Being a researcher in the era of AI feels a bit like being at the bottom of a mountain. You can choose not to go up, but you can’t avoid it.

I’ve been teaching myself much like everyone else, learning as I go what works for me and what doesn’t. I have been blogging a bit about it on my substack.

One tool I have used a lot is Zerve AI. It solves lots of the problems most researchers might have around reproduction of research, data flows and much more. It’s a cool Irish startup based in the MidWest.

The founders Phily Hayes and Jason Hillary are former UL students based in UL’s Nexus Innovation Centre who have helped Immersive Software Engineering out many times, including taking students on residency, so I know they are good people taking on hard problems. AI in science is likely one of the hardest.

When I blogged about their platform they asked me to help out designing a fellowship programme for researchers using Zerve. I was delighted to, and that just went live.

Check it out here.

They also made a video about the fellowship idea.

Apply and be accepted and you get:

• Tens of thousands of euros worth of credits,

• compute support (this is incredibly important when you are doing scaled work),

• research support, and

• direct technical support.

It’s a very cool offer. The call is a rolling one, so apply away.

Apologies for the delay. My day jobs got very, very busy.

I have not come up with a name for it, so I am simply calling it This. If you have a suggestion for what This should be called, suggest it please in the comments.

Right so. My last three posts built an argument for a new kind of institution. Part 1 said the future belongs to people who convert ideas into shipped reality under uncertainty, and that these people are far rarer than they should be, especially in places like Ireland. Part 2 offered a typology of the institutions that shape their supply. Part 3 introduced what I see as a recipe to create these people, which is

Formidability = Responsibility Exposure × Feedback Speed × Network Density × Time Under Uncertainty

The multiplication is very much the point. Weakness in any one term collapses the output. You can’t compensate for near-zero responsibility exposure with abundant feedback. You can’t substitute network density with more time. The inputs are complements, not substitutes, at least below a certain threshold.

This week I want to answer the question all of this is building towards. What would an institution look like if it were designed from this equation rather than by analogy to existing ones?

Tl;dr if you’re busy and don’t feel like reading 2944 words I’m proposing an 18-month residential programme that takes 20 people selected for latent rather than demonstrated ability, and puts them under the four conditions I think actually produce formidable people. It costs €1–1.5 million per cohort. I contend a cohort of 200 truly formidable people after a decade or so would be enough to change the trajectory of the country. We are increasing the state’s ability to act under uncertainty.

Subscribe now

This is not designed by analogy

Right. Buckle up.

Most institutional proposals start as the deviation from some precedent. MIT but in Ireland. UL but without the CAO, etc, etc. This doesn’t. I’m trying to build it from first principles instead. Every structural choice, from cohort size to programme length, selection process to review rhythm to physical location is derived from the equation. If you understand the equation, you understand the institution. If the institution ever loses sight of the equation, it will drift, as institutions always do, into becoming another allocation institution with better branding.

So let me go term by term.

Responsibility exposure

By this I mean, if you don’t do it, it visibly does not happen. The felt difference, the one I described when a student has to present working software to an actual company against a deadline, is the input.

This is structured around live projects with external partners. Real organisations, communities, or markets that have a genuine stake in the outcome. Each resident carries a project that only exists if they exist. There is no module. There is no assignment. There is a thing that either gets done or doesn’t, and the world notices.

The sweet spot, as I said in Part 3, is ‘if this goes wrong, it matters, but you’ll survive.’ Too catastrophic and you freeze. Too inconsequential and you learn nothing. This is designed to keep every project in that zone—real stakes, recoverable failure modes—and to increase the stakes progressively across the programme.

Feedback speed

Feedback arriving too late to change your next action is useless. The Leaving Cert is the pathological case. Two years of work, one number, received after the learning is over. Most professional training follows the same pattern. Grant applications with six-month turnarounds. Performance reviews once a year.

This runs on a weekly public review cycle. Everyone presents progress, not a polished presentation, a genuine account of what happened, what didn’t, and what they’re doing next. External reviewers rotate. Practitioners, operators, people who have shipped things and failed at things, not primarily academics. The standard is not ‘did you do good work’ but ‘did you learn something fast enough to use it.’

This is where AI changes the design in a specific way. AI now allows participants to compress the prototype-to-feedback loop to near real-time in many domains. A policy proposal can be stress-tested against simulated stakeholder responses within hours. A software product can instrument its own failure modes from day one, or weirdly, even before. A designed object can be rendered and user-tested before a single physical component is ordered. The process should explicitly teach participants to use AI to accelerate their own feedback loops, not as a shortcut around judgment, but as a way of getting to the moments that require judgment faster.

Network density

Not the number of people you know, but the combinatorial richness of your environment. David Byrne’s scene at CBGB wasn’t special because of the building. It was special because Talking Heads, Ramones, Television, and Blondie were all playing the same small room and watching each other work.

This runs cohorts of 18 to 22 people. Small enough for genuine density, everyone knows what everyone else is working on, the conversations in the corridor are load-bearing, but large enough for genuine combinatorial richness. The cohort is assembled across domains deliberately and with diversity as a bedrock. A hardware engineer, a policy analyst, a visual artist, a bioscience graduate, a builder from rural Clare. The cross-pollination is not an accident of selection. It is a design choice derived from the network density term.

In a country of slightly over five million people a cohort of twenty formidable people propagates through the national network in a way it simply wouldn’t somewhere ten times the size.

Alumni from previous cohorts are brought back regularly. Not as mentors performing wisdom at junior people, but as peers with relevant recent experience. This becomes self-reinforcing. The alumni network is itself a source of responsibility exposure, feedback, and density for the people who come after.

Time under uncertainty: Keep going without knowing it will work.

Sustained exposure, long enough that you develop coping mechanisms, judgment, and the emotional tolerance to keep going when you genuinely don’t know if something will work.

This runs for 18 months. Not 6, not 12. The length is determined by the production function/recipe, not by academic calendars or funding cycles or the patience of programme administrators. Here is why 18 months specifically might work. At 6 months, the initial excitement has not yet worn off. You are still running on novelty, on the shiny. At 12 months, it has worn off and you face the first real test, which is, keep going without knowing if it will work. At 18 months, you have been through that test and are building something on the other side of it. That is where formidability actually forms. Not at the beginning of the discomfort, but past it.

The three phases

The 18 months are not uniform. The production function inputs should increase progressively, and the institution’s role should decrease. I’m trying to think of it in three phases.

The first four months are scaffolded carefully. Responsibility exposure is real but bounded, feedback comes primarily from the institution, network density is being actively built. By month five the institution steps back. By month thirteen it is mostly a container, not a support. Participants are running real things with real consequences. This is where the interesting things happen. It is also where most existing programmes end, which is precisely the wrong moment to stop.

Who gets in

This is where most institutional proposals make the same mistake, as I’ve said repeatedly. They design a selection process that selects for prior formidability, like, people who are already formidable and use the selection mechanism to be seen. (Good) Will Hunting knew he was special. The MIT hallway competition thing was just him putting his hand up.

This wants something harder to find, latent formidability. People who have the raw inputs but haven’t yet had the conditions to develop them. The production function tells you what latent formidability looks like. It’s not a high Leaving Cert score or a first-class degree. It’s evidence that when someone was given even a partial version of the four inputs, briefly, accidentally, by luck, then something happened. They made something. They took responsibility for something that mattered. They stayed in an uncomfortable situation rather than escaping it.

The selection process has three stages, and none of them look like a standard application.

Stage 1: The brief

Applicants are given a real problem. Not a hypothetical, a genuine open problem with no obvious answer, and perhaps four weeks to do something about it. Not to write about it. Not to research it. To do something. Build a prototype, run a conversation, test an assumption, make a thing. Make that thing.

The output is a five-minute video and a one-page document. What did you do, what happened, what did you learn. No word count. No references. No personal statement. The brief is calibrated to the applicant’s stated domain and starting point, which is where AI enters the selection process directly. AI makes it possible to personalise briefs at national scale, so a 19-year-old from a disadvantaged rural background and a 30-year-old Dublin professional and a 57 year old from Cork are all genuinely stretched and genuinely able to compete rather than one being advantaged by cultural proximity to the problem.s

tage 1 is deliberately accessible. It requires no credentials, no references, no prior track record. It requires initiative. That is the first signal.

Stage 2: The pressure test

The top 60 to 80 applicants from Stage 1 come together for a three-day residential. This is not an interview. It is a compressed version of This itself.

Teams are formed across domains. They are given a real constraint and asked to ship something by the end of day two. Day three is a public review with genuine external stakeholders, and here I’m think not panel members performing scrutiny, but people who actually care about the problem and will say so.

The selectors are watching for specific things, all of which map directly onto the production function. When something goes wrong on day one, and something always, always does, does this person seek feedback or avoid it? Do they pull the network together or retreat into their own lane? Do they stay in the uncertainty or resolve it prematurely into false certainty? Do they take responsibility or distribute it until no one is accountable? These are production function inputs, observable under pressure in ways they never are in a standard interview or application.

Stage 3: The conversation

The final 30 to 35 candidates have a long, unstructured conversation with two selectors. Not a panel. A conversation. The goal is to understand the chain of action in their life to date. Not what they’ve achieved, but what they’ve started, what they’ve broken, and what they did next.

You’re looking for evidence that the chain exists and that it’s been getting longer and harder. A 22-year-old who started something small at 16, learned from it, tried something harder at 19, and is now stuck because they lack the environment, that’s a This candidate. A 45-year-old who left a professional pathway and is trying to start something genuinely difficult, also a candidate. The age range is wide deliberately, because formidability doesn’t have a fixed developmental window.

What the selection process explicitly does not use. Points scores, degree classifications, employer references, personal statements drafted with AI assistance, or any measure that primarily captures prior allocation success. This is not for the people the system has already found and rewarded. It’s for the people the system hasn’t found yet. They may not know they are formidable, or that they can be.

Why now

AI changes the urgency of this proposal, not just the design.

Here is my best argument. AI dramatically compresses the cost of going from idea to first prototype, within specified boundaries. The rate-limiting step in innovation is therefore shifting away from technical capacity and toward the thing This is designed to build. If you can prompt your way to a working prototype in an afternoon, grand. What remains scarce is the willingness to take genuine responsibility for it, to persist under uncertainty when the first version fails, to seek feedback rather than avoid it, to stay in the problem long enough to understand it.

AI raises the return to formidability while doing nothing to increase its supply. If we don’t deliberately build institutions to produce more formidable people, the gap between those who have the capacity to ship things and those who don’t will widen. That is not a technological problem. It is an institutional one.

There is a second point I think. AI lowers the cost of building This itself. Personalised selection briefs at national scale. Near-real-time feedback loops built into project management. Alumni networks that stay genuinely connected rather than drifting into mere LinkedIn courtesy. The institution I am proposing would have been expensive to run well ten years ago. It is not expensive now.

What This costs, and where it should be

Let me be concrete, because vagueness about costs is how institutional proposals die in the corridor after the presentation.

A cohort of 20 participants, stipended (is that a word?) at roughly Irish postgraduate researcher level, so €22,000 to €25,000 per year, costs around €700,000 in stipends over 18 months. Stipends matter for equity. Asking people to give up 18 months of income without support will systematically exclude the people the selection process is specifically designed to find, and these people have both a high opportunity cost and enjoy eating regularly. Add space, programme delivery, and a small permanent staff a director with genuine credibility in building things, a programme coordinator, an operations person, and the total per cohort runs to approximately €1.4 to €1.5 million over 18 months, or roughly €1 million per year on a rolling basis once cohorts overlap.

For comparison, Research Ireland funds about 200 PhD students annually at a total cost of roughly €10 to €13 million. One cohort of This costs about one-tenth of that, and produces people specifically selected and trained to convert uncertainty into action, which no PhD programme, however good, is designed to do. Enterprise Ireland’s Competitive Start Fund gives €50,000 to early-stage startups. This costs about the same per participant, but provides a full formation environment rather than a cash injection into a company that may or may not have a formidable founder. Over ten years, running overlapping cohorts, you spend approximately €15 to €18 million and produce 200 or so people who know each other, trust each other’s judgment, and are working on hard things across every domain. Again, in a country of five million, that is not a rounding error, at all.

The funding model that makes most sense is a blend, a founding philanthropic grant to cover capital and the first two cohorts, state co-funding from year three onward, and, following the credit union model, a structural reinvestment from alumni in successful ventures contributing back to the institution that formed them. Not repayment. Reinvestment.

On location, the production function/recipe has views here too. This needs to be close enough to a meaningful cluster of tech, biotech, policy, arts, and agriculture that external practitioners rotate through easily and project partners are reachable. But it also needs enough friction and distance from ordinary life that residents are genuinely in the thing. If participants commute, if they can retreat to their normal social environment every evening, the time-under-uncertainty input collapses. The Deep Springs model is instructive. The remoteness is very much a feature, not a bug.

In Irish terms, this points toward somewhere rural or semi-rural, within 60 to 90 minutes of a significant city, with existing infrastructure that can be adapted rather than built from scratch.

On validation, a sceptical Irish reader, and there will be many, will ask whether This produces a recognised qualification. The honest answer is that it should, and the path of least resistance is a formal partnership with one or more universities that recognises it as equivalent to a postgraduate award, a professional doctorate by practice, without the university controlling the curriculum, the selection process, or the physical environment, but as a Professor in a University I can promise you that is not going to happen. The moment a university controls those things, you have an allocation institution.

What This is, and what This isn’t

This is not a university programme. It sits outside the CAO system by design, because allocation institutions tend to colonise anything they touch. They ‘must needs’ grow. It is not an accelerator, because accelerators are selection institutions for businesses rather than transformation institutions for people. It is not a mentorship scheme, because mentorship alone boosts one input temporarily while the others remain near zero, what I called diet-formidability in Part 3.

It is an integrated formation institution. The only kind that targets all four inputs simultaneously, keeps them above threshold, and increases them progressively. There isn’t one in Ireland. There are very few anywhere. It should be governed accordingly, not housed in a university, not a department of a state agency, but a chartered independent body with a Board drawn mostly from people who have built things.

All that remains is the will to build it.

I want to end where I began, three posts ago. The things around us aren’t there by accident. Someone made them happen.

The things around us aren’t there by accident. Someone made them happen. The question is whether we can deliberately produce more of those people, or whether we just wait and hope the conditions happen to be right.

The answer is that we can.

Leave a comment

Thank you to the 500+ subscribers to what is really just me talking to myself to see what I think about things.

Part 1: Everyone has ideas. The difference is some people try to make them real. Why?

Part 2: How to Make More People Who Make Things:A typology

Part 3 (this one): A simple production function for formidable people

Part 4: (next week): This is the kind of institution we should create

The last two weeks argued the future is created by people who actually get on with the job of making things. These people are formidable. Formidable means someone who can go from idea to shipped reality under genuine uncertainty. They have justified confidence in their abilities.

We need more of them. I argue there are four kinds of institutions impacting the supply of formidable people.

• Selection institutions-think the Young Scientist—identify latent formidability. (I’m never giving up the em dash folks).

• Allocation institutions—think Schools and Colleges—distribute people across roles.

• Transformation institutions—think the Scouts—increase formidability.

• Amplification institutions—think YCombinator—increase returns to formidability

Ireland today has loads of allocation institutions, patchy selection institutions, thin transformation institutions and uneven amplification institutions.

OK, you’re up to speed. Off we go with the next bit.

Subscribe now

I had a plan to outline an ambitious institutional idea but got stopped by an excellent commenter who asked what I thought the specific process for formidability formation was, which is something I hadn’t really thought through, and which really matters for the next part, so thank you very much for that.

My core argument is formidability is not personality. It is produced by repeated exposure to responsibility under uncertainty with feedback. Therefore institutions, which I think of as repeatable social technologies, can engineer it.

I have something like a production function in my mind, which is a semi-formal way of describing how inputs are transformed into outputs.

OK, enough basic economics. I’m going to suggest you increase formidability in this way. I don’t know how to typeset using substack so I’m going to block quote it.

Formidability equals responsibility exposure x feedback speed x network density x time under uncertainty

In picture terms, at least to ChatGPT, it looks like this:

I’m not here for equations Steve. Explain yourself.

Item by item, by responsibility exposure I mean asymmetric responsibility. If you don’t do it, it visibly does not happen. I don’t mean your name is on something. Think about a typical group project at a university versus what happens in the Immersive programme where a student has to present working software to an actual company on a deadline. One has diffused accountability, the other has nowhere to hide. The felt difference is enormous and that felt difference is the input. The GAA club where a 17-year-old gets picked for the senior team and has to mark someone ten years older. Nobody asked if they were ready. That's responsibility exposure in a weekend. We don’t need to boil the ocean here. Responsibility exposure has to be real but not catastrophic. The sweet spot is "if this goes wrong, it matters, but you'll survive." Deep Springs sending 19-year-olds to manage cattle at 4am is a good example. The stakes are genuine (the cows need feeding regardless of your feelings) but the failure mode is recoverable.

By feedback speed I meant the feedback arrives quickly enough to change your next action. Feedback arriving too late to learn from, or implement, is as useless as a chocolate teapot. Think a coder pushing to production and seeing their error rate in real time versus a PhD student who submits a chapter and waits four months for comments. Same effort, wildly different learning velocity.A market trader knows within hours whether their thesis was right. An academic might wait years. Competitions like the Young Scientist compress feedback loops. In street performance, which Steve Martin references in Born Standing Up, you know within thirty seconds whether your material works. That brutal feedback speed is part of why comedy produces formidable people. A negative example: the Leaving Cert, where you work for two years and receive a single number at the end. This isn't an argument against exams, it's an observation about what they don't produce.

Network density is about the probability that any given conversation leads to a useful connection, collaboration, or challenge to your thinking. Not the number of people you know, but the combinatorial richness of your environment. David Byrne’s scene idea. CBGB wasn't special because of the building. The building was a kip. It was special because Talking Heads, Ramones, Television, and Blondie were all playing the same small room and watching each other. The reason certain university departments produce disproportionate output isn't the curriculum, it's that the people in the corridor are working on adjacent problems, talking to each other constantly, and openly competing with one another. For Ireland specifically I think network denisity is where scale becomes a constraint and an advantage simultaneously. The country is small enough that networks can be genuinely dense — everyone in Irish biotech is two introductions from everyone else and let’s be honest they are likely cousins. But that density is unevenly distributed geographically and, crucially, sectorally. If you're in Dublin or Galway in tech, network density is quite high. If you're in a rural area working on something unusual like say rocketry, it can be close to zero.

Time under uncertainty is the one that separates a weekend hackathon from something that actually transforms people. The input isn’t just exposure to uncertainty, it’s sustained exposure, long enough that you have to develop coping mechanisms, judgement, and the emotional tolerance to keep going when you genuinely don’t know if something will work.

This week I spoke with two founders taking leave from Immersive Software Engineering. We talked about the value of them burning their bridges entirely and just dropping out in order to really go for their start up idea. A weekend hackathon gives you a taste. A twelve-month startup with a do or die demo day gives you the real thing. The difference is that in twelve months you hit the point where the initial excitement has worn off and you still have to make decisions without knowing if they’re right. That’s where formidability actually forms.

Military training is designed around this deliberately, in that you’re not just enduring physical stress but prolonged decision-making under ambiguity. The uncertainty in some sense is the curriculum.

The farmer lives with genuine uncertainty as a baseline. Even with a heavy use of technology, from weather, to prices, to animal health, none of it is under their control, and they still have to act every single day. There’s a reason rural communities have historically produced really resilient people. As someone said to me years ago for a difficult role I was hiring for, “wide boy wanted, country boy preferred”.

Many professional training pathways think the academy, law, medicine, accounting, are designed to reduce uncertainty as quickly as possible through protocols, precedent, and procedure. Norms are everywhere designed to protect the professional. That’s appropriate for the profession (at least before AI) but it means the training itself doesn’t build tolerance for the kind of open-ended uncertainty that formidable people navigate. The person who leaves one of those pathways to start something new often finds they have deep skill but low uncertainty tolerance.

Weakness in any one term in practice collapses the output, so for example you can have someone who is given lots of responsibility, time under uncertainty, and feedback, but with low network density, they don’t really prosper.

Formidable people need somewhere, real or virtual, to go and to be together, and bounce off one another. Network density isn't a nice to have, and when it goes to zero the other inputs can't compensate past a certain point.

One key idea behind the production function that matters for you the reader right now is substitutability.

Some inputs can replace others, at least partially. That ‘elasticity’ is central, and I think often we don’t realise that much of the institutional design we do really impacts on the kinds of substitutability between elements.

Weakness in any one term collapses the output. You can substitute between inputs to some degree, in that sparse networks can compensate with better feedback speed, less time under uncertainty with increased responsibility exposure, but there are limits. When any input approaches zero, the others can't save you.

Share

Notions: Where the Chain Breaks in Practice

Most people in Ireland go from school to college to employment without ever being given genuine asymmetric responsibility for anything that matters. You can be 28, highly educated, professionally qualified, and never once have had the experience of “if I don’t do this, it visibly does not happen.” That’s not a personal failing, it’s a structural one. Our allocation institutions are specifically designed to distribute people into roles where responsibility is managed, supervised, and shared. That’s of course rational for the institution—it reduces risk—but the production function idea I have discussed says near-zero responsibility exposure collapses the output regardless of how much you invest in the other inputs.

Ireland also has a feedback problem. The dominant feedback mechanism in Irish education is summative assessment. You do the work, you get a grade, the grade arrives after the learning is over. The Leaving Cert is the extreme case but the pattern repeats through higher education and into many workplaces. In the public and private sectors we have performance reviews once a year. Grant applications with six-month turnarounds. Planning permissions taking months/years/decades depending on their scale. The feedback exists but it arrives at a speed that makes real-time learning nearly impossible. Compare this to environments that reliably produce formidable people: startups where the market responds daily, sports where the scoreboard updates in real time, performing arts where the audience reacts immediately, art where the artist can see and feel their progress.

Network density is where Ireland’s scale cuts both ways as I’ve said. In Dublin’s tech scene, which I know pretty well, density is genuinely high. The Irish policy scene which I know very well is less than five square kilometres across, with the Department of the Taoiseach at the centre of that square.

Density is extremely uneven by sector, region, and class. If you’re a young person in a rural area fascinated by materials science or game design or fine art, the number of people within reach working on adjacent problems may be literally zero. That’s a collapsed multiplier. And even in dense environments, networks are often sectorally and sub-sectorally siloed. Academia talks to academia, policy to policy, business to business. It gets worse the finer the sub category gets. If you’re in Limerick doing behavioural economics you might well be entirely on your own. The cross-pollination that characterised Bell Labs or Building 20 at MIT is rare here.

Irish culture has a relatively low tolerance for sustained uncertainty as a life choice. (And yes, I know I say this as a tenured Professor). There are strong incentives to resolve uncertainty as quickly as possible. Get the qualification, get the permanent job, get the mortgage, don’t have notions. Each step is individually rational but cumulatively they keep the time-under-uncertainty input artificially low. Even our entrepreneurship supports tend to be pretty short-burst, a weekend here, a six-week or -month accelerator there. These give people a taste of uncertainty but not enough sustained exposure to develop the judgement and emotional regulation that formidability requires.

Failures compound

The deepest issue is that these failures compound. If you never get real responsibility, you never generate the situations that would produce fast feedback. If your network is thin, you don’t encounter the people who would challenge your thinking or offer you responsibility. If your culture incentivises resolving uncertainty quickly, you never stay in the uncomfortable zone long enough for the other inputs to take effect. Environments that provide one input tend to provide the others. Environments that starve one tend to starve all of them. This is why the occasional intervention, a weekend programme here, a mentorship scheme there, rarely produces lasting formidability. It boosts one input temporarily while the others remain near zero. It is diet-formidability.

Which is the argument for Part 4. Each institution that targets one input in isolation is fighting the multiplicative structure of the production function. The only answer is what I'll call an integrated formation institution, something designed from scratch to tune all four inputs simultaneously.

That’s next week.

Leave a comment

Last week’s piece argued for three things. First, a nation’s progress over the longer term is determined by the ability of its people to have ideas, make them real, and sell those ideas to the rest of the world. Second, the supply of individuals who can have these ideas and take them forward is limited as a share of the population. Third, there are ways you can increase the supply of these people deliberately.

These people have lots of characteristics the last piece discusses, but for a finish they all have the same features, in that they are people who can convert uncertainty into action repeatedly in some domain. Sometimes this requires what YCombinator’s Paul Graham calls being formidable, 'someone who seems like they'll get what they want, regardless of whatever obstacles are in the way', and sometimes it requires what Steve Martin, in Born Standing Up, calls staying 'charged up with delusions between moments of inspiration’. Both descriptions are getting at the same thing, which is the capacity to keep moving when the outcome is genuinely unclear.

My own experience setting up innovative new University programmes like Immersive Software Engineering and Immersive Bioscience and Biotherapeutics convinces me the solution to making more of these people lies in institutional design.

We lack sufficient amounts of repeatable social technology aimed at bringing more idea-having-and-making people, who for the sake of it I’ll call formidable individuals, forward. The good news is institutions are some of the cheapest things we can make to transform societies, especially compared to other society-transformers like wars.

So, in this post I’d like to highlight some institutions that actively increase the supply of formidable individuals, and I’d like to suggest a typology of those institutions. Then in the next piece, which is next week, I’m going to outline something new.

Subscribe now

So, hey presto, typology!

I think we need four kinds of institutions to increase the supply of formidable individuals. By institution here I mean a bundle of norms over time, so some repeatable social technology. I’m going to be a bit Procrustean in my examples here because the best institutions span categories (think about the Scouts as you read through, you’ll get the idea). Forgive me, I think it’s worth it.

Selection institutions identify formidable individuals. Usually they take the form of elite competitions like the Mathematics Olympiad, the Young Scientist, the Young Entrepreneur, Trinity’s Scholar programme. America’s Got Talent. The French grandes écoles concours. Military selection pipelines. Conservatory auditions, elite sports academies, national youth orchestras, Olympic development programmes. Sport is often one of the purest selection systems because performance is measurable and the outcomes are deeply desirable socially. (If you’re reading this and don’t believe me, come to Limerick and let’s talk about Hurling). Open source contribution rankings, Kaggle competitions, coding contest platforms like ICPC, Topcoder, Codeforces. Points based visa systems targeting high skill migrants.

Selection institutions don't just identify formidable people, they often create formidable people through the trial itself. The Young Scientist competition probably transforms a meaningful share of participants regardless of outcome. This week I got to meet this year’s winners and their families, and it was lovely. You can’t meet people like that and not be optimistic about the future.

Some selection institutions do have an issue, in that the formidable person they select already knows they are formidable. They just use the selection mechanism to put their hand up and been seen. Will in Good Will Hunting knew he was special. The MIT maths competition in the hallway was just him putting his hand up.

Allocation Institutions which primarily sort and distribute across pre-defined roles. . The dominant function is sorting, certification, and standardised human capital formation with a strong variance reduction norm. This is not a bad thing. Standardised systems produce fairer outcomes, but there is a problem here. In order to cultivate higher numbers of people who can have ideas and make things from them, we need something with uneven responsibility, risk, and failure exposure. The Irish school system is very good at producing capable students. It is less good at producing people who start things. We should not berate these systems for not doing what they are not primarily set up to do. Other allocation systems are the CAO system, professional licensing bodies, civil service graduate schemes, and national qualifications frameworks.

Transformation institutions are normally environments that measurably increase agency, execution capacity, and tolerance for uncertainty among participants. Officer training, apprenticeship and dual systems, startup accelerators, immersive programmes like the ones I have helped start here at UL. Project based schools like Olin College of Engineering, Minerva University, and Deep Springs College. These institutions deliberately design responsibility, ambiguity and interdisciplinary problem solving into the core educational experience. Deep Springs also features vastly more cows per student than other colleges and better outcomes on many measures. Coincidence? I think not. And perhaps the greatest transformation institution of them all, the Scouts. In the language of pedagogy, transformation institutions create identity transitions and not just skill acquisition.

Amplification institutions. These take what is already there and, well, amplify it. So YCombinator took John and Patrick Collison and amplified Stripe. Capital markets amplify smart people who take risk and win or lose from their hypotheses about the world. ARPA and ARIA and Research Ireland amplify innovators, sometimes over decades. Innovation clusters with really dense labour markets in some areas—think Boston and Biotech— help regions grow. (By the way I love an em-dash, have decided to keep using them.) Mechanistically amplification institutions reduce coordination costs by helping people access networks, capital, legal infrastructure, and more. They reduce the downside risks. Economically-speaking they raise the marginal product of the formidable person. Lots of the legal system exists to help here. Bankruptcy law limits downsides. You get the idea.

Share

tl;dr Steve, I’m busy, get to the point

Right. Here’s my typology:

• Selection institutions identify latent formidability

• Allocation institutions distribute people across roles.

• Transformation institutions increase formidability.

• Amplification institutions increase returns to formidability

Our problem as I have defined it is that we do not identify enough latent formidability.

We do however have loads of allocation institutions.

In Ireland we are making more of them on a yearly basis. We shouldn’t be too down on ourselves in this. Most countries over-invest in allocation and underinvest in transformation and amplification. Echoing a point Sinead O’Sullivan made in a recent series of blogs, education reform debates tend to focus on curriculum, not responsibility. Industrial policy debates tend to focus on subsidies, not coordination infrastructure.

We don’t have nearly enough transformation and selection institutions. Ireland specifically underinvests in them systematically. Consider that the major selection institutions like YSTE are voluntary and in some sense ‘outside’ the system, and you’ll get the idea.

If we take seriously the idea that societies can increase the number of people who make things, the obvious next question is design. What would an institution built explicitly to produce formidable people look like?

That’s next week.

Leave a comment

The future is created by people who actually get on with the job of making things. To quote Stripe’s John Collison, the things around you aren’t there by accident. Someone made them happen.

(In the tech-world these folks are said to have agency. I am not sure that quite gets at what I mean, but it is pretty close).

How do these people happen? Are they an accident, some crinkle in the timeline? Is a country given a genetic quota? Is it family or education or training, the economy, the availability of funds to spur action, being at the bend of a technological curve, or something in the broader culture, pure bleeding luck, or something else?

Subscribe now

Whatever circumstances make these people, they seem to share similar sets of traits.

They have ideas. They have the confidence to articulate their idea in some domain, often they have the people skills to develop that idea with others, have the technical skills to make the idea real, the ability to tell the story of the idea to a skeptical wider world, have it believe them, and then have enough left in the tank to ship the whatever it is to the world at increasing scale and complexity.

That is no joke.

That is also how the modern world is made.

If you think I am only talking about founding companies featuring software or hardware, I am not. Read the paragraph above and think about someone writing a book, or building a boat, painting a masterpiece, repairing motorbikes, a master carpenter, a politician, a builder, a developer, a financier.

Despite working across every domain you can think of these people are rare indeed.

My question is, why?

Relatedly, how do we make more of these people who can take ideas and make them real in every domain, from biotech to architecture, art and design, from civil engineering to sports medicine and from agriculture to electronic hardware?

(An AI-digression). I am an AI optimist. I think it is a tool that is going to great things for humanity. I do not think AI replaces people, or supplants the personal or societal processes I have described above. It will definitely not replace these types of people.

I think AI will help, but let’s turn AI off in our minds for a moment. Control for it. AI didn’t help the architects of Notre Dame, the guy who invented the adjustable wrench, the inventor of the theory of elasticity, or even the founders of DeepMind.

Right, back to it. Why are they rare? Can we increase their supply? I think the answer is an emphatic yes.

More concretely, all other things being unequal, how would you structure things so the supply of more people who can have ideas, make things out of those ideas, and sell them to the rest of the world increases?

It starts, I think, with confidence. Everyone has ideas. Everyone. Ideas are not a constraint. Most are admittedly bad ones but this is true of the greats as well as the rest of us. The difference is some people take ideas and try to make them real somehow. They do that sketch, plan that movie, put pen to paper on that book, start training to be a lawyer, run for local office. Whatever. They do it.

Why? They have the confidence to try this scary thing. Where does confidence come from? Prior experience of smaller acts. Small acts lead to early wins and learning from failures, which inspire slightly bigger acts which increases one’s capability to attempt still bigger acts.

Each act is best understood as part of a chain.

What breaks the chain? Three categories come to mind.

1. Permission and environment. A culture or family where permission is harder to get, or and environment favouring compliance and risk avoidance, is going to lower the number of people with the confidence to try new things out. Don’t have notions, don’t stand out, be one of the lads, you can’t be doing that, etc.

2. The gap between ideas and first actions. There is too often a gap between ‘i have an idea’ and ‘I did something about it’. The problem is often proximity to action. You might have a great idea for a product design but if you can’t get near the tools or don’t have the vocabulary to describe what you want and make it, you’re in trouble. You might have an amazing idea for a new kind of shot in your sport, but if you’re not near a pitch, you can’t test it out, the idea dies and you move on. Ideas are best expressed and tested with other people, and having role models, peers, what David calls the scene in his amazing book How Music Works.

3. Legibility and friction. In many domains, the path from ‘interested person’ to ‘person making things’ has become more bureaucratic, or credentialed, or expensive. Music is a good example of the opposite of this. You used to need a studio and a record company, now you can make amazing music from your Dad’s old laptop. Reducing friction at the entry point is one of the highest-leverage interventions. This is why libraries with 3D printers and podcasting studios and LEGO sets you can borrow are pure positive societal goods. They lower the cost of entry for idea-having and idea-making, but if you’re 40 km from a library that has one, you’ll never get to use it.

Every hotel, park, and railway John Collison talks about started as someone's scary idea. The person who made it happen wasn't a different species. They just had the confidence from a prior smaller act, access to the tools, and an environment that didn't talk them out of it before they started. That's not magic. It's infrastructure. Physical, cultural, institutional. We can build more of it. We should.

The good news is that none of this is mysterious. Every hotel, park, and railway Collison talks about started as someone's scary idea. The person who made it happen wasn't a different species. They just had the confidence from a prior smaller act, access to the tools, and an environment that didn't talk them out of it before they started. That's not magic. It's infrastructure. Physical, cultural, and institutional. We can build more of it.

Confidence is built, not born. Chains of action can be started at any age and they compound. Friction can be reduced. Environments can be redesigned. The question isn't whether we can make more of these people, like we obviously can, it's whether we have the will to structure things so that the kid with the idea and the adult with the sketch and the retiree with the prototype all have a shorter distance between thinking it and trying it.

The future, as ever, belongs to the people who make things.

Subscribe now

The only question then is how to make more of them. In the next post, I’ll talk about how some of our institutions accidentally do this, and what we can do to change them now that we have AI to help and hinder us.

Share

tl;dr Intangible assets, from software and brands to organisational systems now matter more for production than physical capital. This shift is to the advantage of large firms.

The numbers in our new paper tell a story of structural change:

• Intangible capital’s contribution to output surpassed physical capital in 2008–2019

• Digital-intensive industries show two times higher benefits from investing in intangibles

• A 10% increase in intangibles boosts top firms’ profit margins by 3.1%

Intangibles are “non-rival”, meaning when you deploy them once, you can use them everywhere. A factory can only be in one place but a new algorithm can be run across your entire global operation at near-zero marginal cost.

What is surprising about the paper is that it is not patents or flashy acquisitions driving most of the advantage. It’s organisational capital, super boring stuff like management systems and IT infrastructure like MS Teams that scales best.

Yes, MS Teams might be a company’s saviour, it’s real moat.

(This post is not endorsed by Microsoft).

So what? Winner-take-all dynamics may be baked into how modern production works. The bigger you are, the more you can leverage intangibles, the bigger you get.

I’ll explain the paper in detail or you can skip down to the end and get the summary, up to you!

Subscribe now

Something fundamental has shifted in how companies create value and it has been happening for decades, largely out of sight.

Our new study spanning fifty years of data reveals a striking transformation. The assets you can’t touch like software, brands, organisational systems, R&D, and intellectual property now matter more for production than the physical factories and equipment that once defined corporate power.

The Numbers Tell the Story

Together with Dr Michael McMahon from Enoda Health and Prof. Eleanor Doyle from UCC, we analysed firm-level data from 1970 to 2019, tracking how the building blocks of production have changed across seven complete business cycles.

Here is what we found.

The output elasticity of intangible capital, essentially how much an additional unit of intangible capital contributes to production, rose steadily from near zero in 1980 to surpass physical capital in the most recent cycle. At the same time the contribution of variable inputs (like materials and direct labour) declined substantially.

We did loads of statistical tests to satisfy ourselves this isn’t just some accounting curiosity. The shift is multi-decadal, and real. Now we have to ask ourslves the Smart Brevity question, which is:

Why Does This Matter?

The properties of intangible assets differ dramatically from physical ones. A factory can only be in one place at a time. But a piece of software, a brand identity, or an organisational system can be deployed across multiple divisions, geographies, and product lines simultaneously, often at near-zero marginal cost. Readers of Ben Thomson’s work will recognise this property in just about everything the modern economy does.

This is what economists call “non-rivalry.” When your competitive advantage comes from assets that can be replicated infinitely within your organisation, the economics of scale change completely.

We found larger firms exhibit lower overhead costs as a share of revenue and higher profit margins. The difference in overhead share between the largest and smallest firms has grown substantially over time. This suggests scale advantages are increasingly tied to intangible-intensive production.

The Digital Amplifier

Digital technologies, which appear in successive waves in our dataset, appear to supercharge these dynamics. Industries Michael classified as “digital-intensive” showed intangible coefficients roughly double those of less digital-intensive sectors in the most recent cycle.

We used a taxonomy capturing multiple dimensions of digital maturity like ICT investment, intermediate purchases of digital goods, online sales, and robot deployment to classify different industries. The pattern is quite consistent: greater digitalisation correlates with greater importance of intangibles.

This makes intuitive sense. Digital assets are particularly non-rival. An algorithm or software system can be copied, stored, and distributed at essentially no cost. The more digitalised a firm becomes, the more its intangible assets can be leveraged across its operations.

Winners Take More, or more accurately, winner takes most

Perhaps the most striking finding concerns who benefits from these shifts.

The marginal impact of intangible capital on profitability is significantly higher for top-performing firms. When the researchers compared the top eight firms by sales in each sector against everyone else, they found that intangibles boost the profits of leaders far more than laggards.

This effect has grown stronger over time. And it’s not just a “big fixed costs” story. Physical capital doesn’t show the same pattern. Something about intangibles specifically favours scale.

Digging deeper ,we decomposed intangible capital into three components:

• knowledge capital (capitalised R&D),

• organisational capital (investment in systems, processes, and capabilities), and

• balance sheet intangibles (primarily goodwill from acquisitions).

It turns out the scalability premium comes almost entirely from organisational capital. The chart below shows the evolution.

So yes, MS Teams FTW.

Share

The Organisational Capital Puzzle

This finding deserves attention. It’s not patents or acquired brands driving the performance gap, it’s the super unglamorous stuff: management systems, processes, IT infrastructure, and hard to measure but important organisational capabilities.

A version of the argument runs like this: Organisational capital is increasingly non-rival due to digitalisation. A financial management system, once developed, can serve an entire global organisation. A set of processes, once codified in software, can be deployed to new markets at minimal cost.

This supports theoretical models suggesting that falling costs of spanning multiple products and markets give larger firms systematic advantages. As digital tools make it easier to coordinate complex organisations, the benefits of scale increase.

(I have just thought about measuring a nation’s soft power as a type of relational capital. We could use the same measurement schema. One for another paper/substack maybe").

What It Means

These findings have implications across several domains.

For competition policy, they suggest that rising concentration may be partly driven by genuine efficiency gains from intangible-intensive production, not just market power. Distinguishing between the two becomes critical and quite difficult.

For corporate strategy, they highlight the importance of organisational capabilities. The firms pulling ahead aren’t necessarily those with the most patents or the biggest marketing budgets, but those who can effectively deploy organisational systems across their operations. This is quite poorly studied, certainly in economics. The work Nick Bloom leads at Stanford might be a rare outlier.

For investors, we suggest looking beyond traditional measures of capital intensity. The ability to scale organisational capabilities may be a better predictor of future performance than physical asset bases. That is extremely hard to measure at the moment.

For workers and policymakers concerned about inequality, the paper points to mechanisms that may be reinforcing winner-take-all dynamics. If intangible-intensive production systematically favours larger firms, we may be in a self-reinforcing cycle of concentration.

Limitations and Open Questions

No study is perfect. Our study relies on publicly listed firms, potentially missing dynamics among smaller private companies. The digital intensity measure captures sectoral differences but not firm-level variation in digital adoption.

Most importantly, and this is key, our research documents correlations and trends rather than establishing clear causal mechanisms. Whether intangible investment causes scale advantages, or whether larger firms are simply better positioned to invest in intangibles, remains difficult to untangle.

The precise channels through which organisational capital creates scalability advantages also need further investigation. What specific types of organisational investments matter most? How do they interact with digital technologies? At what point do coordination costs offset scalability benefits? All this is for further work.

Too long, I didn’t read all of that, what does it mean in 3 and a bit bullet points Steve

• We’re living through a transformation in the fundamental nature of production. The things that make firms competitive are increasingly invisible residing in code, processes, and capabilities rather than bricks and machines. This is where AI’s value is truly coming from.

• This shift has been gradual enough to escape notice in any given year, but the cumulative change over fifty years is dramatic. The economy that young people are entering operates on different principles than the one their grandparents knew.

• Understanding these dynamics seems essential for navigating what comes next, especially when it comes to AI, whether you’re running a company, investing in one, regulating them, or simply trying to make sense of why the economic landscape looks the way it does. That’s what our paper looked at.

The invisible assets have taken over. The question now is what we do about it.

Leave a comment

Why it matters. Dani Rodrik’s trilemma says you can’t fully have democracy, nation‑state sovereignty and deep globalisation all at once. It is the cleanest model for understanding Ireland’s biggest choices right now. We’ve chosen democracy and globalisation. It’s worked brilliantly in a stable world. It becomes a bind if the world fractures. I have three ideas on what we can do about it, policy makers can pick two.

Subscribe now

A model to fit the moment

After a previous post in which I suggested the book Vandalising Ireland lacked a real model to understand the Irish economy, a thoughtful commenter asked what model I would suggest.

Fair enough.

There are a few candidates but the model that most fits the moment is Dani Rodrik’s policy trilemma.

I’m going to explain the whole thing now but you can skip right down to the end of this section if you like and just get the results.

Economists love trilemmas. I don’t know why. We just think they are cool.

Anyway, Rodrik’s framework has three constraints (hence the tri). Rodrik argues democracy, national sovereignty, and deep economic globalisation cannot be fully achieved simultaneously.

In the paper introducing the idea Rodrik projects 100 years into the future when some of these constraints might bind more than others.

Here is the extremely MS Word-drawn diagram Rodrick used to explain his idea:

The top figure is the first big trilemma anyone in economics classes learns. The impossible trinity. Rodrik describes it this way:

If a government chooses fixed exchange rates and capital mobility, it has to give up monetary autonomy. If it wants monetary autonomy and capital mobility, it has to go with floating exchange rates. If it wants to combine fixed exchange rates with monetary autonomy (at least in the short run), it had better restrict capital mobility

This will matter later.

The second super crappy diagram is the one I want you to focus on.

States can choose only two elements: democracy (which he calls mass politics), sovereignty of what he calls the nation state, and globalisation (which he calls integrated national economies).

Ireland has clearly chosen democracy and globalisation, meaning the increased interconnection of markets for capital, labour, and ideas. Rodrik calls that the golden straightjacket. Here is how he puts it:

The essential point is this: once the rules of the game are set by the requirements of the global economy, the ability of mobilized popular groups to access and influence national economic policy-making has to be restricted.

Because your economy needs to integrate into the global one, any government of left or right needs to pursue tighter money, smaller government, lower taxes, more flexible labour legislation, deregulation, privatisation, and openness. Ideology, individual circumstances, indeed even the will of the mass of the polity becomes subservient to making sure the nation state is as integrated as possible.

This is part of Eoin Lenihan’s argument in Vandalising Ireland. He misses the other two sides of the trilemma of course.

Problematic aspects of the golden straightjacket are clear to policy makers and have been for a long time. When his ramblings were clear at all, Trump’s critique of the Davos set last week was how anti-worker it all was. He was correct that low-wage workers in the US lost their jobs and their communities once capital was offshored to Asia. He was correct that the US incurred huge costs to anchor the worldwide systems of defence, finance, and trade. He didn’t bother to mention how much it benefitted from the systems it oversaw of course.

At his speech last week in Davos Mark Carney called the golden straightjacket out as well. Carney’s speech will be the opener for a dozen books in years to come, and I imagine this will be the quote opening most of them.

Over the past two decades, a series of crises in finance, health, energy and geopolitics have laid bare the risks of extreme global integration. But more recently, great powers have begun using economic integration as weapons, tariffs as leverage, financial infrastructure as coercion, supply chains as vulnerabilities to be exploited.

You cannot live within the lie of mutual benefit through integration, when integration becomes the source of your subordination.

States like the US and China have, can, will, and would use their interconnections against other states. Henry Farrell has called this Weaponised Interdependence. Carney’s argument, that states must diversify from their dependence on the US, is very well made and something Ireland should consider seriously, both on its own and as part of the EU. The diversification away from the US as hegemon may end up being perhaps Trump’s greatest legacy.

Rodrik’s ideas have been tested on real world data. Funke and Zhong found strong evidence the trilemma exists. Ito and Aizenmann found the trilemma exists for developing countries, but only a dilemma, between sovereignty and globalisation existed. Their work is shown below. The other papers I found looking at the problem aren’t very good.

That was too long, I’m not reading all that, give me the highlights Steve

• The classic macro trilemma (“impossible trinity”) says you can’t have all three: a fixed exchange rate, free capital movement, and monetary autonomy. Pick two.

• Rodrik’s policy trilemma says you can’t fully achieve democracy, nation‑state sovereignty, and deep globalisationat the same time. Pick two, the third constrains you.

• Ireland today is the corner solution of Democracy and Globalisation called the “golden straightjacket.” Rules of the global game discipline national policy, force tighter money, lighter regulation, credible rules, openness to capital, talent, and ideas.

Finally, evidence, briefly: Empirical work finds the trilemma shows up in real economies; for developing countries it can look like a sharper sovereignty vs. globalisation trade‑off. (There is nuance: results vary by period and method.)

Share

So that’s the framework. Let’s apply it today’s problems.

Today the economic model Ireland have relies on openness to mobile capital, regulatory predictability, and legal alignment with supranational regimes like the EU, WTO, and others.

This is not an accident. It’s a logical consequence of prioritising globalisation within a small open economy. Sovereignty over things like industrial and labour market policy, and macro stabilisation is limited, but it is important to say each limiting took place with the consent of the governed via referendums.

Sovereignty is not eliminated. Ireland still has powers over her taxation policy for example, but can’t choose a set of workers’ rights below what the EU mandates, and can’t subsidise domestic industries beyond what the EU mandates under state aid rules.

We are in the golden straightjacket.

If globalisation falters because great powers balkanise, all of a sudden Ireland faces a forced trade-off between preserving democracy and preserving sovereignty. You cannot keep both in their current form without rebuilding the economic base.

The first stage effects would be fiscal and employment related, and those are obvious.

The second stage would be institutional. We derive exchange rate credibility from the euro (remember the first diagram? Scroll up to it now, take a look a the first crappy diagram. We have chosen fixed exchange rates and capital mobility). We derive industrial policy discipline via EU state-aid rules, corporate governance via global firms. If globalisation weakens, we have to internalise all those constraints, which will be hugely costly.

The third stage is the alteration of our political economy. Growth via openness and highly progressive policies have been the order of the day since the 1990s. Absent that, politics becomes much more zero sum, much less focused on distributional issues. Housing, migration, and regional inequality could harden into identity-linked conflicts. Democratic legitimacy then depends on a shrinking delivery capacity. Any delivery weaker than voters expect pushes them toward either technocratic insulation or populist sovereignty. Both erode the current democratic settlement.

So what should we do about it?

The Rodrik framework gives you three answers, and you can pick two them if you are the policy makers of the day.

1. France in the 1970s. Re-sovereignisation (have I just made that word up?) with managed democracy. Stronger industrial policy, strategic protection, higher domestic taxation, tighter migration control. Democracy remains but with more constraints and elite coordination. Inflation becomes the key variable to manage.

2. Australia in the 1990s. Democratic strain with continued openness. Ireland tries to stay globally integrated despite weaker globalisation by competing much harder on tax, regulation, and labour flexibility. This preserves openness but hollowes out democratic choice as policy becomes credibility signalling to mobile capital. Likely quite unstable.

3. Taking Carney seriously. Economic diversification to relax the trilemma. Reduce exposure to footloose capital by building indigenous scale firms, which is very hard but do-able, deepening EU-level fiscal capacity, and embedding multinationals more locally through supply chains and co-investment. This does not escape the trilemma but shifts the constraint. It is slow, technically demanding, and politically under-rewarded. Likely quite stable if the polity buy into it.

Finally, if globalisation fragments rather than collapses, as Carney suggests it might, the EU becomes the decisive arena. Ireland’s real choice then is not sovereignty versus democracy, but national democracy versus pooled European sovereignty. The trilemma does not disappear, it gets rescaled. The most ambitious Irish people fly to Brussels, and live their lives there. The end.

Next week: Why the economy now runs on organisational code.

Leave a comment

Late last year I read Neal Stephenson’s Seveneves. The book starts with something blowing up the moon. Then the world ends about two years from the day the moon shatters.

Stephenson majors on the orbital mechanics, the engineering challenges of moving thousands of people into space while telling a great story, which includes a lot of politics, and a jump 5,000 years into the future. I highly recommend the book.

What Stephenson doesn’t spend much time on, because I think it would make an already very dark story even darker, is the economics of it all.

That’s where I come in.

Often to see something’s importance you have to imagine its lack. So let’s blow up the world to see the importance of the economic structures that bind us here in the real world.

For science.

Or substack. Or whatever. It’ll be fun.

Subscribe now

Billionaires go boom, or rather, splat

In Seveneves, once the moon shatters, the economy built on tomorrow evaporates.

Much of our prosperity is a story about next quarter and the next 30 years. If the story ends in two, the prices we trust are revealed as beliefs about time, not about needs. That’s a useful mirror for today.

A billionaire, for example, is a pivotal character early on in the narrative. But the day after you credibly know the world is going to end, is a billionaire really a billionaire? After all, they have their wealth tied to the stock of a company, which holds value because of its future performance. Bonds, insurance policies, pension funds, all become worthless overnight. What is the value of stock a day after you come to know the world is going to end? The entire architecture of financial capitalism, built on the assumption of continuity, simply evaporates.

Think about it:

• Billionaires aren’t billionaires if equities have no future cash flows.

• Bonds, pensions, insurance, promises dated beyond T+700 days, turn to dust.

• Currency limps along as a short-term coordinator, then a sack of rice beats a stack of euros.

• Schools will probably empty, except for crash programs feeding the space mission.

The end of education

Education would essentially end for most people. Why would a teenager attend high school if the world ends before graduation? Parents might try to maintain normalcy for young children, but institutional education collapses. Universities would empty except perhaps for those training people for the space mission. Human capital investment becomes pointless unless it serves the immediate survival project.

The death of finance

Essentially every financial product with a time period of more than two years ends, which creates liquidity problems everywhere that will certainly not matter. Who is filing accounts for the year the world is ending? Who cares about national debts or, for that matter, sovereignty itself?

Currency itself becomes interesting. ‘Public’ money derives value from faith in governments and future economic activity. In a two-year scenario, you'd likely see an initial flight to "hard" assets like food, fuel and tools, followed by a strange levelling. Money might retain some function as a coordination mechanism for those first months, but its value would decay rapidly. Governments might try to maintain it through force, but why would anyone work for euros they can't spend in three years?

In Adam Fergusson’s book ‘When Money Dies’ the Weimar Republic's hyperinflation in Germany from 1918 to 1923 was a little bit like this. Essentially no one believed in the state anymore, and so the currency sort of, stopped. In the Irish example of 1970 when banking services stopped for six months, community (or really, pub-based) banking took over, so perhaps systems of credit and debt might devolve, but that would be it. Gold, silver, copper, pretty much every market for anything would freeze.

Share

Survival of the technocrati-est

Just like during COVID, the societal ranking of the most valuable people would shift dramatically. In the normal economy, a derivatives trader or corporate lawyer might earn more than a machinist or farmer. That inverts instantly. Practical skills like growing food, fixing things, building things, providing security would all become paramount. Abstract knowledge work becomes nearly worthless unless directly applicable to survival or the space project.

The only people who really matter are the ones going into space. And who are they? A lot of -ists. Machinists, Physicists, Biologists, agronomists, folks like that. People with real expertise, of working age, with far more women than men needed.

You’d likely see extreme bifurcation in the real economy, because those selected for or working toward the space project would live in a kind of mobilised war economy - highly organised, resource-prioritised, focused on the mission. Everyone else would face three outcomes: a collapse into subsistence survival, violent anarchy, or, just keeping things going because there’s nothing better to do, but I can’t see security guards standing some place earning a wage when they have less than 700 days to live. Perhaps local communities trying to maintain order and dignity in their final months, while others descend into Hobbesian chaos.

The darkest economic question: what happens to anyone not useful to the survival project? In a normal economy, even people who don’t “produce” much receive some care, children, elderly, disabled people, those with chronic illnesses. But in a world ending in two years with limited spots on the international space station, the economic logic becomes brutal. They consume resources but can’t contribute to the only goal that matters. Societies would face horrifying triage decisions, most acutely in who to nominate to send to space.

When I started thinking about it, I quickly realised food production would get particularly grim particularly quickly. Agriculture today is an elegant, fragile machine: diesel, spare parts, fertiliser, credit. Snap those links and you get hoarding, empty shelves, and very likely state control of farms and logistics. Markets freeze; force fills the gap.

Coordination needs a story

This is all really bad if you need everything to keep going, as the governments of the world do in Seveneves. They have to boost as many people as possible into space. Food, resources, and especially space-faring technologies need to keep on being developed and deployed. A world looking forward to nothing would not help do that.

A story is needed, so the governments of the world focus everyone on working for the next two years to preserve ‘our inheritance’. A press conference is organised where the story is spun: We all have to accept we are individually going to die, but that those who make it into space will carry our cultures forward, and we as a species need to maximise the chance of that happening. That is, the governments of the world tell the people a story to keep them behaving more or less as if things are going to keep going as normal, but just for someone else.

This is a very clever part of Stephenson‘s book. To keep factories humming for a future most won’t see, governments sell “our inheritance”. Work now so humans and your culture have a shot.

Narrative becomes the incentive mechanism when profit can’t.

What the apocalypse reveals

The uncomfortable truth is that we’re already living in a slower version of this scenario. Not two years, but we face existential threats operating on the same logic stretched across decades. Climate change, biodiversity collapse, aquifer depletion, and topsoil loss all create the same tension between immediate economic incentives and long-term survival.

The difference is time horizon.

In Stephenson’s world, governments could coordinate around a shared story because the threat was immediate and undeniable. The moon was sitting in the sky, shattered. Everyone could see it.

We lack that forcing function for gradual catastrophes. We know coral reefs are dying, ice sheets are melting, and feedback loops are accelerating, yet we continue pricing 30-year mortgages and quarterly earnings as if continuity is guaranteed. The market, it turns out, prices in everything except the possibility that the market itself might end, which is a curious kind of market failure.

This is at least partly a coordination problem. In the two-year scenario, the governments of the world could credibly promise that working toward “our inheritance” mattered because everyone faced the same deadline. With climate change operating over decades, economic actors rationally exploit resources today because they can’t trust others to restrain themselves and the worst effects likely won’t affect decision-makers anyway. The tragedy of the commons writ global. Why sacrifice when competitors won’t? Why invest in sustainability when shareholders demand returns this quarter?

In The Ministry for the Future, one of the key mechanisms, and without spoilers, a fairly gruesome mechanism, used to stop the climate crisis is making decision makers feel the effects of their decisions.

The thought experiment reveals something deeper about value itself. In the apocalypse scenario, “real” value, food, fuel, working tools, practical and technical knowledge, separates cleanly from paper and book value. All those derivatives, insurance policies, and stock portfolios evaporate because they were always just shared belief backed by assumptions of continuity.

When continuity ends, the belief dissolves.

This should terrify us because much of what we call “the economy” is exactly that: shared belief rather than material reality. As I tell my students in UL every year, a forest being cleared for timber increases GDP. Depleting an aquifer shows up as economic activity. Burning through fossil carbon counts as growth. Our measures of economic health often move inversely to actual wellbeing and ecological stability. We’ve built an entire system that treats the drawdown of natural capital as income. One of the projects I worked on for several years, INCASE, looked at ways to introduce natural capital measurement into national accounting, so if you destroyed an ecosystem making some timber, modified GDP would fall, and not rise.

The people who matter most in Stephenson’s book, the ones who actually keep humanity alive, aren’t the finance bros or the lawyers or the economists. They’re the agronomists, the machinists, the people who know how to make things work.

In our much slower apocalypse, we face the same inversion of value, but it could be argued we’re refusing to acknowledge it. We still pay derivatives traders more than the people who grow our food, repair our infrastructure, or understand our ecosystems. We act as if the paper economy is real and the physical world is just an input, and that is a pricing problem.

What would it mean to organise our economy around intergenerational preservation?

Not as virtue, but as the organising logic. In the two-year version of Seveneves, the reframing happens automatically. The only payoff left is cultural survival so people work toward it. Our timeline is slower, which means we have to engineer the same alignment, by pricing natural capital properly, using policy to ensure today’s decision-makers feel longer-run costs, and rewarding the skills and systems that keep us fed, powered, and repaired.

The real task is separating “real value” from paper value while there’s still time to build something durable. Can we redesign incentives so survival is the rational choice without waiting for a shattered moon to force the point?

Leave a comment

Debates about Ireland’s economic future are rarely disagreements about numbers. They are disagreements about structure, about what kind of economy Ireland actually has, and who it ultimately serves.

The island of Ireland can be thought of as having three interacting economies, and their interactions give the macroeconomy of the island its structure. It is easiest to see the structure of the three economies by looking at employment levels, though plenty of other measures would reveal the same outlines.

The three economies on the island are the economy of Northern Ireland , where employment is dominated by the British state and small enterprises, the domestically-owned economy of the Republic, where employment is dominated by the Irish state and many small firms, and the multinational-owned economy headquartered in the Republic, where employment is concentrated in some of the largest firms on the planet.

Productivity levels vary wildly between the various public and private sectors, and between the domestic and foreign-owned sectors. Taxes paid by different types of sector vary. The fiscal vulnerabilities paper published by the Department of Finance is clear on why and where sectoral strengths translate into macroeconomic vulnerabilities.

The Minister for Finance has been at pains, in official reports going as far back as 2022, 2023, 2024, and 2025, to point these vulnerabilities out. The Programme for Government, on page 18, is very clear about what the government would do if things went wrong. Recently a medium-term fiscal strategy was published setting out a plan for the coming years which discusses these vulnerabilities as well.

The basic policy logic is that the economy’s structure creates immense upside for current and future citizens but also exposes obvious risks. Government plans to bank the upside by running persistent surpluses while still meeting Programme for Government commitments, financing about €7 bn of extra annual spending alongside reforms to deliver those promises.

State spending will roughly triple in the space of three governments. That is unprecedented even with the needed spending for COVID, the Ukrainian war, the migration crisis, and the cost-of-living crisis factored in. What is often missed is the fiscal reality that COVID-related spending is still with us, migration-related spending is still with us, and targeted cost of living supports are still with us.

Policy created this structure. Policy is being used to manage it. When conditions change, policy will change the structures to meet these new conditions.

All of which brings me to a book written by Eoin Lenihan I was given this Christmas called Vandalising Ireland: How the Government, NGOs, Academia and the Media Are Engineering a New Globalist Ireland.

As an academic of 20 years or more who has been appointed to various state boards and bodies and advised government in various ways and even been on the board of a charity, I found the book a fascinating blancmange of some interesting ideas peppered with poorly argued points that, somehow, rang a series of true notes for me.

I’ll describe these below.

Subscribe now

There are a few biographical sections in the book that can be skipped and I recommend you do. The writing is often over-heated, but it is a polemic. I don’t castigate Fintan O’Toole for it, I won’t castigate Eoin Lenihan.

Lenihan’s basic argument is structural, indeed almost Marxist in the Anwar Shaikh school of Marxist accumulation literature.

It goes like this.

Elite structures, from government, to the media, to non-governmental organisations (NGO)s, to academia ,privilege the interests of foreign-owned companies and the cadres who serve them. Staff circulate between NGOs, academia, media, and government advisory roles, producing ideological homogeneity. I guess this is me and those like me. The result is a sort of policy monoculture that justifies whatever it is foreign-owned companies want to do.

The companies’ interests are of course in accumulating capital via the exploitation of labour, necessitating increased inward migration, and encouraging more profits via highly compliant accounting rules. The result is a privileging of their interests over the ‘true’ Irish people, who are not those people, and who suffer resource-constrained lives, especially with respect to housing, with their mother-culture diluted as a byproduct of all this migration.

Here is how Lenihan puts it on page 107, in an argument he repeatedly restates.

For too long, Irish people tried to escape what they saw as the backward elements odf their identity or they simply didn’t need to consciously reflect on it. And now, as Ireland burns amid unprecedented demographic change and political parties close ranks to ensure the country’s rapid integration into the global economy, a group of well-funded NGOs and academics have stepped into the identity gap to forge a new, post-nation globalist identity for Irish people.

How the four pillars supposedly interact with one another

Once activated by the presence of multinational interests, the circuit goes as follows. Academia supplies willing and compliant labour for the multinationals and a tolerated idea set, outside of which one cannot stray or be called far left/far right and perfoce get ignored as a crank. The state supplies a compliant ruleset consistent with the interests of its principle funders, the foreign-owned corporations. NGOs are basically a parallel system for arbitraging policy ideas into the government ruleset without needing to go through the filter of democratic argument. Social justice-type NGOs, funded mostly by the state, are a key target. They are agents not of counter movement or critical perspective but mere agents of a new elite consensus. Media function merely as a narrative enforcement mechanism, again funded mostly by the state, which in turn is funded mostly by the foreign-owned companies.

So, for a finish:

• Lenihan argues the result is ever-weaker democratic oversight, a diluted native-born population and culture, weaker domestic industries, and a loss of sovereignty.

• The structure defines the incentives of groups of actors which shapes the outcome.

• Legitimacy flows forward through institutions, while accountability does not flow back.

• Each actor gains insulation from voters by passing responsibility downstream and justification upstream.

• In the end, the source of funds, the multinational, controls.

The argument as I read it is all about how ideas traverse a circuit between different institutions. These ideas help one group out at the expense of everyone else by enabling a great transformation of the underlying structure of a society and an economy.

I was reading along, ploughing through steaming phrase after steaming phrase, and waiting for a long discussion applying Karl Polanyi’s ideas. He’s the obvious person to ground Lenihan’s ideas in a serious framework where ideas influence everything. For a brilliant modern version, read Cornel Ban’s work. I got none.

Polanyi argued liberal market utopianism undermines democracy by subordinating politics to market imperatives. Lenihan argues more or less the same thing. Democratic choice in Ireland has been hollowed out because in the end, choose who you like, they’ll support the foreign-owned company’s interests at the expense of ‘native Irish stock’. There is a difference in causal attribution. Polanyi blamed the market system and its intellectual defenders. Lenihan blames the four-pillar institutional bloc I described above where NGOs act as legitimising intermediaries between global economic pressures and domestic politics.

Lenihan offers anecdote as data far too often for this thesis to hold up. It’s grand, as I said, as polemic. Mostly I read it as entertainment. But it is not scholarly. There is no organising framework here, most of the references are to blogs and newspaper articles.

If you already think the world is designed in the way Lenihan suggests, then you’ll find lots to confirm your view within the covers of the book. If you don’t think this way you’ll find it objectionable and probably say mean things about the guy, but I have to say the hand wringing I have read about the book is not justified by even a single read of it. It’s a polemic, you can read it as an argument and not end up either a right or left wing headbanger, or as someone who likes or dislikes migrants or whatever. It’s one argument among many. I have read better, and read worse.

Lenihan repeatedly tells us Ireland is burning. That syllogism comes from trying to connect the people who burned the big houses of the landed gentry before during and after the war of independence with the people who, much more recently, burned the centres housing migrants. There have been 38 incidents like that since 2018. Lenihan describes the burnings of a century ago as ‘icons of revolutionary mythology and ‘skeletal trophies’. It’s easy to focus on the buildings. Though most of the old big houses were burned after being abandoned, people like Cork’s Edith Mills died in hers. Though few people have been injured in the burnings in the 2020s, what has happened to some families has been by any standards appalling.

For me the biggest problem in the book, after its framing of migration as the key policy problem, is its partial-equilibrium thinking. I can generate many more circuits of ideas that fit the pattens Lenihan describes. You shouldn’t really be able to do that.

Here is a mutually-enforcing partial equilibrium with the same outcome but without Lenihan’s globalist push. It’s not one I necessarily believe deeply in, it just has the same amount of validity and it is much simpler: Institutional drift under international constraints plus technological change. It goes like this. Post financial crisis Ireland had huge budget deficits and capital deficits to close, creating housing and infrastructure crises. The labour-demand profile of multinationals changed as the underlying technologies enabling them changed. Academia stayed largely the same. Media shrank under sustained pressure from technological change. NGOs, well, NGO’d. They have never had much influence in my experience, and never will. Ditto for Academia, which is too often a study in irrelevance. (Most academics wish they had the influence Lenihan ascribes to them, and most professors wish their students read their syllabi as closely as Lenihan read UCD’s Dr Ebun Joseph’s one on Black Studies and CRT). The media and the political system are deeply entwined but this is not an Irish problem, this is a feature of all political systems. Politicians simply give people what it is they think they’ll want.

Result? Roughly what we see now but the causal circuit starts with the EU constraining policy choices and deep deficits which take lots of time to close. Add in technological change and you have today’s set of outcomes without all the race replacement/dilution stuff.

What I liked

What I found convincing is the description of the set of ‘accepted’ ideas in these groups, which as I have said I have spanned in my time in Ireland. What I did not see was Lenihan’s theory of how these ideas change, or are incorporated, into the system. In my experience most people in Ireland in the elite bundle, myself included, are centre-left, meaning to me at least, all things being equal, we’d plumb for a progressive outcome that expanded the state, especially in social rights terms. This centre-left bias is easily explained by Lenihan’s argument—everyone in that nexus stands to gain personally if the state expands. A little example: I have attended most of the National Economic Dialogues since it was set up as a Rapporteur. Not once did a lobby group call for spending to fall, and there was usually no representative from the private sector funding all of this at the event.

Lenihan misses a trick in misunderstanding the speed at which something can go from ‘not accepted’ to ‘everyone always knew this was true’ because he lacks a theory of how the system changes, beyond simply pivoting to serve the needs of the globalist elite in a given moment.

Another, again simpler thesis, is middle class people think like middle class people. To quote Kevin Kelly, there is no ‘them’ there.

Share

What I did not like

Often the polemic, which is enjoyable enough as such, clashes too much with the facts established in the literature, and Lenihan’s book comes a-cropper. Across most of the substantive chapters I mostly found myself checking the evidence for each statement, and quite often found a blog.

An example. Lenihan argues up until 1997, the FDI economy worked to the benefit of the Irish people. What’s the evidence for that statement other than less of a need for inward migration for workers?

Then on page 251, we find this argument, “[r]ather than providing favourable conditions to multinational companies that squat in Ireland for economic benefits while recruiting from outside, all companies should be invited on the basis that they fit the current worker profile of the nation.”

The basic logic here is that the FDI system did not need excess inward migration to meet its labour market needs, after which something changed and foreign-born workers were required in droves. The thinking is partial, in the sense of partial equilibrium. To get a good sense of the labour market dynamics, there are excellent papers Lenihan could have consulted, but sadly none are cited. The work of Alan Barrett and colleagues from the ESRI looking carefully at migrant labour market outcomes from about 15 years ago would have been really useful to Lenihan’s argument in some cases, but would largely disconfirm it.

A structural explanation of change needs a structural set of variables and a model of how they hang together when things do change. Lenihan’s book has the variables—the NGOs and all that—but lacks the model and the theory of change.

It is why his policy prescriptions are doomed to fail.

Take any citizen of Ireland and offer them a simple policy of managed migration tied to housing supply and they’ll likely say sure thing, sounds great. Why ask someone to come to a country they will struggle to find a place to live? Then suggest that the health and food systems staffed by these people would creak and fall over, and people might think again.

Take any citizen and offer them an Ireland-first industrial policy in a tweet and they’ll say sure thing, sounds great. Then say that means limiting the taxes that power the state and supply them with the services they need, and they might think again. Pruning (his word) ‘the FDI economy to suit the needs of the native workforce and no more’ is a recipe for economic collapse.

Lenihan argues for a laissez-faire model, where the government gets out of the way of the private sector, but wants the same get out of the way government to get in the way of the free movement of capital, labour, and ideas. Naturally leaving the EU is part of the policy prescription, so we can negotiate our own trade deals. We can see the effects that decision is having on our colleagues in the UK, who have lost a decade of increases in living standards already, with more to come.

There is a liquidationist element to the argument, for example.

“By reducing social welfare benefits, by dismantling the multi-billion euro IPAS industry, by reigning in government waste, by dismantling the NGO industrial complex, and its Social Justice Theory conveyor belt in Irish universities, and by offloading payouts to the national press, the government would find itself, despite a reduction in FDI corporation taxes, in a position to take their boot off the neck of domestic industry.”

You get the idea—partial thinking, partial solution.

In reality, government is all about trying to think about general solutions and then make them work politically, administratively, legally. You raise minimum wages, that helps workers but it also makes some businesses less viable. You change building standards, you help builders build faster but some people then have to live in rooms with less light down longer, darker corridors. You push one thing, three other things react.

In the end Lenihan’s book is a good thing for stretching policies out on the frame of the ruling ideas we all live by, but I wouldn’t take it all that seriously as offering a policy menu. It’s grand as far as it goes but for me the ideas in the book, such as they are, are too poorly stitched together to make a coherent offering.

But of course, as a representative of academia/media/government, I would say that, wouldn’t I?

My next post will be about the weird economics of the end of the world.