The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • EA - Fatebook for Slack: Track your forecasts, right where your team works by Adam Binks
    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: Fatebook for Slack: Track your forecasts, right where your team works, published by Adam Binks on May 11, 2023 on The Effective Altruism Forum.
    Announcing Fatebook for Slack - a Slack bot designed to help high-impact orgs build a culture of forecasting.
    With Fatebook, you can ask a forecasting question in your team's Slack:
    Then, everyone in the channel can forecast:
    When it's time to resolve the question as Yes, No or Ambiguous, the author gets a reminder. Then everyone gets a Brier score, based on their accuracy.
    It's like a tiny, private, fast Metaculus inside your team's Slack.
    Why build a culture of forecasting?
    Make better decisions
    Communicate more clearly
    Build your track record
    Trust your most reliable forecasters
    We built Fatebook for Slack aiming to help high-impact orgs become more effective.
    See the FAQs on the website for more info. We'd really value your feedback in the comments, in our Discord, or at [email protected].
    You can add Fatebook to your workspace here.
    Thanks to all our alpha testers for their valuable feedback, especially the teams at 80,000 Hours, Lightcone, EA Cambridge, and Samotsvety.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • LW - Better debates by TsviBT
    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: Better debates, published by TsviBT on May 10, 2023 on LessWrong.
    [Crossposted from.]
    When two people disagree about a proposition even though they've thought about it alot, the disagreement is often hard to resolve. There's a gulf of data, concepts, intuitions, experiences, inferences. Some of this gulf has to be resolved by the two people individually trying to collate and present their own positions more clearly and legibly, so that they can build up concepts and propositions in whoever is receiving the model. Also, most new understanding comes from people working on their own or with others who are already synced up--for the most part they already agree on what and how to investigate, they have shared context of past experience and data, they agree on background assumptions, they have a shared language, they trust each other.
    But still, a lot of value comes from debate. The debaters are forced to make their evidence and logic legible. Ideas are tested against other ideas from another at least somewhat coherent perspective. Analogies and disanalogies are drawn out. Bundles of facts can be sifted, spot-checked, and compared; predictions can be resolved into updates. The case, in being laid out to the debate partner, is also made clearer to the audience. The audience can see where another expert thinks the strong and weak points are in a position.
    So there's value on the table from debates. But debates usually suck. People talk past each other. They waste effort talking about non-cruxy claims. The truth about verifiable claims isn't made available, missing the chance to bring one or the other perspective into contact with reality. People cite statistics and analyses which become meaningless in a context where they can't be investigated and can't take on their full meaning by relating in detail to the discourse context.
    What would a better debate look like? Below is a speculative list of ways to structure a debate so that it finds and explains more truth. As armchair speculation, this list is gesture or inspiration, not tested advice.
    Multiple rounds. Have multiple conversations with the same debaters on the same topic. Debaters sometimes have to pause a thread of conversation and go off to collect data, clarify their case, reevaluate and update, look into counterdata, come up with a more concise response to critiques, pass the other's Ideological Turing Test, or mull over the other's concepts and perspective.
    Seconds. Debaters bring seconds--people who can step in to make certain points, or can (being already versed in the perspective) efficiently work in the background to bring up information for the main debater to use.
    Free-flowing. Debates often have a regimented structure, with a fixed overall length, and alternating statements with prescribed length. Don't do that, let the debaters have back and forth.
    Technical facilitation. Offload cognitive labor from the debaters by having support people who (perhaps on big visible screens):
    Check easily-checkable claims.
    Bring in numbers and diagrams and exact statements from citations.
    Map the argument. Summarize claims and counterclaims, define terms (especially clarifying how each debater uses important words), track sub-points and digressions.
    Track stated cruxes and double-cruxes.
    Remind the debaters of the current thread stack, if needed.
    Filter comments from an audience.
    Discourse facilitation. Have one or two facilitators who more directly guide the discourse:
    Encourage debaters to paraphrase each other's position, and sometimes try to pass their Ideological Turing Test.
    Highlight when the debaters are using terms differently, or are making unshared background assumptions.
    MC the conversation, e.g. sometimes briefly summarizing the conversation so far, and summarizing prior rounds at the beginning of subsequent ro...
    6 min
  • EA - Community Health & Special Projects: Updates and Contacting Us by evemccormick
    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: Community Health & Special Projects: Updates and Contacting Us, published by evemccormick on May 10, 2023 on The Effective Altruism Forum.
    Summary
    We’ve renamed our team to Community Health and Special Projects, in part to reflect our scope extending beyond what’s often considered to be “community health.”
    Since our last forum update, we’ve started working closely with Fynn Heide as an affiliate, along with Anu Oak and Łukasz Grabowski as contractors. Chana Messinger has been acting as interim team lead, while Nicole Ross has been focused on EV US board duties.
    In response to reports of sexual misconduct by Owen Cotton-Barratt, an external investigation into our team’s response is underway, as well as an internal review.
    Other key proactive projects we’ve been working on include the Gender Experiences project and the EA Organization Reform project.
    We are in the early stages of considering some significant strategic changes for our team. We’ve highlighted two examples of possible changes below, one being a potential spin-out of CEA and/or EV and another being a pivot to focus more on the AI safety space.
    As a reminder, if you’ve experienced anything you’re uncomfortable with in the community or if you would like to report a concern, you can reach our team’s contact people (currently Julia Wise and Catherine Low) via this form (anonymously if you choose).
    We can also be contacted individually (our individual forms are linked here), or you can contact the whole team at [email protected].
    We can provide anonymous, real-time conversations in place of calls when requested, e.g. through Google Chat with your anonymous email address.
    The Community Health team is now Community Health and Special Projects
    We decided to rename our team to better reflect the scope of our work. We’ve found that when people think of our team, they mostly think of us as working on topics like mental health and interpersonal harm. While these areas are a central part of our work, we also work on a wide range of other things, such as advising on decisions with significant potential downside risk, improving community epistemics, advising programs working with minors, and reducing risks in areas with high geopolitical risk.
    We see these other areas of work as contributing to our goal: to strengthen the ability of EA and related communities to fulfil their potential for impact, and to address problems that could prevent that. However, those areas of work can be quite disparate, and so “Special Projects” seemed an appropriate name to gesture towards “other miscellaneous things that seem important and may not have a home somewhere else.”
    We hope that this might go some way to encouraging people to report a wider range of concerns to our team.
    Our scope of work is guided by pragmatism: we aim to go wherever there are important community-related gaps not covered by others and try to make sure the highest priority gaps are filled. Where it seems better than the counterfactual, we sometimes try to fill those gaps ourselves. That means that our scope is both very broad and not always clear, and also that there will be plenty of things we don’t have the capacity or the right expertise to have fully covered. If you’re thinking of working on something you think we might have some knowledge about, the meme we want to spread is “loop us in, but don’t assume it’s totally covered or uncovered.” If we can be helpful, we’ll give advice, recommend resources or connect you with others interested in similar work.
    Team changes
    Here’s our current team:
    Nicole Ross (Head of Community Health and Special Projects)
    Julia Wise (Community Liaison)
    Catherine Low (Community Health Associate)
    Chana Messinger (Interim Head and Community Health Analyst)
    Eve McCormick (Community Health Pr...
    12 min
  • LW - New OpenAI Paper - Language models can explain neurons in language models by ViktorThink
    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: New OpenAI Paper - Language models can explain neurons in language models, published by ViktorThink on May 10, 2023 on LessWrong.
    Summary by OpenAI: We use GPT-4 to automatically write explanations for the behavior of neurons in large language models and to score those explanations. We release a dataset of these (imperfect) explanations and scores for every neuron in GPT-2.
    Link:
    Please share your thoughts in the the comments!
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    1 min
  • LW - Mental Health and the Alignment Problem: A Compilation of Resources (updated April 2023) by Chris Scammell
    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: Mental Health and the Alignment Problem: A Compilation of Resources (updated April 2023), published by Chris Scammell on May 10, 2023 on LessWrong.
    This is a post about mental health and disposition in relation to the alignment problem. It compiles a number of resources that address how to maintain wellbeing and direction when confronted with existential risk.
    Many people in this community have posted their emotional strategies for facing Doom after Eliezer Yudkowsky’s “Death With Dignity” generated so much conversation on the subject. This post intends to be more touchy-feely, dealing more directly with emotional landscapes than questions of timelines or probabilities of success.
    The resources section would benefit from community additions. Please suggest any resources that you would like to see added to this post.
    Please note that this document is not intended to replace professional medical or psychological help in any way. Many preexisting mental health conditions can be exacerbated by these conversations. If you are concerned that you may be experiencing a mental health crisis, please consult a professional.
    Preface to the 2nd Edition
    This post was released in April 2022 under the same title. This April 2023 update features new resources in every section, with a particular emphasis on the Alignment Positions and People Resources sections. Within each section, resources have been thematically categorized for easier access.
    Following the large capabilities leaps in the past year, these resources seem more important than ever. If you have suggestions for improving this post, for making it more accessible, or for new resources to add, please leave a comment or reach out to either Chris Scammell or DivineMango.
    We hope you are all well and that you find this update helpful.
    Introduction
    There is no right way to emotionally respond to the reality of approaching superintelligent AI, our collective responsibility to align it with our values, or the fact that we might not succeed. As transformative AI approaches, we must ensure that we have the tools and resources to be okay. Here, the valence of “be okay” is your decision. This question could be rephrased “how can I thrive despite the alignment problem,” “how can I cope with the alignment problem,” “how can I overcome my fear of the alignment problem,” etc. Everyone needs to find their own question and their own answer.
    At its foundation “being okay” is the decision to continue to live facing reality and the alignment problem directly, with internal stability and rationality intact. And as a high ideal, we’re going for some degree of inviolability, of unconditional wellbeing, the kind of wellbeing that holds onto “okayness” even if the probability of solving alignment drops to 0. It can be difficult to stand in some place of positive mental health and stability while facing the alignment problem; but it is a gift if we can do that for ourselves, and a gift if we can share it with others.
    Fortunately, we don’t have to do this alone. Many community members have found ways to make sense of themselves, their work, and their lives in relation to the alignment problem, and they have kindly made their reflections and advice public.
    Resources
    Several resources on this subject (along with summaries) are cataloged below. While there are a number of general mental health resources on LW, the EA Forum, and elsewhere that form a great baseline, this post aims to be more specific by focusing on mental health with respect to the alignment problem. Here, we feature a wide variety of ideas and practices in the hope that you may filter through them to create and discover the approach that works for you. Human brains come in many shapes – we all have different internal subagent dynamics, motivational systems, values, needs, triggers for j...
    34 min
  • LW - Have you heard about MIT's "liquid neural networks"? What do you think about them? by Ppau
    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: Have you heard about MIT's "liquid neural networks"? What do you think about them?, published by Ppau on May 9, 2023 on LessWrong.
    I came across this video by MIT CSAIL.
    Here is the article they are talking about:
    This team claims to have achieved driving tasks that previously required 10000 neurons, while using only 19, by using "liquid neural networks" inspired by worm neurology.
    They say this innovation brings massive improvements on performance, especially in embedded systems, but also in interpretability, since the reduced number of neurons makes the system much more human-readable. In particular, the attention of the system would be much more easily tracked; this would open the door to safety certifications for high-stakes applications.
    Having tried driving and flying tasks in different conditions and environments, they also claim that their system is vastly better at out-of-distribution zero-shot tasks.
    So basically, they believe they have made very substantial steps in pretty much every dimension that matters, both for performance and for safety.
    As far as I can tell these are very serious researchers, but doesn't that sound a bit too god to be true? I have no expertise in machine learning and I haven't seen any third-party opinions on this yet, so I'm having a hard time making up my mind.
    I'd be curious to hear your takes!
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • EA - Continuous doesn’t mean slow by Tom Davidson
    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: Continuous doesn’t mean slow, published by Tom Davidson on May 10, 2023 on The Effective Altruism Forum.
    Note: This post was crossposted from Planned Obsolescence by the Forum team, with the author's permission. The author may not see or respond to comments on this post.
    Once a lab trains AI that can fully replace its human employees, it will be able to multiply its workforce 100,000x. If these AIs do AI research, they could develop vastly superhuman systems in under a year.
    There’s a lot of disagreement about how likely AI is to end up overthrowing humanity. Thoughtful pundits vary from <5% to >90%. What’s driving this disagreement?
    One factor that often comes up in discussions is takeoff speeds, which Ajeya mentioned in the previous post. How quickly and suddenly do we move from today’s AI, to “expert-human level” AI[1], to AI that is way beyond human experts and could easily overpower humanity?
    The final stretch — the transition from expert-human level AI to AI systems that can easily overpower all of us — is especially crucial. If this final transition happens slowly, we could potentially have a long time to get used to the obsolescence regime and use very competent AI to help us solve AI alignment (among other things). But if it happens very quickly, we won’t have much time to ensure superhuman systems are aligned, or to prepare for human obsolescence in any other way.
    Scott Alexander is optimistic that things might move gradually. In a recent ACX post titled ‘Why I Am Not (As Much Of) A Doomer (As Some People)’, he says:
    So far we’ve had brisk but still gradual progress in AI; GPT-3 is better than GPT-2, and GPT-4 will probably be better still. Every few years we get a new model which is better than previous models by some predictable amount.
    Some people (eg Nate Soares) worry there’s a point where this changes. Maybe some jump. could take an AI from IQ 90 to IQ 1000 with no (or very short) period of IQ 200 in between.
    I’m optimistic because the past few years have provided some evidence for gradual progress.
    I agree with Scott that recent AI progress has been continuous and fairly predictable, and don’t particularly expect a break in that trend. But I expect the transition to superhuman AI to be very fast, even if it’s continuous.
    The amount of “compute” (i.e. the number of AI chips) needed to train a powerful AI is much bigger than the amount of compute needed to run it. I estimate that OpenAI has enough compute to run GPT-4 on hundreds of thousands of tasks at once.[2]
    This ratio will only become more extreme as models get bigger. Once OpenAI trains GPT-5 it’ll have enough compute for GPT-5 to perform millions of tasks in parallel, and once they train GPT-6 it’ll be able to perform tens of millions of tasks in parallel.[3]
    Now imagine that GPT-6 is as good at AI research as the average OpenAI researcher.[4] OpenAI could expand their AI researcher workforce from hundreds of experts to tens of millions. That’s a mind-boggling large increase, a factor of 100,000. It’s like going from 1000 people to the entire US workforce. What’s more, these AIs could work tirelessly through the night and could potentially “think” much more quickly than human workers.[5] (This change won’t happen all-at-once. I expect speed-ups from less capable AI before this point, as Ajeya wrote in the previous post.)
    How much faster would AI progress be in this scenario? It’s hard to know. But my best guess, from my recent report on takeoff speeds, is that progress would be much much faster. I think that less than a year after AI is expert-human level at AI research, AI could improve to the point of being able to easily overthrow humanity.
    This is much faster than the timeline mentioned in the ACX post:
    if you’re imagining specific years, imagine human-genius-level AI in the 2030s and world...
    5 min
  • EA - Psychological safety as the yardstick of good EA movement building by Severin
    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: Psychological safety as the yardstick of good EA movement building, published by Severin on May 10, 2023 on The Effective Altruism Forum.
    I recently learned about the distinction between "movement building" and "community building": Community building is for the people involved in a community, and movement building is in service of the cause itself.
    A story I've heard from a bunch of EA groups is that they start out with community building. They attract a couple people, develop a wonderful vibe, and those people notoriously slack on their reading group preparations. Then, the group organizers get dissatisfied with the lack of visible progress on the EA path, doubt their own impact, and pivot all the way from community building to movement building. No funny pub meetups anymore. Career fellowships and 1-on-1s all the way.
    I think this throws the baby out with the bathwater, and that more often than not, community building is indeed tremendously valuable movement building, even if it doesn't look like that at first glance.
    The piece of evidence I can cite on this (and indeed cite over and over again) is Google's "Project Aristotle"-study.
    In Project Aristotle, Google studied what makes their highest-performing teams highest-performing. And alas: It is not the fanciness of degrees or individual intelligence or agentyness or any other property of the individual team members, but five factors:
    "The researchers found that what really mattered was less about who is on the team, and more about how the team worked together. In order of importance:
    Psychological safety: Psychological safety refers to an individual’s perception of the consequences of taking an interpersonal risk or a belief that a team is safe for risk taking in the face of being seen as ignorant, incompetent, negative, or disruptive. In a team with high psychological safety, teammates feel safe to take risks around their team members. They feel confident that no one on the team will embarrass or punish anyone else for admitting a mistake, asking a question, or offering a new idea.
    Dependability: On dependable teams, members reliably complete quality work on time (vs the opposite - shirking responsibilities).
    Structure and clarity: An individual’s understanding of job expectations, the process for fulfilling these expectations, and the consequences of one’s performance are important for team effectiveness. Goals can be set at the individual or group level, and must be specific, challenging, and attainable. Google often uses Objectives and Key Results (OKRs) to help set and communicate short and long term goals.
    Meaning: Finding a sense of purpose in either the work itself or the output is important for team effectiveness. The meaning of work is personal and can vary: financial security, supporting family, helping the team succeed, or self-expression for each individual, for example.
    Impact: The results of one’s work, the subjective judgement that your work is making a difference, is important for teams. Seeing that one’s work is contributing to the organization’s goals can help reveal impact."
    What I find remarkable is that "psychological safety" leads the list. While some factors in EA actively work against the psychological safety of its members. To name just a few:
    EA tends to attract pretty smart people. If you throw a bunch of people together who have been used all their lives to being the smart kid in the room, they suddenly lose the default role they had in just about any context. Because now, surrounded by even smarter kids, they are merely the kid. I think this is where a bunch of EAs' impostor syndrome comes from.
    EAs like to work at EA-aligned organizations. That means that some of us feel like any little chat at a conference (or any little comment on the EA Forum or our social media accounts) also i...
    7 min
  • EA - Why Not EA? [paper draft] by Richard Y Chappell
    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: Why Not EA? [paper draft], published by Richard Y Chappell on May 9, 2023 on The Effective Altruism Forum.
    Hi all, I'm currently working on a contribution to a special issue of Public Affairs Quarterly on the topic of "philosophical issues in effective altruism". I'm hoping that my contribution can provide a helpful survey of common philosophical objections to EA (and why I think those objections fail)—the sort of thing that might be useful to assign in an undergraduate philosophy class discussing EA.
    The abstract:
    Effective altruism sounds so innocuous—who could possibly be opposed to doing good, more effectively? Yet it has inspired significant backlash in recent years. This paper addresses some common misconceptions, and argues that the core ideas of effective altruism are both excellent and widely neglected. Reasonable people may disagree on details of implementation, but every decent person should share the basic goals or values underlying effective altruism.
    I cover:
    Five objections to moral prioritization (including the systems critique)
    Earning to give
    Billionaire philanthropy
    Longtermism; and
    Political critique.
    Given the broad (survey-style) scope of the paper, each argument is addressed pretty briefly. But I hope it nonetheless contains some useful insights. For example, I suggest the following "simple dilemma for those who claim that EA is incapable of recognizing the need for 'systemic change'":
    Either their total evidence supports the idea that attempting to promote systemic change would be a better bet (in expectation) than safer alternatives, or it does not. If it does, then EA principles straightforwardly endorse attempting to promote systemic change. If it does not, then by their own lights they have no basis for thinking it a better option. In neither case does it constitute a coherent objection to EA principles.
    On earning to give:
    Rare exceptions aside, most careers are presumably permissible. The basic idea of earning to give is just that we have good moral reasons to prefer better-paying careers, from among our permissible options, if we would donate the excess earnings. There can thus be excellent altruistic reasons to pursue higher pay. This claim is both true and widely neglected. The same may be said of the comparative claim that one could easily have more moral reason to pursue "earning to give" than to pursue a conventionally "altruistic" career that more directly helps people. This comparative claim, too, is both true and widely neglected. Neither of these important truths is threatened by the deontologist's claim that one should not pursue an impermissible career. The relevant moral claim is just that the directness of our moral aid is not intrinsically morally significant, so a wider range of possible actions are potentially worth considering, for altruistic reasons, than people commonly recognize.
    On billionaire philanthropy:
    EA explicitly acknowledges the fact that billionaire philanthropists are capable of doing immense good, not just immense harm. Some find this an inconvenient truth, and may dislike EA for highlighting it. But I do not think it is objectionable to acknowledge relevant facts, even when politically inconvenient... Unless critics seriously want billionaires to deliberately try to do less good rather than more, it's hard to make sense of their opposing EA principles on the basis of how they apply to billionaires.
    I still have time to make revisions -- and space to expand the paper if needed -- so if anyone has time to read the whole draft and offer any feedback (either in comments below, or privately via DM/email/whatever), that would be most welcome!
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • LW - Naturalist Experimentation by LoganStrohl
    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: Naturalist Experimentation, published by LoganStrohl on May 10, 2023 on LessWrong.
    Ordinarily, when someone is trying to solve a problem, experimentation is where they begin. “Things aren’t going right? Try changing something, and see what happens!” In naturalism, though, experimentation comes at the very end, if it happens at all.
    There’s a balance to be kept in naturalist experimentation, and it can be a tricky one. I said in Getting Started that the naturalist stance deemphasizes most ways relating to the world, and regards the world instead as something that may reveal itself to you as you open up to it. The trick in this phase is to balance between open curiosity and agency.
    By the time you’ve made it to the “Experimentation” phase in a naturalist study, you’ve spent a long time being very hands-off: Only look, do not touch; just observe, do not intervene. In the very beginning of your study, you asked your desperation to wait in the hall, then invited curiosity to take its place, in an attempt to create plenty of space for patient and direct observation. This whole time, you have tried to observe your fulcrum experience in its natural habitat, learning what happens around it by default, when you do nothing to interfere.
    In this final phase, you’ll begin to interfere. You’ll deliberately interact with the systems you’ve been observing. You’ll exercise your agency.
    It can be quite difficult, while getting much more hands-on with the circumstances you observe, to remain in a state of open curiosity all the while. It can be tempting to stop exploring patiently, and start trying to immediately force the outcomes you want.
    I’m not here to tell you that it is wrong to force the outcomes you want, to grab onto the levers of causality that you’ve discovered and use them to guide the world into your preferred configuration. If you’ve made it this far, you’re probably in a much better position to do that wisely than you were when you started.
    In this essay, I only mean to show you that it is possible to choose another way to interact with those levers, if you want to. It is possible to balance agency with patient, direct observation, in a way that causes your understanding of your topic to deepen indefinitely, rather than bringing your study to an abrupt close. Instead of switching to goal-oriented engineer mode to obtain the object-level solution that you wanted at the start, you might prefer to take a much longer path toward deep mastery of the field you’ve really only just begun to explore.
    I call that longer path “naturalist experimentation”.
    What is different about naturalist experimentation?
    There are a couple of things that set naturalist experimentation apart from a more casual approach to exploratory experimentation, and also from scientific experimentation.
    The first is why we experiment. The second is when.
    Why We Experiment
    When I was a kid and our television picked up stations using antennae (which we called "bunny ears"), the picture would sometimes cut out, and we’d try all sorts of things to make it come back. We’d whack the side of the TV, move the bunny ears around, stand in a different part of the room, change the channel. It was all sort of hapless and frantic; we were just changing things largely at random in the hopes that something would solve the problem, whatever it was, and the picture would come back. This is an extreme example, but here it is clear that we experimented to fix things.
    In scientific experimentation, we experiment to improve our models, usually by attempting to disconfirm hypotheses generated by them. This is certainly related to naturalist experimentation, and I’ll come back to the relationship in a moment.
    In naturalism, we experiment to create space for alternative action.
    To get what I mean by this, it may help to think in ter...
    15 min