The Lundy Model provides a practical framework for improving participation in justice processes, especially where decisions can deeply affect a child’s life and well-being. The idea of giving feedback to children about how their views influenced the outcome is also powerful, because it helps make the process feel more respectful, transparent, and child-centred.

Thank you for highlighting this important approach to meaningful participation in the justice system

What an excellent piece – well done! You can note that as of last year we now have in Ottawa what is described as the “rocket docket” which means you get a trial date a year out from one’s first court appearance. That was implemented without any consultation with the Bar and by people who have NO systems design training.

Not the first nor will it be the last.

Another legal historian has referred to Professor Mossman’s earlier book, The First Women Lawyers, as being a ‘magisterial’ work… and I see you have called this one a masterpiece. That’s good enough for me!

And, as you’ve probably heard, it’s insanely hard to say exactly what you want: instead you get “bugs”, and have to test extensively to just keep the number small. Those you debug

With an LLM, you’re trying to write a “prompt”, in an ambiguous human language to compile into a formal argument.

I wouldn’t use LLMs that way unless you have a way to write tests for the correctness of the outputs, and a debugger to fix errors when you find them.

I fear you’ll have to use a formal logic language to write prompts. An example of this, for some some math I did is:

Lower Hull

Given: A finite set of points P = {p₁, p₂, …, pₙ} in 2D space

Output: An ordered sequence of points L = [l₁, l₂, …, lₖ] representing the lower hull

Postconditions (what must be true after execution):

Subset property: L ⊆ P (all points in the lower hull are from the original set)

Ordering property: l₁ has the minimum x-coordinate (leftmost), and lₖ has the maximum x-coordinate (rightmost). For all i 0

Minimality property: L is minimal — no point from P can be removed from L while still maintaining properties 2 and 3. (This ensures you have the actual hull, not just some counterclockwise path)

Completeness property: All points on the lower convex hull boundary are included. Formally: for any point p ∈ P that lies on the lower boundary of the convex hull, p ∈ L

Preconditions (assumptions):

P contains at least 2 points
Points are distinct (no duplicates)
The meaning of lower hull is understood by all

I’m not sure I could write that kind of logic in order to create a court submission (:-)) And I still don’t have a tester or debugger for it.

That idea of “informing the hope for an outcome” is exactly it. Firms put enormous care into selecting the right students, and then, almost unintentionally, shift into a model where success is expected to follow from that decision alone.

And your point about the “back end” really being the beginning is such an important reframing. It’s where capability, confidence, and contribution are actually shaped.

The challenge, as you noted, is that this part of the process competes with everything else that demands attention in a busy firm. So the focus naturally gravitates to the front end, where the decisions feel more immediate and concrete.

But when you step back, it becomes clear that even modest shifts in how firms approach that early period can have an outsized impact, not just on outcomes, but on the experience of those young professionals as they find their footing.

I really appreciate you sharing this perspective – it captures the tension, and the opportunity, exceptionally well.