Too many people in EA underprice the cost of a bad reputation

Summary: Good work helps build a good reputation, but it's not the only factor. Reputation also requires making your work legible, building relationships with stakeholders, and communicating with them consistently over time. I think many people in EA systematically underestimate what a bad reputation costs, largely because those costs are invisible. You can see the people who stick around. You can't see the people who decided not to work with you. I close with some practical suggestions for weighing reputation in decisions.

Notes on writing: This is more of a hot-take, quickly written, with some Claude assistance in shaping it into a full post, but heavily edited and reviewed by me. This was spurred by Manifund's recent announcement but is something I generally have been wanting to write. I've written this in my personal capacity, and it's informed by working 5 years in senior roles at EA orgs.

Good work is necessary, not sufficient

Reputation is everything. It's a cliché for a reason.

Doing good work is one of the main ways to build a good reputation, and there's an implicit belief I've noticed in many people in EA that it's the only thing that should matter. If the work is good, the thinking goes, the reputation will follow — and if it doesn't, that's a failure of the world rather than a failure of ours.

I think that's naive. A good reputation requires making your work understandable to people outside your own context. It requires building relationships with the stakeholders who matter to your goals. It requires communicating your work to them, and then continuing to communicate over time, because trust is built by repetition rather than by a single impressive output.

What a good reputation actually buys you

When people trust you, they help you succeed. They amplify your work. They partner with you. They introduce you to the people who can do the things you can't.

A good reputation isn't a nice-to-have sitting alongside impact. It's one of the inputs to impact.

The converse follows. A bad reputation damages trust and shrinks the set of people willing to work with you. That isn't merely a reputational cost — it can be, and I think often is, a direct reduction in what you're able to achieve.

The invisible costs of a bad reputation

When your reputation takes a hit, what you observe is the trust that remains. You see the people who stick by you, who tell you the criticism was overblown, who keep collaborating. What you don't see is the person who read one article and decided not to reply to your email. And you especially don't see the counterfactual: the partnerships, hires, funders and introductions down the line that would have existed had your reputation been better.

These costs are a known unknown, and I think the general bias towards measurement in EA leads people to believe these costs are smaller than they are.

I can't hand you a number proving I'm right and others are wrong. I'd only point out that the absence of a number isn't evidence in the other direction.

Being weird about this doesn't scale

This is a case where EA should look at how the rest of the world treats something, and be less weird.

Frequently, our ability to do good work depends on partnering with people who are more conventional than we are, like in government, in philanthropy, in academia, in the organisations that actually implement things at scale. Those people take reputation seriously. A culture of not caring much about reputation works fine while you're a small group talking mostly to each other. It doesn't scale to a movement that wants mainstream institutions to act on its recommendations.

And working with the mainstream matters, because the mainstream in turn shapes what the people downstream of it — often the more risk-averse and reputation-sensitive ones — treat as credible.

There's obviously a real cost on the other side of this. Organisations that weigh reputation heavily can end up timid — deferring to conventional opinion, softening claims that ought to be made plainly, declining to say true things because someone will object. That's a failure mode, and one I have seen before in EA. But overall, I'd argue for more focus on reputation than less, as I still think it's systematically underweighted.

The people we lose

I also think it's a mistake to treat reputation-sensitivity and risk-aversion as signs that someone isn't a good fit for effective altruism.

Plenty of thoughtful, careful, high-integrity people care what their colleagues and their professional community think of them. That's usually not irrationality in need of correction. Often it's an accurate read of how their career, their institution and their ability to get things done actually work. When we treat that caution as a deficiency, we don't win the argument — we select for people who already resemble us, and lose the people who could have given us reach we don't otherwise have.

It's easy to mistake a high tolerance for reputational risk for superior rationality, when it may only reflect having less at stake.

What to do about it

Give reputational consequences the same seriousness we already give other predictable costs.

Treat reputational impact as a real input to decisions. When a hire, a grant, a partnership or a public statement is being weighed, reputational effects tend to come up late and get filed under optics. Give them a place in the analysis from the start, and be willing to let them change the answer. A consideration that never changes a decision isn't really being considered.

Widen the list of stakeholders. Most of us model the reactions of the people we already talk to. The ones who matter are often further away: a public servant, a journalist, a prospective hire's partner, an organisation that might want to co-sign something with us in three years.

Model the reaction, or just ask. Predicting how an unfamiliar audience will read something is hard, and we're often bad at it. Show the draft, float the plan, have the conversation in advance. A few phone calls beforehand cost far less than a correction afterwards, and the answers are frequently not what you expected.

Think on a longer timeline. Reputational damage is paid slowly, in emails not returned and invitations not extended, often years later and often by people who only absorbed the impression the original episode left behind. Ask what a decision looks like in five years, to someone hearing about it secondhand.

There's no silver bullet here, and no easy way to magically strike the correct balance, but I think in many cases more proactively taking reputation into account when making decisions will create more good.