Mozilla Foundation

ABSTRACT The 2020 Amazon Fire Season crisis can be summarized by record monthly deforestation numbers and by the negligence of the Brazilian government. Considering the political scenario in Brazil as a fertile ground for false information, we question which discourses on the Amazon forest fires were constructed by legacy media outlets and junk news sources during the 2020 fire season in Brazil. By means of a critical discourse analysis, we observe the ideological structure of the discourse created by both media types, comparing and synthesizing how the Amazon fires were portrayed throughout the fire season. Our findings demonstrate that neither legacy media nor junk news websites framed the Amazon fire season as a sustainability crisis. Instead, coverage overlooked the main goals of environmental journalism, focusing on the economic consequences of deforestation resulting from Brazil’s “bad image” internationally. Our results also indicate similarities and differences between legacy media and junk news discourses and suggest that Bolsonaro was able to instrumentalize the media ecosystem with conspiratorial denialism and populist discourse.

Millions of people worldwide use smartphones every day, but the standard issue QWERTY keyboard is poorly optimized for non-sighted input. In this article, we document the variety of methods blind people use to enter text into their smartphones, and focus on one particular need: sending text messages. We analyze two modern corpora of text messages and contrast them with an older text message corpus, as well as other corpora gathered from news articles, chat rooms, and books. We present a virtual keyboard for blind people optimized for sending text messages called Ally. To evaluate Ally, we conducted two user studies with blind participants. Our first study found increasing speeds and our second study found that half of participants reached comparable speeds to QWERTY, suggesting it may be a viable replacement. We conclude with a discussion of future work for non-sighted text-entry of text messages.

The advancement of text-to-speech (TTS) voices and a rise of commercial TTS platforms allow people to easily experience TTS voices across a variety of technologies, applications, and form factors. As such, we evaluated TTS voices for long-form content: not individual words or sentences, but voices that are pleasant to listen to for several minutes at a time. We introduce a method using a crowdsourcing platform and an online survey to evaluate voices based on listening experience, perception of clarity and quality, and comprehension. We evaluated 18 TTS voices, three human voices, and a text-only control condition. We found that TTS voices are close to rivaling human voices, yet no single voice outperforms the others across all evaluation dimensions. We conclude with considerations for selecting text-to-speech voices for long-form content.

Open access, open data, open source and other open scholarship practices are growing in popularity and necessity. However, widespread adoption of these practices has not yet been achieved. One reason is that researchers are uncertain about how sharing their work will affect their careers. We review literature demonstrating that open research is associated with increases in citations, media attention, potential collaborators, job opportunities and funding opportunities. These findings are evidence that open research practices bring significant benefits to researchers relative to more traditional closed practices.

Electric Agents is a multi-player game that uses augmented reality on mobile phones to facilitate in-room collaboration and promote joint media engagement around interactive educational television content in the home. Joint media engagement describes 'collaborative' consumption of media and has potential learning benefits. Our research examines whether technology can support joint media engagement, through the use of a series of prototypes that combine handheld augmented reality with web-based video. In pilot testing, sibling play patterns showed evidence of the kind of in-room collaboration and joint media engagement that previous research has shown to facilitate children in making sense of media content.

JavaScript performance is often bound by its dynamically typed nature. Compilers do not have access to static type information, making generation of efficient, type-specialized machine code difficult. We seek to solve this problem by inferring types. In this paper we present a hybrid type inference algorithm for JavaScript based on points-to analysis. Our algorithm is fast , in that it pays for itself in the optimizations it enables. Our algorithm is also precise , generating information that closely reflects the program's actual behavior even when analyzing polymorphic code, by augmenting static analysis with run-time type barriers. We showcase an implementation for Mozilla Firefox's JavaScript engine, demonstrating both performance gains and viability. Through integration with the just-in-time (JIT) compiler in Firefox, we have improved performance on major benchmarks and JavaScript-heavy websites by up to 50%. Inference-enabled compilation is the default compilation mode as of Firefox 9.

Gradual type systems offer a smooth continuum between static and dynamic typing by permitting the free mixture of typed and untyped code. The runtime systems for these languages, and other languages with hybrid type checking, typically enforce function types by dynamically generating function proxies. This approach can result in unbounded growth in the number of proxies, however, which drastically impacts space efficiency and destroys tail recursion. We present a semantics for gradual typing that is based on coercions instead of function proxies, and which combines adjacent coercions at runtime to limit their space consumption. We prove bounds on the space consumed by coercions as well as soundness of the type system, demonstrating that programmers can safely mix typing disciplines without incurring unreasonable overheads. Our approach also detects certain errors earlier than prior work.