The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • LW - Who Aligns the Alignment Researchers? by Ben Smith
    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: Who Aligns the Alignment Researchers?, published by Ben Smith on March 5, 2023 on LessWrong.
    There may be an incentives problem for AI researchers and research organizations who face a choice between researching Capabilities, Alignment, or neither. The incentives structure will lead individuals and organizations to work towards Capabilities work rather than Alignment. The incentives problem is a lot clearer at the organizational level than the individual level, but bears considering at both levels, and of course, funding available to organizations has downstream implications for the jobs available for researchers employed to work on Alignment or Capabilities.
    In this post, I’ll describe a couple of key moments in the history of AI organizations. I’ll then survey incentives researchers might have for doing either Alignment work or Capabilities work. We’ll see that it maybe that, even considering normal levels of altruism, the average person might prefer to do Capabilities rather than Alignment work. There is relevant collective action dynamic. I’ll then survey the organizational level and global level. After that, I’ll finish by looking very briefly at why investment in Alignment might be worthwhile.
    A note on the dichotomous framing of this essay: I understand that the line between Capabilities and Alignment work is blurry, or worse, some Capabilities work plausibly advances Alignment, and some Alignment work advances Capabilities, at least in the short term. However, in order to model the lay of the land, it’s helpful as a simplifying assumption to examine Capabilities and Alignment as distinct fields of research and try to understand the motivations for researchers in each.
    History
    As a historical matter, DeepMind and OpenAI were both founded with explicit missions to create safe, Aligned AI for the benefit of all humanity. There are different views on the extent to which each of these organizations remains aligned to that mission. Some people maintain they are, while others maintain they are doing incredible harm by shortening AI timelines. No one can deny that they have moved at least somewhat in the direction of more profit-making behavior, and are very much focused on Capabilities research. So, at best, they’ve stuck to their original mission, but having watered it down to allow a certain amount of profit-seeking; at worst, their overall efforts are net-negative for alignment by accelerating development of AGI.
    OpenAI took investment from Microsoft in January, to the tune of $10b. At the time, they said
    This multi-year, multi-billion dollar investment from Microsoft follows their previous investments in 2019 and 2021, and will allow us to continue our independent research and develop AI that is increasingly safe, useful, and powerful.
    And this seems plausibly like a systemic pressure other AI Capabilities researchers will face, too. Because of the disparate capital available, in order to fund research in AI Safety, any AI research organization will be incentivized to do capabilities research.
    On the other hand, it’s striking that no organizations founded with the goal of AI Capabilities research have drifted towards Alignment research over time. Organizations under this category might include John Carmack’s recent start-up, Keen Technologies, Alphabet, and many other organizations. Systemically, this can be explained by the rules of the capitalist environment organizations work within. If you create a company to do for-profit work, and get investors to invest in the project, they’ll expect a return. If you go public, you’ll have a fiduciary duty to obtain a return for investors. For organizations, Alignment doesn’t earn money (except in so far as it improves capabilities for tasks); Capabilities does. As the amount of money available to investors grows, more an...
    20 min
  • LW - Introducing Leap Labs, an AI interpretability startup by Jessica Rumbelow
    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: Introducing Leap Labs, an AI interpretability startup, published by Jessica Rumbelow on March 6, 2023 on LessWrong.
    We are thrilled to introduce Leap Labs, an AI startup. We’re building a universal interpretability engine.
    We design robust interpretability methods with a model-agnostic mindset. These methods in concert form our end-to-end interpretability engine. This engine takes in a model, or ideally a model and its training dataset (or some representative portion thereof), and returns human-parseable explanations of what the model ‘knows’.
    Research Ethos:
    Reproducible and generalisable approaches win. Interpretability algorithms should produce consistent outputs regardless of any random initialisation. Future-proof methods make minimal assumptions about model architectures and data types. We’re building interpretability for next year’s models.
    Relatedly, heuristics aren’t enough. Hyperparameters should always be theoretically motivated. It’s not enough that some method or configuration works well in practice. (Or, even worse, that it’s tweaked to get a result that looks sensible to humans.) We find out why.
    Aims:
    We must grow interpretability and AI safety in the real world. Leap is a for-profit company incorporated in the US, and the plan is to scale quickly, and to hire and upskill researchers and engineers – we need more meaningful jobs for AI alignment researchers to make progress, nearly as much as we need the researchers themselves.
    Slow potentially dangerous broad domain systems. Public red-teaming is a means of change. Robust interpretability methods make discovering failure modes easier. We demonstrate the fragility of powerful and opaque systems, and push for caution.
    Speed potentially transformative narrow domain systems. AI for scientific progress is an important side quest. Interpretability is the backbone of knowledge discovery with deep learning, and has huge potential to advance basic science by making legible the complex patterns that machine learning models identify in huge datasets.
    Regulation is coming – let’s use it. We predict that governments and companies will begin to regulate and audit powerful models more explicitly, at very least from a bias-prevention viewpoint. We want to make sure that these regulations actually make models safer, and that audits are grounded in (our) state-of-the-art interpretability work.
    Interpretability as standard. Robust interpretability, failure mode identification and knowledge discovery should be a default part of all AI development. Ultimately, we will put a safety-focussed interpretability system in the pipeline of every leading AI lab.
    We are currently seeking funding/investment. Contact us here.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    3 min
  • EA - Model-Based Policy Analysis under Deep Uncertainty by Max Reddel
    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: Model-Based Policy Analysis under Deep Uncertainty, published by Max Reddel on March 6, 2023 on The Effective Altruism Forum.
    This post is based on a talk that I gave at EAGxBerlin 2022. It is intended for policy researchers who want to extend their tool kit with computational tools. I show how we can support decision-making with simulation models of socio-technical systems while embracing uncertainties in a systematic manner. The technical field of decision-making under deep uncertainty offers a wide range of methods to account for various parametric and structural uncertainties while identifying robust policies in a situation where we want to optimize for multiple objectives simultaneously.
    Summary
    Real-world political decision-making problems are complex, with disputed knowledge, differing problem perceptions, opposing stakeholders, and interactions between framing the problem and problem-solving.
    Modeling can help policy-makers to navigate these complexities.
    Traditional modeling is ill-suited for this purpose.
    Systems modeling is a better fit (e.g., agent-based models).
    Deep uncertainty is everywhere.
    Deep uncertainty makes expected-utility reasoning virtually useless.
    Decision-Making under Deep Uncertainty is a framework that can build upon systems modeling and overcome deep uncertainties.
    Explorative modeling > predictive modeling.
    Value diversity (aka multiple objectives) > single objectives.
    Focus on finding vulnerable scenarios and robust policy solutions.
    Good fit with the mitigation of GCRs, X-risks, and S-risks.
    Complexity
    Complexity science is an interdisciplinary field that seeks to understand complex systems and the emergent behaviors that arise from the interactions of their components. Complexity is often an obstacle to decision-making. So, we need to address it.
    Ant Colonies
    Ant colonies are a great example of how complex systems can emerge from simple individual behaviors. Ants follow very simplistic rules, such as depositing food, following pheromone trails, and communicating with each other through chemical signals. However, the collective behavior of the colony is highly sophisticated, with complex networks of pheromone trails guiding the movement of the entire colony toward food sources and the construction of intricate structures such as nests and tunnels. The behavior of the colony is also highly adaptive, with the ability to respond to changes in the environment, such as changes in the availability of food or the presence of predators.
    Examples of Economy and Technology
    Similarly, the world is also a highly complex system, with a vast array of interrelated factors and processes that interact with each other in intricate ways. These factors include the economy, technology, politics, culture, and the environment, among others. Each of these factors is highly complex in its own right, with multiple variables and feedback loops that contribute to the overall complexity of the system. For example, the economy is a highly complex system that involves the interactions between individuals, businesses, governments, and other entities. The behavior of each individual actor is highly variable and can be influenced by a range of factors, such as personal motivations, cultural norms, and environmental factors. These individual behaviors can then interact with each other in complex ways, leading to emergent phenomena such as market trends, economic growth, and financial crises.
    Similarly, technology is a highly complex system that involves interactions between multiple components, such as hardware, software, data, and networks. Each of these components is highly complex in its own right, with multiple feedback loops and interactions that contribute to the overall complexity of the system. The behavior of the system as a whole can then be highly unpredict...
    36 min
  • EA - More Centralisation? by DavidNash
    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: More Centralisation?, published by DavidNash on March 6, 2023 on The Effective Altruism Forum.
    Summary
    I think EA is under centralised
    There are few ‘large’ EA organisations but most EA opportunities are 1-2 person projects
    This is setting up most projects to fail without proper organisational support and does not provide good incentives for experienced professionals to work on EA projects
    EA organisations with good operations could incubate smaller projects before spinning them out
    Levels of Centralisation
    We could imagine different levels of centralisation for a movement ranging from fully decentralised to fully centralised.
    Fully decentralised, everyone works on their own project, no organisations bigger than 1 person
    Fully centralised, everyone works inside the same organisation (e.g. the civil service)
    It seems that EA tends more towards the decentralised model, there are relatively few larger organisations with ~50 or more people (Open Phil, GiveWell, Rethink Priorities, EVF), there are some with ~5-20 people and a lot of 1-2 person projects.
    I think EA would be much worse if it was one large organisation but there is probably a better balance found between the two extremes then we have at the moment.
    I think being overly decentralised may be setting up most people to fail.
    Why would being overly decentralised be setting people up to fail?
    Being an independent researcher/organiser is harder without support systems in place, and trying to coordinate this outside of an organisation is more complicated
    These support systems include
    Having a manager
    Having colleagues to bounce ideas off/moral support
    Having professional HR/operations support
    Health insurance
    Being an employee rather than a contractor/grant recipient that has to worry about receiving future funding (although there are similar concerns about being fired)
    When people are setting up their own projects it can take up a large proportion of their time in the first year just doing operations to run that project, unrelated to the actual work they want to do. This can include spending a lot of the first year just fundraising for the second year
    How a lack of centralisation might affect EA overall
    Being a movement with lots of small project work will appeal more to those with a higher risk tolerance, potentially pushing away more experienced people who would want to work on these projects, but within a larger organisation
    Having a lot of small organisations will lead to a lot of duplication of operation/administration work
    It will be harder to have good governance for lots of smaller organisations, some choose to not have any governance structures at all unless they grow
    There is less competition for employees if the choice is between 3 or 4 operationally strong organisations or being in a small org
    What can change?
    Organisations with good operations and governance could support more projects internally - One example of this already is the Rethink Priorities Special Projects Program
    These projects can be supported until they have enough experience and internal operations to survive and thrive independently
    Programs that are mainly around giving money to individuals could be converted into internal programs, something more similar to the Research Scholars Program, or Charity Entrepreneurship’s Incubation Program
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • EA - 3 Basic Steps to Reduce Personal Liability as an Org Leader by Deena Englander
    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: 3 Basic Steps to Reduce Personal Liability as an Org Leader, published by Deena Englander on March 6, 2023 on The Effective Altruism Forum.
    It's come to my attention that many of the smaller EA orgs are not putting into place basic protection measures that keep their leaders safe. In the world we live in, risk mitigation and potential lawsuits are a fact of life, and I wouldn't want anyone to put themselves at greater risk just because they are unaware of the risk and easy steps to avoid it.
    Rule #1: Incorporate.
    I know most are hesitant to start an actual non-profit since that is more expensive and time-consuming, but at the least, you can form an LLC. That means that any liability accrued by the org CANNOT pass on to you (I think there are a few exceptions, but you can research that). LLCs are easy to start, and are pretty inexpensive (a few hundred to start, and then annually).
    Rule #2: Get your organization its own bank account
    It is NOT a good idea to keep your organization's finances together with your personal ones for many reasons. That increases the risk of accidental fraud and financial mismanagement. If you have your funds and the org's funds together, you run the risk of using the wrong funds and increasing your liability, since it's not clear which activities are personal (not protected by the LLC) or from the org. You also can't really keep track of your expenses well when it's all mixed up. You don't need a fancy bank account - any will do.
    Rule #3: Get general liability insurance
    Basic liability insurance is an expense (mine costs about $1300 USD a year, but that's for my particular services), but if you're providing any type of guidance, mentoring, services, or events, it's a must. I can go into all sorts of potential lawsuits that you hopefully won't have, but if you even have one, your organization will likely go bankrupt if you don't have the protection insurance provides.
    This is not meant to be an in-depth article of all the things you can do, but EVERY EA org that is providing some type of service should have this in place. There's no reason to have our leaders assuming unnecessary risk.
    I don't know what this looks like if you're fiscally sponsored - I'd assume that they assume the liability - but I would love it if someone could clarify.
    I hope we can start changing the standard practices to protect our leaders and organizations. If anyone has any questions about their particular org, please feel free to reach out.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    3 min
  • LW - Do humans derive values from fictitious imputed coherence? 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: Do humans derive values from fictitious imputed coherence?, published by TsviBT on March 5, 2023 on LessWrong.
    [Metadata: crossposted from. First completed November 1, 2022. This essay is more like research notes than exposition, so context may be missing, the use of terms may change across essays, and the text may be revised later; only the versions at tsvibt.blogspot.com are definitely up to date.]
    Humans are born with some elements of their minds, and without many other elements, some of which they'll acquire as their life unfolds. In particular, the elements that we pretheoretically call "values"--aesthetic preferences, goals, life goals, squad goals, aspirations, needs, wants, yearnings, drives, cravings, principles, morals, ethics, senses of importance, and so on--are for the most part acquired or at least unfolded, rather than being explicitly present in a newborn. How does this happen? What generates these mental elements?
    Hypothesis: a human derives many of zer values by imputing coherent agency to zer past behavior, and then adopting the goals of that fictitious agency as actively influential criteria for future action.
    Thanks to Sam Eisenstat for relevant conversations.
    The FIAT hypothesis
    As a shorthand: "the FIAT hypothesis" = "the Fictitious Imputed Adopted Telos hypothesis". ("Fiat" is Latin for "may it happen" or "may it be made", which has some resonance with the FIAT hypothesis in that they both talk about a free creation of goals.) FIAT goals are goals imputed to some behavior and then adopted as goals.
    Human behavior is determined by many things: built-in behavior-determiners such as the instinctive ability to breath, socially learned behavior and values, convergent instrumental goals, and freely created autopoietic goals such as artistic goals. The FIAT hypothesis says that a major determiner of a human's behavior is the process of adopting goals based on interpreting zer past behavior as agentic.
    Ze can be interpreted as asking the question: if my past behavior were the behavior of a coherent agent trying to do something, what would that something be? Then, whatever the answer was, ze adopts it as a goal--a target of more coherent behavior (more effective, more strategic, more orchestrated, more coordinated, more conscious, better resourced, more reflective, more univocal, more wasteless).
    This hypothesis gives a possible answer to the question: how did evolution build something with some substantial level of agentic coherence, even though evolution can't directly program conscious concepts like "avoiding death" or "saving food" or "inclusive genetic fitness" for use as terms in a utility function for an organism to pursue?
    This process could be continuous, with goals becoming gradually more coherent (and then potentially deprioritized, but usually not de-cohered). This process is iterative, starting with built-in behavior-determiners, then adopting new FIAT goals based on past behavior mainly generated by built-in determiners (and also maybe adopting new goals for other reasons), and then adopting new goals based on past behavior influenced by previously adopted goals, including previous FIAT goals, and so on. FIAT goals also come from not just imputing goals to zer own behavior, but also to the behavior of others, such as parents and leaders. Everything gets enshrined, but everything is open to criticism.
    Note that calling this a hypothesis is maybe presumptuous; it's an idea, but since it's abstract and it's about a complex system, there's a lot of ambiguity between FIAT and other explanations or descriptions of behavior, and it's not necessarily obvious how to make different predictions according to the FIAT hypothesis.
    Something left quite unspecified is how the FIAT process picks different possible interpretations of past behavior as servi...
    25 min
  • EA - After launch. How are CE charities progressing? by Ula Zarosa
    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: After launch. How are CE charities progressing?, published by Ula Zarosa on March 6, 2023 on The Effective Altruism Forum.
    TL;DR: Charity Entrepreneurship have helped to kick-start 23 impact-focused nonprofits in four years. We believe that starting more effective charities is the most impactful thing we can do. Our charities have surpassed expectations, and in this post we provide an update on their progress and achievements to date.
    About CE
    At Charity Entrepreneurship (CE) we launch high-impact nonprofits by connecting entrepreneurs with the effective ideas, training and funding needed to launch and succeed. We provide:
    Seed grants (ranging from $50,000 to $200,000 per project)
    In-depth research reports with promising charity ideas
    Two months of intensive training
    Co-founder matching (this is particularly important)
    Stipends
    Co-working space in London
    Ongoing connection to the CE Community (~100 founders, funders and mentors)
    (Applications are now open to our 2023/2024 programs, apply by March 12, 2023).We estimate that on average:
    40% of our charities reach or exceed the cost-effectiveness of the strongest charities in their fields (e.g., GiveWell/ACE recommended).
    40% are in a steady state. This means they are having impact, but not at the GiveWell-recommendation level yet, or their cost-effectiveness is currently less clear-cut (all new charities start in this category for their first year).
    20% have already shut down or might in the future.
    General update
    To date, our CE Seed Network has provided our charities with $1.88 million in launch grants. Based on the updates provided by our charities in Jan 2023, we estimate that:
    1. They have meaningfully reached over 15 million people, and have the potential to soon reach up to 2.5 billion animals annually with their programs. For example:
    Suvita:
    Reached 600,000 families with vaccination reminders, 50,000 families reached by immunization ambassadors, and 95,000 women with pregnancy care reminders
    14,000 additional children vaccinated
    Fish Welfare Initiative:
    1.14 million fish potentially helped through welfare improvements
    1.4 million shrimp potentially helped
    Family Empowerment Media:
    15 million listeners reached in Nigeria
    In the period overlapping with the campaign in Kano state (5.6 million people reached) the contraceptive uptake in the region increased by 75%, which corresponds to 250,000 new contraceptive users and an estimated 200 fewer maternal deaths related to unwanted pregnancies
    Lead Exposure Elimination Project:
    Policy changes implemented in Malawi alone are expected to reach 215,000 children. LEEP has launched 9 further paint programs, which they estimate will have a similar impact on average
    Shrimp Welfare Project:
    The program with MER Seafood (now in progress) can reach up to 125 million shrimp/year. Additional collaborations could reach >2.5 billion shrimp per annum
    2. They have fundraised over $22.5 million USD from grantmakers like GiveWell, Open Philanthropy, Mulago, Schmidt Futures, Animal Charity Evaluators, Grand Challenges Canada, and EA Animal Welfare Fund, amongst others. 3. If implemented at scale, they can reach impressive cost-effectiveness. For example:
    Family Empowerment Media: the intervention can potentially be 22x more effective than cash transfers from GiveDirectly (estimated by the team, 26x estimated by Founders Pledge)
    Fish Welfare Initiative: 1.3 fish or 2 shrimp potentially helped per $1 (estimated by the team, ACE assessed FWI cost-effectiveness as high to very high)
    Shrimp Welfare Project: approximately 625 shrimp potentially helped per $1 (estimated by the team)
    Suvita: when delivered at scale, effectiveness is at a similar range to GiveWell’s top charities (estimated by external organizations, e.g. Founders Pledge, data on this will be available later this year)
    Giving G...
    12 min
  • LW - The Practitioner's Path 2.0: A new framework for structured self-improvement by Evenflair
    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 Practitioner's Path 2.0: A new framework for structured self-improvement, published by Evenflair on March 4, 2023 on LessWrong.
    The Guild of the ROSE is proud to announce the full launch of the Practitioner's Path 2.0, a new framework for structured self-improvement.
    Two years ago, we founded the Guild to provide structure and community for people interested in self-improvement. Since then, we've realized that our old strategy of weekly workshops and community groups needed a revision. People liked our workshops, but more often than not, they struggled to translate abstract ideas into tangible life improvements.
    One idea that had floated around the Guild since its creation was the dream of a Skilltree for life — what if self-improvement was as easy as leveling up a new videogame character?
    The Practitioner's Path 2.0 is our attempt to make that dream a reality. Self improvement is ultimately rewarding, but that doesn't mean it's easy. Our goal is to design systems that will remove the barriers to self improvement, whether by reducing the cognitive load of wondering what to work on, providing extrinsic social motivation, or playing into a completionist challenge-and-reward mindset.
    This article introduces the Practitioner's Path 2.0, starting with the three aspects of the Path:
    Skills: discrete blocks of achievement
    Attributes: long-term, concrete metrics
    Tasks: specific goals you want to accomplish
    Afterwards, we'll briefly discuss the future role of workshops in the Guild and wrap up with some changes to our pricing model.
    Part 1: the Skilltree
    The heart of the Path 2.0 is the Skilltree. The Skilltree is one part map, one part achievement log, and one part reputation system.
    A Map for Self-Improvement
    Each Skill in the Path has a certain number of prerequisites. For example, Basic Life Enhancement I (fixing problems in your life) requires you to acquire Unseen Chances I (identify problems in your life) first:
    Lines and arrows indicate prerequisites.
    The example above is simple, but it illustrates the power of having your self-improvement path laid out before you. Rather than trying to figure out an optimal path (and potentially choice-paralyzing yourself into inaction), you can simply look at the Skilltree and work on the next Skill available.
    The three states of a Skill.
    The style of a Skill indicates whether:
    Unlit: you're missing prerequisites (unlit)
    Dim: you can work on it next (dim)
    Bright + double border: you've acquired it
    An Achievement Log
    Every time you acquire a Skill, it's permanently associated with your Character Sheet. This means you can always open up your Character Sheet and reflect on what you've accomplished — especially since acquired Skills have a section describing what you did to acquire them.
    Skills remember what you did to acquire them.
    Having a concrete record of what you've accomplished is far more motivating than having a vague sense that you did 'something' a few months (or years!) ago.
    A Reputation System
    Our subconscious loves social prestige, and the Path system takes advantage of this by linking self-improvement and social reputation.
    Each Skill has a certain amount of XP associated with it. This is the XP you gain by acquiring the Skill. Thus, XP and doing things are tightly linked. Being high level means you've done much to improve your life ­— everyone on the Path will know that the level 100 Pragmatist is a badass who knows their stuff.
    Each time a member levels up, we announce it to the Guild. (Announcements are batched weekly to avoid spam, similar to how we announced Path 1.0 level-ups.)
    A Character Sheet contains a lot of potentially sensitive information, especially since it remembers what you did to acquire each Skill. In the interest of privacy, Character Sheets will never be made public — only the broad details such ...
    18 min
  • EA - On the First Anniversary of my Best Friend’s Death by Rockwell
    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: On the First Anniversary of my Best Friend’s Death, published by Rockwell on March 6, 2023 on The Effective Altruism Forum.
    Thanks to encouragement from several people in the EA community, I've just started a blog. This is the first post: www.rockwellschwartz.com/blog/on-the-first-anniversary-of-my-best-friends-deathThe title likely makes this clear, but this post discusses death, suffering, and grief. You may not want to read it as a result, or you may want to utilize mental health resources.
    Some weeks back, I had the opportunity to give a presentation for Yale’s undergraduate course, “A Life Worth Living”. As I assembled my PowerPoint—explaining the Importance, Tractability, Neglectedness framework; Against Malaria Foundation; and global catastrophic threats—I felt the strong desire to pivot and include this photo:
    It was taken sometime in 2019 in my Brooklyn basement and depicts two baby roosters perched upon two of my human best friends, Maddie (left) and Alexa (right). One year ago today, Alexa died at age 25. This is my attempt to honor a tragic anniversary and, more so, a life that was very worth living.
    I’m sure you’re curious, so I’ll get it out of the way: The circumstances surrounding their death remain unclear, even as their family continues to seek the truth. I made a long list of open questions a year ago and, to my knowledge, most remain unanswered today. What I do know is that Alexa suffered greatly throughout their short-lived 25 years. And I also know that Alexa still did far more good than many who live far less arduous lives for thrice as long.
    That’s what I want to talk about here: Alexa, the altruist.
    Alexa, my best friend, roommate, codefendant, and rescue and caregiving partner.
    Alexa, cooing in the kitchen, milk-dipped paintbrush in hand, feeding an orphaned baby rat rescued from the city streets.
    Alexa, in a dark parking lot somewhere in Idaho, warming a bag of fluids against the car heater before carefully injecting them into an ill chicken.
    Alexa, pouring over medical reference books on the kitchen floor, searching for a treatment for sick guppies.
    Alexa, stopping when no one else stopped–calling for help when no one else called–as countless subway riders walked over the unconscious man on the cement floor.
    Alexa, hopping fences, climbing trees, walking through blood-soaked streets, bleary-eyed and exhausted but still going, going. Alexa, saving lives.
    Alexa, saving so many lives. Thousands. From childhood, through their last weeks. In dog shelters, slaughterhouses, and the wild. Everywhere they went.
    Alexa, walking the streets of Philadelphia, gently collecting invasive spotted lantern flies before bringing them home to a lush butterfly enclosure, carefully monitoring their energy levels and food. Alexa, caring for 322 spotted lantern flies until they passed naturally come winter.
    Alexa, the caregiver. Alexa, the life-giver.
    Alexa directly aided so many individuals over the years, I don’t think any one person is aware of even half those they helped. Their efforts were relentless but shockingly low-profile. They were far more likely to share a success to spotlight the wonders of the individual they aided than their heroic efforts to bring them to safety. And, painfully, they were also much more likely to dwell on the errors, accidents, or unavoidable heartbreaking outcomes inherent to the act of staving off suffering and dodging death. Alexa’s deep compassion caused them equally deep pain. And when Alexa and I ultimately distanced, it was to evade the deep void of grief too great to bear that lay between us. I know the pain Alexa carried because I do too.
    Sometimes, the pain that binds you to another becomes the pain you run from, and you never get the chance to go back and shoulder their pain in turn.
    Alexa had a bias: Do. And do fearles...
    7 min
  • LW - Why did you trash the old HPMOR.com? by AnnoyedReader
    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 did you trash the old HPMOR.com?, published by AnnoyedReader on March 6, 2023 on LessWrong.
    I am very disappointed.
    Now I'm sure you'd have to see it to believe it, but you see, just today I was rereading my favorite book, HPMOR, for the fourth time, and I clicked to go on to the next chapter, and to my surprise, there was no next chapter! Which is pretty crazy, considering there was a next chapter last time. Instead it just dropped me on LessWrong (on the HPMOR page). As an enjoyer of the old HPMOR.com site I just wanted to express my frustration.
    Now I'm not trying to complain about LessWrong here. I actually like LessWrong, and sometimes I read posts on here. There are many forum and social media sites, and it does seem to me that LessWrong is one of the better ones. I mean if I found out that HMPOR.com was being shut down and instead we would be forced to read it as a series of Facebook posts, I would be much more angry. But still, the best way to run a social media site, and the best way to publish an ebook cannot possibly coincide. (Or, on second thought, perhaps they might, since they're both sort of publishers, but I'm specifically annoyed about this specific merger.)
    The thing is, I already knew about HMPOR being on LessWrong. But I chose to read on HPMOR.com anyway, because that is what I preferred. One annoying thing about LessWrong for this, is that if I scroll up at all (even just slightly) while reading a long post (or chapter?) the annoying banner at the top of the screen shows up, making it difficult to include the specific text on the screen which you wanted. HMPOR.com didn't have that problem. But there are many frustrations. I feel like in the past, I had the choice, and now that choice has been taken away from me.
    I have, at times, been accused of not liking change. (I think it's more a feeling that changes should optional, and people shouldn't be forced into new things they don't want.) But I feel like there are many circumstances where it should be clear change is not a good idea? Like, if you make something that people specifically seek out, instead of other things they could go for, well, among things, from whoever wants it's point of view, it is one of the better ones. Perhaps even one of the best things. And then if it is changed, well, most changes of something so atypical (from any point of view) tend to be reversions to the mean. So if you like something a lot, you shouldn't want it to change, or at least should be worried about potential change.
    I think this is also why sequels tend to be worse than originals. If something is a big deal and people demand a sequel, that thing was probably atypically good. So even with the same author and universe, you should expect to get a reversion to the mean. And you do.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    3 min