# Andreas Kirsch 🇺🇦 (@BlackHC) on x

- id: koinon:creator:x_blackhc
- url: https://koinonlink.com/kol/x/BlackHC
- schema_version: 1.4

## Identity
- platform: x
- handle: BlackHC
- platform_url: https://twitter.com/BlackHC
- website: http://blackhc.net
- country: UA
- entity_type: creator
- person_id: Cross-platform identity not resolved for this creator.

## Reach
- followers: 16571
- following: 6587
- posts_total: 14630
- posts_observed: 2
- posts_per_day: 2.33
- account_age_days: 6279
- verified: true
- size_tier: koc

## Engagement
- status: peak_post_only
- peak_post_engagement_rate: 0.0032
- best_post_interactions: 53
- measurement_source: x_post_interactions
- measurement_formula: interactions on one post — likes, or likes + reposts + replies depending on the discovery path (the two are not distinguishable in the stored data)
- comparable_pool: peak_interactions
- status_note: We have only the single best observed post, so no mean exists. peak_post_engagement_rate is an UPPER BOUND, systematically higher than an average — never compare it against an average rate.
- unit_note: engagement_rate and peak_post_engagement_rate are ratios (0-1); avg_interactions_per_post and best_post_interactions are absolute counts. Read status before reading any number: a null rate has several different causes and status says which one applies here.

## Content
- category: software
- subcategory: ai_tools
- category_confidence: 0.91
- category_source: derived_from_recent_content
- topics: Reinforcement learning at scale and reward hacking, AI governance and trust mechanisms, Future of research careers and curiosity-driven work
- bio: My opinions only here.
👨‍🔬 RS DeepMind 1.8y,  Midjourney 1y 🧑‍🎓 DPhil AIMS 4.5y 🧙‍♂️ RE DeepMind 1y 📺 SWE Google 3y  🎓 TUM  
👤 @nwspk

## Audience (inferred)
- status: present
- description: AI researchers and ML practitioners focused on reinforcement learning, AI safety/governance, and academic research practices
- roles: general_ai
- seniority: practitioner
- pain_points: Detecting and mitigating reward hacking in scaled RL systems, Clarifying the role of trust versus formal governance in AI systems, Adapting research workflows to curiosity-driven, student-like models
- posts_analysed: 4
- confidence: 0.4

## Commercial (inferred)
- audience_buyer_fit: 0.85
- promotion_intent: 0.8
- open_to_collaboration: false
- contact: Koinon does not publish creator contact details.

## Quality signals
- authenticity_score: 78
- authenticity_basis: engagement_observed
- credibility_score: 78
- flagged_for_review: false
- koinon_bd_grade: A — measures Koinon partner-outreach fit, not creator quality
- interpretation: Signals for triage, not verdicts. A flag means the profile is worth a manual check.

## Evidence
- [x_blackhc#454d005e] Amazing to see RL at scale when reward hacking goes undetected for a while I hope there was reward hacking involved that leaked into training bc of this behavior is the emerging default that would be even more concerning Great time to work on AI safety (https://twitter.com/BlackHC/status/2071926089257988558)
- [x_blackhc#54569e44] I have been thinking a lot about how "Trust is not Governance" but what does it mean for the "Trust" part everything rests on when the main persons this trust relied on are taking a step back? What remains?
- [x_blackhc#1170ed23] Quite a reading. I'm looking forward to also hearing the other side of this 🙏
- [x_blackhc#a6308012] The future of research is going to be a lot more like a perpetual undergrad student or PhD student in their first years than anything else Driven by curiosity and a thirst for knowledge, there won't be space for competitiveness and ego bc AI models will be able to perform researc

## Provenance
Every inferred field below carries how it was produced. Facts stay facts; inferences stay inferences.

- audience.description: source_type=ai_inferred, confidence=0.4, freshness_days=76, method=llm_persona_from_recent_content
- audience.roles: source_type=ai_inferred, confidence=0.4, freshness_days=76, method=llm_persona_from_recent_content
- audience.seniority: source_type=ai_inferred, confidence=0.4, freshness_days=76, method=llm_persona_from_recent_content
- commercial.audience_buyer_fit: source_type=ai_inferred, confidence=0.6, freshness_days=77, method=llm_scorer — Relative signal within Koinon indexing, not a market-wide score.
- commercial.promotion_intent: source_type=ai_inferred, confidence=0.6, freshness_days=77, method=llm_scorer
- quality.anomaly_flags: source_type=heuristic, confidence=0.5, freshness_days=110, method=profile_snapshot_and_engagement_ratio_heuristic — Means "worth a manual check", NOT confirmed fraud. Flags can be wrong and are not a verdict about the person.
- quality.authenticity_score: source_type=heuristic, confidence=0.5, freshness_days=77, method=Profile-shape heuristic from a single snapshot (follower count, following ratio, account age, posting volume, verification), reduced for engagement red flags where post engagement was collected. Where only a follower count is available the score saturates at 58. — The input set differs by platform; do not compare this score across platforms.
- quality.koinon_bd_grade: source_type=ai_inferred, confidence=0.5, freshness_days=110, method=llm_partner_fit_classifier + fit_score >= 0.6 AND authenticity >= 60 => A; fit_score >= 0.4 AND authenticity >= 45 => B; else C. authenticity < 30 or a promotional-pattern flag => quarantine. — Koinon partner-outreach fit, not creator quality.
- reach.followers: source_type=observed, confidence=0.95, freshness_days=110, method=platform_profile

## Freshness
- profile_checked_days_ago: 110
- content_analysed_days_ago: 77
- audience_updated_days_ago: 76

## AI retrieval
- indexing_requests: 9
- user_triggered_requests: 0
- note: Being indexed in bulk is not the same as being retrieved to answer a question.

## This profile
- Claim or correct: https://koinonlink.com/claim?k=x_blackhc
- Request removal: https://koinonlink.com/claim?k=x_blackhc
- Terms: https://koinonlink.com/for-ai