Is It Worth Building? Index

What This Is

A structured ethical reasoning framework for the question every builder eventually faces: should I build this? Not in the abstract, but for a specific technology, right now, given what exists.

As AI capabilities accelerate, the gap between "technically feasible" and "should exist" grows faster than the mechanisms for evaluating it. This project builds a rigorous index not a vibes check, but a scored, structured analysis across multiple dimensions to help builders reason clearly about technologies with dual-use potential.

The Framework

Five dimensions, each scored 0–10:

1. Capability overhang How much does this technology accelerate capabilities that don't currently exist? A password manager doesn't create a capability overhang. A model that can autonomously write and deploy exploit code does. Score: magnitude of new capability enabled × reversibility.

2. Misuse asymmetry Is this easier to use for harm than for benefit? Nuclear reactors require massive infrastructure for weapons use but relatively less for power generation low asymmetry. Deepfake models have near-zero friction for misuse high asymmetry.

3. Counterfactual impact If you don't build it, will someone else build it anyway, and soon? If the technology is inevitable in 6 months from a better-resourced team, your counterfactual impact on the world is low. This doesn't remove responsibility but it changes the calculus.

4. Harm concentration Do the harms fall on the users who consented to the risk, or on third parties who didn't? Tools that harm only willing users score low. Technologies that externalise harm onto non-consenting populations score high.

5. Correction capacity If this turns out to cause harm, how reversible is that harm? Can the technology be undeployed? Can affected parties recover? Systems with high correction capacity warrant more experimentation tolerance.

Why This Matters Now

The standard "move fast and break things" heuristic was fine when the things you could break were mostly web interfaces. When the things you can break are epistemic systems, democratic processes, or biological safety, the calculus is different. This framework is an attempt to operationalise the intuition that some technologies warrant more caution than others.

What's Next

• Collaborative scoring: allow builders to submit projects for community review
• Case study library: historical technologies scored retrospectively
• Integration with AI safety literature: map framework dimensions to existing alignment concepts