From D&I metrics to decision integrity: let evidence quality govern whose “diversity” matters
- 12 minutes ago
- 4 min read
This post responds to a segment from a Yah Lah But podcast featuring Crystal Lim-Lange (with a shorter clip circulating on Facebook).
In the segment I watched, the hosts start with a common assumption: Singapore’s system should be able to select the smartest person to lead. Crystal’s response (as I understood it) is that for a system to consistently surface strong leaders, it needs both:
Diversity (so a wider range of talent and perspectives exists), and
Inclusion / psychological safety (so those perspectives can be voiced without fear)
It’s a clean frame, and I understand why it works in a podcast format.
But I think the bigger risk is this: we end up treating “D&I” (often defined along standard demographic dimensions) as the goal, instead of asking what D&I is actually supposed to do for a country—especially a small one with limited room for strategic error.
So here’s the reframing I prefer:
Instead of optimising for D&I for D&I’s sake, we should let evidence quality govern diversity of views and inclusion of dissent.The purpose isn’t to make everyone feel heard. The purpose is decision optimisation—better judgement, better trade-offs, better learning loops, and more durable trust.
Why the stakes are higher now
Singapore is navigating:
a more volatile geopolitical and economic environment, and
serious demographic constraints
In a small, open system, mistakes are costlier and course-corrections have less slack. That doesn’t mean leaders must be “right” all the time—nobody can predict the future. It means Singapore needs a decision approach that is disciplined, testable, and adaptive.
A truly data-driven posture is not “we have data.”It’s: we decide with humility, measure outcomes, learn quickly, and adjust credibly.
The missing middle: D&I doesn’t produce outcomes by itself
Many DEI conversations jump from inputs to outcomes:
Inputs: representation targets, training, policies
Outcomes (hoped): innovation, performance, fairness, cohesion
But in a national context, outcomes aren’t always explicit or aligned:
growth, affordability, security, social mobility, birth rates, openness, cohesion… These objectives sometimes conflict. And when objectives conflict, “more D&I” doesn’t tell us what to do.
So rather than asking “Do we have D&I?”, the more useful questions are:
What are we optimising for?
What trade-offs are acceptable—and for how long?
How do we decide who should be in the room, and how much weight their input should carry?
How do we learn and adjust without losing trust?
A fuller, data-driven model: ETTC
Evidence quality → Trade-off competence → Trust → Communication
1) Evidence quality (and it should govern “diversity” and “inclusion”)
This is the key upgrade: evidence quality should be the filter for inclusion and dissent.
There are two equally important cases:
Case A: when data is availableEvidence quality means we can:
quantify the size of the problem relative to other problems,
prioritise,
estimate consequences of options,
define success measures, and later test whether outcomes were achieved.
Just as important, evidence quality tells us something about “voice”:
Not all viewpoints carry the same signal.
Not all claims deserve the same weight.
Inclusion should protect the right to raise concerns, but influence should be earned through relevance, evidence, and reasoning.
Case B: when data is not available (or not decisive)Some of the most important decisions are values-driven:
“What kind of Singapore do we want?”
“What does resilience mean?”
“How do we balance fairness and competitiveness?”
Here, evidence quality becomes process quality:
clarifying values,
surfacing disagreements honestly,
and building legitimate direction before arguing about tactics.
This is where engagement efforts like Forward Singapore are trying to contribute: not “data” in the narrow sense, but clarity on priorities and tensions.
What this means for D&I
If evidence quality is the governing principle, then “diversity” stops being primarily demographic and becomes decision-relevant diversity:
diversity of incentives (not everyone gains the same way),
diversity of lived constraints (those who pay the costs vs those who get the benefits),
diversity of expertise (domain, delivery, second-order effects),
and crucially, diversity of independence (people who can speak without fear or favour).
Likewise, “inclusion” stops meaning “everyone gets equal weight,” and becomes:
structured dissent that is safe to raise,
but filtered through standards (evidence, logic, track record, accountability).
This also addresses a real risk in advisory ecosystems: if too many voices are perceived to be dependent on the system for continuity or advantage, you may get confirmation bias (or at least the perception of it). A data-driven approach deliberately counterbalances that with credible, independent dissent.
A useful principle here is: invite people who can disagree without being punished—and without being rewarded for agreeing.
2) Trade-off competence
Every decision produces gains for some at the expense of others. The job of leadership is to make trade-offs competently:
speed vs consultation
efficiency vs resilience
fairness vs simplicity
short-term relief vs long-term sustainability
This is where “inclusion” needs structure:
Who is consulted, when, and why?
What counts as a relevant dissent?
Who decides, and how is accountability preserved?
Without this, inclusion can become paralysis—or worse, lowest-common-denominator decisions.
3) Trust and legitimacy
Because nobody can predict the future, good governance must be adaptive:
decide based on the best information now,
measure outcomes,
learn,
adjust.
Trust is what makes adaptation possible.
In a high-trust system, course correction looks like learning. In a low-trust system, course correction looks like backtracking.
So trust isn’t built on “always being right.”Trust is built on whether people believe decisions were arrived at credibly:
evidence was used appropriately,
trade-offs were acknowledged,
dissent could be raised safely,
and changes later are explained as learning, not convenience.
4) Communication under constraint
This is where I agree with Crystal’s instinct: communication matters.
But communication is not just “speaking well.” It’s sensemaking under constraint:
fragmented attention,
social media incentives that punish nuance,
and lower baseline trust.
This is why some leaders are remembered not for perfect accuracy, but for making trade-offs legible—helping citizens understand constraints, choices, and consequences. When people can follow the reasoning, they’re more likely to accept both the trade-offs and the later adjustments.
Closing: upgrade the debate
I’m not arguing against diversity or inclusion.
I’m arguing for a more decision-centric interpretation:
Don’t optimise for D&I as a demographic scoreboard.Optimise for decision integrity—where evidence quality governs whose input matters, how dissent is included, and how trade-offs are made.
Especially for a small country with limited room for error, the goal isn’t to maximise how many voices are present.
The goal is to maximise the quality of judgement—and the credibility of the process by which judgement is formed.
Comments