Four For Friday | Jan 16, 2026
LF203 Open science operating system, lessons from CES, AI vs. civic institutions, data unions + quote of the week. Enjoy!
Welcome to this week’s Four For Friday. Here are four nuggets of interest I’ve picked up this week, plus a quote of the week
1. Reinventing science as an operating system
The way we do science needs an upgrade, and this piece lays out some bold ideas in response to a call from the US government.
Researchers now squander a third of their time chasing grants while funders struggle to make decisions, producing an opaque, sclerotic system where money concentrates predictably at the usual suspects. The fix? Combine a research commons (open standards, transparent decisions) with a systems-investing mindset (portfolio approaches, multiple shots on goal).
The piece proposes a “networked operating system for science”: machine-readable data across all agencies (goodbye, PDF prisons), standardized research “containers” portable between funders, and marketplace mechanics where ideas compete on merit rather than institutional pedigree. At the heart is an “AI Program Manager” directing portfolios like index funds, complete with syndication pooling public and private capital.
DOE’s Small Business Vouchers pilot, which awarded 114 firms access to National Labs, is a step in this direction. Strip away the grant-writing theatre, unleash the signal —> unshackle innovation.
So What? If AI generates hypotheses faster than humans can fund them, the bottleneck isn’t discovery but deployment infrastructure. Fix the plumbing or drown in possibilities.
2. Your Mirror Will See You Now
CES 2026 was quite a buzz fest, but it also saw AI going from gadgetry to infrastructure (and front-line robot workers).
Among the standouts in this summary by Catherine Ball: NuraLogix’s Longevity Mirror transforms a selfie into health indicators and a longevity score and Boston Dynamics’ Atlas robot - addressing labour gaps through human-machine collaboration. Also, Ambient AI for aged care (e.g. Samsung’s platform) deploys passive sensing to extend independence without surveillance.
Australia has a choice: build evidence-based longevity infrastructure or spectate while others commercialise the healthspan economy.
The So What? Prevention infrastructure beats hospital heroics when scaled properly.
3. AI: Killer of Civic Institutions
Boston University professors argue artificial intelligence will demolish democracy’s scaffolding - the civic institutions that built society.
I’m not totally convinced by the arguments, as I don’t think most civic institutions are fit for purpose for the challenges we’re facing, but it’s certainly got some good points. They suggest AI undermines expertise by encouraging cognitive offloading and producing hallucinations (true, but guard rails and RAG help this, and this feels a bit like protecting turf).
Second, it short-circuits decision-making by obscuring moral choices behind algorithmic black boxes, flattening hierarchies that enable accountability. (For sure).
And third, it isolates humans, eroding the interpersonal bonds institutions require for adaptation and legitimacy. From courts outsourcing bail decisions to universities replacing professors, AI transforms functioning institutions into brittle shells.
So What? Technology’s promise of efficiency means we’re swapping institutional resilience for algorithmic convenience, potentially unraveling centuries of social infrastructure.
4. Are we ready for data unions?
The UK’s Nesta thinks 2026 could be the year of the data union, and I hope they’re right.
ChatGPT's meteoric rise has accelerated an already fairly gnarly race to own humanity's digital exhaust. UK households unknowingly shell out £1,000 annually feeding the platform beast through advertising pass-through costs. The First International Data Union has just been launched to wrest control from Silicon Valley by pooling citizen power (but I don’t see folks paying $5/mo for uncertain benefit). But this is what Tim Berners-Lee originally thought of for the web; unfortunately his purity has been derailed by profit seeking.
If it works it would redirect attention rents back to members' pockets while unlocking research goldmines (like Imperial College's ovarian cancer study correlating medication purchases with diagnosis). But data collectives aren’t easy to build, suffering like most networks on the web of a cold start problem. Regulation could help - Australia has shown it can break from the pack and regulate social media and now others are following. Certainly needs more people working on the topic.
So what? The history of unions is workers realizing their power. We’ve all unknowingly become digital serfs for Big Tech, and that could change if we start to self organize.
Quote of the Week:
“Your first task is to find what feels effortless to you. Your second task is to put maximum effort into it.” - James Clear
That’s all for now - happy weekend everyone.
- Stephen