June 2017 Summaries

The text discusses the process of revamping the charts in a video analytics product, focusing on the choice of charting library. Initially, Chart.js was used but later switched to React-specific libraries due to their better compatibility with React components and more flexible design options. Three main libraries were tested: Victory, react-vis, and Recharts. Ultimately, Recharts was chosen as it met most of the requirements and had a strong community support. The author also mentions other honorable mentions like Chart.js, Highcharts, and D3 or SVGs directly.

Philip K. Dick's science fiction story "Do Androids Dream of Electric Sheep?" explored the difficulty in distinguishing living beings from machines, a problem that extends to online video consumption by web crawlers and scrapers. Crawlers are designed to follow links in web content and scan large sections of the internet, while scrapers download entire web pages and their resources. These tools can have a significant impact on video playback metrics, leading to false positives for error rates and other issues. To address this problem, Mux developed an anomaly detection system that automatically finds anomalies in error rates for individual video titles and across entire customer sites. By examining video-view attributes such as browser user-agent strings, IP address blacklists, and data center ranges, the company can identify crawlers and exclude them from their metrics. This helps improve the accuracy of performance metrics for video playback and ensures that genuine errors are not overlooked.

The text discusses the process of revamping the charts in a video analytics product. It highlights five specific problems that were identified with the old charts, including misleading aggregation and line charts, lack of context for data spikes, unhelpful visualizations for low-volume data, difficulty comparing dimensions, and excessive visual clutter from gridlines, axis labels, and tooltips. The text then outlines a solution involving modified line charts with filled areas, large dots for every data point, secondary bar charts to show volume impact over time, and improved responsiveness of charts. The new designs are also made color blind accessible.

Video advertising is a complex technology involving multiple systems and components working together to deliver ads to viewers. The process includes ad request, ad response, ad load, and ad playback stages. Various factors such as targeting rules, frequency capping rules, and value optimization can lead to multiple requests, waterfalls, and redirects, further complicating the process and causing delays in video playback. Mux, a company focused on improving viewer experience, has introduced new metrics under the Startup Time category to help publishers understand the impact of ads on their overall performance. These metrics include Video Startup Preroll Request Time, Video Startup Preroll Load Time, and Requests for First Preroll. By providing actionable information, Mux aims to empower publishers to make data-driven decisions on how they approach ads and improve viewer experience.

Wistia is a video hosting and analytics company that provides a highly customizable white label player optimized for viewer experience with support for high resolution videos and adaptive streaming using HLS. Founded in April 2006, the company uses Mux for A/B testing and validating new algorithms such as their latest improvements to their adaptive bit-rate algorithm. Wistia's customers use data from video viewings to gain insights into audience engagement and preferences. The company is currently focused on improving video quality through newer formats like h.265 or VP9, and building in more fallbacks for playback issues. Max Schnur, Principal Software Engineer at Wistia, suggests that Mux could improve its service by offering features such as an embed code to show individual customer's video performance and normalizing aggregate stats by video.