be popular with writers and journalists too.
Video marketing is another way to depict these same features and benefits. These could be promoted with Youtube SEO to appear in search results and sidebar recommendations.
With a large enough budget, an Alliance could sponsor the creation of a closed source program that's compiled only for Linux and then distributed for free.
If it's technically feasible, is might be made in such a way that makes it far more desirable to run it natively, rather than in a VM or through a compatibility layer.
This might make particular sense as a way to target gamers. Sequels or “enhanced and expanded versions” of indy games with a cult following might be relatively inexpensive and very interesting to a large audience. A small studio who participated in this would benefit both by sponsorship and by the publicity value for the other games in the series.
This completely reframes the problem of how you get people to install Linux. Instead of convincing them that Linux is now their forever operating system – which is both a complicated topic at the logical level and a huge barrier psychologically – they just have to want one particular application enough to install a Linux partition specifically to use it. Once it's there though, they're free to explore it at their leisure.
It could also be a huge driver of publicity. The novelty value would be immense and could merit coverage in all sorts of places – including a few howls of outrage from free software purists.
Reviewers, content creators and other media who want to engage with it would of course have to install a Linux partition too. Having these influencers engage with desktop Linux would be of lasting strategic value.
You could optionally choose to later release part or all of these programs under an open source license after a few years as Linux exclusives.
“Now all we need is a little energon and a lot of luck” - Optimus Prime
It's all fine and well to have grand visions of epic ventures. But how do you foot the bill?
Many businesses would benefit materially by growing desktop Linux.
For some, it means growing their market: who they can sell to, or the size of the audience they can speak to. Then there are tech giants for whom Microsoft's desktop monopoly is either a present disadvantage or a future risk.
For the largest of them, sponsorship would be more or less a rounding error in terms of their bottom line. They've so much more at stake than the sum involved. It doesn't hurt that it could be easily presented in PR-friendly terms as a generous support for open source (which it actually would be, too).
For much smaller businesses with revenues in the six or low seven figures, launching a Desktop Linux Alliance is in large part a value proposition. At this size, it's a bit expensive to foot the bill on your own for an amount of research and advocacy work that would actually move the needle. The trade-off is that you have to share it with other Linux projects. That's a thing they're doing already with their code.
At the smallest level are those for whom desktop Linux is a solo business activity or even a semi-professional pursuit. From these people it's neither realistic nor reasonable to expect more than a token sponsorship. There is still a lot of value to that – far more than the actual sum involved. A small payment builds a feeling of commitment to the cause, a sense of obligation from the Alliance back, and signals to anyone watching that this is a serious project.
For this purpose, we can divide popular Linux distributions into three groups. There are the flagship distros for large businesses that monetise Linux through support services to enterprises. The benefit to these businesses of growing desktop Linux's market share is that they can sell support services here too.
Then there are the explicitly desktop focused efforts the most user friendly offering they can for the broadest audience possible. These tend to have more modest revenue streams from donations, sponsorship, ad revenue sharing.. all things that depend on the size of their user base. Ultimately, the big incentive would be if they could develop a training, certification and support business.
The third group is just everything else, from Hannah Montana Linux to Debian. These distros probably have less to gain directly from a Desktop Linux Alliance.
Linux bloggers, video bloggers and podcasters are already at the coal face in communicating the Linux desktop to new users. They can promote the idea of a Desktop Linux Alliance, build momentum behind it, and get it in front of other key decision makers – who may have to hear about it many times before they take action.
They also have a very direct interest in the Linux desktop's market share. The size of the user base directly relates to the size of the audience they can serve. The more that new users there are, the more traffic they can pull.
Those who feel their work requires a journalistic independence may prefer not to formally join an organisation, even as they're sympathetic to it. Others may feel that this is no problem for the style of content they create.
These businesses can largely be divided into two categories: smaller outfits who specialise in Linux laptops and workstations, and big brands that feature a Linux offering.
For a while it looked like geopolitics might prevent Huawei from selling Windows laptops, but Microsoft was granted an export licence to continue supplying them. Huawei and other Chinese brands should still perceive a Linux marketing effort as a way to mitigate against this ongoing risk to their supply chain.
There is an opinion that the large hardware manufacturers are only interested in Linux as a way to pressure Microsoft to lower the price they pay for Windows. If true, this is in itself a clear commercial reason to get behind this effort. Without some visible shift in market share, that leverage isn't real.
The annual trade in cloud services is currently $200b and growing at 15-20% a year. Tech giants valued at billions or even a trillion dollars, like Google, Microsoft, Amazon, Alibaba and IBM, fight tooth and nail for their piece of this big and growing pie.
Owning the dominant desktop operating system is a massive competitive advantage for Microsoft against the other cloud giants. They manage a Windows ecosystem that includes training, certification, partnership and other commercial programs. This is a sales and marketing channel for cloud services into enterprises, small businesses and managed service providers.
The Stasi level of telemetry in Windows 10 is also the mother of all market research tools. That's large scale, real time data on macro trends in computing that only Microsoft is able to base decisions on.
This is where some of the world's wealthiest, most sophisticated and best connected businesses might view favourably a commercial alliance to bring desktop Linux to the masses.
Valve Corporation is already a big backer of the Linux desktop and has done large amounts of heavy lifting to make it a viable option for gaming. Pure enthusiasm for the technology might be carrying them part of the way, but when a big company offers this kind of sustained support for a cause, they usually perceive a commercial interest as well.
In this instance, it's probably a way to address an exposure to a significant downside risk. Valve's estimated worth of $4-10 billion is depends a great deal on their game distribution platform Steam, which dominates the PC gaming market. 96.3% of Steam customers use Windows.
Microsoft's efforts to promote the Windows Store are a direct threat to Steam. If Microsoft ever succeeds in training users to go here to buy software, this could wipe out billions in wealth for Steam. Microsoft's recent acquisitions binge of games developers only clarifies this threat.
Valve already contributes to the Linux desktop heavily by contracting Codeweavers to develop the Proton compatibility layer. Proton has been a huge leap forward for Linux as a gaming platform. To really get the most back from this investment in code, they'll have to care about the perception, visibility and reputation of desktop Linux as well.
Other digital distribution companies might also see a similar interest in encouraging the Linux desktop.
“You know, when you were a baby in your crib, your father looked down upon you and had one wish: ‘Some day my son will grow up to be a man'. Well look at you now. You just got your asses whipped by a bunch of goddamn nerds. NERDS!! Well, if I were you I would do something about it. I would get up and redeem myself in the eyes of my father, my Maker, and my coach!” - Coach Harris, Revenge of the Nerds
Is this far too ambitious?
Probably.
I'd be ecstatic if this actually took off. I am realistic though about whether dozens of technology businesses around the world would immediately back this idea just because some guy they never heard of wrote 11,000 words about why they should.
My real hope is to seed ideas and open up conversations among those with decision making power or with the platform and influence. I'm urging a wider view of the marketing challenges facing desktop Linux, what could be done about them, and who could work together to make it happen.
To fall back here on crude stereotypes of computer nerds who are brilliant with tech but terrible with people would be both lazy and wrong. On the whole, the Linux world is actually now quite good at speaking to the conscious rational mind. If you're hungry to learn all you can about Linux, there's miles of great explanations available in text, audio and video, and the best of it is vivid, clear and even very funny.
So little wonder then that Linux does so well on servers and embedded systems – that's exactly how they're considering an operating system. It doesn't hurt that these IT pros making these choices are not accountable for them to people who know little about computers – as long as the damn thing works, their choice of OS is invisible. It also doesn't hurt that these applications are a perfect fit with Linux's very technical brand. All of this is surely as much a part of Linux's success here as the licence terms and the code.
Desktop users are different. Their conscious focus is everywhere but here. These attitudes and decisions falls to much lower level “monkey brain” instincts: the automatic intuitions that all of us use for most of our decisions all the time.
Calling this the monkey brain is meant as no insult. It's been fine tuned by countless generations of natural selection and is quite sophisticated in its own way. It's peculiar though.
The monkey brain cares little about choosing perfectly and nothing for objective truth. It's optimised to consistently avoid disaster. It's evolved to reliably make these non-disastrous decisions fast and with little effort in situations of limited information and limited trust.
So it happily draws broad conclusions from surface level information, with a heavy weighting on social cues. The monkey brain then straight away moves on, conserving scarce resources of thought and attention for those few key decisions we choose to focus on.
When we make automatic intuitions quickly and on little evidence, they're routinely wrong. So long as we're wrong in ways not personally disastrous, the monkey brain has worked fine.
If desktop Linux is to reach large new audiences, it has to engage much better with this style of cognition. When you speak to the monkey brain, the context you appear in is at least as much a part of your communication as the actual content of what you say.
At the moment, this context is relentless that Linux belongs to a technical elite. Because almost nobody feels that's their place, they conclude it's not for them.
If they delved further, they might find a desktop experience that traded away the harassment and sprawling confusion of Windows for something genuinely navigable, responsive and visually consistent. The monkey brain doesn't take that step though.
Dealing with that effectively will take more than good ideas. It needs some level of organisation and leadership. There's definitely scope for enthusiastic community members to craft killer content: there are a few busy corners of social media where that's going on already.
There are also a few crucial things here that are quite hard to do well on a strictly volunteer basis. There's a limit to the quality of the market research you'd come up with as a hobby. Securing coverage in top tier media in your spare time is also a big ask.
This doesn't require the marketing spend of Apple and Microsoft. A quite modest amount could have an outsized effect. It's worth a shot anyway. It would be at least a little weird to keep throwing so much at the code while acting as though this other stuff doesn't exist.
You can't code your way through some situations. That doesn't mean they're too hard or not worth trying.
On a more personal note, I wrote this firstly because I guess I'm quite mad. But in part it also felt like it could be my way of giving back to open source. I can code a little bit too but nothing I have to offer there is as good.
So please treat it as you would an open source project. You can come on board with it as it's presented, suggest changes upstream, fork the project or just pilfer what's most useful to you for your own thing. I've tried to make it easy for people to link to specific sections: just copy the hyperlink from the table of contents.
If you make any use of any of this, all I ask in return is that you respect this open ethos with whatever you do with it. I've no licence to enforce that, I'm just asking.
Oh, and if anyone is looking for a wordsmith for their Linuxy product or service, I'm available for hire as a technology copywriter too.
© 2020 James Mawson