The Transform Technology Summits start October 13th with Low-Code/No Code: Enabling Enterprise Agility. Register now!
Google is working on a feature that would recommend articles directly in its browser. The suggestions would appear on the new tab page in Chrome for Windows, Mac, Linux, Android, and iOS.
Google confirmed it is still testing the feature, which is not yet available. “We’re always experimenting with new features in Chrome, but have nothing new to announce at this time,” a Google spokesperson told VentureBeat.
It’s not clear what the new functionality will be called, though multiple tickets on Chromium Code Reviews mention a “Morning Reads” service and a “ChromeReader” feature: 584347, 584278, 586452, 586131, 586129, 587857, 586126, 586121, 584266, 584303, 587822, 584620, 587819, 584346, 584301, 586134, 586132, 586128, and 586136 (via Myce). We went through many of these tickets to piece together a picture of what the feature would entail.
First of all, Google wants to recommend articles based on your most-visited sites, and then build personalization on top of that:
ChromeReader is currently queried using a hardcoded set of parameters that will return unpersonalized results. This needs to be updated to produce personalized results. We also need to check advanced options parameters we can tweak on ChromeReader to get even better results from the service.
At least for the MVP, we’ll provide the list of MostLikely/MostVisited URLs to ChromeReader. I don’t think there’s really room for any personalization on top of that.
Recommendations would naturally be more customized if you are logged into Chrome, but you would still get articles even if you’re not logged in. The feature would display snippets of articles (likely just a few words in addition to the headline), to tease the user:
The MorningReadsService will:
– Schedule the fetching of snippets
– Call the SnippetsService to fetch/update snippets when appropriate
– Vend the snippets to the UI that displays them
Google engineers are still brainstorming how frequently to check for new articles:
First iteration: Simply fetch in fixed intervals (if unmetered connection and (maybe) charging).
Some possible improvements, roughly from simple to advanced:
– Fetch only during certain times of day (e.g. between 6 AM and 9 AM, for *morning* reads).
– Vary the frequency based on charging state (e.g. every 30 mins while charging, every 2 hours otherwise, all only if on an unmetered connection).
– Possibly coalesce with other wake-up events (service worker background sync) – if Chrome is being started anyway, the extra cost is much lower.
– Have fancy prediction for when the user is likely to visit the NTP, and fetch an update just before that.
As you might expect, users would be able to dismiss articles, just as with any of Google’s recommendation services. There is also mention of a “preview view” and a “preview carousel.” Indeed, there is talk of how the feature would work with Google AMP, which launches next week:
For articles with an AMP version, we would need to prefetch that version and store it. We also need to indicate to the UI that it’s an AMP version so a different version of the preview can be shown
Based on what we’ve seen, it appears Google wants to bring something like the article recommendations in Google Now directly into its browser. While it’s definitely interesting to see what Google is playing with, this is still very early days. The feature isn’t even available in Chromium or Chrome Canary, and because development is still ongoing, it could be killed at any time.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
- networking features, and more