On Paper
01It was done, then it stopped02What gets blamed03What would move it

PIECE 03PrecedentLiving document

The question

Has this state ever built anything at this scale before?

Piece7 min read3 chaptersupdated September 2026every figure traceable to a named table

The contract went to Siemens-Schuckert of Berlin, because no Irish firm could do it. It was attacked in the Dáil as reckless and in the press as a German adventure.

The state that signed it was three years old, had just come through a civil war, had no electrical engineering industry, and no obvious means of repayment.

The short versioneverything below this is the working
17.3%

Of one year's state spending went on a single power station in 1925. It worked, and it was still working sixty years later.

56%

Of judicial reviews the planning board loses or concedes. So why is the thing everyone blames the thing that keeps winning?

“It can't be done here” is a claim about history, and history is a dataset. This piece checks the claim twice — once against what Ireland used to build, and once against the mechanism everyone says stops it now.

Your voice — 120 words, at the top

Something built that is now so ordinary nobody remembers it was fought over. What it is, who objected at the time, and what it is used for now by people with no idea there was a fight.

  • This makes Ardnacrusha a category rather than a museum piece, before the reader files a 1925 story under nostalgia.
  • The temptation is triumphalism, and this piece cannot afford any.

One · It was done, and then it stopped1925, 1975, now

A three-year-old state spent a sixth of its budget on one power station.

The Shannon Scheme at Ardnacrusha cost about £5.2 million. Irish government expenditure that year was about £30 million.

That ratio is difficult to feel at a distance. Counted out in hundredths of a single year’s spending, it looks like this.

One year of state spending, in hundredths

Each square is one per cent of Free State government expenditure in the year the Ardnacrusha contract was signed.

How to read thisOne hundred squares, each one per cent of what the state spent that year. The rust squares are the power station.

the Shannon Scheme, c. £5.2m of c. £30meverything else the state did that year

Contemporary government accounts. We compute 17.3%, not the “fifth of the budget” usually quoted — see About the data.

The numbers behind this chartTable & CSV
What the spending went onShare of that year's state spending (%)
the Shannon Scheme, c. £5.2m of c. £30m17
everything else the state did that year83

Contemporary government accounts. Squares are whole percentage points; the share computed in the text is 17.3% — see About the data.

Download CSV

Within a decade the scheme was supplying the great majority of the country's electricity, and the ESB — created to run it — went on to electrify rural Ireland over the following thirty years.

17.3% of current Irish gross voted expenditure would be roughly €25 billion on a single project. That is arithmetic for scale, not a proposal and not a costing of anything.

The constraint in 1925 was not money, expertise or consensus. The state had none of the three.

And this is the part that matters more than the anecdote: it was not a one-off. Fifty years later, local authorities were still building at a rate that would look like a national emergency programme today.

Then 1975 — and then nothing

Population grows, so absolute counts mislead. Per thousand people is the honest denominator, and on that measure the direction of travel is not what the announcements suggest.

What went missing between those two years is not money. It is capacity. A local authority that has not run a construction programme in forty years has no design team, no quantity surveyors, no standing relationships with contractors and nobody left who remembers doing it.

Rebuilding that takes a decade and cannot be announced. It is the least glamorous item on any list of what would fix Irish housing, and it is close to the top of the list.

Local authority and social housing output per 1,000 people

Two years, on the same denominator.

How to read thisEach bar is one year’s output for every thousand people living in the country. The longer bar is the older year.

Dept of the Environment and Dept of Housing statistics, over CSO population estimates. Not like-for-like — see About the data.

The numbers behind this chartTable & CSV
Year and programmeHomes per 1,000 people
1975 — local authority2.77
2025 — social housing1.68

Department of the Environment and Department of Housing statistics over CSO population estimates. Not like-for-like: 1975 is local authority new-build completions, 2025 is social housing output on the current, broader definition, which includes acquisitions and long-term leasing.

Download CSV
Known gap in our own evidence

There is no clean, publicly available local authority completions series before 2005 on a consistent definition. Everything earlier has to be assembled from annual departmental reports.

The year that is called a record

−40%

Social housing output per thousand people in 2025, against what councils were doing routinely in 1975.

1.68 per 1,000 against 2.77. And the 2025 definition is the broader of the two — it counts acquisitions and leasing as well as new build. On a like-for-like basis the gap is wider.

Two · What gets blamed insteadthe payoff

Everyone knows what stopped it. The records disagree.

The mechanism being blamed and the mechanism being measured are different things.

Roughly 150 planning decisions a year are challenged by judicial review, out of tens of thousands made. When they are challenged, the planning board loses or concedes more often than it wins.

The claim this piece tests — stated onceNothing gets built in Ireland because anyone can object and tie it up in the courts for years.
The error

A vivid case generalised into a rate. The handful of long-running judicial reviews that make the news are real, and they are a tiny and shrinking share of planning decisions. The measure that would show the system seizing up — how long a case takes to dispose of — has been falling sharply.

If the courts were the bottleneck, disposal times would be rising and the state would be winning. Neither is true.

The three numbers behind that sentence, each from the body that keeps it:

~150planning decisions challenged by judicial review in a year, out of tens of thousands madeAn Bord Pleanála annual reports
56%of those challenges the board loses or concedesAn Bord Pleanála annual reports
406 → 151days to dispose of a case — the trend is down, not upCourts Service statistics

A majority of successful challenges is not evidence of vexatious objectors. It is evidence of decisions that were not made properly the first time — which points the repair at the planning authority rather than at the people appealing it.

Objection

"The risk is the point. Developers do not need to lose a case, they just need to fear one."

Fair, and we cannot measure a chilling effect from the outside. What we can say is that the mechanism being blamed and the mechanism being measured are different things, and that a 56% loss rate points at the quality of the original decision rather than at the people appealing it. Fixing the first would remove most of the grounds for the second.

Three · What would move itnow what

Optimism, but the boring kind.

The argument that Ireland is structurally incapable of building at scale is contradicted by Ireland. A state with less money, less expertise and a great deal more chaos did it, twice, within living memory of people still alive.

What changed is not the national character. The capacity was dismantled. The incentives were rearranged so that scarcity became profitable for people who already owned things. And the requirement stack grew one reasonable condition at a time until the thing could not be delivered.

Every one of those is a decision, which means every one of them has a reversal.

That is why this publication ends pieces with mechanisms rather than villains. A villain is a story that finishes. A mechanism is a thing you can go and change.

1
Fix the decision, not the appealThe board loses or concedes 56% of the challenges brought against it. Publish that rate per authority, annually. A planning system that lost fewer cases would face fewer of them.
2
Rebuild a standing state build capacityNot a scheme with a sunset — the thing that existed in 1975. Measured on completions per 1,000 people, which is the metric those programmes were actually run on.
3
Publish the pre-2005 seriesIreland cannot argue about whether it used to build more because the consistent series does not exist in public. The Housing Agency holds it. Publishing it costs nothing and settles a decade of argument in both directions.
About the data

The Ardnacrusha ratio. £5.2m against roughly £30m of Free State government expenditure in the year the contract was signed, from contemporary government accounts. We compute 17.3%; the figure usually quoted in Irish public life is "a fifth", which is rounded in the direction that flatters the argument, so we round it back.

The 1975 / 2025 comparison is not like-for-like, and the difference runs against us. 1975 is local authority new-build completions. 2025 is social housing output on the current, broader definition, which includes acquisitions and long-term leasing. A like-for-like new-build comparison would show a wider gap, not a narrower one.

Judicial review figures are from An Bord Pleanála annual reports and Courts Service statistics. "~150 a year" is an order of magnitude, not a precise annual count, and is written that way deliberately.

Provenance tiers. Tier 1: recomputed by our own pipeline from a named table. Tier 2: taken from a published secondary source we have not independently verified — the 1975 local authority figure is Tier 2. Tier 3: unchecked, and never published.

Sources, in full
  1. Shannon Scheme cost and Free State expenditure — contemporary government accounts and ESB historical material.
  2. Local authority housing completions 1975 — Department of the Environment annual housing statistics. Tier 2, FOI outstanding.
  3. Social housing output 2025 — Department of Housing, Local Government and Heritage quarterly output reporting.
  4. Judicial review volumes, outcomes and disposal times — An Bord Pleanála annual reports and Courts Service statistics.
  5. Population denominators — CSO Population and Migration Estimates.
What changed on this page
Restructured from four chapters to three, and reframed. The judicial review section used to read as a digression; it is now the payoff, and the piece builds towards it.
Added a "what would move it" list. This piece did not have one, which broke a promise made on the homepage.
Correction, against our own interest. The Ardnacrusha share of state spending is routinely quoted as a fifth 17.3%. We were repeating the rounded version.
What this page is still waiting on

Where is the consistent local authority housing completions series before 2005? There is no clean public series. Everything earlier has to be assembled by hand from annual departmental reports, which is why the most important chart in this piece has two data points instead of ninety.

An FOI request is in to the Housing Agency for the underlying series. When it lands it goes on this page whether or not it supports the argument above — and if 1975 turns out to be an unrepresentative peak, that is the correction and it goes at the top of the homepage.

Read next · Piece 04

Are immigrants living on the dole and getting free housing?

82% of everyone on a welfare payment in Ireland is an Irish national. The claim has two real edges and we print those too.

Read it →

In the pipeline

Who dismantled the thing that used to build houses, and when?

This piece is the sketch. That one is the autopsy: local authority completions per 1,000 people since the 1930s, and the year the capacity left the building. Waiting on the FOI above.

Blocked on the FOI

Draft notes