◎ OireachtasDB

FAQ

Common questions

Short answers. If something here is wrong or missing, it should be — this page is meant to grow.

What is this?

A search engine over the official record of what is said in Dáil Éireann and Seanad Éireann, and how members voted.

Updated on 01:00 GMT Monday.

Where does the data come from?

Directly from the Houses of the Oireachtas open data API, which publishes the Official Report. Nothing here is typed in or edited by hand — it is the published record, reorganised so it can be searched.

What period does it cover?

2020-01-21 to Present. Earlier material exists in the source archive going back to 1919, and the pipeline can be pointed at it, but it is not loaded here.

Why does a speaker's party sometimes look out of date?

Because it is deliberately the party they held on the day they spoke, not the party they are in now. Members change party, and most archives quietly relabel their old speeches. A contribution from 2020 here carries its speaker's 2020 party.

What is not included?

Three things worth knowing:

  • Written parliamentary answers. These are not in the debate record at all — the Oireachtas publishes them separately.
  • Committee debates. Committee votes are here, but their transcripts are not.
  • Anything said in private session, which is not published by the Oireachtas in the first place.

How do I search?

By default every word you type has to appear in the same contribution.

mica redress Both words, in the same contribution
"mica redress" That exact phrase
housing OR homes Either word
housing -dublin Housing, but not anything mentioning Dublin
hous* Starts with: housing, housed, householder

What do Tá and Níl mean?

Yes and no, in Irish. A vote in the Dáil or Seanad is called a division: Tá is a vote in favour, Níl against, and Staon is an abstention.

Are the debate summaries written by AI?

No, and this is worth being precise about. Each summary is built by scoring every sentence in the debate on how distinctive its wording is, then quoting the highest-scoring sentences back word for word and in the order spoken. Nothing is paraphrased or generated. If a sentence appears in a summary, somebody said it in the chamber.

That is a deliberate choice for a record people may want to cite. A language model writing its own prose could smooth over or invent detail, and here the text is the evidence.

What are the "most used terms"?

A count of how often each word appears in that debate, ranked by how distinctive it is to that debate rather than common to all parliamentary language. That is why "housing" or "mica" rises to the top while "deputy" and "amendment" do not.

Why does a vote sometimes say the result was derived?

For a small number of older divisions the Oireachtas did not record whether the motion carried, but the vote counts survived. Where those counts differ, the result follows arithmetically and is shown — labelled as derived, so you know it was calculated here rather than published.

Why is there no age or gender breakdown on votes?

Because the source data does not contain it. The Oireachtas API publishes no date of birth, and returns an empty gender field for every member in this period. Showing such a breakdown would mean inventing it.

Are the party colours the official ones?

No. They are chosen so that parties stay distinguishable to colour-blind readers and in greyscale. Several parties genuinely share a colour family — Fianna Fáil, Sinn Féin, the Greens and Aontú are all green — so the abbreviation on each chip, not the colour, is what identifies a party.

Can I quote or reuse this?

Yes. The Official Report is published by the Houses of the Oireachtas under a CC BY 4.0 licence, which asks only that you credit the source. Every download from this site includes the record's address and the exact time it was fetched — the Report can be revised after publication, so a citation without a timestamp cannot be checked later.

How accurate is any of this?

Measured rather than asserted. 99.6% of contributions resolve to a named member, the search has been evaluated against a hand-checked question set, and the gaps in the source data are documented rather than hidden.

See the workings →