WorldTracking
Can Talarico overcome narrowed lead to beat Paxton?
Public tracking started; no independent probability estimate yet.
Watching: 民调显示塔拉里科领先但优势缩小，具体差距和选举日结果存在不确定性；特朗普的警告和辩论结果可能改变选情。
Question tracking
Ask a question. Follow the answer.
Get a current estimate, save it, and see how it changes with new information.
Tracked questions
TechnologyTracking
Can Anthropic's independent evaluation pledge deliver?
Anthropic has granted METR employee-level access, but no report has been published; independence remains uncertain due to confidentiality redactions and commercial interests.
Watching: METR 尚未发布任何基于该访问权限的评估报告，其实际独立性和披露程度未知；Anthropic 的删减权范围（如模型架构细节）可能涵盖关键安全信息，但具体执行情况不明。
TechnologyTracking
Gemini 4 Argon doubles price after launch—how will AI market competition shift?
Gemini 4 Argon launched globally on Oct 13 with pricing at one-fifth of rivals, doubling after Nov 30, likely intensifying price competition and reshaping the AI market.
Watching: 报道未说明价格翻倍后用户是否会流失，也未提及谷歌是否计划推出替代定价或促销活动。此外，性能对比基于特定基准，实际应用中的表现可能因任务而异。
WorldTracking
Pentagon livestream of Hasan execution: deterrent or division?
Latest reports say Trump halted the live broadcast of the execution, but the death sentence remains; deterrence effect lacks evidence and may fuel violence.
Latest development: An Oct 16 report confirms Trump halted the live-streaming part while the execution proceeds, updating the earlier plan.
TechnologyTracking
Kling 4.0 launch: how will its open API reshape the video generation market?
Kling AI released Kling 4.0 on Sept 28, 2026, supporting 30-second native video; Flash version began limited testing, full version planned for October, API details not disclosed.
Watching: Kling 4.0的API具体开放时间、定价策略、实际生成质量（如30秒片段的连贯性）以及企业采用率均未公布，且竞争对手（如Runway、Pika）可能推出类似功能，市场影响存在变数。
TechnologyTracking
Can Anthropic's internal access for METR set a new standard for AI safety evaluation?
Anthropic granted six METR evaluators employee-level access; METR can publish independently but with redaction limits. Whether this becomes an industry standard remains uncertain.
Watching: METR评估的实际效果未知，Anthropic的删节权可能限制透明度，其他公司是否跟进尚不确定。
TechnologyTracking
What financial or operational concerns drive CoreWeave's 28% bankruptcy probability?
The market's elevated bankruptcy probability for CoreWeave reflects concerns over high leverage, debt costs exceeding returns, large negative free cash flow, and customer concentration risks.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
WorldTracking
Press secretary claims attorney without bar—impact on trust?
Zacharia claimed to be a lawyer without a license, prompting complaints and media scrutiny that may erode public trust in White House information credibility, though no direct polling data exists.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
SpaceX bundles X Premium, Grok, Cursor: can it shift paid AI market?
SpaceXSI launched The X Subscription bundling X Premium, Grok, and Cursor in October 2026, but reports provide no data on user acquisition or competitive impact.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
WorldTracking
Will Musk's donation reshape campaign finance?
Musk donated over $276 million to Republican groups, becoming the largest individual donor in the midterms, but reports do not mention new regulatory discussions.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
Question trackingTracking
Ruling halting East Potomac tree removal: impact on renovation and judicial review
A federal judge barred further tree removal at East Potomac Golf Course until Oct 26 and required three business days' notice before irreversible changes; the renovation is halted pending a hearing next Monday.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
Apple TV 4K remote-free voice in noise and privacy
The new Apple TV 4K has three far-field microphones for hands-free voice, but current reports lack noisy-environment test data or specific privacy concerns.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
Why isn't Apple selling HomeView speaker in the EU?
Apple is not selling HomeView in the EU because Siri AI is not yet available on EU iOS; it says it is in constructive talks with the EU. Reports give no EU launch timeline or commitment to adjust strategy.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
After DeepSeek's $8B+ funding, how will its AI development reshape Chinese tech giants' competition?
DeepSeek's over 8 billion yuan financing will intensify China's AI competition, raising valuations, concentrating resources among leaders, and boosting domestic compute ecosystems, though specific impacts remain to be seen.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
WorldTracking
What political shifts does Centre Party's 50% chance of winning most seats reflect?
The Centre Party boycotted and abstained in the 2026 presidential election, showing opposition to the ruling coalition, but the 2027 parliamentary landscape remains unclear.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
Eight states sue OpenAI over ChatGPT harm to minors: what's next for AI regulation?
Eight states filed separate suits against OpenAI on Oct 14, 2026, alleging ChatGPT harms minors; OpenAI says it will cooperate with states on pragmatic, industry-wide AI regulation.
Latest development: New fact: OpenAI has publicly stated willingness to cooperate with states on pragmatic regulation, emphasizing policies should apply to the whole AI industry, not a single company.
WorldTracking
How does UK's IRGC plot claim affect UK-Iran relations?
UK accused Iran's IRGC of plotting attack on RAF Fairford and sanctioned two Iranians; Iran summoned UK ambassador in protest; 29 countries backed UK; UK-Iran ties strained, future policy likely tougher but specifics unclear.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
MiniCPM-V 4.7 performance on 3GB devices?
ModelBest released MiniCPM-V 4.7; the 35B-A3B version runs in ~3GB memory with 13-15+ token/s decoding, but evidence lacks direct comparison with cloud LLMs or verified industry impact.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
Meta open-sources Muse Spark Open model: how will developers use it?
Meta released open-weight model Muse Spark Open on Oct 15, 2026, with performance below its closed-source frontier models. Developers can deploy and customize it, but adoption cases and third-party benchmarks remain to be seen.
Latest development: New evidence confirms Muse Spark Open was released on Oct 15, 2026, as an open-weight model with performance below its closed-source frontier counterparts, including Artificial Analysis intelligence index comparisons.
TechnologyTracking
Meta's open 816B multimodal model: can consumer hardware run it efficiently?
Muse Spark Open has 816B total params with 42B active per token and is optimized for Mac Studio, but no public benchmarks on consumer hardware exist yet, so its impact on multimodal app adoption remains unverified.
Latest development: New info: Meta has deeply optimized Muse Spark Open for Mac Studio and emphasizes coding/agent tasks, but no specific consumer-hardware inference performance data is provided.
TechnologyTracking
How will ByteDance's new model alignment and safety department affect large-model iteration and compliance?
ByteDance Seed's new safety-alignment department, plus its strategy to abandon third-party distillation, may slow LLM iteration, but specific compliance-strategy changes are not yet disclosed.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
WorldTracking
Can NATO's deterrence posture effectively deter Russia?
Evidence shows NATO's 2026 land-war concept and eastern-flank buildup aim to deter Russia, but does not prove deterrence will stop an attack; key uncertainty remains future actions.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
TechnologyTracking
Anthropic's 500MW Osaka data center off-grid: impact on power markets?
Anthropic and NTT Data are in final talks for a 500MW Osaka data center, initially considering on-site generation and fuel cells, but full grid bypass is unconfirmed and direct impact on residential electricity prices lacks evidence.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.
WorldTracking
Putin leaving office by end-2026 at 4.65%: why so confident?
The market prices a low probability of Putin leaving office before end-2026 because the next presidential election is scheduled for 2030 and he recently began a new term, indicating political stability.
Latest development: The assessment representation was corrected; evidence and earlier records are retained.