MIT May Have Just Given the Web a Huge Speed Boost

The web can be dog slow, but some researchers at MIT are now tackling the issue head on.

An open source tool may help tame the hobbled mess of autoplaying videos and advertising trackers that is the modern web.

Announced on Wednesday morning, Polaris is a software framework that attempts to speed up the web by addressing how web browsers and web servers interact. Developed by a team of researchers at MIT's Computer Science and Artificial Intelligence Lab, the framework may reduce the amount of time you wait for webpages by as much as 34 percent.

Normally, when you type something like CNN.com into your address bar, the browser will contact the server and then fetch objects like HTML, JavaScript, and images from the server. Where things can get bogged down is when these objects then trigger additional dependences—a snippet of JavaScript code, say, that's needed to then fetch additional images in a photo gallery.

Given that many of today's websites are absolutely loaded with all sorts of different elements, from banner ads to autoplaying video to social media buttons begging you to share or like a story, it's no wonder that websites can be dog slow to load, particularly on mobile devices.

"This really adds up for mobile page loads," research lead Ravi Netravali, who's a PhD candidate at MIT's Computer Science and Artificial Intelligence Lab, told me. "The way that we wanted to accelerate the page loads was to somehow either reduce the number of times we contact the server and/or reduce the delay that you actually incur. We decided to take the former approach."

Polaris works by mapping out all of these dependencies beforehand so the browser can make data requests more efficiently. Doing so helps ensure that the browser isn't haphazardly requesting data like images and JavaScript in an unoptimized sequence. "Our main goal," said Netravali, "was to say, How can we reduce the number of round-trips the browser has to make when fetching data?"

This is a fundamentally different approach than some mobile browsers like Opera take to speed up the mobile web, where the name of the game is data compression and shrinking the size of files before they reach your device. "The bandwidth for cellular networks is quite high," said Netravali, noting the widespread availability of 4G networks (with 5G in development as well). "Compression is useful when you have low bandwidth because you can only support sending so much data at a given time, but it's not going to reduce page load times when data size isn't the limiting factor to begin with."

So what's next for Polaris? The researchers plan to open-source the framework, opening up the possibility for developers to create browser extensions based on the technology. And asked if he'd welcome the idea of working with companies like Google to more directly integrate Polaris into their browsers, Netravali did not hide his interest.

"I would love such a discussion!" he said.