Matthew Waddington

Getting it right first time: Making lawful decisions

🔔 𝐆𝐞𝐭𝐭𝐢𝐧𝐠 𝐢𝐭 𝐑𝐢𝐠𝐡𝐭 𝐅𝐢𝐫𝐬𝐭 𝐓𝐢𝐦𝐞: 𝐌𝐚𝐤𝐢𝐧𝐠 𝐋𝐚𝐰𝐟𝐮𝐥 𝐃𝐞𝐜𝐢𝐬𝐢𝐨𝐧𝐬
📅 4, 6, 11 & 14 Nov | Zoom

4 webinars for decision-makers! In partnership with @legalactiongroup.bsky.social & @39essexchambers.bsky.social.

Book now: https://shorturl.pulse.ly/mltceu8krj

Very encouraging to see "Public sector automated decision-making" will be one of new projects at Law Commission (E&W)
lawcom.gov.uk/news/law-com...
We published Margaux McQuilton's pre-print report on ADM osf.io/a7jwc
We see logic-driven systems taking human decision-makers to the right questions

Looking for a really simple explainer?

The #ECHR and #HRA on post-it notes.

In which we bid farewell to Professor Michael Rush. A kind, generous and gentle man. More notes from @anyaso.bsky.social and I:

ukparliament.github.io/ontologies/m...

the professionalisation of legislative drafting (only 150 years or so in the UK) and the unlikely nature of any oversight.
Implied repeal is an argument of last resort. If made it indicates someone has no real arguments. Hard to identify the last time such an argument was successful.

This is the first time I have seen L4 running in the wild, which is exciting.

I have some thoughts...

Very sorry you see it that way. I can only repeat that I don't see what we're doing as in any way at all taking away from what you (or others) are doing - it isn't either/or. I do want benefits beyond better drafting - to me it seems sensible to push towards that on different fronts at the same time

I would like to have a call with you &/or Martin if you have some time, ideally before the RaC Guild - I'm sure it would be more useful than talking across each other on here. What do you reckon?

We are not trying to teach drafters any of the full systems - I will leave that to you (& Martin P) for Blawx, & the Legalese team for L4, & whoever for whatever. We are trying to do something different, which really doesn't clash with any of what anyone else is doing. We're all on the same side???

I am saying that to introduce many drafters, on their territory, to what this can do for drafting, it is a good idea to do that with propositional. We have not found anyone to do that for us, but the L4 guys were prepared to let us hobble their system and we leapt at that.

Seems like I am still not getting this across - I absolutely am not saying you have to use our approach to learn L4 - and I really am not interested in any point-scoring between Blawx, L4, Logical English, Catala or whoever.

I do have the interest, but many people need intermediate steps.
Also not sure what you mean on alternative tool - we both know no tool could automatically reliably convert real legislative text into code/logic, even just propositional. Getting drafters to analyse their text is a key purpose for us

For the next step I don’t necessarily mean flowcharts. Just a simple way to go from the text to something that captures something beyond propositional and can then show that something can be done with it. Stepping stones for some people, jump straight into learning full Blawx/L4/etc for others

By the way my next mad plan is to look into what would be the simplest minimum we could add to produce a marginally less hobbled (but still brazenly hobbled) version that takes us a minimum step beyond propositional into the shallowest & most drafter-friendly end of predicate. If that sinks so be it

Do bear in mind this is us jumping on one tool they had produced, and us asking them to help us hobble it back down to just propositional logic, for our purposes. I would love to persuade you (or Martin P) to help us do an equivalent with Blawx too - could you be open to discussing that?

We are also presenting this to drafters as worth the small effort partly because it improves drafting, by helping drafters to think yet more rigorously about what they are doing. Again this opens the door to seeing scope for further improvements in drafting too.

I honestly don’t think it will obscure anything. Visualising is an impactful way to get people started, & this starts from where legislative drafters are. Once you have them over that hill it is far easier for them to picture the potential benefits of a more sophisticated system.

I really don’t think it’s an either-or. I see this as part of helping to get more of one specific audience to start down a road that could then just as easily take them to Blawx as anywhere else. Will you be at RaC Guild on 3 Sept to see us? Maybe talk before?

A wee bank holiday treat for you guys. More week - well, fortnight - notes from @anyaso.bsky.social and I. In which we approached ramming speed, only to collide head-on with documentation buffers. Rescued in part by the kindness of strangers
ukparliament.github.io/ontologies/m...

Thanks Jason - the point of the propositional approach is to give legislative drafters, with no tech experience, an easy & familiar way in to seeing what we mean by capturing the logic and what that can deliver. So all your points are valid but we are trying to do something else

Here are some results from our work with the L4 team at SMU - getting legislative drafters to produce basic visualisations by showing how to make propositional logic versions of their drafts
#RulesAsCode

If we needed new draft legislation to criminalise burglary, surely it shouldn’t fail the tests just because we know some people will still burgle, or even because some police forces will not investigate and deter?
Or is this just me not really knowing what test-driven means?

We might.
But this one doesn’t look like a problem with the drafting, which requires “highly effective” age measures & tells OFCOM to give guidance. Isn’t this about either providers not following OFCOM or OFCOM’s guidance being too lax? Or policy wrongly assuming affordable effective tools exist??

The last day of ICAIL had a lot of papers directly relevant to #RulesAsCode, and at least three papers about getting LLMs to do the encoding.

A lot of unanswered questions, still, but it seems like some of the design choices I made about Blawx v2 to help codegen were well founded.

NLLP Workshop 2025 The seventh workshop on Natural Legal Language Processing (NLLP 2025) explores methods and applications of Natural Language Processing for the Legal Domain by focusing on legal text and text with lega...

☀️ We’re here too - getting ready for the 7th NLLP workshop (8/Nov @emnlpmeeting.bsky.social ) 🚀✨
The updated Call for Papers is coming soon — stay tuned! 📣📄

💙 Follow us & spread the word about NLLP🌍💬

🔗 Keep an eye on the website for updates! 🖥️🔍 nllpw.org/workshop/

#NLLP #legaltech #nlproc #nlp

You mean as in encoding - L4, DataLex, Blawx, Catala, LegalRuleML, AkomaNtoso etc? Not as in Napoleonic codification?

Pia Andrews - Regional Innovation and Digital team, UNDP Bangkok Regional Hub | LinkedIn Experience: Regional Innovation and Digital team, UNDP Bangkok Regional Hub · Location: Canberra · 500+ connections on LinkedIn. View Pia Andrews’ profile on LinkedIn, a professional community of 1 bi...

Mr Pants is right - Pia spread “Rules as Code” from NZ to Oz & beyond - www.linkedin.com/in/pia-andre...

This afternoon we will host a Work-in-Progress session with visiting PhD candidate Filipa Paes (University of Oxford), who will be presenting a paper on ‘Juridical Bivalence’. The event is pre-read, so make sure to get in touch if you’d like to receive a copy! www.law.ed.ac.uk/news-events/...

For those who don't have access to the final published version, there is now a free preprint version here osf.io/preprints/so...

Start With Imagination, Not Code Lessons from Building a Shared Legislative Platform

This is a really good interview with the fabulous Matt Lynch on Lawmaker - start with imagination not code, he says. modparl.substack.com/p/lawmaker