Decisions, Decisions: Our Case Law Database, or Why We Only Keep Full Written Judgments

By our estimate, the portals of the eight courts the case law database draws on publicly hold over 200,000 decisions. That figure is an estimate and not a count: we did not set out to tally every record on every portal, only to find the full written judgments among them. The courts' full record is larger still: a great deal of it has never been digitised in a public manner, though some may have been by external publishers or by others who sourced it directly from the courts. The database holds 45,000+ full written judgments. The difference is not a backlog. It is a choice about what belongs in a research collection, and this entry sets out that choice and the reasons for it.

What the portals publish

A court portal publishes what the court decides, and much of what a court decides is not a written judgment. Short orders record an outcome. Dismissals in limine and half-page refusals dispose of a matter without reasons. Interim orders adjourn, direct or stay. All of them bind the parties before the court, and all of them have their uses: an advocate tracing the history of one particular case may need exactly that order.

Those records are already available, on the same portals as the written judgments. Copying them into a second place adds nothing to what an advocate can already reach.

What the database keeps

The database keeps written judgments: the decisions in which a court sets out the facts, the law and its reasoning. These are the decisions that carry legally valuable content, the content that can be cited, distinguished and relied on in another matter.

That choice is what makes the rest of the database possible. Every judgment is categorised by court, area of law and case type, and every judgment carries a summary, so you can decide from a short read whether the full text is worth your time. Categories and summaries of that kind are only meaningful over decisions that contain reasoning. Laid over a short order, they would restate what the portal already shows.

Each summary is checked to be sound and coherent before it is served, so it can be read with confidence. It is still a guide to the judgment, not a replacement for it, and we encourage reading the judgment itself before relying on it.

Why it matters for Research Lab

When a question is put to Research Lab, the models we use read through a large body of material: provisions from many statutes and judgments from many courts, gathered over several searches for that one question. One useful written judgment buried among five or ten short orders, whether from the same case or another, does not help that reading. The models can often tell the difference from context, but not always, and every short order takes room that a judgment with reasoning should have had.

The law itself is easier to work with. The statutes library holds 3,800+ statutes, covering the law of the federation and of its four provinces, the capital included. Statutes are organised material: titled, sectioned and grouped by subject. Judgments are far harder to categorise, even when they are full written ones. Keeping the collection to written judgments is what lets Research Lab's answers stay accurate and precise.

Everything ties back to the first source

Every judgment in the database links to the original judgment on the court's own portal, so the summary and the categories can always be checked against the source they came from.

The statutes are different, and so is the reason. A statute is served in full, exactly as published, without summary and without anything added. What you read is the text itself, so there is nothing to trace back.

We also do not take judgments from databases hosted elsewhere, including collections that hold the same judgments the court portals publish. We do not have permission to use them, but that is the smaller reason. The larger one is that every judgment in the database must lead back to the court that issued it.

Why some written judgments are still missing

The database aims to hold every written judgment the eight courts make publicly available, and it holds almost all of it. It does not claim to hold every one, for reasons that have nothing to do with the choice above:

• Some are not easily reachable. Older judgments in particular are not always available for download from the court's own portal.
• Some links do not work. A court portal can reorganise its addresses, so that a judgment once at one address moves to another and the old link stops working. Others are genuinely damaged files, or are not open to the public.
• Some files are hard to read. Most source files carry clean, readable text. The rest are scanned pages, some of them of low quality, and not every one of those yields a judgment that can be relied on.

Coverage grows as more written judgments are published and as more become reachable.

Courts, not tribunals

The database draws on the Supreme Court of Pakistan, the five High Courts, the Federal Shariat Court and the Federal Constitutional Court. It does not hold decisions of tribunals or of courts below the High Courts.

That is a separate decision from the one about short orders, and it rests on precedent. Judgments of the superior courts are the ones that bind and are cited. The Supreme Court sits at the top, and after it the Lahore and Sindh High Courts are the most likely to be cited. Tribunal decisions are a very small share of what is cited at all. The reasoning on binding force is set out in The Binding Force of Precedent Under Articles 189 and 201.

Leaving tribunals and lower courts out for now is not a judgement that their decisions are without value. They can matter in a narrow jurisdiction or as a historical record. For the time being they sit outside the collection, for the reasons given here.

For what the database covers court by court and how to search it, see The Case Law Database.