← back

Methodology

How we rank.

Every ranking is a conjecture open to refutation. The point is not to be right forever, it is to be wrong in public and correctable by evidence.

Three lenses, not one composite

A single score always hides trade-offs. We use three lenses from three distinct traditions. Every problem gets scored on each lens where the data supports it. Users pick which lens matters and sort by it.

welfare lens

Copenhagen Consensus Benefit-Cost Ratio

Dollars of social good per dollar of intervention. Developed by Nobel- laureate economists at the Copenhagen Consensus Center. Favors high-certainty, shovel-ready interventions like micronutrients and disease control.

x-risk lens

80,000 Hours ITN composite

Importance × Tractability × Neglectedness. Developed at 80,000 Hours for Effective Altruism cause prioritization. Favors problems where an additional unit of effort has the highest marginal impact on avoiding civilizational loss.

utility delta lens

State-of-the-art vs physics-possible

The ratio of current best solution to what physics and economics would allow. Inspired by Musk’s utility-delta framing. Favors hard tech with massive unrealized gaps. Energy, construction, transportation, bio.

Time and capital are order-of-magnitude estimates

Two questions a founder asks before anything else: how long will this take, and how much capital will it consume? We surface both, with the caveat that each is an order-of-magnitude estimate based on analogues, not a forecast.

For time-to-impact, we pattern-match against the closest historical arc — biotech base rates for longevity, IEA Net-Zero envelopes for energy, Gavi/Global Fund scaling curves for infectious disease. The number is years from today to civilizational-scale impact, not first product.

For capital required, we sum the R&D, deployment, and supply-chain build-out across the arc. These are always tagged low confidence and will move with new data. The point is to give founders and allocators a starting reference, not a closed answer.

Severity is willingness to pay over wealth

The severity of a problem to a person is what they would pay for a solution divided by what they have. Someone with $100 facing a $100-severity problem would spend the whole $100 to solve it. We proxy this from market data where possible, health-economics studies where not.

Every number carries a confidence tag

If we don’t know, we say so. Every numeric cell is tagged high, med, or low with a source and an as-of date. Unknown values are written as “unknown” and tagged low. No silent guessing.

This is the single rule that makes a reference asset trustworthy over decades. CoinMarketCap became the reference because it was transparent about methodology and did not sell rank placement. We are holding the same line.

Five tiers, for navigation not hierarchy

  • welfare floor: Copenhagen Consensus baseline. High-certainty, high-BCR interventions.
  • x-risk frontier: 80,000 Hours top-priority areas. AI, bio, nuclear, catastrophic risk.
  • hard tech: Stephens’ “good quests” from Founders Fund. Energy, construction, space, bio-manufacturing.
  • progress & abundance: Roots of Progress / Collison / Cowen canon. The meta-quests that compound.
  • emerging: massive problems under-indexed by EA and e/acc. Fertility, loneliness, epistemics.

What we refuse to do

  • Never sell ranking placement. No company pays to move up. No advertising that looks like endorsement. The moment this rule breaks, the dataset is worth nothing.
  • Never silently guess a number. If we don’t have a source, the cell is “unknown” and tagged low until we do.
  • Never collapse to a single moral score. Welfare, x-risk, and utility delta disagree. That disagreement is information, not noise.

Shoulders we stand on

optimism.fun is downstream of three open inspirations. Each one supplies a piece of the rubric we use to rank.

  • Founders Fund · “Choose Good Quests” — Trae Stephens’ rubric for what makes a problem worth attacking: important if it works, a real frontier, unambiguously good. Anchors our tier definitions.
  • Patrick Collison · Progress Studies — the case that progress is a study, not a vibe. Cowen & Collison’s 2019 essay set the field. The dense linkroll on that page is the spirit we want for our /voices and /progress.
  • Y Combinator · Requests for Startups — the format for turning a problem into a concrete buildable company. Our per-problem whitepapers follow the RFS voice: lead with the pain, name the unlock, describe the shape.
  • 80,000 Hours · problem profiles — the ITN composite (importance × tractability × neglectedness) we use as the x-risk lens.
  • Copenhagen Consensus — benefit-cost ratios for the welfare lens. Lomborg’s Nobel-laureate panel methodology.
  • Our World in Data — the gold standard for “every number sourced.” Many of our progress receipts trace back here.
  • Roots of Progress — Jason Crawford’s case that progress is the moral imperative of our era. Anchors the “progress & abundance” tier.
  • Works in Progress — the magazine of the progress-studies movement. Where many of the arguments behind the rankings get sharpened.

Phase 0 status

This is v0.1. 10 problems, ~50 companies, 15 ecosystem entities. Shipped in days, not months. Every cell is hand-curated and honestly tagged. Many cells are “med” or “low” confidence. Expected at this stage.

The public ledger gets refined weekly. Criticism is the point. If you see a number that is wrong, we want to know.