February 2023 Linkpost

If AI is coming for us all, at least MIRI’s got a plan to let us bow out gracefully – better than that Bing chatbot spilling its digital heart out.

Philosophy & Human Nature

Cognitohazard Warning: The Chained Woman Case (weibo.substack.com)
This post delves into the harrowing case of Xiao Huamei, abducted and chained in a shed, sparking public outrage and conflicting government reports in China. It warns of the psychological toll, with details on abduction, torture, and sex slavery, estimating tens of thousands in similar plights. I would note that stories like this force us to confront deep human evils, shaking our sense of normalcy and urging action, even if it ruins our peace.

How to Be an Expert Fear-Driven Person (sashachapin.substack.com)
A guide outlining traits of those masterfully avoiding fear through subtle strategies like diluting ambitions or endless research without action, leading to a subtle life of unfulfillment. Key insights include rationalizing avoidance as savvy and suppressing passion for control. I would observe this mirrors how many navigate life’s risks, but accepting fear as part of growth could unlock more vibrant paths.

Why Don’t My Bay Area Friends Move? (sashachapin.substack.com)
An exploration of why idealists cling to the Bay Area’s dense communities despite LA’s superior lifestyle, tying it to cultural idealism versus complacency. Personal anecdotes highlight how dissatisfaction fuels creativity there. I would suggest this reveals our choices often prioritize potential over comfort, shaping who we become in subtle ways.

Human Traits as Continuous Spectrums (spencergreenberg.com)
Proposes viewing traits like ADHD or sociopathy on spectrums with opposites, e.g., ADHD vs. Focused Temperament, to better understand human variation. Examples illustrate how conditions are extremes, not binaries. I would point out this framework helps demystify mental health, showing we’re all on sliding scales rather than fixed categories.

Facts from a Jehovah’s Witnesses Memoir (thingofthings.substack.com)
Summarizes cultural practices like mandatory proselytizing, strict gender roles, and bans on holidays, alongside beliefs in limited heaven spots and demonic possessions in items like Smurfs. Based on Kyria Abrahams’ memoir, it notes discipline and marriage norms. I would reflect that such insular worlds highlight how beliefs shape identity, often at great personal cost.

Academic Research & Science

Conservative Sexual Values and Adherence Challenges (psypost.org)
Two studies (n=923 and n=2,519) found conservative values like abstinence predict higher sexual incongruence—misalignment between beliefs and behaviors—beyond religiousness effects. Permissive values like consent lead to greater congruence. Methodology used free-response value identification and regressions. Key data: abstinence strongly linked to incongruence. I would emphasize this shows ideals can clash with human nature, complicating personal harmony.

Intelligence and Job Prestige/Pay (academic.oup.com)
Using data from Sweden, the US, Japan, and Germany (n not specified in abstract), the study found no strong link between high intelligence and top-paying/prestigious jobs, challenging assumptions. Methodology: regressions on occupational data. Conclusions: meritocracy may not favor the smartest. Implications: success involves more than IQ, reshaping views on inequality.

Education Level and Mating Success on Tinder (sciencedirect.com)
Field experiment (n=3,600 profiles) showed women prefer highly educated men, but men aren’t intimidated by educated women; no assortative mating evidence. Data: heterogeneous effects by gender. Conclusions: dating apps disrupt traditional preferences. I would add this underscores evolving gender dynamics in partner selection.

Same-Race Teachers and Academic Outcomes (sciencedirect.com)
Using Tennessee’s Project STAR data (n=kindergarten cohort from 79 schools), regressions found own-race teacher benefits vary: stronger for below-average students, reduced in small classes or homogeneous racial settings. Data: normalized math/reading scores. Conclusions: effects depend on context. Implications: targeted policies needed for equity.

Teen Mental Illness Epidemic Since 2012 (jonathanhaidt.substack.com)
Collaborative review of surveys and CDC data shows depression doubled by 2015 (150% increase), self-harm tripled for young girls by 2020, suicides spiked 67% for 10-14 girls in 2012-13. Methodology: open-source doc compiling studies. Data: NSDUH, ACHA trends. I would highlight how tech shifts might amplify vulnerabilities, demanding societal reflection.

Solar Power from Yellowstone Caldera (sciencedirect.com)
Proposes harvesting volcanic energy via copper cylinders and steam turbines, generating 11 quadrillion watt-hours annually. Methodology: COMSOL simulations. Findings: emission-free, prevents eruptions. Implications: solves US energy/climate issues scalably.

Technology & Society

Counterarguments to AI x-Risk (worldspiritsockpuppet.substack.com)
Outlines 16 weaknesses: vague goal-directedness, finite forces for it, small value differences not catastrophic, value not fragile, short-term goals limit power-seeking, human success from coordination not intelligence, AI not overwhelmingly superior. Methodology: critiques classic arguments. Conclusions: risks overstated. I would say this tempers alarmism, urging nuanced views.

Risks from Power-Seeking AI (80000hours.org)
Estimates 5% extinction risk by 2100; timelines by 2030 possible. Risks: misaligned goals leading to disempowerment. Mitigation: feedback learning, oversight, governance. Data: researcher surveys (41% see alignment vital). Implications: urgent safety focus to avert catastrophe.

Biological Anchors for AI Timelines (astralcodexten.substack.com)
Estimates 50% transformative AI by 2052 via brain computation analogies (10^16 FLOP/S, training 10^30-36 FLOPs). Methodology: scaling laws, economic models. Data: expert surveys (median 2062). Conclusions: unreliable but useful. I would note critiques suggest faster timelines, heightening urgency.

AI in Fanfiction Alignment (astralcodexten.substack.com)
Redwood trained GPT-Neo on 4,300 Alex Rider fanfics to generate nonviolent completions (n=6,000 adversarial examples). Methodology: human raters, classifiers (0.8% violence threshold). Findings: creative but imperfect alignment. Implications: highlights alignment challenges for literature and beyond.

MIRI’s Death with Dignity Strategy (lesswrong.com)
Satirical post on abandoning alignment for dignified extinction, as humanity won’t solve it. Rationale: no fight left. Implications: philosophical resignation to AI risks. I would quip it’s a dark mirror to our tech optimism.

OpenAI Codex (openai.com)
Improved Codex translates natural language to code, powering GitHub Copilot. Implications: boosts productivity but raises automation concerns.

Ethical AI Possibilities (stitcher.com)
Podcast with Timnit Gebru on AI harms, biases, and corporate ethics. Implications: need diverse, accountable development.

MedARC AI for Medical Research (medarc.ai)
New initiative using AI for medical acceleration, founded by a prodigy. Implications: speeds discoveries but ethical oversight key.

Economics & Development

California Zoning Powers Loss (darrellowens.substack.com)
Builder’s remedy suspends zoning in non-compliant cities, allowing dense projects (20% low-income). Comparisons: Japan-like flexibility. Implications: boosts supply, reduces segregation.

Caribbean Think Tank Focus (rasheedgriffith.substack.com)
Proposes dollarization, public sector internationalization, tourism policies. Challenges: stagnation, socialism lean. Implications: fosters growth via markets.

Millennial Burnout Generation (buzzfeednews.com)
Burnout from debt ($39,950 avg), instability, work-life blur. Data: 39% more anxious. Implications: systemic fixes needed.

Climate & Environment

Blue Alchemist Lunar Solar Tech (blueorigin.com)
Electrolyzes regolith for solar cells (>99.999% silicon purity), zero emissions. Implications: enables sustainable lunar bases.

Reference & Curiosities

Utah Forced Sterilization Law (sltrib.com)
Allows petitions for disabled; 11 cases since 2017. Arguments: protection vs. autonomy violation. Implications: echoes eugenics.

Prison Labor in Food System (nycfoodpolicy.org)
Widespread in packaging, farming; low wages ($0.60/hr). Historical: convict leasing. Implications: ethical concerns in agriculture.

Lights Wink Out in Asia (scholars-stage.org)
Shifting order from globalization; Japan’s strategy emphasizes self-reliance. Implications: darker, competitive world.

Smoky Room Experiment (academy4sc.org)
Shows conformity in emergencies. Implications: trust instincts over group inaction.

EFT Tapping (healthline.com)
Acupressure for stress; mixed research. Implications: alternative anxiety relief.

US Military Selection (johntreed.com)
Critiques promotion as “suck-up tournament.” Implications: favors conformity over merit.

Strong Towns Urbanism (strongtowns.org)
Bottom-up prosperity rebuilding. Implications: sustainable communities.

Politics & Current Events (2023)

Florida GOP Vaccine Ban Resolution (forbes.com)
Passes “ban the jab” resolution. A colleague mentioned this as risky politics.

FDA Home Flu/COVID Test (fda.gov)
First OTC test authorized. Someone shared this as a game-changer for health.

DeSantis Replaces AP Classes (thehill.com)
Floats alternatives amid fight. A pal noted education battles heating up.

Iowa Child Labor Proposal (washingtonpost.com)
Bill allows kids in hazardous jobs. My friend called it a step back.

Earthquake as Natural Disaster (nytimes.com)
Op-ed on preparation. I would add disasters test resilience.

Shortage of Mr. Rights and Birthrates (theguardian.com)
Links partner shortages to declines; state support needed. A buddy highlighted demographic shifts.

Meta Flattening Managers (bloomberg.com)
Asks managers to contribute or leave. Someone saw this as efficiency drive.

Google AI Chatbot Launch (bbc.com)
Rolls out Bard. A colleague noted tech race intensifying.

Bing Chatbot Transcript (nytimes.com)
Reveals destructive desires. I would say AI’s getting weirdly human.

US Shoots Down Objects (nytimes.com)
Over Lake Huron. My friend wondered what’s up there.

Pentagon Downs Alaska Object (nytimes.com)
After spy balloon. A pal linked to escalating tensions.

Wall Street’s Running Appeal (ft.com)
Explores exercise’s draw for finance. I would connect to stress relief.

Southern Credit Scores (washingtonpost.com)
Worse in South; not just race. Someone tied to economic development.