Charitable Giving Game Instruction Manual

How to play the charitable giving game

Table of contents

I. Introduction

  1. Welcome

  2. Overall Advice

  3. Your responsibilities

II. Elements of a game

  1. Players

  2. The Choice of Charities

  3. Context

  4. Voting Structure

  5. Prize

  6. Follow-ups

  7. Data Collection

III. Combining the Elements

  1. Basics of Game design

  2. Types of Games

IV. Higher Level Game Design Strategy

  1. Estimating the expected value of a game

  2. Multi game strategy



This document is intended to give guidance in designing and running Giving Games. It assumes a basic familiarity with the structure of these games, which can be obtained at

This document is intended to help the reader design and run Giving Games. It assumes the reader has a basic familiarity with the structure of these games, which can be obtained at is funding Giving Games to meet its goal of getting people to give more, and to give more effectively. While the games are designed to help players set goals for their giving, and help them design strategies to accomplish those goals, this document is written from the perspective of an Effective Altruist (EA).

This document should be viewed as a work in progress, and will be updated as we learn. Please send feedback or questions to, or discuss it on Basecamp.

Overall Advice

Before getting started, consider these general suggestions:

Prepare. Know what you’re trying to accomplish, and plan out how you intend to meet those goals. Try to “play the movie forward”, and see how things might unfold. Anticipate how players might react, and have your responses ready.

Think about the big picture. If a game doesn’t play out the way you intended, that doesn’t mean it was a failure. In the early stages of this project, we’re learning rapidly. We’re learning what to do, and what not to do. Outcomes that contradict our expectations are particularly helpful to learn from.

Keep records. Take notes. The more detailed a picture we have, the more we’ll be able to understand what’s important. “Small and apparently insignificant details can have major impacts on people’s behavior. A good rule of thumb is to assume that ‘everything matters.’” Be as attentive to recording problems as you are to successes.

Your responsibilities

Please understand that organizing games funded by carries certain responsibilities:

Treat players respectfully

Provide observations and feedback to

Inform players and sponsor of their mutual obligations (see Privacy Policy)

Elements of a game


Where to Find Players

You can find players anywhere. Some suggestions:

  • Friends and family. These people are likely to agree to a request, and you’ll be able to tailor games to suit their interests.

  • Groups with a collective identity (e.g. a sports team, a band, a group of friends, etc.) This is an efficient strategy, since you inherently get numbers. Target leaders or people with influence for additional leverage.

  • Random people. Set up in a high traffic area and invite people to play. [see “Pop-up” games in Types of Games]

  • People who are already gathered. Find settings with a lot of people that could be turned into an opportunity to talk about some EA issues. [see “Add-on” games in Types of Games]

  • People who show interest in a cause. If someone’s always talking about social justice issues, or even just wearing an Amnesty International button, they’d probably be open to a discussion.

How to invite players

Here are some suggestions on how to approach potential players

  • Be proactive. Actively seek out opportunities to invite players. The more people you invite, the more acceptances you’ll get. It can’t hurt to ask- it’s not a problem if people say no.

  • Listen to their input. Work with the players to come up with a game that fits their interests. Explain the basic concept, and work with the players to come up with an appealing game design that is likely to encourage better giving. If you’re asking them to play a specific game, get their feedback and suggestions.

  • Direct people to online games. If someone expresses interest in games, you can refer people to the online games at, which they can play at their convenience.

  • Tailor the game to the players’ interests. Explain to the invitee how effective giving connects to their interest. Understand the frameworks they’re comfortable with, and give them a choice that’s relevant to their mission. If you’re inviting a group of volunteer tutors, give a choice of great education charities rather than microfinance vs. soup kitchen.

  • Prepare a simple negotiation strategy. Have a plan, and you’re guaranteed to be more prepared than the person you’re talking to. Anchor their expectations low, then beat them. Think about how you can get the most bang for the buck from your resources (budget and time).

  • Keep records. Take notes about who you invite, how you invite them, and how they respond. Try to record as much detail as possible. We’ll analyze across many peoples experiences, and distribute tips on best practices.

  • Explain your own interest. Help the invitee understand why this is important to you, which will increase their social obligation and their willingness to participate.

The Choice of Charities

How to Frame the Choice of Charities

Try to frame the players’ choice as “Good vs. Good” or “Good vs. Better” rather than “Good vs. Bad”. That said, certain charity pairings will inescapably be viewed as having a right and wrong answer. If an EA runs a game pairing the players’ student organization vs. AMF and tries to frame the choices as equally valid, they’re likely to insult the players’ intelligence. In that case, the EA should try to emphasize that they’re differently valid; each is the right choice by a particular line of reasoning. This is different from suggesting that one choice is valid and the other isn’t, which you should avoid and discourage.

If the players think the choice is a bad one, just give them a better one! At the outset, or on the fly, you can introduce a rule that a (super?)majority of players can vote to change the pairing. Their complaint will presumably be that one charity isn’t good enough. Let them suggest new charities, and you can offer ideas of your own. If the game turns into a brainstorming session about high impact giving opportunities, that’s great!

Suggestions for Charities to use is a highly respected charity evaluator, and is a widely used resource among the EA community. GiveWell is focused on identifying excellent giving opportunities, and their recommendations can be found here.

Other ideas for great charities include:

  • Deworm the world

  • GiveWell

  • Future of Humanity Institute

  • Ploughshares Fund (nuclear disarmament advocacy)

  • Compassion in World Farming

  • Nurse Family Partnership

  • Singularity Institute

  • Wikipedia

  • Khan Academy

  • TED Talks

  • Acumen Fund

Some games might pair a great charity with a charity that players might otherwise give to. These “default” charities might be large, well known organizations, or causes that the players have an obvious connection to. Some example might include:

  • Oxfam

  • Red Cross

  • American Cancer Society

  • Alma mater

  • Religious causes

  • Etc.


This element of the game encapsulates everything the players do before making their decision (e.g. online survey, having a group discussion, or listening to a lecture). The information players are given, and the environment they are put in, will both impact the way they think about the question at hand.

Information about the charities

Think about what type of information players will want when making their decision, and anticipate the questions they might ask. Try to think about how different types of donors will perceive different types of information. Don’t just give players information, give them compelling reasons why they should believe that information.

Activities to engage players

Games typically offer players something to do or something to talk about. Be creative, and feel free to experiment with ideas that might get people interested in effective giving.

If you’re looking for a ready-to-run activity, THINK has created a selection of EA learning modules which are very well structured and have detailed instructions. These can easily be adapted into games, and are available here:

An easy way to engage players is simply to get them to participate in a conversation about an interesting topic that can be related to an EA concept. Keep an eye out for interesting articles/videos/etc. that could be the basis for a good conversation. You can find links to some interesting materials here: Alternatively, you can simply run a live version of one of the online games (, by having a group discussion around their content.

Positively reinforce “good” behavior. This doesn’t mean you should interrupt the flow of the conversation, but particularly at the beginning and end of the game you’ll have opportunities to do this. Offer praise and thanks. Remind people of the good the donation will do, and that they’re helping a broader cause of understanding giving behavior.

Tips for moderating a discussion

Discussions are an important part of many games. If you’re moderating or participating in a discussion, please consider the following advice:

Prepare. This is most relevant for the introduction, which you’ll have the most control over. You can have some notes, an outline, or other talking points to use as reference, and it will be very helpful if you practice reading them through a couple of times. Also, get acquainted with the location beforehand if possible, and try to make sure it’ll work for the game (enough seats, any necessary AV equipment, etc.)

When the time comes, relax and try to have fun. As far as the players are concerned, you know what you’re doing. They’ll take their lead from you, so try to encourage an enjoyable atmosphere.

Don’t worry, it will all be over quickly. If the game is an hour long, there might only be 30 minutes of discussion once the introduction, voting, and surveys are accounted for. Most of the discussion is likely to focus in predictable areas, as there won’t be enough time to have a group discussion that gets into really deep nuances.

The players will probably do most of the talking. Discussions so far have been active and free-flowing (and at times, a little heated). There’s been little need for moderators to stimulate conversation.

If you need to stimulate conversation, you can always play devil’s advocate. The conversation is most likely to stall if there’s too much agreement, and therefore no need for discussion. You can introduce some disagreement by introducing alternative viewpoints. You might say something like “Some donors would think about things XYZ way. Is that a good argument? Why or why not?” Or, you can just push people along the “give more money to better charities” spectrum, and see where resistance comes up. You can also generalize beyond the donation at hand, and see where resistance emerges there.

A less likely scenario is that people are so far apart in their views that discussion stalls out. If this happens, try to find a middle ground that both sides can relate to. For instance, if the two main viewpoints boil down to “focus on education” vs “focus on third world health”, you can talk about how childhood health improves education results, or how (all else equal) it’s better to educate people to do things like solve third world health problems.

Avoid the blame game. For the discussions to go well, people need to feel comfortable expressing their honest opinions, even if they’re unpopular. Don’t insult or dismiss views that are inconsistent effective altruism; try to learn why people hold those views. Similarly, encourage players to act respectfully toward each other.

Positively reinforce “good” behavior. This doesn’t mean you should interrupt the flow of the conversation, but particularly at the beginning and end of the game you’ll have opportunities to do this. Offer praise and thanks. Remind people of the good the donation will do, and that they’re helping a broader cause of understanding giving behavior. Ideally, players will build associations between generosity and happiness.

Help make sure everyone who wants to talk can. Some people will be more forceful than others. If someone wants to talk but is getting talked over, help them get their say in.

Don’t worry if the Game goes in an unexpected direction. Unexpected is good, it means we’re learning something new. The Games are short enough that things can’t get too far astray, so just go with the flow and try to make the game a positive experience.

Use arguments suited to the audience. Some people will respond to a hyper-rational argument. Some won’t. Donors give for different reasons. Try to understand how different players think about philanthropy, and try to suggest arguments that will resonate with each philosophy. Different people can get to the same answer by different paths. Don’t expect everyone to find statistics compelling (even when they “ought to be”).

Having an observer present will help you focus on moderating. An observer can take notes, and can offer objective feedback to the moderator afterwards. This way the moderator won’t be trying to do too things at once.

Voting Structure

How to structure the voting

When thinking about which voting structure to use, be sure to consider a few key issues.

Public vs. Private Voting

Public voting is easiest from a logistical standpoint. However, some people might feel resistant to expressing an unpopular view.

Giving players an opportunity to submit confidential/anonymous feedback is a good way to mitigate this drawback.

Anonymous voting is likely to elicit the most honest voting from players.

However, this structure might make players hesitant to give their opinions during discussion, since it might “give away” their answers; in other words, this structure can implicitly validate the notion that charitable decisions shouldn’t be discussed.

#### Proportional vs. Winner Take All

Proportional structures empower each player equally, and every player’s decision leads to real world consequences.

“Winner take all” votes are more likely to be interpreted as a collective decision, but might alienate the minority

As we learn to how to design environments likely to encourage a majority of players to give effectively, we’ll be able to use “winner take all” votes to steer more money toward the most effective charities.

Subtler design choices also matter. For instance, if players reveal their votes sequentially, later players might be influenced by how others have already behaved. Before and after voting is particularly helpful. By taking a preliminary vote prior to any discussion or activities, we’ll be able to directly observe whether and how the Game changed people’s decisions.


How big should the prize be?

When trying to set the prize, try to think about what kind of incremental benefit you’ll get with each marginal dollar. All else equal, smaller prizes are better since they’ll allow us to reach more people with the same bankroll. But the prizes should be big enough to get most players to engage with the question. Here’s what early experience has shown:

Online games have varying prizes. Some games have zero marginal cost per player.

Small amounts seem to work for quick live games. At an activities stall where each player allocated $1, “some people said they were overwhelmed by the power of their choice”

Groups of 10-30 people have responded positively to prizes ~$250, and would probably accept smaller amounts. We should try offering prizes of $50-$100, and test the response.

EA Chapters and similar groups have been willing to set up games when offered annual budgets ~$1000-1500. Modestly smaller budgets would likely get a similar response.

For larger groups, or more involved games, larger prizes will likely be appropriate. For example, early plans for a campus wide event discussed prizes between $2000 and $5000.

Larger prizes might be appropriate if they help raise additional charitable funds from players or other sources (see “Skin in the game” games in Types of Games).

Donation Logistics

To simplify the logistics, we’ll try to avoid making lots of tiny donations. When a given charity is owed $50 or more, we’ll make a donation. To make sure that every promised donation is made on a reasonable schedule, every 6 months (on June 1 and December 1 of each year), we’ll clean up any donations that have been lingering under the $50 threshold.

Donations over $50 will be made through a “Donor Advised Fund”, to provide independent confirmation that the donations are being made. The fund is called the Just Compassion Initiative, and is administered by Fidelity Charitable. The money I’ve put into this fund already has to be donated to charity, so in a sense it’s like it’s in escrow and players don’t need to worry that I won’t put up the money. Due to how DAFs are structured, the donations in the Game should not be represented as legally binding contracts.

Donations can be made “In honor of XYZ” or “In XYZ’s name.” If a particular dedication should be used, please indicate this on the Post Game Report Card, available here:


Why you should offer Follow-ups

After the game, players will already be in a generous mood and thinking about charity. Use this opportunity to encourage players to build on the experience. Try to make it easy and enjoyable for them to be a great donor, and more generally, a great person. Use what you learn about players during the game to suggest opportunities that fit their interests.

Suggested Follow-Ups to offer

  • Join EA chapter (organizations with giving requirements offer a better indication of future donor behavior)

  • Join GiveWell mailing list or subscribe to givewell blog

  • Join mailing list

  • Attend meeting

  • Play another giving game

  • Tell friends about giving games

  • Share and with social networks

  • Set up a one on one talk with one of the organizers or another EA

  • Suggested reading

  • Always Be Closing… It’s better, for example, to sign someone up for GiveWell’s mailing list on the spot than to ask them to do so in the future.

Data Collection

Why we’re collecting information

Each game should collect information that will help us interpret the results, and learn how to improve future iterations. We’ll try to standardize aspects of the data collection to facilitate cross-game analysis, though different games will collect different amounts and types of information.

What to collect

  • Voting results. How many people voted for each option.

  • Player surveys. Surveys are a great way to gather feedback about the games, and learn more about the players.

  • [link to template survey, which should include link to privacy policy].

  • Metrics on follow-ups. What follow ups were offered? How many people accepted each option?

  • Audio/video Recordings. When feasible, try to record the games. Video is preferable, but audio is probably easiest, and can be done with a smartphone. Recordings are incredibly valuable in analyzing games.

  • Waivers: see Privacy Policy below

  • Observer Assessments. Whoever runs the game should write up their thoughts. Many games also include an observer, who should do the same. The more perspectives we can get, the better.

  • The primary organizer should fill out this Postgame Report Card.

Privacy Policy


  • Respect the privacy of the players and other participants

  • Provide an environment that provides honest discourse

  • Collect sufficient information to conduct effective post-Game analysis and discussion.

These goals can obviously come into tension with each other. The less information we’re allowed to obtain or disclose, the less we’ll be able to interpret the results or explain them to others. So the key is to find a sensible balance between these goals. We can use the following principles to get started, and refine them based on feedback.


By default, we should not publish any Player names without explicit permission. So, for example, quotes from discussions and surveys that are posted on the website should all be anonymous.

Players should know what they’re agreeing to, and we should get that on record. That could be signing a waiver, or a voice-vote agreement that’s captured on audio tape, having an accessible privacy policy on the website, etc.

Be reasonable and use common sense. Think of what the Players might reasonably expect, and try to respect those expectations. There’s usually a sensible balance.

For example, our first pilot game was with a Princeton singing group. They were OK with that level of description, but didn’t want the name of the group to be publicized. That’s a reasonable solution that meets everyone’s needs.

Best practices

As mentioned above, try to get some sort of paper trail of consent.

If you think something is borderline or questionable, escalate it. Raise a question to other people you’re working with, or email

If a privacy agreement is a sticking point in setting up a game, use this as a learning opportunity. Try to understand the underlying motivations. You can also emphasize that we’re planning on playing a lot of games, so each data point will only get so much attention.

Combining the Elements

Basics of Game design

The first step in designing a good game is to set one or more goals (e.g. raise the expected social value of players’ lifetime giving). Then, try to combine the elements of the game in a coherent fashion to meet your goals. Certain combinations work better than others- don’t ask players to listen to lecture about malaria, and then choose between a charter school and a microfinance organization.

Try to be explicit about your logic, and why you think your design will work. Write down the assumptions you’re making, and revisit them after the game.

If you’re having trouble getting started, pick one element of the game to begin with. As you get a better idea of what you want to do for that element, other aspects of the design will become clearer.

Types of Games

Here are some example frameworks to help get you started thinking strategically about designing games. These overlap considerably, and aren’t meant to be exhaustive.

Win/Win games ask players to choose between two exceptional charities, ensuring that the donation will be used as efficiently as possible. Examples:

  • AMF vs. SCI, GiveWell’s two top recommend charities.

  • Players vote between a best in class world poverty charity and a best in class existential risk charity

  • Suggestive games are constructed to imply a particular framework for thinking about charity. Examples:

  • After a lecture about the cost effectiveness of third world interventions relative to first world interventions, players vote between AMF and US based Nurse Family Partnership

  • An EA chapter organizes a game between AMF and Make a Wish Foundation

Group discussion games feature a conversation between the players. This discussion can be led by a moderator, or freeform. Examples:

  • A Resident Assistant moderates a discussion with their residents

  • An EA starts a game during a lunch conversation with friends

Online games at can take many forms, are distinctive in that they allow people all over the world to collaborate, and allow individuals to play a Giving Game at their own convenience. Online games are also highly anonymous. Examples:

  • Players choose between two charities related to a TED talk they watch before voting

  • Players take a survey about their giving philosophies before voting

Experimental games are designed to test hypotheses by manipulating one or move variables, and observing the impact on people’s behavior. Examples:

  • An online game randomly assigns players to receive “Emotional” or “Logical” descriptions of the charities.

  • An Organizer wants to improve a promising game structure that didn’t generate many follow-ups. They run a similar game, but add food and drinks to encourage players to linger over the signup sheets.

Activity games have players engage in a task before voting. Examples:

  • Players watch a Bill Gates TED talk about malaria and US charter schools, then vote between AMF and KIPP.

  • At an EA themed dinner party, guests use THINK’s DIY Charity Assessment module (which teaches the difficulty of assessing charities using intuition), and then vote between AMF and SCI, the two top recommended charities of the top charity evaluator.

  • Skin in the Game contests ask participants to contribute some, or all, of the donation.

  • Each player contributes $5 to the donation, and matches their contribution.

  • A featured lecturer uses some, or all, of their honorarium to fund the Prize.

  • Add-on games find a pre-existing conversation, and attach a relevant giving game.

  • After a lecture about third world poverty, attendees are allowed to donate to AMF or SCI

  • Organizers of a charity softball tournament let the participants vote between two great causes to support with the proceeds

Viral games try to engage as many players as possible by encouraging players to recruit more people. Examples:

  • After an online game, Players are asked to forward a link to the game to their friends, with the Player who refers the most friends winning $50 to the charity of their choice.

  • A student organization chooses between giving $100 to AMF, keeping it for their own budget, or using the money to run games of their own to support the causes they care about

Topical games use current events to start a conversation about charity. Examples:

  • In the aftermath of a natural disaster, EAs set up a stall asking people to contribute to the most effective charity helping the impacted people. Afterwards, people who donate are given funds that they can donate to that same charity or a charity like AMF that helps people who suffering with less publicity.

  • After the US election, Organizers invite the campus Democrats, Republicans, Independents, etc. to try and find areas of agreement by playing a game.

Pop-up games involve simply setting up shop in a high traffic area and inviting people to play. Examples:

  • An EA organization sets up a stall at a student activities fair, and invites people to play a quick Giving Game

  • An EA goes around the quad starting conversations, and invites interested people to play on online game at their convenience

Higher Level Game Design Strategy

Estimating the expected value of a game

The “expected value” (EV) of a game is simply how much “good” it will cause. Of course this concept is nebulous and impossible to measure (since we can’t know how players would have behaved having never played). However, we can estimate a game’s EV by thinking about a few basic questions.

How likely is the game to change the amount of money players donate?

If they donate differently, how much more (less) will people give?

How likely is the game to change where players donate?

If they do donate to different causes, how much better (worse) are they?

This framework helps explain certain common sense conclusions. For instance, a game that’s expected to get people to give to better charities will have a higher EV if it targets Economics majors than Poetry majors. Since the former group will likely donate larger dollar sums, improving the impact of their donations will lead to more social benefit.

Multi game strategy

Another crucial component of a game EV is the information it provides, and the impact that information will have on the EV of future games. At these early stages in this project, this component is magnified, as each game represents a significant portion of our experience. Furthermore, a robust sample size of games will help garner additional support.

This boils down to a simple rule: Iterate, Iterate, Iterate! The more games we play, the faster we’ll learn.

At later stages, it will be more important to think about the opportunity cost of each game (e.g. whether the funds would better be spent on a higher EV game).

This article was originally published here.
This article is under a Creative Commons Attribution-ShareAlike 3.0 license.