Photo: YouTube Screenshot / wesolveforx
Google has a research team of more than 400 scientists working on fundamental problems that will advance the fields of computer science and engineering.They aren’t building self-driving cars, or better smartphones. They are making us all better computer programmers and engineers.
Doing deep research is part of Google’s DNA. Its cofounders Larry Page and Sergey Brin met while at Stanford. They’ve always devoted resources to solving big problems.
Research area: Algorithms and Theory
Research papers published: 205
Here's a problem you might not think about: how long should your computer actually be powered on?
When you just have one computer, it isn't a big deal. But when you have tens of thousands of servers, that suddenly becomes a very relevant power problem.
That's just one problem that Google's Algorithms and Theory team is working on.
Google still has to operate as efficiently as possible, which means its servers need to do as little extra work as possible. It has an entire team dedicated to finding new ways to become more efficient at serving advertisements and keeping its power consumption down.
Research area: Artificial intelligence and machine learning
Research papers published: 232
Much of Google's services are actually programmed to learn on their own. They acquire tons of new information on a daily basis and process that to improve algorithms.
One of the best examples is its speech recognition and translation algorithms. The more data those services receive, the more accurate they become over time.
Still, Google has to develop algorithms that will help their services learn on their own. So Google has an entire team devoted to making those learning processes even more efficient.
Google has more data than has ever been produced in history — and it's researching how to manage it.
Research area: Data management
Research papers published: 60
Google has acquired and processed more data in the past year than has been produced throughout much of the history of the human race.
They have to be able to not only record and process it, but also access it and allocate the correct information depending on what its users are seeking.
That's a gigantic problem -- so Google hires a bunch of scientists to make those processes faster and more efficient.
Google is constantly developing new processes to find interesting information in all that data, too.
Research area: Data mining
Research papers published: 83
All that data that Google is collecting also has an incredible amount of value. You can figure a lot out about your users if you can process it.
Unfortunately, the problem is that you have to process it. Petabytes (that's about a thousand terabytes -- likely 2,000 times larger than your current computer's hard drive) at a time.
Luckily, Google has a crack team of scientists developing new ways to dive into that data and divine characteristics about its users and their behaviour, and teach its existing services (like language processing) how to be better.
That helps Google make its services even more efficient -- and ends up making them more money.
Google has thousands of servers — and it's building new ways to make sure they play nice with each other.
Research area: Distributed systems and parallel computing
Research papers published: 67
Google doesn't run on a single super computer.
Come on, that's just silly.
Instead, Google has thousands of servers that it spins up and shuts down on a regular basis. Getting a single process to run across multiple servers (which ends up making that process faster) is actually an incredibly complicated problem.
Google has to develop algorithms that call upon multiple servers at once to complete processes. So it has a bunch of scientists devoted to making that process more seamless.
Google processes millions of auctions a day. It's developing algorithms to process millions more, more quickly.
Research area: Economics and electronic commerce
Research papers published: 38
Google runs auctions with millions of advertisers and users on a regular basis.
Yes, you read that right: millions.
So the faster and more efficiently you can process those advertising auctions, the more money you can make from advertisers.
There's also the side benefit of actually deriving new computer science methods that speed up and improve the rest of Google's services.
Research area: Education innovation
Research papers published: 9
Google also employs a bunch of scientists that are trying to make people more aware of computer science. They do this by producing applications in educational context.
As in, Google is trying to produce more and better computer scientists and engineers.
Google naturally has something to gain here: the more computer scientists there are in the world, the more great Google employees there are.
Research area: General science
Research papers published: 42
Almost all of Google's projects end up impacting the advancement of mathematics and computer science in some way.
Google's internal projects are solving some of the most important and complicated problems that already exist.
Research area: Hardware and architecture
Research papers published: 25
It's easy to say that Google is now a hardware company after buying Motorola. But Google was already investing a lot of time and money in hardware research.
That's because Google deploys a huge number of servers, and they all have to operate at a level of efficiency that most other companies could only dream of.
When you're absorbing and processing as much data as Google, though, it becomes a necessity. Except now it has to do the same thing with phones and other gadgets, like the Google Glasses.
Research area: Human-computer interaction and visualisation
Research papers published: 126
Google has invested a ton of money and effort in figuring out the best way for you to interact with your computer.
That can be a design website, or a literally new way of interacting with a computer. Like interacting with a search engine with your voice, for example.
These are extremely hard problems to solve, but they also have applications in other areas -- like making computers more accessible for blind or deaf people.
Research area: Information retrieval and web
Research papers published: 117
Google's original mission was an attempt to organise the web into something searchable. That's still a big deal for Google, which has more or less pioneered search technology.
Google still employs a ton of scientists that are trying to find new ways to find what the user is looking for as quickly as possible. They just end up solving problems about relevance and ranking along the way.
Research area: Machine perception
Research papers published: 135
Most data in the world comes in the form of video, images and music -- not bits and numbers.
But Google is still able to index all of those in a relevant way for search results and access.
That's because Google employs a ton of scientists that find new ways for computers to understand visual and aural media. The better computers can understand a picture. the easier it is for the computer to deliver that to a user.
Research area: Machine translation
Research papers published: 36
Translation is rapidly finding ways to translate one language into another. This is a hard problem to solve because there are thousands of semantic rules that vary from language to language.
It's also very hard to divine the exact meaning of a phrase, because often times it has more than one meaning.
But it's a problem that Google wants to solve -- because it spends a lot of time making sure its translation applications are perfect. Luckily it has a bunch of scientists working on that problem.
Research area: Mobile systems
Research papers published: 13
Much like Bill Gates and Steve Jobs sought to do in the earliest days of computing, Larry Page and Google are committed to building an operating system.
Only, this operating system is powered by remote servers and wireless technology -- a 'cloud' server.
It was already a really hard problem to solve on just a simple piece of hardware, but once you start mucking around in the cloud you have to worry about a lot of things. Signal quality. Interference. Speed.
Google has only in recent years began exploring this heavily, and now it's deploying a fleet of scientists to solve some of the hardest problems to make it a reality.
Research area: Natural language processing
Research papers published: 154
Most mobile devices are moving in a direction in which the operating system is literally powered by your voice. This is done through a process called natural language processing.
It's an extremely hard problem. You have to use tone, speed, and audio quality to determine exactly what the user is trying to say. You have to turn a frequency into a meaning.
Mathematically, that's an incredible problem to solve. And now Google has a ton of scientists working on ways to adequately translate that sound wave into something that makes sense to users.
Research area: Networking
Research papers published: 57
The government is largely clueless about how the Internet works.
Wireless and broadband networks aren't like circuit-switched telephone networks. They are much, much more complicated.
That's why Google employs a bunch of scientists to find ways to measure and audit the Internet and present what is happening to the public in a way relevant to them.
It's to make sure the public stays informed and the government doesn't do something stupid, like shut it down.
Research area: Security, cryptography and piracy
Research papers published: 119
For as wonderful and incredibly innovative the Internet is, it's also really, really unsafe. It's really easy to crack databases and obtain unauthorised access to data.
That's a big deal for a company that is holding most of the world's data and powering a huge chunk of the email market.
So Google has to employ scientists to find ways to make sure all the data it holds stays safe.
Research area: Software engineering
Research papers published: 34
What do you do when the computer language you are using won't do what you need it to do?
You build your own.
That's what Google does over in Mountain View. It has languages like Dart and Go. But on top of designing its own languages, it also finds ways to speed up the infrastructure and ecosystems of existing computer languages like C++ and Python.
Research area: Software systems
Research papers published: 66
When it comes to the Internet, even a microsecond is too long.
So Google is constantly finding new ways to deploy hardware, fibre and connectivity between its servers that will reduce that load time more and more.
When you search for something, it already feels instantaneous. But that's still too slow for Google.
At least, when it comes to Google's scientists.
Research area: Speech processing
Research papers published: 31
Google's speech processing team has two goals: make speaking to your phones and computers ubiquitous, and make videos on the web accessible and searchable.
Those are both really hard problems to solve.
For example, you have to teach a computer what is going on with any sound file -- whether it's embedded in a video or coming from a mouth in front of a microphone.
Google has a bunch of scientists working on this problem. It appears to be slightly different from its natural language processing division -- but that's OK. The more brains working on complicated problems, the better.