Second Horizon in Data
• REPORT
What do London's 18th-century gin crisis and today's data economy have in common — and what does the answer tell us about where AI, cybercrime, and the collapse of state power are taking us next?
London, 1736. A series of measures designed to disincentivise the consumption of French brandy have backfired, leading to a rapid expansion of the domestic production and consumption of gin. A combination of falling food prices and the availability of cheap spirits had led the English poor to a choice: spend much less getting as drunk as they always had, or spend the same and get a great deal drunker. They went overwhelmingly for the latter, causing a moral panic and what seems to have been a genuine and massive expansion of crime and neglect.
The Government failed to learn the lesson of its distant ancestor King Canute and attempted to change behaviour with legislation, in this case by imposing ruinously expensive licenses and offering hefty boni for informers who led the licensing authorities to illicit stills. The result was an upsurge in violence against the informers, those suspected of being informers, and those who had offended someone and could be accused of being an informer. In an age before even the police, this was a catastrophe, and the law was quickly abandoned.
Almost anywhere on earth, 2026. The price of data drops virtually to zero, giving the population of most of the world a choice: consume the same amount of data for much reduced cost, or increase data consumption to unprecedented levels. Deepfakes, echo chamber algorithms and generational exclusivity produce an environment of rapid, bewildering change in which platforms, means of expression, payment and even belief come and go, with the constant shock of the new. Governments forget Canute and the hapless excise men of the Eighteenth Century and attempt to impose a regulation system conceived and executed at analogue speed on this digital maelstrom, with predictable results. Pearl clutchers of the world unite in another moral panic.
How did we get here, and where is it going? While the AI excitement creates a lot of hype (and some really useful tools), the underlying causes and trends remain, and will continue to shape the world of data and digital culture.
The story so far.
Firstly, (and this is the gin lesson), as prices fall, people gorge themselves. A lake of data has emerged as people have been and are careless of what they share. The back story of the modern data economy (and associated attention- or dopamine- economies) is the modernised gin story. From about the early 1990s, the cost of connections and data transport over telecommunications networks fell, and fell, and fell. By the end of the 1990s, ‘the Death of Distance’ described a world where the real cost of communications tended to zero, and where prices were then competed down too.
At the same time, the internet really took off, creating a supply of content and a demand for it, juicing up a new economy based on what quickly became ubiquitous data and information. There were some big technological stepping stones through this process: fibre optic cabling and very efficient data compression systems meant the internet really could double in size every two years and not get clogged. Smartphones untethered people, and social media provided an addictive scrollable diet of distraction, and worse.
The new economy started to supplant the old, and whole generations now grow up knowing only the convenience of the digital world. Data has been democratised, up to a point; many complain that services have then hollowed out – the process known as enshittification. It’s a story for another day.
Network effects (scale pays off) meant the new economy runs on a few super-large platforms. These cannibalise their competitors (buying them up before they become a threat). The biggest companies are American, exploiting English-language strengths and a favourable legal and regulatory homeland, and systematically offering convenience and a comprehensive though rather bland service. In recent years, cloud storage (nothing to do with the weather: it’s cheap storage in bulk at very low prices) has enabled much more data to be stored and aggregated.
Convenience has a price: identity and privacy went. More of this later.
The lake of data is now open to greater and greater analysis. The result is personalisation, or the appearance of personalisation – so we all think we’re special, even when it’s commodity stuff we’re consuming. Personalisation of data – meaningful disaggregation of your stuff and mine without losing the context - is a real opportunity in healthcare and the real threat to anything that aggregates risk, like insurance. The key point though is that the data lake is a private asset. It’s data about you, and from you, but not owned by you, and it never goes away.
Monetisation has always been the goal, through monopoly control and network effects. Just as some banks are too big to fail, the big platforms have probably reached that stage too – companies like Amazon Web Services and Microsoft’s Azure are now too big to fail, too.
Smartphones. These have proved to be the sugar on the pill of the digital economy, offering the user access to the whole entertainment-education-distraction-obsession warren of rabbit holes and associated advertising ecology, all while funnelling valuable personal data back. Social media plus smartphones has become the opiate of the masses (as Marx would have said; they’ve certainly replaced religion!).
Against this foundation, we get AI. AI only works because it builds on these foundations, using new at-scale deployment of ‘high-end’ chips to draw inferences and patterns from the existing lake of data and then apply really clever statistical techniques to produce results that are literally predictable, increasingly plausible, and capable of refinement and improvement at scale.
Scale. That’s the key. Data at scale and now computing power at scale. The result is an astonishing race to invest in storage and computational capacity. Financially, that looks like a massive bet, but given the growth in underlying demand for storage over time, it’s probably less of a bet than it looks.
So, that’s how we got here. What next? And what could go wrong?
Firstly, terminology: “AI” replaces the hitherto ubiquitous “computers” as the cause and answer to every problem, real and imagined. AI becomes harder to define as the term is used more and more as an accelerant to any idea, innovation or invention.
But some things are clear. Firstly, the fear that it’s all a bubble. Markets react as they always have, by spread betting on companies whose tech could change the real or digital world. The basic equation is, as it always was, that of the casino: a bet on a single company to be the next Apple or Google is the longest of shots - risky, maybe existential for the individual, but all to the good of the impersonal markets. Only a few will survive, and fewer thrive. The luckiest will get bought by incumbents. That’s all known; it’s really part of the first horizon. Today’s news.
But one more lesson of history also endures. Big change happens slowly then quickly, Inflection points between the slow and fast phases are like the elephants of proverb, easier to recognise than to describe. And just as hard to catch. So what is on the Second Horizon?
Beyond Moore’s Law: Not really; the principle will remain the same - processing power and storage will tend to the infinite and unit cost will tend to the infinitesimal. The lake of data will grow, and the ability to extract value from that lake will improve. That value will be shared between the incumbent tech companies and the rest of us, probably unequally. But almost everyone will get something. We can already see changes in employment (both the distribution and the intensity of human work), and maybe in productivity. Much of that is good.
That’s what we can see now, up to a point. Beyond that, as the digital landscape spreads, and darkens:
Cyber security will get a lot tougher. Online crime is already a royal road to wealth. The rapid analytical and near-instant coding power unleashed even now by ‘mainstream’ AI is going to make cyber attacks of all sorts easier, more lucrative, and harder to detect. It sounds hyperbolic to say that any system connected to the internet will be vulnerable. That’s true now; it will be much more obvious more quickly. Crime will move from ransomware and (old fashioned) denial of service to include more disruption, destruction and falsification of data. Nothing is safe now. Soon, we will all know it from experience.
Electricity systems are a huge weakness now and will remain so. Electricity grids are vulnerable and operationally fragile. Bad weather and bad actors threaten them. And the economics of scarcity gives their owners every reason to keep supply tight, costs down and prices up. Big destructive blackouts beckon. But electricity companies of all sorts will prosper.
Generally, states will continue to get weaker. As our lives and economies are lived online, states’ inability (or unwillingness) to police the internet will mean they retreat to try to protect their own data and core assets, and so much of our data and online experience will either be in ungoverned digital ‘marches’, or confined to the protected ‘walled gardens’ of big platform companies. States that try to police the internet themselves will struggle. However states will try to get into the robotics game to deliver social care to aged populations.
The big incumbent tech platforms will try to co-opt states to help buttress their collective power. This is especially true of the US, where these platforms are all based. Having set out the turn back the clock on trade in goods (through tariffs), the US will favour ‘free speech’ (ie free trade in data), because it suits them. Other countries chafe under this pressure and as they come to fear their dependence on US technology. But it will take a real crisis for anyone to actually choose – or be forced – ‘turn off the internet’ as we know it and go it alone. Possible, but it’d be desperate.
Identity issues will become important, as digital identities will be shown to be unreliable. Expect more in-person exams and interviews (and dating). The harm done by social media addiction affects everyone, but protections will initially focus on children. Countries that seek to enact these protections first (like Australia) will soon have strained relations with the US. But others will follow.
Tacit knowledge will emerge as a strategic asset. That’s the hands-on stuff you learn-by-doing, or from a teacher or supervisor. As digital delivery swamps everything, the places where you can learn rare aspects of surgery, or car mechanics, or actual cooking will be come valuable islands of analogue skill. They are likely to be at the margins geographically and economically. The ability to repair things will soar in value.
States will – to some extent – start to merge with cyber criminals and also with the big legitimate cyber platforms (think Palantir). Russia already tolerates cyber crime gangs who target the West; the US has announced it will ‘licence’ private actors to engage in approved ‘offensive’ operations. Piracy will start to become mainstream, and a licenced activity.
What next?
Robot sentience. Neither of these words is easy to design. Robotics is stretching between care of the elderly, three dimensional and judgement free porn provided by life size sex dolls, and postal delivery systems; through to the AI which is colonising the battlefield kill chain in Ukraine, and is part of Chinese military doctrine.
Sentience is a problematic concept because - that elephant again - it’s easier to describe or exclude than to define. We will return to this. But for this scan of the Second Horizon, we can assume that nothing we call a robot is yet sentient like humans; and that robotic sentience may always be different from human. Future armies may want to keep a leavening of human soldiers to deal with the analogue stuff, margin cases, and perhaps repairs (although AI plus 3-D printing may obviate that). The interesting question will be the extent to which humans get left in the decision-making loop simply to take the blame when things go wrong – expendable moral fuses.
We also need to note the extent to which humans project personality on to inanimate objects. It’s not that long since we worshiped the Sun, and the numbers of young men in the West with LLM based “AI girlfriends” suggest an high level of emotional credulousness (and desperation). We anthropomorphise dogs and cats, and embrace and celebrate the apophenia which has long since lost its evolutionary value.
Warfare is a whole topic, even beyond Ukraine (where we have already seen so much change so quickly). Here the tactical – manging weapon systems as they interact with the other side’s systems – will merge with the strategic, like knocking out satellites, or using nuclear warheads to generate Electro-magnetic pulses that fry electronics. This is a whole topic by itself, to which we will return.
Finally, the context: The emergence of the digital and robotic world with all its opportunities and risks is one of the two great shifts going on now that actually merit the adjective “tectonic”. The other is our aging or aged populations, plateauing and then falling (increasingly fast). This too will be changing politics, employment and of course the shape of demand in every society and economy. We live at the point where these meta-forces interact, and at a time when all the old political, and social certainties of the past few hundred years are eroding fast. The future won’t look like the past.
