About This Publication

A small publication about legal technology, litigation records, and the public datasets that sit underneath both. It sells nothing and represents nobody.

What it is not

It is not a law firm. It does not practise law, does not take instructions, and nothing published here is legal advice for any particular situation. It is also not a software vendor, and it does not review products in exchange for anything.

That matters for reading the articles about model evaluation in particular. The questions posed there are the ones a buyer should ask a supplier. Nobody here is selling the answer.

A note about this domain

This domain name previously carried the website of a legal technology company that is no longer trading. That business and this publication are unconnected. No part of the former site has been republished here, and none of the people involved in it are associated with what appears now. The name was kept because the subject matter genuinely continues, not to inherit anything the earlier site claimed.

Anything the old site asserted about the performance of its software should be treated as gone rather than merely unstated. Those figures were never independently verifiable and they are not repeated anywhere here.

How sources are handled

Every article closes with the primary documents behind it, each labelled with the body that issued it. A summary of a regulation is not the regulation, and a news article about a dataset is not the dataset, so aggregators are not cited as sources.

Each article also carries the date its sources were last confirmed to resolve. Public documents move. During the writing of the dataset guides, one federal landing page turned out to be missing at three separate plausible addresses, and a statistics index published at a documented path returned a 404. A citation is a claim about the world at a moment in time, and the honest thing is to date it.

Where a figure exists but could not be confirmed from a primary source, the article says what is not known and explains how a reader can look it up for their own case. That is less satisfying than a clean number and it has the advantage of not being wrong.

Authorship

Articles publish unattributed under the publication name. There is no byline because inventing one would be a fabrication, and a fabricated writer with fabricated credentials is the same class of thing as a fabricated statistic. If a piece is ever contributed by a named person, the name will be theirs and it will say so.

Corrections

Corrections are made on the page itself rather than quietly. If something here is wrong, or a cited document has moved, write to editor@pythialegal.com and say which page and which claim. Reports that identify the primary source get handled fastest, because that is the only thing that settles it.

For a sense of the range of subjects covered, the rest of this publication is the shortest route in.