
This week was busy with a big event at Hitachi EMEA where we presented our HARC Agents and then I wrote a blog article about agentic skills launched by Anthropic. But elsewhere, where’s the resume of this week interesting stuff to read: a scalpel-sharp dispatch from the operating theatre where Donna Vatnick trails a donor liver from midnight procurement to dawn rebirth, stitching together medicine’s craft with policy’s messy trade-offs; Felix Stocker’s candid field note on “talent” that nudges us to pick games our wiring can actually win; a Wall Street Journal reality check on AI’s compute hunger outpacing America’s wires, pushing data centres into a duct-taped “bring your own power” era of turbines, fuel cells and contingency planning; Bruce Schneier’s sober map of political actors arming up with AI – from fundraisers to citizen bots – while regulation naps; and John Messerly’s crisp Frankfurt refresher reminding us that confident ignorance corrodes faster than lies, so organisations need truth-first habits before the volume knob goes to eleven.
In this issue
5 piecesDonna Vatnick’s long-form essay “A Liver on Ice” documents the journey of a donor liver from procurement to transplant, weaving vivid observation with the medical, historical, and ethical complexities of organ transplantation

. Centred on Dr. Johanna Lee’s surgical team and recipient “Dorothy,” it traces the logistical precision and moral ambiguities involved in turning death into renewed life. Alongside detailed clinical description, it charts the evolution of liver surgery, from ancient grafts to modern immunosuppressants and allocation protocols shaped by science and policy.
This one floored me. It’s a scalpel-sharp blend of science and humanity; all adrenaline and ethics in one breath. I’ve always thought of hospitals as sterile places, yet this piece hums with messy compassion. I once toured an anatomy lab and couldn’t forget the smell; Vatnick captures that and turns it into awe. It exposes the fragile infrastructure underpinning transplantation, from donor consent laws to allocation algorithms, and illustrates how administrative frameworks shape outcomes as much as surgical skill. The piece also situates transplant medicine within an evolving moral economy: between altruism and market logic, public trust and private inequity. The upside? It restores faith in medicine’s craft, where each stitch is an act of grace. The downside? It reminds us how much modern life still depends on tragedy. Hard to read, harder to forget - like all honest writing about life and death.
Felix Stocker reflects on “talent” through prolific creators and traders, arguing that natural aptitude shapes both output and ease.

Citing newsletter writers like Matt Levine, mathematician Paul Erdős, and Scott Alexander’s “Parable of the Talents,” he contrasts work that feels “essentially free” with toil that demands willpower. He extends the idea to finance, quoting profiles of Bill Gross, Steve Cohen and John Meriwether to show that calm under pressure, inscrutability and obsession often look innate rather than trained. Stocker then turns personal. His writing arrives in intense bursts, not daily cadence, and his years striving at rugby - bulking, bleeding, still mediocre - taught him a hard boundary: effort can’t always bridge raw ability.
This is a bracing, useful memo to pick games you’re wired to win. I like its honesty, especially the rugby confession, because it punctures the comforting myth that grind always closes the gap. Stocker’s thesis implies leaders should optimise for “energy-cost” arbitrage, assigning work where behaviours are intrinsically cheap for the performer. In practice: interview for default cadence (burst vs metronome), stimulus response (calm vs adrenaline), and communication latency (Donovan-style responsiveness vs deliberate batching). Build teams by complement, not uniformity. His trading vignettes map to selection under stress; organisations can simulate these conditions to observe trait expression without waiting for a crisis. Talent realism can excuse avoidable plateaus. One question lingers: how much discomfort is useful before it’s waste?
This Wall Street Journal piece reports that AI-fuelled data-centre growth is outpacing the U.S. grid’s ability to supply power, pushing operators into “bring your own power” solutions.

Permitting delays, equipment shortages and transmission bottlenecks mean many sites can’t secure grid connections until the 2030s. Developers are therefore deploying on-site generation - predominantly natural-gas turbines, reciprocating engines and fuel cells - sometimes as a temporary bridge, sometimes by design. The article quotes forecasts that the U.S. should add ~80 GW of capacity annually to meet demand but is adding <65 GW, widening a near-term shortfall.
This is pragmatic, not pretty. I like elegant, system-level fixes; this feels like duct tape with turbines. BYOP keeps build schedules on track and avoids multi-year interconnection queues. But it normalises fossil infrastructure just as decarbonisation gets harder. Sites designed to flip from on-site gas to grid or cleaner fuels without a rebuild. This is markets outrunning institutions. AI is a classic demand shock colliding with a sclerotic U.S. permitting and transmission regime. Capital flows to the fastest firm power - modular gas, fuel cells -because interconnection queues and transformer shortages are binding constraints, not price signals. Policy oscillation; tariffs, changing tax credits, executive orders, adds risk premia, tilting portfolios toward assets with short lead-times and reversible commitments.
Bruce Schneier and Nathan E. Sanders warn that artificial intelligence is reshaping American politics faster than regulation can keep up.

They describe how campaign professionals now use AI for fundraising, targeting, and voter analysis; organisers leverage it for mobilisation, deliberation, and even symbolic protest; and citizens wield it both to amplify activism and automate misinformation. They note that AI is already embedded in political processes, from AI robocalls to avatar candidates, and predict its decisive influence in the 2026 U.S. midterms amid almost no oversight.
I’m with Schneier on the core anxiety: AI isn’t corrupting politics. it’s mirroring it, only faster. I like tools that scale civic voice, but this piece reminds me how easily optimisation turns into weaponisation. I once built a campaign email A/B test and felt that faint thrill of “playing god with words”; AI multiplies that tenfold. So, more access and participation for those long priced out of politics, come truth and accountability get crushed by volume and velocity. The urgent move is transparency, which require disclosure on all AI-generated political content. Otherwise, democracy risks becoming a generative prompt farm. Hard problem; zero time left.
John G. Messerly revisits Harry Frankfurt’s distinction between lying and bullshitting. Both liars and bullshitters want to be believed and to “get away with something,”

but liars knowingly deny truths, whereas bullshitters neither know nor care about the truth and may ignore the truth–falsity distinction altogether. Messerly summarises Frankfurt’s worry that bullshit is the greater enemy of truth because it pays no attention to truth. He also cites Frankfurt’s causal claim: bullshit flourishes when people must speak beyond their knowledge.
This is a governance memo for organisations. The Frankfurt–Messerly distinction helps leaders design controls around communication risk. Lying is a compliance breach; detectable against known facts. Bullshit is a cultural failure; speaking beyond knowledge, therefore a management problem. The fix differs: for lying, enforceable policies, audits, and sanctions; for bullshit, capability and cadence, raise knowledge floors, slow publication clocks, and build “truth workflows” (source logging, second-reader checks, and explicit “unknowns”). Messerly’s emphasis on truth’s primacy implies incentives should reward correction and uncertainty, not volume or virality. I’m allergic to confident ignorance; I keep a notes file to check claims before I opine. The liar-versus-bullshitter binary can hide mixed motives and the role of incentives. In short: a compact map for navigating noisy discourse.
Last Week in AI
- Coral NPU: A full-stack platform for Edge AI | research.google
- Claude and your productivity platforms | anthropic.com
- Microsoft wants you to talk to your PC and let AI control it | theverge.com
- Equipping agents for the real world with Agent Skills | anthropic.com
- Spotify partnering with multinational music companies to develop ‘responsible’ AI products | theguardian.com
- Microsoft will test Copilot Actions and Manus in Windows 11 | cnbc.com