What People Do: How We Score

Please read this first

What People Do is an experiment, not a verdict. Its real goal is to explore whether the good and harm a person’s actions do to all life on Earth can be measured at all, and to start a conversation about what counts as positive and what counts as negative.

I am not trying to say anything good or bad about anyone. I’m experimenting with ways such a measurement might be made, if one is even possible, and it very likely is not.

  • The effects of actions are hard to know. They spread out over decades, mix with what millions of other people do, and can turn from good to bad, or bad to good, in ways no one can foresee. In the long run, judging them may be impossible.
  • Every rating is a judgment call. Two careful people could rate the same action differently.
  • The formula is a simplification. It squeezes complicated, uncertain effects into a single number.
  • So please don’t take the numbers seriously as facts or as a measure of anyone’s worth. They’re a starting point for discussion, nothing more.

Facts and opinions are kept apart. Statements of fact on these pages come from the published sources cited next to them, so you can check them and draw your own conclusions. The ratings, scores and lists are opinions: the result of applying this experimental method. The people discussed have no connection to this site and have not reviewed or endorsed it.

Corrections are welcome. If you think something is wrong, missing or unfairly rated, please say so, with a source if you can, and the page will be updated.

Every What People Do post scores a person’s major actions the same way, so people can be compared fairly. This page explains how each action is chosen, scored and sorted, and how the graph at the bottom of each post is made.

What we mean by “all”

We judge each action by its effect on all life on Earth: people, other animals, plants, whole ecosystems, and the climate they all depend on, now and in the future. An action that helps people but harms ecosystems is not counted as purely good, and the reverse is true too.

Which actions we include

  • Major actions only. An action must affect, or realistically could affect, at least millions of people or an area the size of a country. That means a scale score of 3 or more (see below).
  • Things the person actually did. This means actions they led, funded, founded or directly controlled, including through companies or foundations they ran. Actions of a company after the person left it don’t count.
  • Up to 100 actions per person, from the most recent to the oldest, starting with the most important. Lists grow as research continues.
  • Every action needs a source. Each one links to at least one reliable public source, listed under References on the post.

The six ratings

Each action gets six ratings:

RatingQuestionValues
Direction (D)Overall, does it help or harm?−1 clearly harmful · −0.5 mostly harmful · −0.25 slightly harmful · 0 mixed or unknown · +0.25 slightly helpful · +0.5 mostly helpful · +0.75 strongly helpful · +1 clearly helpful
Scale (S)How many lives, or how much of the planet?1 thousands / a town · 2 hundreds of thousands / a region · 3 millions / a country · 4 hundreds of millions / a continent · 5 billions / the whole planet
Duration (T)How long do the effects last?1 under a year · 2 a few years · 3 about a decade · 4 a generation · 5 a century or more, or permanent
Breadth (B)Which kinds of life does it affect?Count of: people · other animals · plants and ecosystems · climate (1 to 4)
Certainty (C)How sure is the evidence?1.0 well documented · 0.8 mostly documented · 0.6 disputed, early or uncertain
Role (R)How much of the outcome is this person’s own doing?1.0 led it · 0.75 a major part · 0.5 one of several · 0.25 a small part

The score

First we work out how big the action is, its impact, from 0 to 1. Scale counts the most. Duration and breadth count equally after that:

Impact = (S ÷ 5) × ( ½ × T ÷ 5 + ½ × B ÷ 4 )

Then direction, certainty and role turn that into a score from 0 to 100, where 50 is neutral:

Score = 50 + 50 × D × Impact × C × R

The best possible score is 100, for something clearly helpful, planet-wide, permanent, good for all kinds of life and well documented. The worst is 0, for the same kind of action when it is clearly harmful. Small, short or uncertain actions stay near 50 whichever way they point.

Why role matters: big outcomes usually have many causes. A donor who gave a few percent of a program’s money shouldn’t get full credit for everything it achieved, and a celebrity who endorsed a decision made by others shouldn’t get full blame for it. Role scales each score toward 50 by the person’s share. Judging that share is itself uncertain, which is one more reason to treat the numbers as rough.

Which list an action goes on

  • Positive: score of 55 or higher
  • Neutral: score between 46 and 54. The effect is mixed, small for the person’s share, or not yet clear.
  • Negative: score of 45 or lower

Within each list, actions are shown newest first, with the score in a colored badge next to each one.

Worked example

Bill Gates: launching Gavi, the Vaccine Alliance. Clearly helpful (D = +1). Reaches children across the world’s lower-income countries, billions of people in total (S = 5). Lives saved last a generation (T = 4). Mainly affects people (B = 1). Well documented (C = 1.0). His foundation founded and seeded Gavi, but governments and others now fund most of it, so he played a major part (R = 0.75).

Impact = (5 ÷ 5) × (½ × 4 ÷ 5 + ½ × 1 ÷ 4) = 1 × (0.4 + 0.125) = 0.525
Score = 50 + 50 × 1 × 0.525 × 1.0 × 0.75 = 69.7 → 70 (Positive)

How the graph is made

  • Across (x axis): the 100 years from 1926 to today.
  • Up (y axis): score, from 0 (most harmful) to 100 (most beneficial). The dashed line at 50 is neutral. Green shading (55 and up) is the Positive zone, and red shading (45 and down) is the Negative zone.
  • Each action is “active” from the year it began to the year it ended, or to today if it is still going.
  • Each dot is the average score of all actions active that year. For example, if three actions were active in 2005 with scores 76, 75 and 46, the dot for 2005 sits at (76 + 75 + 46) ÷ 3 = 65.7.
  • Bigger dots mean more actions were active that year. Years with no active actions are left blank, so the line has gaps.
  • Hover over the graph, or tap it on a phone, to see the actions behind any year and their scores.

The graph shows the average character of what someone was doing over time, not their total impact. A person doing many helpful things and one harmful thing will show a high average. The lists above the graph show the full picture.

Limits and honesty

  • Ratings are careful judgments, not exact measurements. Reasonable people could rate some actions a step higher or lower.
  • Some start and end years are approximate. This is noted in the action’s description when it matters.
  • Effects on animals, plants and climate are often less studied than effects on people. When the evidence is thin, we lower certainty instead of guessing.
  • We update scores when better evidence appears. Suggestions with sources are welcome.