The competition regime in 2050
Would we die from shock if we saw it?
Last week I gave a talk to the Competition and Markets Authority, the watchdog charged with protecting competition in the UK economy.
I’m a big fan of the CMA’s work and it was a pleasure to test some ideas with such an expert audience. I took the opportunity to speculate about the digital revolution and the distant future of competition policy, asking: what will the competition regime look like in 2050?
Since the talk ties in well with the theme of this blog — how do we govern the future? — I’ve posted my remarks below.[1] This is shared in the same spirit as the speech — it was a provocation intended to open up some big questions rather than resolve them, and of course my comments don’t represent the views of the CMA.
The competition regime in 2050
What will the competition regime look like in 2050? Or, rather, what will it need to look like in order to cope with a digital world?
Reflecting on this question over the weekend, I was reminded of an idea from the blogger Tim Urban — the ‘die unit of progress’.
The idea is that if you time-travelled far enough into the future, you’d probably die from shock at the social and economic changes you’d see when you arrived. And because change has been speeding up over time, the die unit of progress has been getting shorter.
My hunch is that, when it comes to competition policy and economic regulation, and I say this only half tongue in cheek, we’ll see a die unit of progress between now and 2050.
Or rather, I should say we’ll see a die unit of change. Because while big changes to the competition landscape now feel inevitable, what remains unclear is whether these changes will be progress or an unravelling of the progress we’ve made in the past.
The good scenario is that we’ll find the courage to reform the competition regime proactively, taking the reins of change so that we are in charge of where we end up. The bad scenario is that we’ll get stuck in incrementalism or dogmatism, leaving us on the receiving end of another kind of change: the ramifications of an increasingly outmoded regulatory regime.
In a sense, that’s the question facing all policymakers in an exponential age. Change is coming, whether we like it or not. The choice is: will we lead the change or be on the receiving end of it?
Recent history
In the interests of finding our way to the good scenario, I’m going to speculate about three flavours of reform we’ll need to see in competition policy between now and 2050. But first let’s set the scene with some more proximate intellectual context.
In the last 5–10 years there’s been a decisive shift in economics. I might even go so far as to say we’re seeing the start of the first true paradigm shift in economics in our lifetimes.
Actually, it might be more accurate to say there’s been a shift not in economics as such, but in the flavour of economics that gets applied in mainstream public policy, including in economic regulation. After all, economics in the academy has long been more diverse than economics as applied in public policy.
This broadening of economics in public policy is long overdue since one of the most striking features of post-war intellectual history has been the way public policy, and our entire debate about the role of the state, has leaned on an ever narrower intellectual base.
In the second half of the 20th century, public policy became ever more narrowly reliant on the discipline of economics, as opposed to other disciplines like history of psychology. And at the same time, public policy came to draw from an ever narrower subset of economic ideas, until it leaned almost all of its weight on the sharp point of an orthodoxy that was in essence a reductive reading, or perhaps really a mis-reading, of Econ 101.
Together, these two developments — over relying on economics and then discarding most of economics — made for a kind of double stretching of the intellectual source material of public policy. We trimmed back the cloth of economics to discard all but the most orthodox ideas and then we stretched that small cloth over an ever broader domain of our social and economic lives.
I’m going to call this economic orthodoxy Standard Economics, in honour of Standard Oil, because of the way it came to monopolise public policy debate. And the good news is that Standard Economics has now started to lose its monopoly as more diverse branches of economic reasoning, from behavioural economics to complexity theory, are being more routinely applied.
Why is this broadening happening? I see two main reasons, which economists might label endogenous and exogenous.
The exogenous or external driver for change, which is the focus of my remarks today, is advances in technology. Or, more specifically, the revolutionary practice of production that has been enabled by, and in a sense necessitated by, the emergence of digital technologies, most obviously the internet and ubiquitous high speed computing. The thing we call the digital revolution.
The effects of digital technology have of course been clear for decades. But I’d argue that it’s only quite recently — only really in the last 10 years — that the full, breathtaking implications of those revolutionary practices have become clear.
Only really in the last decade has it become indisputable that a digital economy functions in a qualitatively different way to the industrial or managerial economy of the early to mid 20th century. i.e. it’s become clear that the digital revolution is best thought of as a discontinuity in economic history, and one that is as significant as the transition to industrial capitalism. It’s a change to the economy’s logic.
For our purposes today, what matters most about this shift is that it undermines the keystone assumptions of Standard Economics. Claims that were critical to the structural integrity of Standard Economics no longer hold true. Or, to avoid getting into a scrap about what ‘true’ really means here, these claims are no longer useful ways to think about the world.
As an audience of economists and policymakers, you’ll be familiar with these crumbling premises, so I won’t labour them here. For anyone who’s not, I recommend Diane Coyle’s book Cogs and Monsters, which explores how economics needs to change for the 21st century. This includes the need to:
- Recognise that there are often increasing returns to scale
- Ditch the idea of consumers as isolated rational agents
- Accept the pervasiveness of network effects
And, in the context of competition policy, we could add a point made by Lina Khan, now Chairperson of the US Federal Trade Commission, who says we can no longer view antitrust policy mainly through a lens of consumer welfare. The rewards of securing a future monopoly over digital infrastructure are so big that we might see situations where consumers gain for years or decades, as firms give away their services away for free, and only later will the full political and economic costs of those monopolies become clear. So what we need is a deeper structural analysis of the dynamics at play.
This profound and, in a sense, eerily quiet loss of structural integrity reminds me of a house I once viewed and didn’t buy after the survey revealed pervasive dry rot. We had seen the rot in the floorboards but it turned out to run deep into the joists and beams that kept the whole thing standing.
The issue for policymakers is that we’ve rested the weight of our policy settlements on the joists and beams of Standard Economics, and this includes the settlements of economic regulation and competition policy. (Side note: those joists and beams also support the weight of labour market policy and welfare policy and they underpin the delivery models and choices we make about education and health, so you’re not alone.)
So that’s the exogenous reason economics is changing. Some of its core concepts are losing their utility, and we need a new intellectual structure on which to build a 21st century policy settlement.
This takes me to the endogenous driver for change. Because before I’m accused of being unfair to economists, it’s important to say that change has also come from within the discipline. Reforming thinkers have questioned the legitimacy of Standard Economics to act as the main representative of economics in policy debates.
Like the digital revolution itself, this change-from-within has roots that go back decades; in fact it dates back further than digital technology.
What’s odd about the intellectual history of economics in the 20th century is that it’s not nearly as narrow or homogenous as it seems if you encounter economics mainly through policy debates. In fact if you look at the list of Nobel Prize-winners for Economics in the 20th century, it’s in large part a list of people who challenged Standard Economics, including work on:
- The role of the firm
- Market imperfections
- Understanding the role of the minimum wage
- Behavioural economics
- Ways to think about complexity and uncertainty
The list would be even longer if the Nobel Committee hadn’t overlooked women like Joan Robinson.
So really what we’ve seen is a long-running under-utilisation of what economics has to offer, which amounts to a tragic waste of insight.[2]
And so as we stand here confronted by those daunting changes in technology, we have lots of material to draw on.
Turning to the future
How far have we got in the task of rebuilding the state so that it works for a digital economy? My personal view is that, when it comes to both the digital revolution and our policy response, we’re still in the opening credits of the movie.
We talk a lot about the disruption caused by digitally-native business models like Uber or Airbnb. But of course the daunting thing about the digital revolution is not that we’ve seen some disruption already, but that the vast majority of economic activity is still undisrupted. What I mean is that most economic activity still runs to a pre-digital logic, and so the latent potential for a ground-up reimagination of that activity is just sitting there, as yet unrealised, like a swimming pool filled with petrol.
Just last week I was at a dinner with investors and experts in industries like solar and synthetic meat; industries that are currently climbing the steep part of an S-curve of cost reduction and uptake. When you hear what it feels like to be on the rock face of that S-curve, and when you see how much of our economy will soon be climbing similar S-curves as this thing works itself through, you get vertigo. It’s not so much the primary effects, which are stomach-churning in themselves, but the unknowable second order social and economic ramifications, some of which will set off their own chains of exponential change. i.e. we haven’t seen anything yet.
Meanwhile, of course, our policy response is still only just starting out. It speaks volumes that we’re still at the point of founding new entities like the Digital Markets Unit — all of which is welcome and really promising, but also a reminder that we’re at setup stage.
So as I turn to 2050, let’s keep all of this in mind. The emblems of the disruptive age, Airbnb and Uber, were founded in 2008 and 2009, so we have more than two Airbnb/Uber lifespans between us and 2050, and when progress is exponential, that’s a die unit of change.
Three speculations
Here’s where things get speculative, and I’m going to hide behind Keynes. He once said that at decisive moments of history like this, even economists have to work in intuition. He felt there was a role for two types of economics: speculative economics, which works in broad brushstrokes of intuition to find new ways to make sense of the world, and synthetic economics, which draws insights together and uses data to turn big ideas into workable theories and models.
I’m going to share three thoughts, all firmly in the speculative camp, about the competition regime in 2050.
1. Back to the future
I’m going to call the first speculation ‘Back to the Future’. My basic point here is that, when it comes to digital platforms like Google and Facebook, these markets and the regulatory regime that governs them, will be structurally unrecognisable by 2050.
More specifically, I think the competition policy settlement of 2050 will have a cyberpunk feel to it, in that it will be built from a mashup of historic and futuristic materials.
To start with the historic materials, it feels clearer by the day that we’re going to have to dust off old policy tools from previous eras of market power. i.e. concepts that last saw lots of use when a previous technological revolution led to dangerous monopolies, with the mid-to-late 19th century rise of industries like the railways.
Digital platforms feel, formally speaking, more reminiscent of infrastructure like railways than they do of monopolies like Standard Oil. And so, from a policy perspective, my sense is that we’ll need to lean less on breakups and merger policy as a way of protecting competition in these markets, as we did with the oil or banking giants, and more on tools that prevent the exploitation of market power in situations where competition cannot sensibly or sustainably function. In other words, we will need to draw on the concept of natural monopolies and the toolkit of public utility regulation.
Part of the challenge we face today is that we haven’t used these muscles for a long time, and they’re stiff. It aches intellectually even to think of Google or Facebook, or aspects of Amazon’s business empire, as public utilities. It just feels uncomfortable. But we will need, I think, to push past this discomfort.
What Google and Amazon and Facebook have built is digital infrastructure in the same way the railway pioneers built physical infrastructure. The story is just so similar. The first railways were built in a free-for-all, and they went on to secure market power, and only then did we realise that what they’d built was in fact physical infrastructure of strategic economic and social significance, and we also realised that there were such benefits to scale that it made sense to have one such infrastructure, and we came to regulate this infrastructure in line with those insights.
I’m not saying of course that digital infrastructure is exactly the same as physical infrastructure. I’m just saying it’s formally similar.[3]
In fact, here’s one big difference: the costs of uncontrolled monopoly power are much higher with digital infrastructure than they were with physical infrastructure. Because digital platforms operate at a higher level of economic and social abstraction than water pipes or railway tracks, their monopolisation has more pervasive and profound social and economic effects.
What kind of effects? I won’t go too deep into this here because an audience of competition economists hardly needs an explainer on the costs of monopoly. But just to say it’s important, I think, to distinguish between different types of internet platform, since they each carry different risks and will require different policy treatments.
Social networks, for example, are one type of platform. Their distinctive business model is to sell to other companies the ability to change human behaviour. This means they (a) want to understand what motivates our behaviour, which they do by capturing what we call data but what is becoming, as data gets richer, something closer to kidnapping our digital twin. They also (b) want as much of our attention as they can get, which they do by building addictive services, since data plus attention equals the power to change our behaviour.
Social networks have already changed our lives. But their moonshot goal promises far greater change. They want ultimately to be the default ecosystem, working in augmented and virtual reality, so that most of our interactions with each other and with reality pass through their platform as a mediating layer.
Let’s not spend too long on the others. (I find the prospect of unchecked, monopolistic social networks enough on its own to make the point.) But just to note two other prominent types of platform:
- There are digital marketplaces like Amazon. They want to be the market, controlling its terms of entry and rules, charging for access, and of course using the data this gives them to launch their own products to compete in that market as a kind of producer of first resort.
- And then there are digital services like Netflix, which are, in a sense, less disruptive. But still we see the same winner-takes-all effects, not least because of the role of data.
Plus of course, for all these platforms, we see the profound power of big data when it’s married to iterative product development. This means these companies are unprecedentedly good at ‘optimising’ their products/platforms, i.e. making them addictive and hyper-effective at shaping our choices. This seems to me a development of profound significance since it undermines one of the most important ideas in economics — the idea that in free markets consumers make choices that reveal their underlying preferences, so that the interests of producers and consumers align.
And of course the biggest of these platforms want to operate in all three of these categories at once, while also working in this iterative way, and some are pulling this off, which further enhances their power.
The basic point is simply that competition policy is more important than ever.
But let’s pick up where we left off: this all explains why we need to push through the discomfort and shake out the stiff muscles of public utility regulation.
To be more specific, as Lina Khan has explored in the US, this suggests we might need to be more assertive using policy tools like:
- Nondiscrimination rules (requiring that digital platforms not favour their own products)
- Common carrier requirements (requiring that digital platforms host other business on equal and non-exclusionary terms, and not capriciously control access to their ecosystems)
None of this is to say breakups have no use in a digital age. It does seem, for example, that the model in which firms both run a digital marketplace and also compete in that marketplace is problematic, just like it was when railway companies tried to run coal companies and kept blocking other companies’ coal from moving on their tracks — something we eventually stopped them from doing.
I said the settlement of 2050 will also feature futuristic elements. This must be true because although digital platforms have familiar features, they’re still basically a new type of entity.
My hunch is that the futuristic elements of a new settlement will include:
- A whole new policy settlement to support interoperability. One of the main roles for economic regulation will be to make sure the world’s major technology ecosystems work together and don’t become private walled estates. This is a bit like the standard gauge we eventually set for railway tracks (but much more complicated and a much bigger deal).
- A whole new policy architecture and institutional settlement around open data in the private sector, requiring big tech platforms to open up their treasure trove of data for entrepreneurs and researchers to use. We will also need ways to steward this data like a global commons and to work out how to determine things like usage/access rights and who pays for maintenance, as well as ways to make sure innovation in the data layer is justly rewarded.
This all speaks to the basic (and, I admit, rather reductive) tagline to this section of my book, End State:
“If we want to harness the world’s technology giants for good, we shouldn’t break them up — we should open them up.”
Let me squeeze in one final point.
Although some people might find this all alarmingly radical, my hunch is that, on a 2050 timescale, what I’ve said so far will turn out to be hopelessly under ambitious.
I think you could reasonably argue that everything I’ve just said will need to happen by 2030, let alone 2050, and indeed a lot of these policy wheels are already in motion.
So to cast further ahead, the question that makes my blood run cold is: how much of the economy will be subject to these same platform-based business models by 2050?
The examples I’ve given are all of existing platforms like Google because of course these are the ones we know. But how many other sectors of our economy will have been transformed by the platform play by 2050, and will therefore be subject to those same winner-takes-all effects? Obvious candidates include:
- Robotaxis and the AI and data that sits behind autonomous vehicles
- Payment processing and financial services
- Sharing/resale platforms for high value/second-hand goods
- Labour market matching platforms
- Logistics and autonomous drone delivery systems
All of which is to say: we’re headed for a die unit of change.[4]
2. The CMA as a platform
My second speculation — and I promise the next two are much shorter — I’ve titled ‘the CMA as a platform’.
What I want to talk about here is the institutional form of the CMA, and of economic regulators in general, in 2050.
What I mean by ‘institutional form’ is what regulators will look like as organisations in 2050. I.e. how will they be structured, how will they operate and make decisions, and what kind of culture will they have?
I’m thinking not just of the CMA here but of economic regulation in general, so a better title might be ‘competition policy as a platform’. Or maybe even ‘regulation as a platform’.
Let’s start with some context.
One of the biggest consequences of the digital revolution is that it has changed the dominant institutional form in our economy and society. If the dominant institutional form of the 20th century was the modern corporation, then the dominant form of the 21st century is the internet platform.
Spotify or Netflix or Google look nothing like 20th-century corporations (the archetype of which was General Motors in the 1950s). From their org charts to their cultures to the way they make decisions, today’s tech giants are unrecognisably different beasts.
History helps put this in context. Because the shift we’re now seeing, with the transition to the internet platform as a new dominant institutional form, is similar to the shift that happened when the corporation itself emerged as the dominant organisational form in the early 20th century in response to that era’s technological revolution, the rise of mass production.
I won’t get too deep into this history here, but just to say it’s fascinating to read the literature from this earlier time, such as the work of Peter Drucker. And indeed if you want to understand the digital revolution, one of the best things you can do is read about the revolutionary impact of the rise of the corporation; it throws so much light on the changes we’re seeing today.
The modern corporation emerged for a reason: it was designed to take advantage of a new frontier of technology and manufacturing practice; General Motors was able to make and sell lots of cars really efficiently. And internet platforms are now replacing the corporation for an analogous reason: they are better placed to take advantage of today’s new frontier of practice.
In what ways do internet platforms differ from modern corporations? There’s a whole literature on this, with endless management books with titles like The Age of Agile, so I won’t rehash this all here. But in essence:
- Planning isn’t waterfall, it’s iterative/agile
- Decision-making isn’t hierarchical, it’s flat/autonomous
- Structures aren’t siloed; people work across disciplines
Those differences aren’t an accident either. They’ve emerged partly because old-school corporations turned out to be incapable of managing the complexity of software development; their attempt to apply waterfall planning methods, hierarchy, and silos to digital work is what triggered the software crisis of the 1960s. And new methods have also emerged partly because they help firms exploit the power of digital technologies, such as by making them operationally capable of responding quickly to live user feedback at scale.
Here’s the awkward issue I’m getting at: the CMA and other economic regulators are still modelled on 20th century corporations. And in case this sounds unfair, this isn’t a criticism of the CMA or of regulators — the same is true of every government department, and every big charity, and every legacy bank or energy company. We’re all in the middle of (well, at the beginning of) a Werewolf-style transformation.
The simple lesson to draw from technological history is that when the commercial sector adopts a new dominant institutional form, the state has to follow suit. This is one of the themes that runs through my book End State: our institutional settlement has to run to the same organising logic as the economy it governs.
So what would the CMA look like, or what would economic regulation look like more generally, if it had the form of a platform?
The answer is partly in those management books and it’s articulated for the public sector in GDS’s concept of the government as a platform. I also think it’s there, in nascent form, in the FCA’s transformation agenda, as they shift to a new outcome-based model of regulation.
To be specific, it means regulators doing things like:
- Being much more fleet of foot and iterative in the way they work, not having big long processes upfront before action is taken, but acting and learning as they go
- Being more outcome-based and driven by live data
- Breaking down silos so that regulators work in cross-discipline teams, perhaps even with teams from other organisations
- Being much flatter in the way regulatory decisions are made by giving autonomy to teams, so that regulators can be as agile as the markets they are governing
This is all really hard work; it’s the coal face of digital reform. And no doubt this work feels even harder in an economic regulator, where I know change feels risky. The point to keep in mind of course is that, in an exponential age, the riskiest thing is failing to change.
3. A Collaboration and Markets Authority
This takes me to my final point, which if anything is even more speculative. The question I want to ask here is: do we think enough about collaboration? As a provocation, I’m going to title this final section ‘a collaboration and markets authority’.
Collaboration might feel like a strange word to use in a talk to a competition authority. After all, isn’t collaboration the opposite of competition? And hasn’t collaboration historically been something of a fig leaf for anti-competitive behaviour/cartels?
Maybe, historically. But my hunch is that by 2050 policymakers will need to think much more about collaboration than we do today. And while I definitely don’t think we would ever want to replace the Competition and Markets Authority with a Collaboration and Markets Authority, I do think we will want to develop what I’d call a collaboration regime to complement our competition regime.
Why do I say this? One reason is that we can see more and more cases where it could be valuable to collaboratively develop shared platforms or at least shared standards (e.g. for interoperability) or data layers for an industry to use. Examples might include:
- The datasets, standards, and digital platforms that underpin urban transport systems like shared bike/scooter schemes
- The patterns and findings that sit behind new medical treatments and other forms of frontier scientific research
- The data and digital infrastructure that sits behind logistics networks, or future autonomous transport networks, or future electricity/EV-charging networks
The basic challenge in these cases is how to harness the collective ingenuity of a group of people, often working for different companies, to build and continue to improve digital infrastructure in a way that all participants support and can draw value from.
At the moment our only real model to develop digital infrastructure is the one I described earlier: there’s a race to the line until one company wins, and then a panic as we work out how to restrain the market power of that winning company. In other words, firms try to build the first best digital infrastructure, and take all the winnings, and the state ends up grappling with how to respond.
Is there another way? What if we had a regime of collaboration, so that there were ways for players in a nascent market to pool their ingenuity and be rewarded for their contribution to shared digital infrastructure, while also competing in other domains?
Or what if we had, at minimum, ways to collaboratively develop technical standards for interoperability, so that different companies’ ecosystems talked to each other and customers could move between them? Could we even start to think of this as a kind of regulation 3.0, running to a similar decentralised and collaborative logic to Web 3.0?
My sense is that, whether we like it or not, we’re going to have to explore this terrain. After all, the models I discussed above, based on public utility regulation, come with huge risks of their own.
It’s hard enough to regulate monopoly infrastructure when it’s physical things like water pipes or electricity networks. In the UK, regulators like Ofgem and Ofwat face a daunting task each time they go through a price settlement negotiation, trying their best to make sure electricity network companies and water companies invest adequately in their physical infrastructure. And of course water pipes and electricity cables are orders of magnitude less complex and slower-moving than digital infrastructure. Imagine if AWS became a regulated utility. How on earth could a regulator prevent exploitative prices while also making sure innovation continues at pace?
What’s promising is that collaboration problems are familiar to the software community. They’re the logic behind the open source movement, in which software can be collaboratively developed and forked into new products by a loose knit community. And they speak to the value of permissive intellectual property regimes like Creative Commons, and of mechanisms for rewarding contributions to a shared digital project. So perhaps a collaboration regime of the future, alongside a competition regime, will draw on these approaches.
Wherever this lands, I suspect we’re back to my opening point: the regime of 2050 will feel scarily different to where we are today.
Conclusion
I hope, if nothing else, this has all sparked a few thoughts. I don’t know about you, but personally it all feeds my wider sense that our public policy settlement is at an inflection point — and we don’t yet know if the line of progress is now going to tick up or down.
As I read over these remarks this morning, I was reminded of one of my favourite accounts on Twitter, which is called liminal spaces. It shares photos of places with a weird liminal vibe — i.e. places that feel transitional, as if they sit at a threshold between two different worlds.
The more I thought about these pictures of liminal spaces — deserted airport lounges, empty motorway off-ramps, flooded stairwells — the more it reminded me of the vibe in many of today’s policy debates.
It feels like public policy — and maybe economics too — is in an in-between place. We know our current settlement can’t last but we don’t yet know what will replace it. It’s like we’re at the airport and we don’t know when our flight leaves, or even where it’s going.
It would be understandable, I think, if we got a bit freaked out by this, and felt paralysed as a result. But of course the last thing you want to do in a liminal space is hang around. And for me that’s the lesson to keep in mind; we need to get a move on. No doubt 2050 will be scarily different from today, whether we like it or not, so our best bet is to lean into the change and define the future ourselves.
This post is part of a year-long series exploring how we govern the future. To read along, you can follow me on Medium here or support the project for £3 a month on Substack here. For the big story behind all this, from Victorian sewers to digital dragons, buy my book End State.
Footnotes
- On the day I made some cuts for brevity, so not all of this was delivered.
- Just in case anyone gets the wrong end of the stick, I’m not talking about public ownership anywhere here. I don’t think anyone sensible thinks a government-run Facebook would be less scary than a Zuckerberg-run Facebook. What I’m talking about is how we regulate/govern these entities, not who owns them. (Side note: Having said this, there are interesting debates about whether certain parts of the stack, such as identity management or data layers of significant public value, like healthcare data, should be built or maintained by public sector entities, but that’s a topic for another day.)
- For what it’s worth, I think we’ve made much less progress on that other tendency, of leaning too heavily on economics compared to other disciplines. Although if any discipline is going to challenge the primacy of economics in the next decade, my money would be on design.
- One criticism of my argument in this section of the speech would be to say: ‘this is all so Web 2.0’. And on one level I think this would be fair. The scary scenario I’m describing here could be seen as the explosive finale of Web 2.0 the movie — i.e. it’s a story in which our lives on the web (and therefore our lives in general) are mediated ever more fully through massive centrally-owned platforms. It would be reasonable to point out that this is precisely the problem that Web 3.0 hopes to solve with its decentralised ownership model. My basic response to this is: ‘maybe, but are you sure?’ Because unless we’re really, really sure that Web 3.0 will save us, then it must be at least prudent to assume that the momentum of Web 2.0 will keep running, and that we’re headed for more winner-takes-all dynamics, and that we should therefore at least work out how our competition regime will cope with this world.