March 2017 Summaries

Sunil Vemuri, Product Manager for Actions on Google, announced new tools to enhance game development on the Actions on Google platform, particularly for the Google Assistant. These tools include a new sound effect library designed to make games more engaging and fun, as well as a guide for creating interactive text adventures, allowing developers to revive classic games for modern users. Furthermore, the platform has expanded its network of tool and consulting companies, now including partners like Pullstring, Converse.AI, Solstice, and XAPP Media, to help developers bring their gaming visions to life. This initiative aims to inspire developers to create both new and classic gaming experiences, encouraging participation in the Google+ community for further discussion and collaboration.

Google has introduced a new issue tracker for G Suite developers, aligning with the system used internally by the Google Cloud Platform team to enhance collaboration and transparency. This new platform allows developers to report bugs and request features for various G Suite APIs and developer tools, with each possessing a unique "component" number for easier searching. Existing issues were migrated from previous systems to this upgraded tracker, which requires users to log in with their Google credentials to access and update issues. The change aims to provide better issue visibility and facilitate closer cooperation between Google and developers.

Google has launched a new website, opensource.google.com, which consolidates its various open-source initiatives, showcasing both the vast array of projects it has released and providing insights into its open-source practices. This site reflects Google's long-standing commitment to open-source software, highlighting major contributions like TensorFlow and Kubernetes, as well as smaller, experimental projects. The launch also includes a directory of Google's open-source projects, with plans to expand it with additional information regarding project lifecycle and usage within Google. The site aims to share Google's approach to open source, emphasizing community involvement and the unique challenges faced by large corporations in this space. Additionally, Google is sharing its internal open-source documentation to offer transparency into its processes, while acknowledging that its methods may not suit everyone, but could provide valuable insights to others interested in open-source practices.

Google announced a live stream event on May 23rd, 2017, where Sridhar Ramaswamy, Senior Vice President of Ads and Commerce, would reveal the latest innovations in Ads, Analytics, and DoubleClick. Scheduled to last one hour, the keynote aimed to showcase new tools designed to enhance consumer reach, simplify campaign measurement, and boost productivity. Additionally, the presentation promised a preview of how brands are integrating Google Assistant to enhance customer experiences. Interested viewers were encouraged to register for the event and stay updated through social media platforms like Twitter, Google+, Facebook, and LinkedIn for upcoming previews.

XLA (Accelerated Linear Algebra) is a compiler developed by Google to enhance the performance of TensorFlow by optimizing data flow graphs through JIT compilation techniques. It analyzes TensorFlow graphs at runtime, specializes them for actual dimensions and types, and fuses multiple operations to produce efficient native machine code for devices like CPUs, GPUs, and custom accelerators such as Google's TPU. XLA retains TensorFlow's flexibility while addressing performance concerns by minimizing kernel launches and memory allocations through efficient operation fusion. It also offers significant executable size reductions for models in restricted-memory environments, such as mobile devices, by leveraging ahead-of-time compilation (AOT) with tools like tfcompile. Although still in its early stages, XLA supports the addition of new device backends with less implementation effort compared to re-implementing all TensorFlow operations, thus paving the way for TensorFlow to run efficiently on a broader range of hardware.

Python Fire is an open-source library that simplifies the creation of command line interfaces (CLIs) from Python code by using the Fire function, which automatically transforms Python objects such as classes, functions, or modules into CLIs without requiring users to define arguments or write main functions. Available on PyPI and GitHub, Python Fire allows developers to effortlessly turn their Python programs into command line utilities with features like tab completion and dynamic documentation, ensuring the CLI remains up-to-date as the code evolves. This utility is particularly beneficial for developers experimenting with Python libraries, as it allows them to execute code from the command line without modifying the main method. Google engineers use Python Fire for various applications, including an image manipulation tool with the Python Imaging Library and an experiment management tool within Google Brain, demonstrating its ability to manage tasks efficiently from both Python and Bash. Additionally, Python Fire offers an interactive mode via the `--interactive` flag, launching an IPython REPL for enhanced interaction with CLI results and predefined variables, making it a versatile tool for diverse projects.

Mobile gaming is evolving into a significant business opportunity, with game developers projected to capture nearly half of the $189 billion market for in-app purchases and advertising by 2020. At the 2017 Game Developers Conference, Google announced several innovations aimed at enhancing monetization and user engagement for mobile games. Among these are new playable ad formats that allow users to try games directly from ads, increasing the likelihood of qualified installs and long-term engagement. Additionally, Google introduced an auto-flip feature for video ads, which optimizes horizontal videos for vertical viewing, resulting in higher click-through rates. The integration of Firebase Analytics offers developers deeper insights into player behavior and game performance, enhancing their ability to build sustainable gaming businesses. These innovations reflect the increasing sophistication of mobile advertising and analytics tools, supporting developers in reaching and engaging their audiences more effectively.

Google's Launchpad Accelerator program has expanded to include tech startups from Africa and Europe, alongside those from Asia and Latin America, offering an equity-free opportunity for growth-stage innovative companies. Applications are open until April 24, 2017, for startups based in specific countries, with the program set to begin on July 17, 2017, at the Google Developers Launchpad Space in San Francisco. Participants will receive two weeks of intensive, all-expense-paid training, including mentorship from over 20 Google teams and experts from leading tech companies and venture capitalists in Silicon Valley. The program focuses on startups that have a proven product-market fit and are addressing significant challenges in their local markets, with an emphasis on leadership and knowledge sharing to benefit their local ecosystems. Selected startups will benefit from credits for Google products, PR support, and continued collaboration with Google in their home countries over a six-month period.