Louth For Ever ★ writing in Medium November-2026.
A friendly reply to our comrade Muiris Ó Súilleabháin — and a case for where the actual votes are sitting.
![]() |
| Photo by K. Mitch Hodge on Unsplash |
What the Figures Really Establish
A few weeks ago, Muiris’ article ‘The Fragmentation Of Unionism And The Illusion Of Republican Progress’ did something that has become unfashionable in this debate: it looked beyond the headline statistic to ask what actually sits underneath it. Muiris wrote that unionist fragmentation “has not translated into any sustained increase in support for Irish unity since the signing of the Good Friday Agreement in 1998.” Worth testing against the record, since most of what follows agrees with him. It is correct that polls and elections are not measuring the same thing, and it is also right that no one should be permitted to fuse them into a single number and declare the matter closed.
Where I part company is in treating that warning as a basis for distrusting the polling itself. LucidTalk (@lucidtalk.bsky.social) pinned the 2022 Assembly election vote shares to within 1.1%, and has predicted every Northern Ireland election since 2015 within a few points, including the EU referendum. Whatever else may be wrong with using a survey as though it were a ballot, “the pollsters are guessing” is not a viable criticism, at least not against that track record.
Still, that record of accuracy cannot resolve the constitutional issue itself, because there has never been an actual vote on that question to compare it against. The pollster has plainly proved dependable on the matter it has been tested against. The issue that really counts remains untested, for the moment, in the only way that would finally settle it. That is something worth pausing on rather than hurrying past.
What the election record does establish, is this. Sinn Féin and the SDLP won 39.6% in 1998. In 2022, they won 38.1%. That’s the 24 years we’re talking about here. In Belfast, unionism shed seven council seats from 2014 to 2023; nationalism added precisely one. The ballot box is changing far more slowly than the polls, and that is not because the polls are mistaken. Elections and referendums pose different questions, and that difference cuts in both directions: a steady nationalist vote share does not show that support for unity has plateaued, just as a declining DUP vote share does not demonstrate that the Union is weakening. I suspect Muiris would be more ready than most to accept that extension of the reasoning.
Still, that record of accuracy cannot resolve the constitutional issue itself, because there has never been an actual vote on that question to compare it against. The pollster has plainly proved dependable on the matter it has been tested against. The issue that really counts remains untested, for the moment, in the only way that would finally settle it. That is something worth pausing on rather than hurrying past.
What the election record does establish, is this. Sinn Féin and the SDLP won 39.6% in 1998. In 2022, they won 38.1%. That’s the 24 years we’re talking about here. In Belfast, unionism shed seven council seats from 2014 to 2023; nationalism added precisely one. The ballot box is changing far more slowly than the polls, and that is not because the polls are mistaken. Elections and referendums pose different questions, and that difference cuts in both directions: a steady nationalist vote share does not show that support for unity has plateaued, just as a declining DUP vote share does not demonstrate that the Union is weakening. I suspect Muiris would be more ready than most to accept that extension of the reasoning.
The Ceiling, in Model Form
None of that undermines the point Muiris made from the outset, and it is the part of his analysis I think will endure longest. Unionism can lose and still preserve the Union. Sinn Féin can win and still fail to deliver unity. This week’s polling, when run through a proper seat model, demonstrates the mechanism his argument was reaching for instead of merely stating it.
The DUP and TUV are level on first preference vote share, and both trail the UUP. Using the same poll, the ‘Salmon Of Data’, Peter Donaghy’s (@peterdonaghy.bsky.social) probability model places the nationalist bloc on 40.9 seats and unionism on 38.5, out of 90, with the remaining 10.5 split between Alliance and the Greens. That is a genuine nationalist seat lead, but only a slight one, and the psephologist’s own warning is important to preserve honestly: many of the DUP-TUV marginals are “impossible to call,” so that exact division could swing either way with a small turnout change.
If you run the median case, Sinn Féin reaches 31 seats, up four, while the DUP falls to 14, down eleven, and TUV climbs to 13, up twelve, close enough to exchange places as the larger unionist party after the next bad week. Now place the formula that actually distributes ministries against that same projected increase. 27 seats versus 31 earns Sinn Féin nothing extra: four ministries out of ten, exactly what it holds now, because d’Hondt reduces a party’s claim according to the number of posts it has already secured, and a party already near the front of the order sees sharply diminishing returns from adding more. A larger mandate. The same executive. That is the ceiling Muiris identified, and it remains in place whether one looks at the trust poll, the vote share, or the seat model.
Donaghy also tested the scenario unionists have been quietly talking about for two years: a single merged unionist party in place of the DUP, UUP and TUV. It would “definitely” be the biggest individual party, at 39.5%. Yet the bloc totals hardly shift: nationalist 40.1 versus combined unionist 40.0, a difference of one tenth of a seat, down from roughly two and a half. A merger fixes the appearance and changes almost nothing about the numbers. If Sinn Féin’s ceiling is this restrictive even while winning, then unionism’s floor is this low even while losing, and even while fully united. Muiris’ instinct that electoral strength and constitutional power have drifted apart turns out to be true on both sides at once, more completely than his own article demonstrated.
What Neither Indicator Settles
Three quarters of unionists say they could not accept a pro-unity outcome, no matter what process produced it. Eight in ten DUP voters say so. Among TUV voters, the figure is closer to nine in ten. That is not a question of preference. It is a statement about consequence, and it means many people on the losing side of this issue have already made clear that a democratic defeat will not settle it for them.
Where the Votes Really Sit
The illusion isn’t that Sinn Féin’s progress is real, it’s that nationalism’s own electorate is where a majority will come from. Here is the argument, and I do not think either of us had fully assembled it before this week.
The republican & nationalist bloc is not expanding. It was 39.6% in 1998 and 38.1% in 2022. Sinn Féin’s growth has come almost entirely from the SDLP rather than from fresh ground, and Donaghy’s own seat model shows the same limit in projected form: four ministries either side of a four seat gain. A border poll fought on the current nationalist coalition does not win. Muiris has already shown that, and I think it is the most important thing either of us has written about this all year.
So the real movement is elsewhere, and the model shows where it is headed. Among voters who have decided, Alliance and Green supporters now back unity over the Union by roughly two to one, and 56% want a border poll within the next decade. Donaghy’s median case has Alliance losing ten of its seventeen seats, the steepest proportional decline of any party in the projection, even greater than the DUP’s. His own unprompted interpretation is that this may be “a post Drumcree effect with the declining centre ground,” with the middle being squeezed as the poles harden. If that is right, then the bloc containing the voters this movement needs is being pulled apart by the same polarisation everyone is watching elsewhere in the DUP, yet almost nobody is discussing it.
I am not asking Alliance to endorse unity. Their neutrality is not hesitation; it is the entire product. Naomi Long has said that sitting out a referendum campaign “might make sense,” cast as a matter of individual conscience rather than party discipline, and that is exactly why the party can draw from both traditions at once. Force a declaration and you do not clarify Alliance; you end it, and you probably drive its soft unionist third back toward the UUP. Nor is that the cause of their recent polling decline. That is incumbency fatigue, a leader whose ratings have softened after another rocky term in office, and the Greens and PBP taking bites out of the left while Drumcree dominated the news cycle. I am not going to invent a better narrative where an accurate one already exists, and Muiris, of all people, is the last person who would thank me for trying.
What I am saying is this: a party whose own voters lean two to one toward unity, while the party officially aligns with neither side and is currently losing seats in both directions at once, is holding a gap it may never close. It does not need to close it. The voters in that gap do not need Alliance’s permission. They are already counted, and Donaghy’s own model is already showing where they are dispersing. The strategic mistake has been to watch DUP collapse and call that the story, when DUP collapse merely redistributes a bloc that was never going to yield a majority on its own. The route to one runs through people who presently vote Alliance, not through squeezing a nationalist vote that peaked before the ‘Millenial’ generation of voters was able to cast one.
I want this poll won the first time it is held. The Agreement permits a second attempt seven years after a loss, which is a mercy and a trap in the same clause, because any movement that can survive defeat will plan for it. There should not need to be a second attempt. The votes for the first one are not waiting inside the nationalist electorate. They sit inside a party that has spent fifty years defining itself by refusing to say where it stands — The Alliance Party — and the model says that party is breaking apart faster than the one everyone has been watching. That is not a reason to leave it alone. It is the reason not to. Muiris’ piece led me to the right numbers. I just think they point somewhere he has not yet named, and I suspect he will arrive there before I do next time.
Sources & References
Muiris Ó Súilleabháin, “The Fragmentation Of Unionism And The Illusion Of Republican Progress,” The Pensive Quill, 31st October 2026.
LucidTalk polling for the Belfast Telegraph, fieldwork 29–31 October 2026, 3,989 respondents — the largest poll held in Northern Ireland to date. Figures on unionist leader trust, party vote share, and attitudes to a pro-unity referendum result are drawn from this release, reported by Suzanne Breen, Belfast Telegraph.
LucidTalk, lucidtalk.co.uk — forecasting accuracy record, including the 2022 Northern Ireland Assembly election vote share forecast (to within 1.1%) and election forecasts since 2015.
Peter Donaghy, Salmon of Data (salmonofdata.com), seat projections and scenario modelling built on the above poll, including the probability range projection, the median scenario seat chart, and the hypothetical combined unionist party scenario. Also shared directly via his Bluesky account, @peterdonaghy.bsky.social.
1998 Northern Ireland Assembly election and 2022 Northern Ireland Assembly election results, Northern Ireland Assembly / ARK.
2014 and 2023 Belfast City Council election results.
Naomi Long MLA, on Alliance’s position regarding a future referendum campaign, reported by the News Letter and the Irish News.
⏩Louth For Ever writes on Irish politics and constitutional change. Follow for analysis of Ireland’s democratic future as it’s constructed by those actually engaged in the work.
















