Status: stub to get me thinking before an EA retreat—please suggest more and I will add them here
People in the effective altruism community (and adjacent communities) often use terms that others do not use often. There are reasons this can make sense, but it’s important to clarify what they mean if talking to someone who might not know, especially if they’re new to EA.
Since it’s easy to forget how uncommon these terms are outside EA, I’ve tried to jolt my memory, so that it’s easier for me to remember to clarify them if I use them. People can Google terms and see EA Forum definitions when these are in writing, but it’s hard to do that during a conversation.
Method: I set a 5-minute timer and wrote as many terms I could think of for each list that, in my experience, are
(a) somewhat commonly used in EA and adjacent communities,
(b) used outside them much less commonly,
(c) not extremely obvious (e.g. ‘high-status’).
Then I set another timer and tried to find more.
I’ve included terms common in specific academic disciplines (especially economics and philosophy) but not outside. I’ve excluded terms coined by EA (e.g. ‘X-risk’), since I find it more obvious to clarify these.
Non-technical
Agentic/agenty
Upside risk
Not obvious to everyone because ‘risk’ is traditionally negative
Aversive
High-friction
Technical but not necessarily coined by EA
Epistemics
Counterfactual
Info-hazard / info-hazardous
Update / updated / updating—in the (pseudo-)Bayesian sense
Prior—in the (pseudo-)Bayesian sense
Watch out especially for ‘On priors, …’
Weakly x-er—in the sense of ‘as x as, or x-er’, e.g. weakly better, weakly greater
Risk-neutral
Scope-insensitive / scope-sensitive
Calibrated / well-calibrated—in the sense of making predictions such that, for all x, x% of predictions on which an agent puts x% probability actually happen
Terms to clarify when discussing EA
Status: stub to get me thinking before an EA retreat—please suggest more and I will add them here
People in the effective altruism community (and adjacent communities) often use terms that others do not use often. There are reasons this can make sense, but it’s important to clarify what they mean if talking to someone who might not know, especially if they’re new to EA.
Since it’s easy to forget how uncommon these terms are outside EA, I’ve tried to jolt my memory, so that it’s easier for me to remember to clarify them if I use them. People can Google terms and see EA Forum definitions when these are in writing, but it’s hard to do that during a conversation.
Method: I set a 5-minute timer and wrote as many terms I could think of for each list that, in my experience, are
(a) somewhat commonly used in EA and adjacent communities,
(b) used outside them much less commonly,
(c) not extremely obvious (e.g. ‘high-status’).
Then I set another timer and tried to find more.
I’ve included terms common in specific academic disciplines (especially economics and philosophy) but not outside. I’ve excluded terms coined by EA (e.g. ‘X-risk’), since I find it more obvious to clarify these.
Non-technical
Agentic/agenty
Upside risk
Not obvious to everyone because ‘risk’ is traditionally negative
Aversive
High-friction
Technical but not necessarily coined by EA
Epistemics
Counterfactual
Info-hazard / info-hazardous
Update / updated / updating—in the (pseudo-)Bayesian sense
Prior—in the (pseudo-)Bayesian sense
Watch out especially for ‘On priors, …’
Weakly x-er—in the sense of ‘as x as, or x-er’, e.g. weakly better, weakly greater
Risk-neutral
Scope-insensitive / scope-sensitive
Calibrated / well-calibrated—in the sense of making predictions such that, for all x, x% of predictions on which an agent puts x% probability actually happen
Marginal—in the economics sense
Ex ante
Prisoner’s dilemma
Pascal’s mugging
Megaproject
High-EV / Low-EV / EV-positive
Less commonly:
Unilateralist’s curse
Optimiser’s curse
Utilitarian
Game-theoretic
Efficient markets hypothesis