Models

/*
Generated by Squiggle AI. Workflow ID: 754c8ed2-937c-4ed9-9ab3-ee7f907341a4
*/
import "hub:ozziegooen/sTest" as sTest

// Probability analysis for insect sentience
// This model evaluates the likelihood that arthropods can experience
// valenced conscious states like pain and pleasure in morally relevant ways

@name("Evolutionary Arguments")

// Assumptions
total_campaign_benefit = beta({mean: 0.07, stdev: 0.04}) / 2 // div by 2 because we are only increasing half the voters' turnout
labor_share = beta({mean: 0.01, stdev: 0.01})
labor_benefit = total_campaign_benefit * labor_share
smallest_margin = normal(0.005, 0.003)
election_won_by_one_state = beta({mean: 0.07, stdev: 0.04})

// Calculations
n = 1000
labor_samples = sampleN(labor_benefit, n)

fmap(f) = {
  n = 100
  g(x) = SampleSet.make(List.map(sampleN(x, n), {|z| f(z) }))
  g
}
curry(f, x) = {
  g(y) = f(x, y)
  g
}
fmap2(f, x) = fmap(curry(f, x))

// cost–benefit analysis of career paths for a cs freshman
// comparing mechanistic interpretability research vs interstellar ai-safety outreach

import "hub:ozziegooen/sTest" as sTest

// ===== inputs =====

@name("ai timeline assumptions")
aiTimelineInputs = {
  @name("year range when agi is developed")

/*
Estimate AGI timeline
*/

a = normal(2, 5)

// CO2 vs Methane climate damage comparison
@name("Parameters")
params = {
  vulnDecline = normal(3.25%, 1.08%) + normal(0, 1%)
  damageExp = 1 to 2
  horizon = 250
  delay = 0
  co2Delay = delay
  methaneDelay = delay
  maxWarm = 1.9 to 2.8

/*
Generated by Squiggle AI. Workflow ID: e5aa49ca-df5a-4fa6-a890-d09f015959fb
*/
import "hub:ozziegooen/sTest" as sTest

// Economic Growth Incentives (EGI) Impact Model
// This model simulates how EGI affects DOT utilization, staking rates, and overall ecosystem dynamics

@name("Base Parameters")
baseParams = {

// Innovation data with correlated lognormal distributions
techData = [
  {name: "SHR Geothermal", weight: 117, min: 0.43, max: 8.82},
  {name: "SMR", weight: 966, min: 0.29, max: 7.64},
  {name: "EGS", weight: 170, min: 0.22, max: 4.41},
  {name: "Plant-Based Protein", weight: 153, min: 0.18, max: 4.04},
  {name: "Cultivated Protein", weight: 117, min: 0.00001, max: 0.07}, // Avoid zero for lognormal
  {name: "Nat Gas + CCS", weight: 313, min: 0.21, max: 5.21},
  {name: "Solar", weight: 497, min: 0.15, max: 4.19},
  {name: "Green Cement", weight: 136, min: 0.13, max: 2.42},

// Can we build TC tractability from a series of comparisons?

// General uncertainty about TC tractability
base_uncertainty(t) = 0.25 to 4

// Easier to defend than create
defend_bonus = 1 to 4

// Easier to boost innovation system than pursue structural changes
nonstructural_bonus = 1 to 4

/*
Generated by Squiggle AI. Workflow ID: f3cd6f14-d436-4dc8-bacd-70ec010d7d53
*/
import "hub:ozziegooen/sTest" as sTest

// Model to estimate the impact of plant-based meat on animal welfare

@name("Market and Cost Parameters")
@doc(
  "Key parameters for market dynamics, costs, and growth of plant-based meat alternatives"

/*
 Generaed code with Google Gemini as a first try at modeling animal welfare outcome for development and consumption of plant based meats

Here is a Squiggle notebook that models the impact of plant-based meat on animal welfare based on your specifications.
*/

// a = normal(2, 5)


// == Inputs ==

/*
1. Following Gelman et al., I model the vote margin as a normal distribution
1.1. NB: I don't think primary elections are actually well modeled as having normally distributed vote margins? Often there is one very distant frontrunner. But maybe highly contested primaries are well modeled like this and this will be a highly contested primary? caveat emptor
2. Dollars per vote: I loosely based this on a document from Dan Eth about Jay Shooster's campaign but kind of made it up. Better numbers appreciated
3. Primary: 
3.1 Default odds of winning: 40%, per Jay Shooster
3.2 Sd of margin: made up by me saying it will be 10% of total votes
4. General:
4.1 Default odds of winning: Jay suggests that the democrat will win this district iff democrats control the house, and probability for the latter is currently ~75% on prediction markets
4.2 Sd of margin: last year was 2.5k, so I vaguely guessed 5k

/*
BOTECs to estimate the Tractability of various Theories of Change
*/

// SYS_INN BOTEC
// Use historical example of Breakthrough Energy (BE), which influenced increasing ambition in innovation through the DoE around 2022.

peakEffectBE = 1% to 4%
shareOfUSInnovationPotentialDoE = 10% to 50%
effectDurationBE = 1 to 5

// Climate damage model with temperature, vulnerability, and damage over time
@name("Delay (years)")
@doc("Number of years to delay the start of the intervention")
delay = 20

@name("Model Parameters")
inputs = {
  @name("Vulnerability Decline Rate (% per year)")
  @doc("Annual percentage decline in vulnerability, based on decline in global catastrophe mortality rate over time, plus noise component")
  @format(".1%")

/*
DEPRECATED. See https://squigglehub.org/models/AI-safety/usa-pause

Is it a good idea for US actors to accelerate an AI arms race with China?
*/

// Probability that an aligned AI has good values regardless of what its creator wants, i.e., an aligned Chinese-built AI will refuse to let China take over the world.
p_aligned_ai_enforces_good_values = 0.1

// Probability that China would take over the world if it could.

/*
Generated by Squiggle AI. Workflow ID: 637a2d2d-11df-4ccb-8e5b-76a5886d3be8
*/
import "hub:ozziegooen/sTest" as sTest

// Anthropic Revenue Model with Bio Lab Projection
// This model projects Anthropic's revenue from 2025-2030 with the ability to condition on
// whether a biological research division is created in 2027, as well as other key variables.

// == Inputs ==

/*
This model compares the expected impact on future AIs' values from two interventions:

1. Conventional animal activism — persuading people to care more about farm animals / eat less meat
2. Creating a friendliness-to-animals LLM benchmark and advocating for AI companies to use it in post-training.

My numbers are intentionally overly optimistic toward conventional activism and overly pessimistic toward an LLM benchmark because I was pretty sure the latter was going to look better, and indeed it still looks better by >2 orders of magnitude.

I came up with the input parameters by doing approximately three seconds of research, I'm sure they could be much better.

/*
If USA unilaterally pauses at a national level and doesn't coordinate with China, is that good or bad?

"pause" doesn't have to be a literal pause, could also be a meaningful slowdown. What I have in mind is a significant pause or slowdown to put serious effort into solving AI safety, not just a short pause/slight slowdown.

Model was written manually with point probabilities and then run through Squiggle AI to generate boilerplate + documentation, then I manually edited the documentation to make it better. Squiggle AI also turned all my point probabilities into ranges, which is fine I guess.
*/

// == Part 1: P(doom) ==

/*
Generated by Squiggle AI. Workflow ID: 60eefacd-2bb4-4371-b8e0-277627606855
*/
import "hub:ozziegooen/sTest" as sTest

// LearnToGive: Cost-Benefit Analysis Model
// Calculates the impact of a nonprofit providing scholarships, computer equipment, and education in Thailand

@name("Key Input Parameters")
inputs = {

/*
Calculate a CBA of my nonprofit, LearnToGive, which raises money for 500 baht scholarships for students in Northern and Northeastern Thailand via tutoring of other students. We have raised 40k baht, enough for nearly 60 scholarships for severly underfunded and undersupported students. Additional money raised will be given to the school to buy computers (only two in the school are working), and I will be spending two weeks teaching english at the school myself (as they do not have actual english teachers). Calculate, considering PPP and the impact on ourselves as tutors with Feynman technique, impact on students we teach, and impact on scholarship students, the overall CBA and impact of the nonprofit.
*/

a = normal(2, 5)