SWITCHED ON
The daily technology series nobody asked for but everyone needed
Thirty Episodes Later: What the Second Season Taught Us
Nuclear reactors to grid batteries. Semiconductor geopolitics to elder care robots. Gene drives to space junk. If there is a single thread running through all thirty episodes, this is where we try to name it.
Every technology in this season promised to solve a problem. Almost every one of them, examined closely enough, turned out to solve part of the problem while quietly generating a new one — usually a governance problem, sometimes a distribution problem, occasionally both at once. This is not cynicism about technology. It is the actual pattern, documented thirty times over, and understanding it is the most transferable thing this season can leave you with.
Yesterday we closed the loop on the climate technology arc that has run through both seasons — grid-scale storage as the genuine bottleneck standing between abundant cheap renewable generation and an actually decarbonised grid, iron-air batteries and pumped hydro and thermal storage each solving part of the duration problem, none of them yet at the scale required. That episode, in retrospect, was a fitting second-to-last entry, because it exemplified almost everything this season turned out to be about: a technology that is real, that is improving, that is not yet sufficient, deployed against a deadline that does not care whether we are ready. Today, as we did at the close of Season One, we are stepping back from the individual technologies to ask what thirty episodes actually add up to.
01 — The Second Pattern
Season One's closing episode identified a pattern: technology deployed before governance exists, harms discovered in retrospect, regulation arriving years late and imperfectly fitted to the problem. That pattern held throughout Season Two — it appeared in deep-sea mining's regulatory gap, in cognitive enhancement's unenforced fairness questions, in space debris accumulating faster than removal technology can address it, in gene drives being developed with more governance care than most technologies precisely because the field learned from the pattern Season One documented.
But Season Two surfaced a second pattern that deserves equal attention: technology solving the visible half of a problem while leaving the harder, less visible half largely untouched. Nuclear energy solves the low-carbon dispatchable generation problem while the waste storage problem remains politically paralysed in most countries. AI diagnostics solve the pattern-recognition half of medical diagnosis while the health equity half — who actually gets access to the AI-equipped hospital — goes largely unaddressed. Early warning systems solve the prediction half of disaster response while the last-mile delivery and shelter infrastructure half remains chronically underfunded. Care robots solve the physical assistance half of eldercare while the loneliness and pension-system halves remain fundamentally human and economic problems that no robot addresses.
The visible half of a problem is usually the half that technology is good at solving, because it is the half that can be specified, measured, and engineered against. The harder half is usually distributional, institutional, or governance-shaped — who benefits, who decides, who pays — and those halves do not yield to better engineering. They yield to politics, and politics moves considerably slower than product development cycles.
02 — What Actually Changed My Mind This Season
In the spirit of intellectual honesty that this series has tried to maintain throughout, a few things covered this season shifted the way I think about specific technologies in ways worth naming directly. Nuclear energy's rehabilitation is more substantive than I expected going in — the AI electricity demand driver is real, SMR economics remain genuinely uncertain but the underlying case for keeping nuclear in the portfolio is stronger than the post-Fukushima consensus suggested for over a decade. Quantum sensing's practical lead over quantum computing surprised me — it is doing real, deployed work while quantum computing remains genuinely years from most of its promised applications, and it deserves more of the attention that quantum computing monopolises.
The gene drive episode was the hardest to write in either season, and writing it changed how I think about the entire category of irreversible interventions — the humanitarian case for eliminating malaria is not a rhetorical trick, it is genuinely one of the strongest moral arguments in this entire series, and the precautionary case against irreversible ecological intervention is equally genuine. Sitting with two strong arguments that do not resolve cleanly is uncomfortable and, I think, correct. Most of the technologies in this series have a right answer if you think about them long enough. That one might not.
03 — The Thread That Actually Held
If Season One's thread was "deploy fast, govern late, pay slowly," Season Two's thread, running underneath nuclear reactors and grid batteries and gene drives and care robots alike, was something closer to: every technology in this season was trying to compensate for a resource that is running short — energy capacity, care workforce, water, orbital space, time before a disaster hits, freshwater, cognitive labour, the working-age population relative to the dependent population. Season Two, more than Season One, was a season about scarcity and the technologies racing to manage it.
This framing matters because it clarifies what these technologies can and cannot do. A technology that manages scarcity well — that allocates a limited resource more efficiently, extends its effective supply, or substitutes for it partially — is doing something genuinely valuable, and this season documented many technologies doing exactly that with real skill: AI leak detection recovering lost water, precision irrigation reducing agricultural water demand, quantum sensing finding underground infrastructure without excavation. What none of these technologies can do is make an actually scarce resource actually abundant. Water efficiency technology cannot create rainfall. Care robots cannot create care workers. Grid storage cannot create more hours of sunlight. The technologies that work best in this season are the ones that were honest about operating within that constraint rather than promising to transcend it.
04 — What Two Seasons Add Up To
Fifty-five episodes, spanning artificial intelligence, biotechnology, energy, geopolitics, psychology, ethics, infrastructure, and demography, is not a comprehensive account of technology's trajectory — no daily series could be, and this one has never claimed otherwise. What it has tried to be is a consistent method applied across a genuinely wide range of subject matter: take the technology seriously, take its stated benefits seriously, take its documented harms seriously, resist the temptation toward either uncomplicated enthusiasm or uncomplicated dread, and try to identify the actual, specific, often unglamorous governance or distributional question that determines whether the technology's considerable potential translates into genuine human benefit or merely genuine technical achievement.
Applied consistently across fifty-five episodes, that method produces a worldview that is neither optimistic nor pessimistic in the conventional sense. It is something closer to: the technology is usually real and usually genuinely useful within a specific, definable scope. The scope is usually narrower than the promotional coverage suggests and broader than the dismissive coverage suggests. The gap between the technology's potential and its actual benefit to the people who need it most is almost always a governance, distribution, or political gap rather than a technical one. And closing that gap requires the unglamorous, patient, often invisible work of building institutions, writing regulations, funding the boring infrastructure, and making the distributional choices that determine who actually benefits — work that receives a small fraction of the attention that the technology itself receives, and that matters considerably more.
05 — Thank You, and What Comes Next
Writing fifty-five daily episodes across two seasons has been the most sustained piece of technology writing I have produced, and it has changed how I think about nearly every subject it touched — which is, I think, the honest test of whether a project like this was worth doing. If you have read even a handful of these episodes, thank you for the attention, which this series has spent considerable time this season arguing is the scarcest and most valuable resource of all.
Switched On will return. There is no shortage of material — the topics list that seeded this season barely made a dent in what deserves coverage, and the technologies covered across both seasons continue to develop in ways that will need revisiting. Between seasons, stay informed, stay appropriately sceptical of both the hype and the doom, and remember the thing that closed Season One and remains true at the close of Season Two: the choices that determine how these technologies actually affect human life are political and institutional, made by people, and therefore capable of being made differently by people who understand what is at stake and choose to act on it.
Turn on two-factor authentication. Question anyone promising a technology has no downsides. Question anyone insisting a technology has no upsides. And keep paying attention, because that turns out to be the whole game.
This has been Switched On, Season Two. Fifty-five episodes down. See you next season.
Switched On is a daily technology series covering the ideas, systems, and arguments shaping the digital world. Two seasons. Fifty-five episodes. Still going.
