The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • LW - Big Mac Subsidy? by jefftk
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Big Mac Subsidy?, published by jefftk on February 23, 2023 on LessWrong.
    I was reading a
    Faunalytics report on people who stopped being vegan or vegetarian,
    via the EA Forum, and was surprised to see:
    by some estimates, a Big Mac would cost $13 without subsidies and a pound of ground meat would cost $30.
    This seemed much too high to me: I know we subsidize meat heavily, but that's a much bigger ratio than I remembered seeing. Clicking through to the source, and 2022 AEIR (angle: anti-subsidy) blog post I see:
    Research from 2015 shows this subsidization reduces the price of Big and the price of a pound of hamburger meat from
    Macs from $13 to $5
    $30 to the $5 we see today.
    Clicking through to their source, a 2015 Berkeley Entrepreneurship & Technology
    Technical Report (angle: anti-climate change) I see:
    The U.S government spends $38 billion each year to subsidize the meat and dairy industries, but only 0.04 percent of that (i.e., $17 million) each year to subsidize fruits and vegetables. A $5 Big Mac would cost $13 if the retail price included hidden expenses that meat producers offload onto society. A pound of hamburger will cost $30 without any government subsidies.
    This is uncited, and not at all plausible. In 2013 a Big Mac cost $4.56, so they're claiming $8.44/burger. There are about 550M Big
    Macs sold annually, so the total Big Mac subsidy would be $4.6B. That's 12% of the entire $38B. And Big Macs are only 0.4% of US beef consumption (25.5B lb of beef, 1/5lb each, 550M/y) let alone all the other subsidized foods.
    I did a bit more looking and found David Simon claiming in his 2013 book Meatonomics that the retail price of a big mac would be $12. This is close enough that it might be the source of the claim. He breaks it down as:
    $0.38 for cruelty. A total of $20.7 billion in cruelty costs is imposed on Americans each year. (Extrapolated from a study in which auction participants bid to end cruel farming practices.)
    $0.67 in environmental losses. This is a small piece of the $37.2 billion in annual environmental costs related to U.S. animal food production each year. The figure includes the costs of soil erosion, climate change, damage from pesticides and fertilizers, devaluation of real property, and manure remediation.
    $0.70 in subsidies. Toss in a few coins from the $38.4 billion in government subsidies that American taxpayers pay to fund the meat and dairy industries each year.
    $5.69 in health care costs. The biggest slice of the pie is a chunk of the $314 billion in health care costs incurred by Americans each year to treat those cases of cancer, diabetes heart disease, and food poisoning related to meat and dairy consumption.
    Now, health care costs are not what I'd normally consider a subsidy, and Simon doesn't claim it is one. But it also looks like each of these national numbers was converted into a per-Big Mac number by assuming Big Macs are responsible for 1% of the total:
    550M / ($20.7B / $0.38) = 1%
    550M / ($37.2B / $0.67) = 1%
    550M / ($38.4B / $0.70) = 1%
    550M / ($314B / $5.69) = 1%
    While 1% is not as implausible as 12%, that's still oddly high. If we look just at the health claim, I don't know where they're getting
    $314B but I see Springmann 2018 claiming $71B for red meat. This would make Big
    Macs responsible for 6.5% of the total health impact of red meat despite being only 0.4% of beef consumption (see above).
    In short, this statistic is junk.
    Comment via: facebook, mastodon
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    5 min
  • EA - Faunalytics Analysis on Reasons for Abandoning Vegn Diets by JLRiedi
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Faunalytics Analysis on Reasons for Abandoning Vegn Diets, published by JLRiedi on February 22, 2023 on The Effective Altruism Forum.
    Nonprofit research organization Faunalytics has released a new analysis on reasons people abandon vegan or vegetarian (vegn) diets, looking at the obstacles former vegns faced and what they would need to resume being vegn.
    Although causes for lapsing have been analyzed to an extent, a deeper analysis that considers people’s reasons in their own words is necessary to not only understand why people give up their vegn goals, but to find the best ways to help people stick with their commitment to vegnism and even lure back some of the lapsers.
    Read the full report here:
    Background
    People have a variety of motivations for switching to plant-based diets, yet not all people who begin the transition to a vegan or vegetarian (collectively called vegn) diet maintain it long-term. In fact, Faunalytics’ study of current and former vegns (2014) found that the number of lapsed (former) vegans and vegetarians in the United States far surpasses the number of current vegns, and most who lapse do so within a year. Are these people the low-hanging fruit for diet advocates? They could be—there are many of them and they’re clearly at least somewhat willing to go vegn, so maybe more attention should be paid to the lapsers.
    That’s one possibility. The other, more pessimistic possibility, is that when we as advocates think our diet campaigns are successful, these are the people we think we’re convincing. That is, we see the part where they go vegn, but not the part where they later lapse back. This interpretation is one that a lot of people made when our study of current and former vegns released, but we don’t have strong evidence either way.
    This analysis, in which we looked at the obstacles faced by people who once pursued a vegn diet and what they would need to resume being vegn, aims to shed a bit more light on these questions. Although causes for lapsing have been analyzed to an extent, a deeper analysis that considers people’s reasons in their own words is necessary to not only understand why people give up their vegn goals, but to find the best ways to help people stick with their commitment to vegnism and even lure back some of the lapsers.
    Research Team
    This project’s lead author was Constanza Arévalo (Faunalytics). Dr. Jo Anderson (Faunalytics) reviewed and oversaw the work.
    Conclusion
    Diets Are More Than Food
    Food plays an important role in our lives. More than just nutrition, food is a very personal yet social experience, a cultural identity, and at times, a religious or spiritual practice or symbol. Naturally, a good-tasting diet is important—especially when the idea is to maintain it long-term. However, lapsed vegns’ answers suggested that food dissatisfaction, although a very common struggle, was not the most crucial obstacle to overcome to return to vegnism. Instead, having access to vegn options, as well as the time and ability to prepare vegn meals (often alongside non-vegn meals for family), were much more common must-haves.
    Additionally, people’s feelings of healthiness while on their diet, seemed to hold a lot of weight. Many lapsed vegns who had faced issues managing their health named this as their main reason for lapsing. Similarly, Faunalytics (2022) found that people who felt unhealthy when first trying out a vegn diet were significantly more likely to lapse within the first six months than people who felt healthier. This was the case even if their initial motivation for going vegn wasn’t health-related.
    Seeking professional medical advice while pursuing a vegn diet (ideally from a doctor who understands and has experience with vegn diets) is the best way to manage any major concerns and get information about the vitamins and nutriti...
    12 min
  • LW - Building and Entertaining Couples by Jacob Falkovich
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Building and Entertaining Couples, published by Jacob Falkovich on February 22, 2023 on LessWrong.
    Cross-posted, as always, from Putanumonit.
    A friend told me recently that she’s looking for a husband to settle have kids, but she’s facing a problem: most of the men she goes on dates with fail to excite her intellectually in the first hour of conversation. They’re not totally boring — my friend is interesting enough to meet interesting men — but if she’s going to settle down for life shouldn’t she wait for the most exhilarating guy she knows?
    No, I don’t think she necessarily should. Especially not if her goal is to have the best husband to build a family with. And it’s not because interesting men make for bad fathers due to some law of conservation of husbandly quality. I think it’s couples who tend to be of one kind or another: a couple that builds together, or a couple that entertains each other.
    Before giving more detail on those, it’s interesting that most people intuitively get it. It’s a rare Twitter poll that doesn’t have several people in the comments complaining that the given four choices don’t capture the full gamut of human experience, yet over 90% of respondents in a relationship picked one of the two choices.
    I suspect that this dichotomy is much less salient for people not currently in a relationship. They often imagine their future partner being anything and everything for them, a questionable hope that I discussed at length before. But the longer people spend in a relationship the more it tends to become oriented towards one or another.
    Characteristics
    “Entertaining” couples measure the relationship by the quality of time spent together. The most important aspect of their partner is that their company is always better than being alone, and these couples spend more time together and do more fun things together like date nights and vacations. People in these relationships focus more on their appearance, humor, conversation skills, and sex. They prefer quick conflict resolution, agreeing to disagree, and make-up sex. These relationships work better for partners who share similar sensibilities and enjoy the same lifestyle and roles.
    The quintessential examples of this are two high-powered career individualists, think Jean Paul Sartre and Simone de Beauvoir. The failure mode for “entertainment” relationships is lack of commitment leading to frequent breakups, as each partner chases the next exciting thing. The success mode is the party-throwing couple at the center of all their friends’ social life, never failing to delight everyone but mostly each other.
    “Building” couples measure the relationship by the fruits of its enterprise, whether it’s happy children, a successful business, or a lovely house. They care about commitment and contribution first and foremost, and are happy knowing that their partner is committed and contributing even if they’re not in their immediate presence. People in these relationships focus on cultivating skills and long-term projects. They value reaching a consensus and hashing issues out thoroughly. These relationships can work for two very different people whose strengths and preferred roles cover for their partner, like a stay-at-home parent and a wage earner.
    The failure mode for “building” couples is being stuck in boring drudgery, slowly building resentment without the courage to disrupt their routine. The successful exemplar is the couple whose house all their children’s friends want to have play dates and sleepovers in. These couples are often respected locally, almost never famous globally.
    An Aside on Cucumbers and Grapes
    There’s a concept in polyamory literature of “relationship cucumbers” and “relationship grapes” inspired by the famous video about monkeys and fairness. Cucumbers are the mundane day-to-day acts of ma...
    7 min
  • LW - Cyborg Periods: There will be multiple AI transitions by Jan Kulveit
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Cyborg Periods: There will be multiple AI transitions, published by Jan Kulveit on February 22, 2023 on LessWrong.
    It can be useful to zoom out and talk about very compressed concepts like ‘AI progress’ or ‘AI transition’ or ‘AGI timelines’. But from the perspective of most AI strategy questions, it’s useful to be more specific.
    Looking at all of human history, it might make sense to think of ourselves as at the cusp of an AI transition, when AI systems overtake humans as the most powerful actors. But for practical and forward-looking purposes, it seems quite likely there will actually be multiple different AI transitions:
    There will be AI transitions at different times in different domains
    In each of these domains, transitions may move through multiple stages:
    DescriptionPresent day examplesHumans clearly outperform AIs. At some point, AIs start to be a bit helpful.Alignment research, high-level organisational decisions. Humans and AIs are at least comparably powerful, but have different strengths and weaknesses. This means that human+AI teams outperform either unaided humans, or pure AIs.Visual art, programming, trading.AIs overtake humans. Humans become obsolete and their contribution is negligible to negative.Chess, go, shogi.
    Stage
    [ = more powerful than]
    Human period:
    Humans AIs
    Cyborg period: Human+AI teams humans
    Human+AI teams AIs
    AI period:
    AIs humans
    (AIs ~ human+AI teams)
    Some domains might never enter an AI period. It’s also possible that in some domains the cyborg period will be very brief, or that there will be a jump straight to the AI period. But:
    We’ve seen cyborg periods before
    Global supply chains have been in a cyborg period for decades
    Chess and go both went through cyborg periods before AIs became dominant
    Arguably visual art, coding and trading are currently in cyborg periods
    Even if cyborg periods are brief, they may be pivotal
    More on this below
    This means that for each domain, there are potentially two transitions: one from the human period into the cyborg period, and one from the cyborg period into the AI period.
    Transitions in some domains will be particularly important
    The cyborg period in any domain will correspond to:
    An increase in capabilities (definitionally, as during that period human+AI teams will be more powerful than humans were in the human period)
    An increase in the % of that domain which is automated, and therefore probably an increase in the rate of progress
    Some domains where increased capabilities/automation/speed seem particularly strategically important are:
    Research, especially
    AI research
    AI alignment research
    Human coordination
    Persuasion
    Cultural evolution
    AI systems already affect cultural evolution by speeding it up and influencing which memes spread. However, AI doesn’t yet play a significant role in creating new memes (although we are at the very start of this happening). This is similar to the way that humans harnessed the power of natural evolution to create higher yield crops without being able to directly engineer at the genetic level
    Meme generation may also become increasingly automated, until most cultural change happens on silica rather than in brains, leading to different selection pressures
    Strategic goal seeking
    Currently, broad roles involving long-term planning and open domains like "leading a company" are in the human period
    If this changes, it would give cyborgs additional capabilities on top of the ones listed above
    Some other domains which seem less centrally important but could end up mattering a lot are:
    Cybersecurity
    Military strategy
    Nuclear command and control
    Some kinds of physical engineering/manufacture/nanotech/design
    Chip design
    Coding
    There are probably other strategically important domains we haven’t listed.
    A common feature of the domains listed is that increased capabilities in...
    10 min
  • EA - Announcing the Launch of the Insect Institute by Dustin Crummett
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Announcing the Launch of the Insect Institute, published by Dustin Crummett on February 22, 2023 on The Effective Altruism Forum.
    The Insect Institute, a fiscally-sponsored project of Rethink Priorities, is excited to announce its official launch. I (Dustin Crummett) am the executive director of this new initiative.
    The Insect Institute was created to focus on the rapidly growing use of insects as food and feed. Our aim is to work with policymakers, industry, and other relevant stakeholders to address key uncertainties involving animal welfare, public health, and environmental sustainability. As this industry evolves over time, we may also expand our work to other areas.
    While we don’t currently have any open positions, we do expect to grow our team in the future. If you are interested in working with us, please feel free to submit an expression of interest through our contact form, or sign up for our email list (at the bottom of our home page) to stay up to date on future developments and opportunities.
    I look forward to seeing many of you at EAG Bay Area—please come say hello if you’d like to chat! I’m also very happy to field questions via DM or via email to [email protected].
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • LW - The Open Agency Model by Eric Drexler
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: The Open Agency Model, published by Eric Drexler on February 22, 2023 on LessWrong.
    Notes on AI for complex, consequential problems
    Eric DrexlerCentre for the Governance of AIUniversity of Oxford
    Introduction
    This document argues for “open agencies” — not opaque, unitary agents — as the appropriate model for applying future AI capabilities to consequential tasks that call for combining human guidance with delegation of planning and implementation to AI systems. This prospect reframes and can help to tame a wide range of classic AI safety challenges, leveraging alignment techniques in a relatively fault-tolerant context.
    Rethinking safe AI and its applications
    AI safety research is too varied to summarize, yet broad patterns are obvious. A long-established reference-problem centers on prospects for rational superintelligent agents that pursue narrow goals with potentially catastrophic outcomes. This frame has been productive, but developments in deep learning call for updates that take account of the proliferation of narrow models (for driving, coding, robot control, image generation, game playing.) that are either non-agentic or act as agents in only a narrow sense, and that take account of the rise of more broadly capable foundation models and LLMs. These updates call for reframing questions of AI safety, and call for attention to how consequential tasks might be accomplished by organizing AI systems that usually do approximately what humans intend.
    Two frames for high-level AI
    The unitary-agent frame
    From its beginnings in popular culture, discussion of the AI control problem has centered around a unitary agent model of high-level AI and potential AI risks. In this model, a potentially dominant agent both plans and acts to achieve its goals.
    The unitary-agent model typically carries assumptions regarding goals, plans, actions, and control.
    Goals: Internal to an agent, by default including power-seeking goals
    Plans: Internal to an agent, possibly uninterpretable and in effect secret
    Actions: Performed by the agent, possibly intended to overcome opposition
    Control: Humans confront a powerful, potentially deceptive agent
    The typical unitary-agent threat model contemplates the emergence of a dominant, catastrophically misaligned agent, and safety models implicitly or explicitly call for deploying a dominant agent (or an equivalent collective system) that is both aligned and powerful enough to suppress unaligned competitors everywhere in the world.
    The open-agency frame
    Recent developments suggest an alternative open agency model of high-level AI. Today, the systems that look most like AGI are large language models (LLMs), and these are not agents that seek goals, but are generative models that produce diverse outputs in response to prompts (in a generalized sense) and random-number seeds. Most outputs are discarded.
    Trained on prediction tasks, LLMs learn world models that include agent behaviors, and generative models that are similar in kind can be informed by better world models and produce better plans. There is no need to assume LLM-like implementations: The key point is that generation of diverse plans is by nature a task for generative models, and that in routine operation, most outputs are discarded.
    These considerations suggest an “open-agency frame” in which prompt-driven generative models produce diverse proposals, diverse critics help select proposals, and diverse agents implement proposed actions to accomplish tasks (with schedules, budgets, accountability mechanisms, and so forth).
    Goals, plans, actions, and control look different in the open-agency model:
    Goals: Are provided as prompts to diverse generative models, yielding diverse plans on request
    Plans: Are selected with the aid of diverse, independent comparison and evaluation mechanisms
    Actions: Incr...
    9 min
  • AF - EIS XI: Moving Forward by Stephen Casper
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: EIS XI: Moving Forward, published by Stephen Casper on February 22, 2023 on The AI Alignment Forum.
    Part 11 of 12 in the Engineer’s Interpretability Sequence.
    So far, this sequence has discussed a number of topics in interpretability research, all building toward this post. Its goal is to explain some approaches that may be valuable moving forward. I plan to work on some of the ideas here soon. Others, I may not work on soon, but I would love to discuss and support such work if I am able. I hope that this post can offer some useful ideas for people entering or continuing with interpretability research, and if you would like to discuss anything here more, feel more than free to email me at [email protected].
    What are we working toward?
    First, it seems useful to highlight two points that are uncontroversial in the AI safety community but important nonetheless.
    Our goal is a toolbox – not a silver bullet.
    As AI safety engineers, we should neither expect nor try to find a single ‘best’ approach to interpretability that will solve all of our problems. There are many different types of interpretability tools, and many of the differences between them can be described as enforcing different priors over what explanations they generate. So it is trivial to see that there is not going to be any free lunch. There is no silver bullet for interpretability, and few tools conflict with each other anyway. Hence, our goal is a toolbox.
    In fact, some coauthors and I recently found an excellent example of how using multiple interpretability tools at once beats using individual ones (Casper et al., 2023).
    This doesn’t mean, however, that we should celebrate just any new interpretability tool. Working in unproductive directions is costly, and applying tool after tool to a problem contributes substantially to the alignment tax. The best types of tools to fill our toolbox will be ones that are automatable, cheap to use, and have demonstrated capabilities on tasks of engineering-relevance.
    Don’t advance capabilities.
    As AI safety engineers, we do not want to advance capabilities because doing so speeds up timelines. In turn, faster timelines mean less time for safety research, less time for regulators to react, and a greater likelihood of immense power being concentrated in the hands of very few. Avoiding faster timelines isn’t as simple as just not working on capabilities though. Many techniques have potential uses for both safety and capabilities. So instead of judging our work based on how much we improve safety, we need to judge it based on how much we improve safety relative to capabilities. This is an especially important tradeoff for engineers to keep in mind. A good example was discussed by Hendrycks and Woodside (2022) who observed that there is a positive correlation between the anomaly detection capabilities of a network and its task performance.
    Some work may improve safety capabilities but if it does so by continuing along existing trendlines, we don’t get more safety than the counterfactual. For the full discussion of this point, see Hendrycks and Woodside (2022).
    What types of existing tools/research seem promising?
    Before discussing what topics may be important to work on in the future, it may be valuable to reflect on examples of past work that have introduced interpretability tools that seem to be able to competitively provide engineering-relevant insights. Here is a personal list that is somewhat arbitrary and undoubtedly incomplete. But hopefully it is still valuable. Consider this an engineer’s interpretability reading list of sorts.
    Some works have competitively done engineering-relevant things with methods for making novel predictions about how a network will handle OOD inputs.
    ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and ...
    16 min
  • EA - Cyborg Periods: There will be multiple AI transitions by Jan Kulveit
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Cyborg Periods: There will be multiple AI transitions, published by Jan Kulveit on February 22, 2023 on The Effective Altruism Forum.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    1 min
  • AF - Cyborg Periods: There will be multiple AI transitions by Jan Kulveit
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Cyborg Periods: There will be multiple AI transitions, published by Jan Kulveit on February 22, 2023 on The AI Alignment Forum.
    It can be useful to zoom out and talk about very compressed concepts like ‘AI progress’ or ‘AI transition’ or ‘AGI timelines’. But from the perspective of most AI strategy questions, it’s useful to be more specific.
    Looking at all of human history, it might make sense to think of ourselves as at the cusp of an AI transition, when AI systems overtake humans as the most powerful actors. But for practical and forward-looking purposes, it seems quite likely there will actually be multiple different AI transitions:
    There will be AI transitions at different times in different domains
    In each of these domains, transitions may move through multiple stages:
    DescriptionPresent day examplesHumans clearly outperform AIs. At some point, AIs start to be a bit helpful.Alignment research, high-level organisational decisions. Humans and AIs are at least comparably powerful, but have different strengths and weaknesses. This means that human+AI teams outperform either unaided humans, or pure AIs.Visual art, programming, trading.AIs overtake humans. Humans become obsolete and their contribution is negligible to negative.Chess, go, shogi.
    Stage
    [ = more powerful than]
    Human period:
    Humans AIs
    Cyborg period: Human+AI teams humans
    Human+AI teams AIs
    AI period:
    AIs humans
    (AIs ~ human+AI teams)
    Some domains might never enter an AI period. It’s also possible that in some domains the cyborg period will be very brief, or that there will be a jump straight to the AI period. But:
    We’ve seen cyborg periods before
    Global supply chains have been in a cyborg period for decades
    Chess and go both went through cyborg periods before AIs became dominant
    Arguably visual art, coding and trading are currently in cyborg periods
    Even if cyborg periods are brief, they may be pivotal
    More on this below
    This means that for each domain, there are potentially two transitions: one from the human period into the cyborg period, and one from the cyborg period into the AI period.
    Transitions in some domains will be particularly important
    The cyborg period in any domain will correspond to:
    An increase in capabilities (definitionally, as during that period human+AI teams will be more powerful than humans were in the human period)
    An increase in the % of that domain which is automated, and therefore probably an increase in the rate of progress
    Some domains where increased capabilities/automation/speed seem particularly strategically important are:
    Research, especially
    AI research
    AI alignment research
    Human coordination
    Persuasion
    Cultural evolution
    AI systems already affect cultural evolution by speeding it up and influencing which memes spread. However, AI doesn’t yet play a significant role in creating new memes (although we are at the very start of this happening). This is similar to the way that humans harnessed the power of natural evolution to create higher yield crops without being able to directly engineer at the genetic level
    Meme generation may also become increasingly automated, until most cultural change happens on silica rather than in brains, leading to different selection pressures
    Strategic goal seeking
    Currently, broad roles involving long-term planning and open domains like "leading a company" are in the human period
    If this changes, it would give cyborgs additional capabilities on top of the ones listed above
    Some other domains which seem less centrally important but could end up mattering a lot are:
    Cybersecurity
    Military strategy
    Nuclear command and control
    Some kinds of physical engineering/manufacture/nanotech/design
    Chip design
    Coding
    There are probably other strategically important domains we haven’t listed.
    A common feature of the domains listed is that increased ca...
    10 min
  • EA - Consider not sleeping around within the community by Patrick Sue Domin
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Consider not sleeping around within the community, published by Patrick Sue Domin on February 22, 2023 on The Effective Altruism Forum.
    Posted under pseudonym for reasons I’d rather not get into. If it’s relevant, I’m pretty involved in EA. I’ve been to several EAGs and I do direct work.
    tldr I think many more people in the community should consider refraining from sleeping around within the community. I especially think people should consider refraining from sleeping around within EA if they have two or more of the following traits- high status in EA, a man who sleeps with women, and socially clumsy.
    I think the community would be a more welcoming place, with less sexual misconduct and less other sexually unwelcome behaviour, if more EAs chose to personally refrain from sleeping around within EA or attempting to do so. Most functional institutions outside of EA, from companies to friend groups to extended families, have developed norms against sleeping around within the group. We obviously don’t want to simply unquestionably accept all of society’s norms, but I think in this case those norms address real problems.
    I worry that as a group, EAs run a risk of discarding valuable cultural practices that don’t immediately make sense in a first principles way, and that this tendency can have particularly high costs where sex is involved (Owen more or less admitted this was a factor in his behaviour in his statement/apology: “I was leaning into my own view-at-the-time about what good conduct looked like, and interested in experimenting to find ways to build a better culture than society-at-large has”).
    Regarding sleeping around within a tight-knit community, I think this behaviour has risks whether the pursuer is successful or not. Failed attempts at sleeping with someone can very often lead to awkwardness or uncomfortability. In EA, where employment and funding may be front of mind, this uncomfortability may be increased a lot, and there may be no way for the person who was pursued to realistically avoid the pursuer in the future if they want to without major career repercussions. Successful attempts at sleeping around can obviously also cause all sorts of drama, either shortly after or down the road.
    Personal factors that may increase risks
    I think within EA, the risks of harm are increased greatly if the pursuer has any of the following three traits:
    High status within EA- this can create bad power dynamics and awkward social pressure. First, people generally don’t like pissing off high status people within their social circles as there may be social repercussions to doing so. Second, high status people within EA often control funding and employment decisions. Even if the pursuer isn’t in such a position now, they might wind up in one in the future. Third, high status EAs often talk to other high status EAs, so an unjustified bad reputation can spread to other figures in the movement who control funding or employment. Fourth, many EAs consider the community to be their one best shot at living the kind of ethical life they want, raising the stakes a bunch. Fifth, the moralising aspect of EA may make some people find it more uncomfortable to rebuff a high status EA.
    A man pursuing a woman (such as a heterosexual man or a bi-/pansexual man pursuing a woman)- this factor can sometimes be an elephant that people dance around in discussions, but I’ll just address it head on. On average men are more assertive, aggressive, and physically intimidating than women. On average women are more perceptive about subtle social cues and find it more awkward when those subtle social cues are ignored. My sense is these factors are pretty robust across cultures, but I don’t think it matters for this discussion what the cause of these average differences are. Add to all that, the EA communit...
    9 min