California’s Governor Signs Landmark Online Child Safety Bills - The New York Times
California’s Governor Signs Landmark Online Child Safety Bills The New York Times
Latest News Archive - Page 14
Garry Tan wants US open-weight AI labs to 'distill' frontier models, too
Article URL: https://techcrunch.com/2026/09/11/y-combinators-garry-tan-wants-u-s-open-weight-ai-labs-to-distill-frontier-models-too/ Comments URL: https://news.ycombinator.com/item?id=49685253 Points: 187 # Comments: 81
Dell stock jumps on RBC initiation, now up nearly 350% in 2026
Dell sold about $16.4 billion of AI servers in its second quarter, RBC said.
Lawyer fined $5K over AI-hallucinated witnesses in a murder case
New Mexico's Supreme Court is punishing a lawyer for including AI-fabricated witnesses and fake police testimony in an appeal for his client's murder conviction, according to a report from Reuters. In a filing on Wednesday, the court fined Stephen Aarons $5,000 and held him in contempt for failing to "verify the factual claims and legal […]
Tragic death of woman who testified against father in notorious mafia murder of mother
Denise Cosco testified against the powerful 'Ndrangheta clan and her own father over the murder of her mother in Italy.
One week left to book your exhibit table at TechCrunch Disrupt 2026
Only one week left to secure your exhibit table. Tables are limited and can sell out before the September 18 deadline.
Final, final, final call for TechCrunch Disrupt 2026 Side Events
The absolute last chance to apply to host an official Side Event during TechCrunch Disrupt 2026 is tonight, September 11, at 11:59 p.m. PT.
How shofars became the spiritual soundtrack of Rosh Hashanah
Rosh Hashanah, the Jewish New Year, begins at sunset Friday. For many Jews around the world, a highlight of the holiday will be hearing the shofar: an animal's horn blown like a trumpet.
Zelensky calls for stronger US sanctions against Russia, says Ukraine must engage in peace talks with Moscow - The Kyiv Independent
Zelensky calls for stronger US sanctions against Russia, says Ukraine must engage in peace talks with Moscow The Kyiv Independent
Roundtables: Could AI really kill us all?
Employees at the world’s leading AI labs are saying there’s a real possibility that advanced AI could destroy humanity. Are they right? Or is this more scaremongering and hype? Join MIT Technology Review executive editor Niall Firth for a conversation with senior AI editor Will Douglas Heaven and AI reporter Grace Huckins unpacking AI extinction…
Auto-resolution and analysis updates in Copilot code review
Copilot code review now resolves its own comments once you address them and writes smart commit messages for you when you apply its code suggestions. Behind the scenes, Copilot now… The post Auto-resolution and analysis updates in Copilot code review appeared first on The GitHub Blog .
Astera Labs vs. Marvell Technology: Which AI-Driven Tech Stock Is a Better Buy in 2026? - The Motley Fool
Astera Labs vs. Marvell Technology: Which AI-Driven Tech Stock Is a Better Buy in 2026? The Motley Fool
Lightning Houthi advance in Yemen may bring dangerous new dimension to Iran war
The capture by the rebels of a stretch of Yemen's strategic coastline is the group's most impressive military coup for several years.
Amid war and sanctions, many Iranians are losing their jobs and struggling to get by
Many Iranians lost their jobs because of the war and the U.S. blockade. Now some are even losing their backup jobs as sanctions tighten.
5 Best Edge Computing Stocks for 2026 and How to Invest - The Motley Fool
5 Best Edge Computing Stocks for 2026 and How to Invest The Motley Fool
DockFix 5.0
Replace the macOS Dock with one that is truly yours. Discussion | Link
Iran-supported Houthis in Yemen chanted "Death to America, death to Israel" as they took control of Mocha, a port city in the Red Sea. | via ANC 24/7 Link to the full story in the comments section - facebook.com
Iran-supported Houthis in Yemen chanted "Death to America, death to Israel" as they took control of Mocha, a port city in the Red Sea. | via ANC 24/7 Link to the full story in the comments section facebook.com
Altimeter's Gerstner blasts researchers voicing AI extinction warnings, questions 'political agenda'
Brad Gerstner responded to the flurry of AI stories this week after a researcher quit his job at Anthropic, accusing AI companies of "gambling with our lives."
Kubernetes v1.37: Native Histograms Graduates to Beta
I'm excited to announce that native histogram support for Kubernetes metrics is graduating to Beta and is enabled by default in Kubernetes v1.37! Native histograms (previously introduced as Alpha in Kubernetes v1.36 under KEP-5808 ) bring high-resolution, low-cardinality observability to Kubernetes metrics. By adopting Prometheus Native Histograms , Kubernetes components now expose latency and duration metrics with far greater accuracy while significantly reducing telemetry storage and scraping overhead. Why move beyond classic histograms? Since the early days of Kubernetes observability, duration and latency metrics (such as API server request latencies or scheduling durations) have relied on classic Prometheus histograms . Classic histograms require metric authors to define a static list of cumulative bucket boundaries ( le labels), such as 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10 . While familiar, this approach introduces three major challenges: The Bucket Guessing Game : If a workload's latency profile changes, for example, shifting into microsecond ranges or experiencing long-tail tail latencies beyond the highest bucket, the histogram loses visibility. Specifying bucket boundaries upfront requires knowing the distribution before observing it High Cardinality & Storage Cost : With classic histograms, each bucket boundary is exported as a separate time series ( _bucket{le="..."} ). A histogram with 10 buckets across multiple labels multiplies the number of time series by 10, increasing memory consumption in Prometheus and inflating time series database (TSDB) storage costs Interpolation Error in Quantiles : Calculating percentiles using histogram_quantile() relies on linear interpolation between static bucket boundaries. When bucket spans are coarse, quantile calculations can suffer from significant estimation error What are Prometheus native histograms? Prometheus Native Histograms replace static user-defined buckets with dynamic, exponential buckets . Instead of emitting a separate time series for every single bucket boundary, a native histogram is stored as a single time series containing a rich schema of positive and negative spans, zero thresholds, and exponential scaling factors. High Resolution Automatically : Exponential buckets dynamically adjust to any value range — from nanoseconds to hours — without requiring pre-configured bucket boundaries Up to 90% Fewer Time Series : By consolidating buckets into structured spans within a single time series, scraping and storage overhead are dramatically reduced Accurate Quantile Calculation : Quantiles can be calculated with mathematical bounds on error (≃5% worst-case relative error under default settings) across the entire spectrum of observations How native histograms work in Kubernetes In Kubernetes, native histogram support is implemented directly inside the shared metrics subsystem ( k8s.io/component-base/metrics ). Figure 1 illustrates how native histogram metrics are processed and exposed across Kubernetes components. Figure 1. Native histogram processing and dual exposition flow in Kubernetes. 1. Dual exposition for zero breaking changes A primary design requirement for KEP-5808 was zero disruption for existing observability stacks. When the NativeHistograms feature gate is enabled, Kubernetes components use dual exposition : Classic buckets ( h.Bucket ) are still emitted alongside native spans. Existing Prometheus servers, dashboards, and alerting rules that rely on traditional text scraping or classic bucket labels continue to work unmodified Native spans ( h.Schema , h.PositiveSpan ) are included in the same Protobuf payload for collectors that understand native histograms 2. Tuned default exponential configuration When NativeHistograms is enabled, the k8s.io/component-base/metrics package automatically applies standardized exponential options to all histogram metrics: BucketFactor: 1.1 : Configures exponential buckets where each bucket is at most 10% wider than the preceding one. This guarantees a mathematically bounded worst-case relative error of at most ~5% for quantile calculations regardless of whether an operation takes 1 millisecond or 10 seconds. MaxBucketNumber: 160 : Caps the maximum number of buckets per histogram to 160. Following OpenTelemetry SDK recommendations for base-2 exponential histogram aggregation, this limit protects component memory usage even under extreme outlier distributions. 3. Broad component support Because native histograms are integrated into component-base/metrics , all major Kubernetes control plane and node components inherit support automatically, including: kube-apiserver (e.g., apiserver_request_duration_seconds , authentication/authorization metrics, validation latencies) kube-scheduler (e.g., scheduler_plugin_execution_duration_seconds , scheduler_scheduling_algorithm_duration_seconds ) kubelet (node-level container runtime and pod lifecycle metrics) kube-controller-manager and kube-proxy How to scrape native histograms The simple answer: upgrade to Kubernetes v1.37, and it works. Because Kubernetes v1.37 enables NativeHistograms by default, your cluster is already emitting dual-exposition metrics. How you configure Prometheus to scrape native histograms depends on your Prometheus version: 1. Prometheus scrape configuration by version Prometheus 3.0+ (Recommended) : Use explicit per-job configuration in your scrape_configs rather than global flags (the global --enable-feature=native-histograms flag is deprecated in Prometheus 3.9+): scrape_configs : - job_name : 'kubernetes-apiservers' scrape_native_histograms : true always_scrape_classic_histograms : true # Recommended during transition You must read the caution in Migrating dashboards and alerts in the Native Histograms documentation. In summary: always set always_scrape_classic_histograms: true during your transition period. Without this setting, Prometheus will only ingest the native format and stop ingesting classic _bucket , _count , and _sum series. Setting always_scrape_classic_histograms: true ensures existing dashboards ( histogram_quantile(..._bucket...) ) and alerts continue to work while you migrate them to native histograms. Prometheus 2.40 – 2.x : Enable Native Histograms globally by starting Prometheus with the feature flag: prometheus --enable-feature = native-histograms Note that in Prometheus 2.x, this is an all-or-nothing setting for all scrape targets. 2. Verify Protobuf dual exposition Standard Prometheus text scraping ( application/openmetrics-text or plain text format) only transfers classic buckets. When scrape_native_histograms is enabled, Prometheus automatically negotiates Protobuf format with Kubernetes endpoints. You can verify that a Kubernetes component is exporting native histograms using curl with an Accept header specifying Protobuf. For example: ## THIS IS NOT SECURE. ONLY DO THIS IN A TEST CONTEXT. curl --insecure \ -H "Accept: application/vnd.google.protobuf;proto=io.prometheus.client.MetricFamily;encoding=delimited" \ --header "Authorization: Bearer $( cat /var/run/secrets/kubernetes.io/serviceaccount/token ) " \ https://localhost:6443/metrics When decoded, the returned MetricFamily for histogram metrics (like apiserver_request_duration_seconds ) will contain both traditional bucket entries and populated schema / positive_span fields. Querying native histograms in PromQL Once native histograms are ingested into Prometheus, you can query them using standard PromQL histogram functions without needing static le bucket labels or _bucket suffixes: # 1. Calculating P99 latency for a single target: # Classic histogram (requires _bucket suffix): histogram_quantile ( 0.99 , rate ( apiserver_request_duration_seconds_bucket [ 5m ] )) # Native histogram (operates directly on the metric name): histogram_quantile ( 0.99 , rate ( apiserver_request_duration_seconds [ 5m ] )) # 2. Aggregating across multiple instances (e.g., all API servers): # Classic histogram (requires sum by (le) to preserve bucket boundaries): histogram_quantile ( 0.99 , sum by ( le ) ( rate ( apiserver_request_duration_seconds_bucket [ 5m ] ))) # Native histogram (no grouping by le required!): histogram_quantile ( 0.99 , sum ( rate ( apiserver_request_duration_seconds [ 5m ] ))) With native histograms, functions like histogram_quantile() operate directly on the dynamic exponential spans inside the time series, producing highly accurate quantiles without static bucket interpolation error. For official documentation on querying Native Histograms in PromQL, see: PromQL histogram_quantile documentation PromQL histogram_fraction documentation PromQL histogram_sum documentation PromQL histogram_count documentation PromQL histogram_count documentation PromQL histogram_stddev and histogram_stdvar documentation Dashboard migration & rollback strategy Recommended migration workflow To safely transition your monitoring infrastructure to Native Histograms without breaking existing alerts or dashboards, I recommend a four-step migration workflow: Enable Both Formats : In your Prometheus 3.x scrape config, set scrape_native_histograms: true AND always_scrape_classic_histograms: true so both formats are collected safely during transition Migrate Queries : Update your Grafana dashboards and Prometheus alerting rules from classic quantile queries ( histogram_quantile(..._bucket...) ) to native histogram queries ( histogram_quantile(...) ), and replace references to classic _count and _sum series with histogram_count(...) and histogram_sum(...) Verify in Staging/Production : Validate that all dashboards and SLO alerts fire and graph correctly using the new native histogram queries Unlock ~10x Storage Savings : Once migration is complete, set always_scrape_classic_histograms: false . Prometheus will stop ingesting the static _bucket , _count , and _sum time series, reducing your histogram time series count by up to 90%! Opt-out and rollback flexibility Because native histograms are dual-exposed, using them is entirely opt-in from a collector perspective: Instant Collector Rollback : If you need to stop ingesting native histograms, simply set scrape_native_histograms: false in your Prometheus job configuration. No Kubernetes restart is required, and Prometheus will immediately resume scraping only the classic format without data loss Component Feature Gate Rollback : Administrators can also disable the feature gate on Kubernetes components using --feature-gates=NativeHistograms=false (requires component restart) What's next & how to get involved As native histograms progress toward General Availability (GA) in future Kubernetes releases, SIG Instrumentation will continue evaluating ecosystem readiness, performance characteristics, and long-term plans for eventually deprecating static classic buckets once native histogram adoption becomes ubiquitous across the monitoring community. Read the KEP-5808 page or the KEP GitHub issue to learn more. Read the Prometheus Native Histograms specification and PromQL querying functions documentation Get involved with SIG Instrumentation on Slack in #sig-instrumentation or join the weekly SIG meetings Acknowledgements A huge thank you to contributors across SIG Instrumentation and component owners who collaborated on the design, implementation, testing, and review of native histograms in Kubernetes!
I spent $220 on Google app ads and 60% of the installs were robots
Article URL: https://dayzlegame.com/blog/google-ads-bot-farm/ Comments URL: https://news.ycombinator.com/item?id=49662990 Points: 423 # Comments: 218
Copycat Tech: Why the AI Bubble Might Burst - Center for European Policy Analysis (CEPA)
Copycat Tech: Why the AI Bubble Might Burst Center for European Policy Analysis (CEPA)
At least 46,000 displaced by escalating Yemen fighting: UN
The International Organisation for Migration (IOM) warns that the figure is rising 'by the hour'.
Lockdown and its (Digital) Afterlives: Media, Infrastructure, and Everyday Life in Regenerative Perspective - Annenberg School for Communication
Lockdown and its (Digital) Afterlives: Media, Infrastructure, and Everyday Life in Regenerative Perspective Annenberg School for Communication
AI gives worse workplace advice to employees with less power, study finds | - contxto.com
AI gives worse workplace advice to employees with less power, study finds | contxto.com