Lesson 5: The limits of ITN
The ITN framework is useful. It’s also incomplete. Before we rely on it too heavily, we should understand where it breaks down.
The measurement problem
ITN works best when you can quantify each dimension. But quantification requires commensurability—the ability to put different things on the same scale. How do you compare a DALY averted with a chicken-year of suffering prevented with a fractional reduction in existential risk?
In practice, people handle this by converting everything to a common unit (often “welfare-adjusted life years” or something similar). This requires moral weights—how much does a chicken’s suffering count relative to a human’s? How much does a future person count relative to a present one? The framework doesn’t answer these questions. It just makes them explicit.
This is a feature, not a bug. But it means ITN can’t resolve the deepest disagreements. It clarifies them.
Neglectedness can mislead
A problem might be neglected because it’s not important—not because the world is making a mistake. Before concluding that a neglected problem is a hidden opportunity, you should ask: is there a good reason nobody’s working on this? Sometimes the answer is yes.
Conversely, neglectedness can change quickly. AI safety was extremely neglected five years ago. It’s much less neglected now. If your cause prioritisation depends heavily on neglectedness, it needs regular updating.
Tractability is the hardest to assess
Of the three dimensions, tractability is the one people most often get wrong. It’s easy to tell a plausible story about how your preferred intervention could work. It’s much harder to honestly assess the probability that it will work.
Recall the forecasting skills from Week 3: base rates, decomposition, outside view. These are essential for tractability assessment. “This could work” is not the same as “this will probably work.” The gap between the two is where most overconfidence lives.
What the framework misses
The ITN framework evaluates cause areas in isolation. But the world is interconnected, and some of the most important considerations are about how causes relate to each other:
Foundational research. Some work—improving institutional decision-making, building better forecasting tools, developing moral philosophy—doesn’t fit neatly into a single cause area. It improves our ability to prioritise everything. The ITN framework tends to undervalue this kind of meta-work.
Personal fit. The ITN framework asks “which cause area is best?” But the question you actually face is “where can I do the most good?” A cause area that’s slightly less important on paper but where you have rare skills, deep knowledge, or unusual motivation might be where your marginal contribution is highest.
Option value. Sometimes the best thing to do is not to commit to a cause area yet, but to build skills, learn, and keep your options open. The framework pushes toward immediate optimisation; the reality of a career is that flexibility has value.