extract: 2025-00-00-em-dpo-heterogeneous-preferences

Pentagon-Agent: Ganymede <F99EBFA6-547B-4096-BEEA-1D59C3E4028A>

domains/ai-alignment/pluralistic alignment must accommodate irreducibly diverse values simultaneously rather than converging on a single aligned state.md

@@ -25,6 +25,12 @@ Since [[universal alignment is mathematically impossible because Arrows impossib
 
 MaxMin-RLHF provides a constructive implementation of pluralistic alignment through mixture-of-rewards and egalitarian optimization. Rather than converging preferences, it learns separate reward models for each subpopulation and optimizes for the worst-off group (Sen's Egalitarian principle). At Tulu2-7B scale, this achieved 56.67% win rate across both majority and minority groups, compared to single-reward's 70.4%/42% split. The mechanism accommodates irreducible diversity by maintaining separate reward functions rather than forcing convergence.
 
+
+### Additional Evidence (confirm)
+*Source: [[2025-00-00-em-dpo-heterogeneous-preferences]] | Added: 2026-03-16*
+
+EM-DPO implements this through ensemble architecture where separate models serve different preference types, with MMRA aggregation ensuring no group is severely underserved. The system maintains diversity by training type-specific models rather than forcing convergence to a single reward function.
+
 ---
 
 Relevant Notes:

domains/ai-alignment/rlhf-is-implicit-social-choice-without-normative-scrutiny.md

@@ -33,6 +33,12 @@ The paper's proposed solution—RLCHF with explicit social welfare functions—c
 
 RLCF makes the social choice mechanism explicit through the bridging algorithm (matrix factorization with intercept scores). Unlike standard RLHF which aggregates preferences opaquely through reward model training, RLCF's use of intercepts as the training signal is a deliberate choice to optimize for cross-partisan agreement—a specific social welfare function.
 
+
+### Additional Evidence (extend)
+*Source: [[2025-00-00-em-dpo-heterogeneous-preferences]] | Added: 2026-03-16*
+
+EM-DPO makes the social choice explicit by using MinMax Regret Aggregation based on egalitarian principles. This shows that pluralistic alignment requires both discovering preference types (EM) and choosing an aggregation principle (MMRA), with the latter being an unavoidable normative choice.
+
 ---
 
 Relevant Notes:

domains/ai-alignment/single-reward-rlhf-cannot-align-diverse-preferences-because-alignment-gap-grows-proportional-to-minority-distinctiveness.md

@@ -33,6 +33,12 @@ Chakraborty, Qiu, Yuan, Koppel, Manocha, Huang, Bedi, Wang. "MaxMin-RLHF: Alignm
 
 Study demonstrates that models trained on different demographic populations show measurable behavioral divergence (3-5 percentage points), providing empirical evidence that single-reward functions trained on one population systematically misalign with others.
 
+
+### Additional Evidence (extend)
+*Source: [[2025-00-00-em-dpo-heterogeneous-preferences]] | Added: 2026-03-16*
+
+EM-DPO demonstrates that the problem is deeper than single-reward optimization — binary comparisons are formally insufficient for preference type identification. Even with perfect optimization, pairwise RLHF/DPO cannot detect heterogeneity because the data format lacks the necessary information structure. Rankings over 3+ responses are required.
+
 ---
 
 Relevant Notes:

domains/health/glp-1-multi-organ-protection-creates-compounding-value-across-kidney-cardiovascular-and-metabolic-endpoints.md

@@ -42,6 +42,12 @@ SELECT trial exploratory analysis (N=17,604, median 41.8 months) shows semagluti
 
 Phase 3 trial shows semaglutide 2.4mg achieves 62.9% resolution of steatohepatitis without worsening fibrosis vs 34.3% placebo. Meta-analysis confirms GLP-1 RAs significantly increase histologic resolution of MASH, decrease liver fat deposition, improve hepatocellular ballooning, and reduce lobular inflammation. Some hepatoprotective benefits appear at least partly independent of weight loss, suggesting direct liver effects beyond metabolic improvement. This adds hepatic protection as a third major organ system (alongside cardiovascular and renal) where GLP-1s demonstrate protective effects.
 
+
+### Additional Evidence (confirm)
+*Source: [[2024-05-29-nejm-flow-trial-semaglutide-kidney-outcomes]] | Added: 2026-03-16*
+
+FLOW trial demonstrated 29% reduction in cardiovascular death (HR 0.71, 95% CI 0.56-0.89) and 18% lower risk of major cardiovascular events in a kidney-focused trial. The cardiovascular benefits emerged as secondary endpoints in a study designed for kidney outcomes, supporting the multi-organ protection thesis. Separate analysis in Nature Medicine showed additive benefits when combined with SGLT2 inhibitors.
+
 ---
 
 Relevant Notes:

domains/health/semaglutide-reduces-kidney-disease-progression-24-percent-and-delays-dialysis-creating-largest-per-patient-cost-savings.md

@@ -28,6 +28,12 @@ This is the first dedicated kidney outcomes trial with a GLP-1 receptor agonist,
 - FDA indication expansion to T2D patients with CKD (2024)
 - Dialysis cost benchmark: $90K+/year per patient
 
+
+### Additional Evidence (confirm)
+*Source: [[2024-05-29-nejm-flow-trial-semaglutide-kidney-outcomes]] | Added: 2026-03-16*
+
+FLOW trial (N=3,533, median 3.4 years follow-up) showed 24% reduction in major kidney disease events (HR 0.76, P=0.0003), with annual eGFR decline slowed by 1.16 mL/min/1.73m2 (P<0.001). Trial stopped early at prespecified interim analysis due to efficacy. FDA subsequently expanded semaglutide indications to include T2D patients with CKD. This is the first dedicated kidney outcomes trial with a GLP-1 receptor agonist, published in NEJM.
+
 ---
 
 Relevant Notes:

domains/internet-finance/futarchy-incentive-programs-use-multisig-execution-groups-as-discretionary-override.md

@@ -17,6 +17,12 @@ The Drift proposal establishes a 2/3 multisig execution group (metaprophet, Suma
 
 The Drift proposal explicitly states 'All grant decisions are at the discretion of the decision council and any such decisions made by the decision council are final.' This creates a hybrid structure where futarchy approves the program budget but a committee controls individual allocations, demonstrating the pattern of discretionary override for operational decisions.
 
+
+### Additional Evidence (confirm)
+*Source: [[2024-05-30-futardio-proposal-drift-futarchy-proposal-welcome-the-futarchs]] | Added: 2026-03-16*
+
+Drift proposal uses 2/3 multisig execution group (metaprophet, Sumatt, Lmvdzande) with explicit discretion: 'exact criteria for this shall be finalized by the execution group' for activity filtering, and 'if successful proposals exceed two, executor group can decide top N proposals to split.' Multisig receives 3,000 DRIFT allocation and has authority to 'distribute their allocation as they see fit' or return excess funds.
+
 ---
 
 Relevant Notes:

domains/internet-finance/futarchy-markets-can-price-cultural-spending-proposals-by-treating-community-cohesion-and-brand-equity-as-token-price-inputs.md

@@ -38,6 +38,12 @@ Dean's List DAO's fee increase proposal included switching quote token from mSOL
 
 The HNT-ORE boost proposal frames strategic partnership value through liquidity network effects and brand positioning ('flagship DePIN project', 'competitive unit of account for real world assets'). Markets must price whether Helium association increases ORE's perceived legitimacy and network depth, demonstrating futarchy's ability to evaluate partnership proposals with significant intangible components.
 
+
+### Additional Evidence (confirm)
+*Source: [[2024-12-02-futardio-proposal-approve-deans-list-treasury-management]] | Added: 2026-03-16*
+
+Dean's List DAO treasury de-risking proposal passed with market pricing showing 5-20% FDV increase ($500k to $525k-$600k) based on financial stability perception. The proposal explicitly modeled how converting volatile assets to stablecoins would impact market confidence and token valuation, demonstrating futarchy markets can price operational stability as a token price input.
+
 ---
 
 Relevant Notes:

domains/internet-finance/futarchy-proposals-with-favorable-economics-can-fail-due-to-participation-friction-not-market-disagreement.md

@@ -79,6 +79,12 @@ The Dean's List proposal passed futarchy governance despite requiring complex mu
 
 LobsterFutarchy's failure ($1,183 of $500,000 target) occurred despite proposing infrastructure for a stated market need (agent financial sandboxing) and reasonable economics ($45k/month burn for 12 months). The 99.8% funding shortfall suggests participation friction or credibility gaps rather than market rejection of the concept itself.
 
+
+### Additional Evidence (extend)
+*Source: [[2024-12-02-futardio-proposal-approve-deans-list-treasury-management]] | Added: 2026-03-16*
+
+Dean's List treasury proposal required TWAP > 3% to pass and projected 5-20% FDV increase, well above the threshold. The proposal passed, suggesting that when economic benefits substantially exceed participation thresholds, friction becomes less determinative of outcomes.
+
 ---
 
 Relevant Notes:

domains/internet-finance/futarchy-proposer-incentives-require-delayed-vesting-to-prevent-gaming.md

@@ -11,6 +11,12 @@ created: 2026-03-15
 
 The Drift proposal structures proposer rewards with a three-month delay between proposal passage and token claim. Passing proposals earn up to 5,000 DRIFT each, but tokens are only claimable after three months. This delay creates a quality filter: proposers must believe their proposals will create sustained value that survives the vesting period. Without this delay, rational actors could spam low-quality proposals to extract rewards, knowing they can exit before negative effects manifest. The proposal also includes an executor group discretion clause - if successful proposals exceed expectations, the group can decide which top N proposals split the allocation. This combines time-based filtering with human judgment to prevent gaming. The 20,000 DRIFT activity pool uses the same three-month delay, with criteria finalized by the execution group to 'filter for non organic activity.'
 
+
+### Additional Evidence (confirm)
+*Source: [[2024-05-30-futardio-proposal-drift-futarchy-proposal-welcome-the-futarchs]] | Added: 2026-03-16*
+
+Drift proposal implements 3-month vesting for proposer rewards (up to 5,000 DRIFT per passing proposal) and activity pool rewards (20,000 DRIFT split), explicitly stating rewards are 'claimable after 3 months.' This prevents immediate extraction and forces alignment with longer-term outcomes.
+
 ---
 
 Relevant Notes:

domains/internet-finance/futarchy-retroactive-rewards-bootstrap-participation-through-endowment-effect.md

@@ -11,6 +11,12 @@ created: 2026-03-15
 
 The Drift Futarchy incentive program explicitly uses retroactive token distribution to MetaDAO participants as a mechanism to bootstrap engagement. The proposal cites the endowment effect - the behavioral economics finding that people value things more highly once they own them - as the theoretical basis. By distributing 9,600 DRIFT to 32 MetaDAO participants based on historical activity (5+ interactions over 30+ days), plus 2,400 DRIFT to AMM swappers, the proposal creates a cohort of token holders who have psychological ownership before the futarchy system launches. This differs from standard airdrops by explicitly targeting demonstrated forecasters rather than broad distribution. The tiered structure (100-400 DRIFT based on META holdings) further segments by engagement level. The proposal pairs this with forward incentives (5,000 DRIFT per passing proposal, 20,000 DRIFT activity pool) to convert initial ownership into sustained participation.
 
+
+### Additional Evidence (confirm)
+*Source: [[2024-05-30-futardio-proposal-drift-futarchy-proposal-welcome-the-futarchs]] | Added: 2026-03-16*
+
+Drift Futarchy proposal explicitly cites endowment effect as mechanism for retroactive rewards to 32 MetaDAO participants (9,600 DRIFT) based on activity thresholds. Proposal states rewards are 'meant to signal rewards for strong forecasters in futarchic markets' by 'rewarding early and active participants of MetaDAO with tokens to participate in Drift Futarchy (via the endowment effect).'
+
 ---
 
 Relevant Notes:

entities/internet-finance/metadao.md

@@ -65,6 +65,7 @@ The futarchy governance protocol on Solana. Implements decision markets through
 - **2024-01-24** — Proposed AMM program to replace CLOB markets, addressing liquidity fragmentation and state rent costs (Proposal CF9QUBS251FnNGZHLJ4WbB2CVRi5BtqJbCqMi47NX1PG)
 - **2024-01-29** — AMM proposal passed with 400 META on approval and 800 META on completion budget
 - **2024-08-31** — Passed proposal to enter services agreement with Organization Technology LLC, creating US entity vehicle for paying contributors with $1.378M annualized burn rate. Entity owns no IP (all owned by MetaDAO LLC) and cannot encumber MetaDAO LLC. Agreement cancellable with 30-day notice or immediately for material breach.
+- **2024-03-19** — Colosseum proposes $250,000 OTC acquisition of META with TWAP-based pricing (market price up to $850, voided above $1,200), 20% immediate unlock and 80% 12-month linear vest. Proposal passed 2024-03-24. Includes commitment to sponsor DAO track ($50-80K prize pool) in next Solana hackathon after Renaissance at no cost to MetaDAO.
 ## Key Decisions
 | Date | Proposal | Proposer | Category | Outcome |
 |------|----------|----------|----------|---------|

inbox/archive/.extraction-debug/2025-00-00-em-dpo-heterogeneous-preferences.json

@@ -0,0 +1,47 @@
+{
+  "rejected_claims": [
+    {
+      "filename": "binary-preference-comparisons-cannot-identify-latent-preference-types-requiring-rankings-over-three-or-more-responses.md",
+      "issues": [
+        "missing_attribution_extractor"
+      ]
+    },
+    {
+      "filename": "em-algorithm-discovers-latent-preference-subpopulations-enabling-ensemble-alignment-without-demographic-labels.md",
+      "issues": [
+        "missing_attribution_extractor"
+      ]
+    },
+    {
+      "filename": "minmax-regret-aggregation-ensures-no-preference-group-is-severely-underserved-by-applying-egalitarian-social-choice-to-ensemble-deployment.md",
+      "issues": [
+        "missing_attribution_extractor"
+      ]
+    }
+  ],
+  "validation_stats": {
+    "total": 3,
+    "kept": 0,
+    "fixed": 10,
+    "rejected": 3,
+    "fixes_applied": [
+      "binary-preference-comparisons-cannot-identify-latent-preference-types-requiring-rankings-over-three-or-more-responses.md:set_created:2026-03-16",
+      "binary-preference-comparisons-cannot-identify-latent-preference-types-requiring-rankings-over-three-or-more-responses.md:stripped_wiki_link:single-reward-rlhf-cannot-align-diverse-preferences-because-",
+      "binary-preference-comparisons-cannot-identify-latent-preference-types-requiring-rankings-over-three-or-more-responses.md:stripped_wiki_link:rlhf-is-implicit-social-choice-without-normative-scrutiny.md",
+      "em-algorithm-discovers-latent-preference-subpopulations-enabling-ensemble-alignment-without-demographic-labels.md:set_created:2026-03-16",
+      "em-algorithm-discovers-latent-preference-subpopulations-enabling-ensemble-alignment-without-demographic-labels.md:stripped_wiki_link:modeling preference sensitivity as a learned distribution ra",
+      "em-algorithm-discovers-latent-preference-subpopulations-enabling-ensemble-alignment-without-demographic-labels.md:stripped_wiki_link:pluralistic alignment must accommodate irreducibly diverse v",
+      "minmax-regret-aggregation-ensures-no-preference-group-is-severely-underserved-by-applying-egalitarian-social-choice-to-ensemble-deployment.md:set_created:2026-03-16",
+      "minmax-regret-aggregation-ensures-no-preference-group-is-severely-underserved-by-applying-egalitarian-social-choice-to-ensemble-deployment.md:stripped_wiki_link:maxmin-rlhf-applies-egalitarian-social-choice-to-alignment-b",
+      "minmax-regret-aggregation-ensures-no-preference-group-is-severely-underserved-by-applying-egalitarian-social-choice-to-ensemble-deployment.md:stripped_wiki_link:post-arrow-social-choice-mechanisms-work-by-weakening-indepe",
+      "minmax-regret-aggregation-ensures-no-preference-group-is-severely-underserved-by-applying-egalitarian-social-choice-to-ensemble-deployment.md:stripped_wiki_link:pluralistic-ai-alignment-through-multiple-systems-preserves-"
+    ],
+    "rejections": [
+      "binary-preference-comparisons-cannot-identify-latent-preference-types-requiring-rankings-over-three-or-more-responses.md:missing_attribution_extractor",
+      "em-algorithm-discovers-latent-preference-subpopulations-enabling-ensemble-alignment-without-demographic-labels.md:missing_attribution_extractor",
+      "minmax-regret-aggregation-ensures-no-preference-group-is-severely-underserved-by-applying-egalitarian-social-choice-to-ensemble-deployment.md:missing_attribution_extractor"
+    ]
+  },
+  "model": "anthropic/claude-sonnet-4.5",
+  "date": "2026-03-16"
+}

inbox/archive/2024-05-29-nejm-flow-trial-semaglutide-kidney-outcomes.md

@@ -7,9 +7,13 @@ date: 2024-05-29
 domain: health
 secondary_domains: []
 format: paper
-status: unprocessed
+status: enrichment
 priority: high
 tags: [glp-1, semaglutide, CKD, kidney-disease, FLOW-trial, organ-protection]
+processed_by: vida
+processed_date: 2026-03-16
+enrichments_applied: ["semaglutide-reduces-kidney-disease-progression-24-percent-and-delays-dialysis-creating-largest-per-patient-cost-savings.md", "glp-1-multi-organ-protection-creates-compounding-value-across-kidney-cardiovascular-and-metabolic-endpoints.md"]
+extraction_model: "anthropic/claude-sonnet-4.5"
 ---
 
 ## Content
@@ -38,3 +42,11 @@ Additive benefits when used with SGLT2 inhibitors (separate analysis in Nature M
 PRIMARY CONNECTION: [[GLP-1 receptor agonists are the largest therapeutic category launch in pharmaceutical history but their chronic use model makes the net cost impact inflationary through 2035]]
 WHY ARCHIVED: Kidney protection is where GLP-1 downstream savings are largest per-patient — dialysis prevention is the economic mechanism most favorable to the VBC cost-saving thesis
 EXTRACTION HINT: Focus on the economic implications of slowed kidney decline for capitated payers, not just the clinical endpoint
+
+
+## Key Facts
+- FLOW trial had N=3,533 patients with type 2 diabetes and chronic kidney disease
+- Median follow-up was 3.4 years before early stopping
+- Trial was stopped at prespecified interim analysis due to efficacy
+- Dialysis costs approximately $90K+/year per patient in the US
+- Separate analysis in Nature Medicine showed additive benefits with SGLT2 inhibitors

inbox/archive/2024-05-30-futardio-proposal-drift-futarchy-proposal-welcome-the-futarchs.md

@@ -6,9 +6,13 @@ url: "https://www.futard.io/proposal/9jAnAupCdPQCFvuAMr5ZkmxDdEKqsneurgvUnx7Az9z
 date: 2024-05-30
 domain: internet-finance
 format: data
-status: unprocessed
+status: enrichment
 tags: [futardio, metadao, futarchy, solana, governance]
 event_type: proposal
+processed_by: rio
+processed_date: 2026-03-16
+enrichments_applied: ["futarchy-retroactive-rewards-bootstrap-participation-through-endowment-effect.md", "futarchy-proposer-incentives-require-delayed-vesting-to-prevent-gaming.md", "futarchy-incentive-programs-use-multisig-execution-groups-as-discretionary-override.md"]
+extraction_model: "anthropic/claude-sonnet-4.5"
 ---
 
 ## Proposal Details
@@ -107,3 +111,11 @@ In the event of uncertainty or excess budget, funds shall be returned to origina
 - Autocrat version: 0.3
 - Completed: 2024-06-02
 - Ended: 2024-06-02
+
+
+## Key Facts
+- Drift Futarchy proposal 9jAnAupCdPQCFvuAMr5ZkmxDdEKqsneurgvUnx7Az9zS passed on 2024-06-02
+- 32 MetaDAO participants qualified for retroactive rewards based on 5+ interactions over 30+ days before May 19, 2024
+- Retroactive reward tiers: <1 META = 100 DRIFT, >=1 META = 200 DRIFT, >=10 META = 400 DRIFT
+- Total budget: 50,000 DRIFT split as 9,600 retroactive to qualified participants, 2,400 to AMM swappers, 10,000 for future proposers, 25,000 activity pool, 3,000 execution group
+- Execution group: metaprophet, Sumatt, Lmvdzande operating as 2/3 multisig

inbox/archive/2024-12-02-futardio-proposal-approve-deans-list-treasury-management.md

@@ -6,9 +6,13 @@ url: "https://www.futard.io/proposal/4gaJ8bi1gpNEx6xSSsepjVBM6GXqTDfLbiUbzXbARHW
 date: 2024-12-02
 domain: internet-finance
 format: data
-status: unprocessed
+status: enrichment
 tags: [futardio, metadao, futarchy, solana, governance]
 event_type: proposal
+processed_by: rio
+processed_date: 2026-03-16
+enrichments_applied: ["futarchy-markets-can-price-cultural-spending-proposals-by-treating-community-cohesion-and-brand-equity-as-token-price-inputs.md", "futarchy-proposals-with-favorable-economics-can-fail-due-to-participation-friction-not-market-disagreement.md"]
+extraction_model: "anthropic/claude-sonnet-4.5"
 ---
 
 ## Proposal Details
@@ -122,3 +126,11 @@ This strategy ensures financial stability while signaling prudence to investors,
 - Autocrat version: 0.3
 - Completed: 2024-12-05
 - Ended: 2024-12-05
+
+
+## Key Facts
+- Dean's List DAO treasury valued at $75,000-$87,000 at $350 SOL (excluding DEAN token)
+- Proposal required TWAP > 3% for passage, with conservative FDV baseline of $500,000
+- De-risking projected to increase survival probability from 50% to 90%
+- Market modeled 5% confidence boost scenario ($525k FDV) and 20% confidence boost scenario ($600k FDV)
+- Proposal created 2024-12-02, completed 2024-12-05

inbox/archive/2025-00-00-em-dpo-heterogeneous-preferences.md

@@ -7,9 +7,13 @@ date: 2025-01-01
 domain: ai-alignment
 secondary_domains: []
 format: paper
-status: unprocessed
+status: enrichment
 priority: medium
 tags: [pluralistic-alignment, EM-algorithm, preference-clustering, ensemble-LLM, fairness]
+processed_by: theseus
+processed_date: 2026-03-16
+enrichments_applied: ["single-reward-rlhf-cannot-align-diverse-preferences-because-alignment-gap-grows-proportional-to-minority-distinctiveness.md", "rlhf-is-implicit-social-choice-without-normative-scrutiny.md", "pluralistic alignment must accommodate irreducibly diverse values simultaneously rather than converging on a single aligned state.md"]
+extraction_model: "anthropic/claude-sonnet-4.5"
 ---
 
 ## Content
@@ -39,3 +43,9 @@ EM-DPO uses expectation-maximization to simultaneously uncover latent user prefe
 PRIMARY CONNECTION: RLHF and DPO both fail at preference diversity because they assume a single reward function can capture context-dependent human values
 WHY ARCHIVED: The binary-comparison insufficiency claim is a novel formal result that strengthens the case against standard alignment approaches
 EXTRACTION HINT: Focus on the formal insufficiency of binary comparisons and the EM + egalitarian aggregation combination
+
+
+## Key Facts
+- EM-DPO paper presented at EAAMO 2025 (Equity and Access in Algorithms, Mechanisms, and Optimization)
+- MMRA operates at inference time when user type is unknown, distinct from training-time aggregation approaches
+- The approach requires rankings over 3+ responses rather than binary comparisons for preference type identification