On Doing Things That Look Bad
Recently, the effective altruism organization Nonlinear was revealed to have exploitative business practices, including:
Socially isolating employees by encouraging them not to spend time with “low-value people.”
Unclear and messy finances which left employees financially dependent on their bosses (who could arbitrarily decide not to pay expenses).
Pressuring employees into breaking the law.
Pressuring employees into becoming monogamous.
Employers living with their employees in a group house as they travelled the world.
Requiring employees to do all the chores at the group house because the employees’ time was less valuable than the employer’s.
Effusive and manipulative compliments.
Threatening former employees for speaking out.
Fortunately, as far as I’m aware, the effective altruism community in general has responded well to these accusations; Nonlinear is no longer an organization in good standing.
The thing I find interesting about the Nonlinear situation is how many aspects of it are similar to many positive situations I know of. Many effective altruists live in group houses with their coworkers. It’s not uncommon for Bay Area effective altruists to “couch” their friends who are in bad living situations or looking for jobs: that is, to allow their friend to live in their house, to cover the friend’s food and some living expenses, and often to ask the friend to do chores or childcare. Effective altruists often pay their friends to do tasks for them, especially childcare, errands, housework, and general support work. Many effective altruists do gig work for other effective altruists using informal/handshake contracts. Many effective altruists regularly break some laws, such as drug laws.1
Of course, these situations are not the same as what Nonlinear did. Living with a coworker is different from putting your entire organization in a single house. It’s fine to couch your friend; it’s exploitative to put out a job ad and then try to pay your employee by couching them. Even handshake contracts should be written down somewhere in order to prevent misunderstandings; they should not be purely verbal. Ordering gray-market modafinil yourself is very different from pressuring your employee to smuggle drugs. And, of course, there’s no excuse for threatening people who say true negative things about you.
But I think that the more positive situations offer cover. It wasn’t necessarily clear that Nonlinear was doing something different from what everyone else was doing. For employees, when you see other people using informal contracts or living in group houses with their coworkers, it’s easy to assume that this is “just how effective altruism works.” Outsiders might round off situations like Nonlinear to ordinary couching and informal work, because you have to dig in a bit to find out about the financial dependence, poor boundaries, and manipulation.
In addition: small shifts in normal behavior cause a huge increase in extreme behavior. I think this is easiest to see with a visualization:
If Overall Sketchiness is normally distributed, then a relatively small increase in population Overall Sketchiness doubles or triples the amount of Overall Sketchiness three or four standard deviations out from normal. Therefore, if effective altruists are somewhat more sketchy than average, we should expect far more extremely sketchy situations than the general population has.
I believe the general case of this observation explains many things about human behavior.
I’m not saying, to be clear, that we should stop doing things that look bad. People do things that look bad for good reasons. In fact, the reason that it’s difficult is that doing things that look bad is load-bearing, sometimes even life-saving. I know people who do valuable direct work or work six-figure programming jobs who never would have been able to get their start without six months of someone paying their rent. Similarly, I know people who were able to leave abusive living situations because a friend could host them. Many depressed people aren’t able to buy antidepressants licitly, and breaking the law is how they get antidepressants at all.
Some things that look bad are less important but still make people’s lives better. Parents want their children to be taken care of by someone they trust, not a random person off Urban Sitter. Informal contracts create a lot less overhead than formal contracts; requiring formal contracts would make a fair amount of gig work unprofitable. It’s unreasonable to ask people to move out of their homes if they get hired at a company their housemate also works at.
One solution is being more clear about what counts as bad behavior. For example, effective altruists could more clearly set norms like:
Nonprofits should follow generally accepted accounting principles and basic professionalism norms (e.g. knowing who is an employee and who isn’t).
Even informal contracts should be written down somewhere like an email, where one side of the conversation can’t delete their messages.
If you have significant power over someone, you should not ask them to break laws, even if the law is stupid and it’s morally fine to break it.
You should not ask people you’re not dating to change their relationship style to make you more comfortable, especially if you have significant power over them.
Many ways that are fine to treat a friend whom you hired to do some casual work for you are not fine ways to treat a stranger you hired through a formal process.
But in general I’m not sure how best to handle the tradeoffs here, and I want to lay this out as a point for discussion.
Not just conventional recreational drugs, but also e.g. buying modafinil from a gray-market pharmacy without a prescription.
I think this general principle is very applicable to kink/BDSM scene. Because we play with pretty sketchy stuff there (non-consent, explicitly unequal power dynamics, physical violence and pain, mental pain and possibly violence, etc etc etc) the tail end will contain a lot of extremely objectionable stuff.
This kinda suggests (and I'm probably unhappy about the implication but it's intriguing as a concept) that a small shift LEFT (towards more focus on safety, accountability and maybe even overdoing the "performative virtue" and performative consent) could result in a significant reduction of the Rare Truly Bad Shit at the end of the right tail.
On the other hand, we could see the Rare Bad Shit as outliers not being part of the general pattern (I have not checked the data but I suspect that serial killers don't follow normal crime patterns to do with poverty, social anomie, etc).
When I read the title I first thought this was going to be about that UK-based EA group that bought a castle because it was cheaper in the long run to host their conferences and seminars there than constantly hire event spaces.
What I read here reminds my of a post I read a while ago that I'm pretty sure was also on thingofthings (or at least linked from there) but I can't find just now, the argument was that if you're making a list called "10 signs you're in an abusive relationship" then you should list 10 features of abuse, not 10 features of relationships. Group houses on their own are not a sign that anything is wrong; how and why those group houses are run might be. Cults tend to want their members to interact only with the "right" people, but not every group building a filter bubble is a cult.
I think the sketchiness argument comes down to this: there are heuristics that work well for normal situations, but fail badly on data points outside of their design space. For example in a monogamous relationship, your partner getting intimate with someone else is, as a heuristic, A BAD SIGN. If most relationships in the world are mono, then someone might round off to the heuristic that it's a bad sign in relationships in general. But that heuristic completely fails when applied to the poly community! There's a clue in the name!
I would add a point to the bullet list that when a community wants an exemption from a generally useful social heuristic, it's on them to construct their own alternative rules, which means understanding what aims the heuristic is trying to achieve in the first place (the Chesterton's Fence argument) and being satisfied, at least among people inside the group, that either the alternative achieves the same aims or that the group explicitly disagrees with those aims. The poly community seems to be quite far down that path already, and there's parts of the kink community as far as I'm told that are really good at this too (and other parts that, a while back, ended up on the YesMeansYes blog in a post called "Your Kink is Rape" IIRC).
I'm happy to grant the EA community their group houses and their own way to allocate chores, if they have their own standards of behavior to replace the usual ones, and that cover the same aims as the "don't date your managers/reports" rules in non-EA organisations. Or, if you want an exemption from suspicion because you pattern-match a properly researched list of "10 signs you might be in a cult", that's fine as long as your group has some rules in place that prevent the bad kind of stuff people normally mean when they say "cult", like the head honcho having the undisputable right to sleep with anyone they want including the children, or the mass suicide thing. For a group that's not actually a cult, this should be a workable compromise.
Generally accepted accounting principles for anything that touches money seems fairly non-negotiable to me. One of the aims of those principles is to prevent the kind of thing SBF got up to; that fence should really really have stayed up. This applies to all startups including non-EA ones.