Google is one of the large technology companies is something we all have clear. What started as a simple project of two students at Stanford University has become a corporate giant with millions of dollars profit, and has led to several products beyond a simple web browser.
But what we did not have much information (at least until now) is how Google has worked in these products up to turn them into global success tools, used by millions of users worldwide. In the original website recently launched Google puts an end to the mystery explaining some curious stories about some of its most popular products.
The site in question is bar.foo, and it can find information about Google Docs, Voice Search, Gmail and YouTube. Each section dedicated to these products tells a centered on the figure of a history software engineer involved in the project concerned, gathering their ideas and opinions. It also explains some of the main obstacles that Google teams met in the development of these products, and how they managed to solve them.
So Google software is developed
The stories included in bar.foo give some idea of how it works in Google; they explain some curious anecdotes and also tell how they deal with developmental problems. In fact, the web itself is an invitation to join Google, if you convinces you read it and find an offer for you.
Logically it is a rather technical text, and it is in English. However, even if you do not understand programming or you are an engineer, it is curious to take a look at these stories “behind the scenes” and be aware of how much it costs to develop tools like Gmail, YouTube or Google Docs.
The real-time editing of Google Docs, a side project
In 2010, two years before the launch of Google Docs, Google engineers were already working in one of its most valuable features: collaborative editing between multiple users on the same document in real time. This feature began as a simple side project of John, star of this part of the story – one of those projects that Google engineers can spend one day a week on their own work hours.
To finish solving this, the engineers had to rewrite from scratch of the code core of Google’s servers. “There are no sacred cows in Google” explains John. “If the best solution to a problem requires server code is rewritten, or ride one from scratch, we will.”
The most fun comes when John account how was the test of their leaders with the first working prototype of this feature of Google Docs. “I remember the first time the director and head coach were testing it and editing the same document at the same time. They started making fun of each other in the document, deleting the changes that made each of them, writing insults and jokes.. . and there sweating thinking that something could be broken.”
Fortunately, not only it did not break, but now millions of people can easily collaborate on documents, presentations and spreadsheets from anywhere in the world.
Voice learning three teachers
When we ask something Google Now, probably to stop and think of all the technology behind it. In fact, as we read in bar.foo, the Google voice recognition is based on reality in three models: the sound, the pronunciation and language. Each of these models allows Google to analyze the wavelength of the sound received, calculate percentages of probability of phonemes included in the words, compare the sound received with huge databases of sequences of phonemes, and use the vocabulary and frequency words in each language to select the correct one . All this, in just seconds.
“When I joined Google in 2005, the voice recognition was only an experimental field,” says Françoise, protagonist of this section. “Think this was before smartphones, and the potential of speech recognition was still not so obvious.”
Françoise also tells us some of the main problems they have faced in the development and improvement of speech recognition, the system captured as background sound and become a meaningless word. For example, “kdkdkd” appeared often when someone used this service Google on a train – was the way Google played the sound of the car on the tracks. Only continuous improvement of recognition algorithms, models and learning pronunciation machine managed to solve these situations.
“But there are many more challenges in sight,” says Françoise. In addition, what if any. For example, the microphone on an Android smartwatch (or even an autonomous car) is not attached to the mouth, and it is easier to have background noise when in use. But this does nothing but feed his passion. “Speech recognition is a fantastic technology have been working on it for over 20 years and still amuses me every day. And of course, see how something you’ve worked from the beginning has evolved and has impacted the lives of millions of people … is a huge satisfaction.”
It seems only a window, but there is more
Although only see a blank page waiting to write an email in her window to compose new message Gmail has a huge design work, engineering and development, as Maria tells us, the protagonist of this section. His idea was to improve the productivity of Gmail, something he began to consider how he used most people email. “We saw that most fairly short messages sent, but also some who write long compositions rich format, and wanted to make Gmail easy and enjoyable for both types of user.”
Create a new composition window is not as easy as it seems, and soon encountered difficulties – especially those relating to make it work in different browsers with the same user experience. “We had to write a lot of code so that the text is maintained without jumps, uncut and formatting changes as was writing,” says Maria. Such simple details as the names of the recipients who become editable when clicked, or the appearance of a scroll bar came to give some headaches to the engineers.
Once solved all, it was to add features to this new mode of composition. “Most people think that this window is only a text box where you type your message, click send and ready. But there is much more” , recalls Maria. Yes, because a draft can be labeled or print, you can add emoticons or format the text, you can insert photos … and there are also a handful of features in Gmail Labs that give you even more potential. “We’re perfectionists, and we wanted an elegant and seamless experience result. It was worth the work.”
More videos, less time waiting
YouTube is one of the products that Google engineers, as Vivek have been working in search of continuous improvement – in this case, avoid buffering video that can reach both despair.
Vivek tells us in this part of the web its history trying to improve the quality of experience when someone is consuming videos on YouTube. His project started by a thorough analysis of all elements that may affect this experience, to give the so – called Quality of Experience System: a system that helped engineers Google to measure the effects on the user experience of any internal change. However, what is striking about the story is that this QoE System also served as a measure to analyze in detail what parts of the process of streaming video from the server to the user, had more impact on transmission speed – and the appearance too frequent buffering.
Thanks to the work of Vivek and his team in this system and the development of the necessary algorithms, finally able to find a solution that also allowed the ISPs outside Google find and fix potential bottlenecks in their facilities. In fact, in some countries like Australia, Canada, New Zealand and the United States can see how work certain Internet providers with YouTube .
Although it does not stop here much less. The steady growth of YouTube traffic undoubtedly brings new challenges to Vivek and his team. The first, apply the knowledge gained to also improve streaming video on mobile and adapt the use of its QoE System to emerging markets.
In short, a website for Google users more curious, they want to learn a little more about what’s behind the tools they use every day in your browser. For more visit http://daypowermedia.com/.