Week 11: Week 11 lecture: The ethics of alignment

Two different questions that get collapsed into one — was the process fair, and was the outcome fair — and why a decision can pass one test and fail the other.

3 May 2027

Two fairness questions, not one#

“Was that fair?” usually means two different questions wearing one word. Procedural fairness asks whether everyone with a stake got a genuine chance to be heard, by a process applied consistently. Distributive fairness asks whether the outcome itself distributes benefit and burden in a way that can be defended. A decision can pass one test and fail the other in either direction — and the two failures do not always arrive as a clean, separable pair.

The 13 March Working Group is a case in point, and it is not the tidy “procedurally fine, distributively questionable” example it can look like at first glance. Sanjay Okoro’s question about the sequence was asked on the record and never actually answered before the group moved on — that is a person with a stake not getting a genuine hearing, a real procedural gap, not just something the minutes happened to compress away. The phase order that resulted — the highest-traffic team going first, with no stated reason on the record — raises a separate, distributive question of its own. Both are worth naming here; filing this meeting under only one heading would understate what actually happened at it.

To see the two axes pull apart cleanly, it helps to use a hypothetical rather than the case itself: imagine a working group that answers every question raised, on the record, by every stakeholder present — including a full, candid explanation of why the ranking puts the highest-traffic team first — genuinely procedurally fair, no gap like Okoro’s, and no question about the ordering itself left unaddressed. The outcome can still be hard to defend: the team going first happens to be the one with the least standing in the room to push back, and being heard on a rationale is not the same as that rationale being a good one. Consulting everyone at length and answering what they ask does not, by itself, make the outcome defensible; it only makes the process that reached it fair. The reverse failure is just as real: a consultation that visibly could never have changed the result is not procedurally fair merely because it happened — a hearing that cannot affect the outcome is closer to a procedural failure wearing the shape of one, not a clean case of “fair process, unfair result” either.

That hypothetical is still incomplete, and deliberately so: “answers every question raised by every stakeholder present” only scores fairness among the people already in the room. Week 2’s own invite list names two seats this test never reaches at all — Dr Femi Adisa and LSS’s frontline staff, neither invited, both affected by what the Steering Committee confirms. A process can be flawless toward everyone it sat down with and still be procedurally unfair to someone it never sat down with in the first place; “was everyone present heard” and “was everyone affected present” are two different questions, and a working group can pass the first while the second was never even asked.

The Alignment Lab’s hot-desking worked example makes the same split visible on paper: the Staff Representative’s constraint field requires taking any agreement back to affected staff before it holds (a procedural guarantee), while the minimumAcceptableOutcome field — named, reserved desks for documented accessibility needs — is a distributive claim about what the outcome itself must guarantee, regardless of how the conversation that reaches it goes. A group could satisfy the first and still fail the second, or the reverse.

Leverage asymmetry is not automatically unfair#

Week 9 introduced leverage as a fourth axis alongside decision rights. This week’s harder question is when an asymmetry in leverage becomes an ethical problem rather than just a structural fact. A Finance and Risk Lead’s sign-off requirement over $10,000 is a large, formal piece of leverage that nobody in the process is likely to call unfair — it exists for a reason everyone can name, applies consistently regardless of who holds the seat, and is visible to everyone in the room. An asymmetry becomes harder to defend when it is invisible to the people it affects, applied inconsistently, or exists for no defensible reason beyond who happened to get there first. The test this week teaches is not “is there an asymmetry” — there almost always is — but whether that asymmetry could survive being explained, out loud, to the person with less of it.

Applying this to your own Lab record#

Your whole Alignment Lab submission — decision record, negotiation log and individual autopsy — was due today at noon, before this session, so the material for this test is already written, not still to come. Somewhere in your own group’s process there was very likely a leverage asymmetry that shaped the outcome. The question worth sitting with now, rereading what you already submitted, is not whether that asymmetry existed — it did, in some form, in every group — but whether it could have been said out loud in the room without anyone objecting to its legitimacy.

A little further reading#

The linked entry is a broad philosophical survey of justice in general, not a text about meetings or organisations — its procedural/substantive (there, “distributive”) distinction is the same one this week draws, but pitched at the level of political and social institutions generally, so most of the entry is more abstract than this week’s argument needs. Read it for one purpose: find where the entry states that a procedurally fair process does not, by itself, guarantee a substantively fair outcome — that is the one claim this week’s lecture borrows directly. Everything else in the entry (competing theories of what makes an outcome just — desert, equality, need) is optional background, not required reading for this week’s own argument.

Before Week 12#

Nothing new to read. Start thinking about which single element of your own group’s process — not the outcome, the process — you would change first if you were redesigning it, because that is where Week 12’s assessment begins.