-
Notifications
You must be signed in to change notification settings - Fork 0
/
max_imp_model.squiggle
60 lines (44 loc) · 2.6 KB
/
max_imp_model.squiggle
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
// //PARAMETERS
// mi_outcomes(num_of_rounds) = {
// number_of_research_reports: 5 to 20*num_of_rounds^2,
// // quality_of_reports: something to estimate how aligned they are with EA + how accessible they are to donors
// money_invested_in_publicity: 1K to 20K,
// // publicity_effectiveness: something to estimate how effective the publicity is
// // publicity_effectiveness = quality_of_reports * publicity_effectiveness (or something like that, this is generated by Copilot)
// }
// donor_attitude = { // weighted by money, not by number of donors
// // metrics for donor familiarity with effectiveness of charities
// heard_of_GW: 1e-3 to 3e-2,
// // metrics for donor motivation to donate effectively
// cause_neutral: 1e-5 to 1e-3,
// agrees_with_EA: 1e-4 to 1e-2,
// importance_of_evidence: 1e-3 to 1e-1,
// // metrics for donor opportunity to donate effectively
// legally_locked_in_IL: 1e-6 to 1e-3,
// availability_of_info_on_effectiveness: 1e-3 to 1e-1,
// }
// effect_GW = 8 to 10 //normalize in "GiveDirectly" effectiveness units (effectiveness GD = 1)
// effect_IL = 0.2 to 1 //based on https://80000hours.org/2023/02/how-much-do-solutions-differ-in-effectiveness/
// total_money_IL = 19.5B to 22B
// gw_share = (12.3/20.7) * 0.1 // this number comes from share of Israel based donors from total donations in Il
// //MODEL
// total_effective_donations = donor_attitude * total_money_IL
// total_effective_donations_GW = total_effective_donations * gw_share
// total_effective_donations_IL = total_effective_donations * (1 - gw_share)
// total_impact_IL = effect_IL * total_effective_donations_IL // TODO: Change to an integral over all charities and cap at room-for-funding
// total_impact_GW = effect_GW * total_effective_donations_GW
// --------------------------------------------------------------
// estimating the cost-effectiveness distribution of IL charities
// --------------------------------------------------------------
// PARAMETERS
// log-normal distribution parameters for the cost-effectiveness of all charities
// parameters are normalized
lmu_all = 0 // mean after taking log
lsigma_all = 1 // standard deviation after taking log
apriori_multiplier_for_developed_countries = 10 to 100
apriori_multiplier_for_low_evidence = 1 to 4
amount_of_il_charities_health_and_poverty = 2k to 5k
// MODEL
charity_effectiveness = lognormal(lmu_all, lsigma_all)
gw_threshold = 10 // 10 times as cost-effective as GiveDirectly
actual_charity_effectiveness = truncateRight(charity_effectiveness, gw_threshold) // everything above this level is fully funded