Year: 2018

NPR, WNYC Studios, WBEZ Chicago and This American Life announced today that they’ve acquired Pocket Casts, a podcast app created by Australian developer Shifty Jelly.

That might sound like a lot of owners for one app, but the idea is to run Pocket Casts as a joint venture. And while former iHeartRadio executive Owen Grover is becoming CEO, NPR says the existing Pocket Casts team will remain in place, with founders Philip Simpson and Russell Ivanovic holding leadership roles in the company.

All four of the acquirers have released their own apps already, but buying Pocket Casts should give them another way to become more involved in distribution and reach listeners directly. (This seems to be a growing concern among all public radio organizations — in 2016, public radio marketplace PRX spun out a for-profit company called RadioPublic to focus on mobile apps.)

At the same time, NPR says Pocket Casts will continue to offer podcasts from a wide variety of producers.

Laura Walker, president and CEO of New York Public Radio (which includes WNYC Studios), said in the announcement:

Public radio has been at the vanguard of audio innovation and podcasting, bringing in new listeners, experimenting with new forms and topics, fostering engagement and community, and cultivating new talent in the industry. And yet despite this remarkable renaissance, the listening experience — particularly around discovery — has remained virtually unchanged. Pocket Casts will enable us to forge a closer relationship with our listeners, provide audiences with more ways to enjoy audio programming, and create a more tailored discovery experience that helps listeners find their next must-listen podcast.

Yes, it’s that time again — password changing time. On Thursday, Twitter revealed that a bug caused the platform to store user passwords in unmasked form. Normally, sensitive personal data like passwords would be stored in hashed form using a mix of letters and numbers to protect the content of the password itself. In this instance, it sounds like plain text passwords were stored openly without any hashing on Twitter’s internal logs.

Twitter notes that it currently has “no reason to believe password information ever left Twitter’s system” or that these unprotected passwords were accessed by hackers, but the risk of the unknown remains. The company has advised users to change their passwords as a precautionary measure.

Here’s what Twitter says happened:

“We mask passwords through a process called hashing using a function known as bcrypt, which replaces the actual password with a random set of numbers and letters that are stored in Twitter’s system. This allows our systems to validate your account credentials without revealing your password. This is an industry standard.

Due to a bug, passwords were written to an internal log before completing the hashing process. We found this error ourselves, removed the passwords, and are implementing plans to prevent this bug from happening again.”

We’ve reached out to Twitter for more details on the bug and additional information about how this could have happened.

It’s pretty unusual for a company of this size to make such a basic security mistake, but that’s just another reason for users to take password protection into their own hands on all services using protections like two-factor authentication and a password manager like LastPass or 1Password.

I/O doesn’t actually kick off until next Tuesday, but Google’s been steadily trickling out news this week. Perhaps it’s priming the pump for next week’s big event, or maybe the company just had more news than it could cram into a couple of keynotes. Whatever the case, today brought some new additions to the wearable operating system formerly known as Android Wear.

It’s hard to say how much energy Google is going to put into Wear OS in the coming week, but in the meantime, it’s getting some solid Assistant updates. The pairing of the two offerings is a no brainer, of course. Siri’s proven an important driver for Apple Watch, and a workaround for the whole small screen issue.

Google added its own Assistant to Android Wear last year, and is continuing to refine the experience with some key updates. At the top of the list is smart suggestions, which offers followup questions based on context. Ask Google what the weather is, and it will offer up followups for additional days’ forecasts.

Assistant’s wrist worn counterpart can also offer up voice answers through a pair of connected headhones. The update, which is rolling out over the next week or so will also bring Actions to the wearable OS, meaning much more more third-party control. Now users can, say, preheat an LG smart oven from the comfort of their own wrist.

There’s nothing really earth shattering here, but it’s nice to see Google continuing to give Wear OS a little love ahead of I/O. The operating system has failed to make much headway as the wearable category has seemingly plateaued for everyone who isn’t Apple at this point. Hopefully the company will have even more to show off on that front next week.

A report from the New York City Comptroller’s office asserts that New York residents are paying hundreds of millions in extra rent linked to the effects of Airbnb . Naturally, the company bitterly rejects these findings.

The report, which you can read here, is fairly straightforward. It looks at hundreds of neighborhoods and their various demographics and characteristics, along with how much their rents rose over the last ten years or so. It finds that when controlling for other variables, Airbnb contributes to a part of the rise on its own:

We find that as the share of units listed on Airbnb goes up by one percentage point, rental rates in the neighborhood go up by 1.58 percent, after controlling for neighborhood level demographic and economic changes. The result is statistically significant at the 1-percent level.

By the researchers’ calculations, the total cost of these increases across the city amounted to about $616 million. That came from running their numbers with Airbnb rentals set to zero instead of the actual tens of thousands of listings and seeing what rents would be in that alternative universe.

The amounts of rent increases and the number of Airbnb listings are tightly and reliably correlated, the Comptroller’s office explained. They were careful to control for other factors, for instance a neighborhood becoming trendy or new housing changing the supply. The hypothesis is that Airbnb listings, contrary to the company’s assertions, do in face reduce housing supply, which has a knock-on effect on rent in remaining rentals.

The increases, the report and its accompanying press release say, are concentrated in midtown and lower Manhattan, where 20 percent of the rent increases were attributed to Airbnb effects. The effect was much weaker in the outer boroughs, as you might expect, where density is lower and fewer listings are available.

Airbnb, of course, calls it a pack of lies and takes the opportunity to pontificate a bit (which, to be fair,  Comptroller Stringer did too):

Unfortunately, this report is wrong on the facts, falsely asserting that middle class New Yorkers who share their space are responsible for the rising cost of housing in New York… Pandering to the powerful by attacking middle class families won’t do a thing to make New York more affordable. It’s time to stop the scapegoating and work with us on a solution.

Its criticisms are a mix of reasonable objections and straw men. It rightly points out, for instance, that Airbnb hosts are most frequently just renting out a room a few nights a month for some extra income, which logically should improve affordability, not harm it. Then there’s the idea that Airbnb units tend to pop up in fashionable areas, which are inherently more likely to see rents increase.

Some of Airbnb’s gripes are less reasonable, however.

“The notion that less than 1 percent of housing stock — much of which is only occasionally shared with short-term renters — is the sole source of New York’s housing affordability challenge is simply not credible,” the company writes. That’s true — which is probably why the report say anything like that.

“We never blamed the whole increase in housing costs on one factor – we quite clearly said that Airbnb was just one factor and explained that it’s 9.2% of the increase,” wrote Tyrone Stevens, the Comptroller’s press secretary, in an email to TechCrunch.

Airbnb also criticizes using 2009 as the starting year for its data, saying the financial crisis changed the whole housing market. Then it cites a report showing that a one bedroom in Williamsburg cost the same in 2018 as it did in 2011.

“We picked 2009 as our base year because that was the year prior to Airbnb’s presence in NYC, and it made sense to measure the full impact of the rise of Airbnb. But the choice of base year is utterly irrelevant to measuring the contribution of Airbnb to rent increases,” Stevens wrote. “And randomly picking an industry report on 1-bedroom apartments in one neighborhood over a different time period doesn’t refute a citywide analysis.”

The fact that rents are leveling out lately doesn’t hold much water, either — arguably that might have occurred sooner but for Airbnb’s alleged contribution to their increases in the first place.

Ultimately the difference comes down to whether or not Airbnb effectively removes housing from the market. The company swears up and down that isn’t the case, and cites user numbers to support that position. The city says there’s little other possible explanation for the close correlation between listings in certain neighborhoods and the specific rent increases it sees in them.

Both, however, seem to agree that the lack of regulation puts everyone at risk. Hopefully that common ground will lead to a fruitful collaboration on new rules in the future.

Get ready to shop the ‘Gram. Instagram just stealthily added a native payments feature to its app for some users. It lets you register a debit or credit card as part of a profile, set up a security pin, then start buying things without ever leaving Instagram. Not having to leave for a separate website and enter payment information any time you want to purchase something could make Instagram a much bigger player in commerce.

TechCrunch reader Genady Okrain first tipped us off to the payment feature. When we asked Instagram, a spokesperson confirmed that native payments for booking appointments like at restaurants or salons is now live for a limited set of partners.

One of the first equipped is dinner reservation app Resy. Some of its clients’ Instagram Pages now offer this native payment for booking. And in the future, Instagram says you can expect direct payments for things like movie tickets through the app. Instagram initially announced in March 2017 that “we’ll roll out the ability to book a service with a business directly from their profile later this year,” but never mentioned native payments.

We’ve confirmed that the payment settings are now visible; some, but not all, users in the U.S. have it while at least some in the U.K. don’t. A tap through to the terms of service reveals that Instagram Payments are backed by Facebook’s Payments rules.

With its polished pictures and plethora of brands, shopping through Instagram could prove popular and give businesses a big new reason to advertise on the app. If they can get higher conversion rates because people don’t quit in the middle of checkout as the fill in their payment info, brands might prefer to push people to buy via Instagram.

Facebook started dabbling in native commerce around 2013, and eventually started rolling out peer-to-peer payments through Messenger. But native payment for shopping is still in closed beta in the chat app. It’s unclear if peer-to-peer payments might come to Instagram, but having a way to add a credit or debit card on file is a critical building block to that feature.

It’s possible that the payments option will work with Instagram’s “Shoppable Tags,” which first started testing in 2016 to let you see which products were in a post and tap through to buy them on the brand’s site. Since then, Instagram has partnered with storefront platforms BigCommerce and Shopify to get their clients hooked up, and expanded the feature to more countries in March. For now, though, none of Instagram’s previous shopping feature partners like Warby Parker or Kate Spade let you checkout within Instagram, and still send you to their site.

But the whole point of Instagram not allowing links in captions is to keep you in a smooth, uninterrupted browsing flow. Getting booted out to the web to buy something broke that. Instagram Payments could make impulse buys much quicker, enticing more businesses to get on board. Even if Instagram takes no cut of the revenue, brands are likely to boost ad spend to get their shoppable posts seen by more people if the native payments mean more of them actually complete a purchase.

Instagram isn’t the only one who sees this potential. Snapchat started testing its own native payments and checkout feature in February.

French cloud hosting company Scaleway announced new servers for its cloud offering. While the company’s offering was already quite cheap, Scaleway is going one step further with prices starting at €1.99 per month (around $2.40 per month).

For this price, you get an x86 server with 1 core, 1GB of RAM, 25GB of SSD, 100Mbit/s of bandwidth with unlimited transfer. For twice that price, you get twice the specs, etc. There are four tiers for this new Start1 cloud server lineup.

And for the first time, Scaleway is using NVMe for its SSD storage. NVMe is a storage protocol that is supposed to be much faster than SATA when it comes to handling read and write instructions on your storage drive. It was designed for SSDs, and that’s what the iMac Pro is using now. Scaleway is also using DDR4 to improve RAM performance.

Overall, Scaleway says that you should see a 50 percent performance increase per core. These new servers are available in the data center in Paris and soon in Amsterdam.

In addition to those new servers, Scaleway is launching hot snapshots in beta. With that feature, you can take a snapshot of your virtual machine while it’s running. You can then boot up another server with the exact same snapshot for instance. It opens up a lot of possibilities for automation and backup.

The company also recently promised a better ImageHub with more updates. For instance, the new Long Term Support version of Ubuntu that was released a few weeks ago is now available for most servers. Scaleway also supports Terraform to orchestrate your cloud infrastructure.

Existing x86 cloud servers are getting phased out. High-end x86 cloud servers and ARMv8 cloud servers are sticking around. Bare-metal ARMv7 and x86 servers are also still available.

Scaleway is the cloud hosting division of Iliad, one of the main internet service providers in France. The division recently got a new influx of cash to go to the next level. After those product updates, you can expect internationalization moves in the coming months.

One of the bigger pieces of product news at Facebook’s F8 developer conference this week was the announcement that Facebook will soon turn on a new dating feature. And it turns out that it quietly made significant hire to help build it.

Dr Steve Carter, a data scientist who helped design and build the psychometric and relationship models that became the basis of eHarmony — the dating site where he was a founding member and worked for nearly 20 years — is working at the social network, out of its offices in Los Angeles.

His LinkedIn profile notes that he joined Facebook in August 2017. Back in January, there was a brief mention of Carter doing work for Facebook, but at the time it was not known that he’d actually left eHarmony months before that. It seems that he’s only recently updated his profile with that information, coinciding with Facebook now finally making its dating intentions public. (We have reached out to Facebook and the company declined provide a comment for this story.)

Carter’s profile on LinkedIn describes him as a data science manager, “helping Facebook get even better at making meaningful and beneficial connections between people and communities.”

But his expertise lies squarely in building matchmaking algorithms. Before this, he held a number of roles at eHarmony, starting as a research analyst on the founding team in 1999, before eHarmony had even launched, working his way up to chief scientist at the company, which has had backing from the likes of Sequoia and TCV in its history.

In that time, he also picked up four patents in online social matchmaking, including one for employment matching. Those patents are all owned by eHarmony.

Facebook, which is still feeling the heat from controversies that have arisen around some of its more scaled-up services like programmatic advertising and viral political content, has shifted its focus to communities, and also exploring more direct and discreet ways that its social graph can be leveraged. Dating is one of the more obvious applications of that concept.

“One in three marriages in the US start online,” CEO Mark Zuckerberg said in his presentation F8 this week. “[But] today we haven’t even built any features to help people find partners,” he said. “There are 200 million people on Facebook who list themselves as single, so clearly there is something to do here. And if we’re focused on helping people build meaningful relationships then this is perhaps the most meaningful one of all.”

There hasn’t been too much written about how eHarmony’s algorithm works, but in a lecture in May 2017 (two months before officially leaving eHarmony), Carter opened up about it to describe it a little.

It turns out that while there may be a lot of questions that get asked of a user’s preferences, ultimately eHarmony threw out everything except for the most extreme likes or dislikes in order to avoid creating what Carter refers to as “a universe of one.” 

This is an interesting idea when you apply it to Facebook, and you can see what the attraction might be for a data scientist in wanting to use the platform to build a dating service. (That’s to say nothing of the fact that Facebook sort of started as a dating site of sorts, where people prominently displayed their relationship statuses on their profile pages, popularizing “it’s complicated” as the short story for all of love’s actual complications.)

As we noted yesterday, when Facebook launches its service, it wants to make it as noninvasive as possible.

You get the option to create a dating profile. And that profile won’t be visible to your friends; only to others who have also opted in, and only if they fit your criteria and match your profile, and only if they are not already your friends and connections on the network. Applying Carter’s concept of matching data between users on eHarmony, Facebook already has a huge trove of information about you, based on what you post and like and share. Facebook can therefore make its dating profile questions — and its need to rely on them — less heavy. (That format could also provide a good contrast to other sites like Tinder, which require you to swipe across many profiles to create shortlists of people who might match with you.)

In hiring Carter, Facebook has picked up someone who has worked on relationships beyond those you forge through dating.

Last year, his former employer eHarmony divested a business that Carter had developed called Elevated Careers, which was acquired by a startup called Candidate.Guru.

Elevated Careers sounds a lot like a dating service, except with ’employer’ and ’employee’ replacing the two people looking for love, which reminds you just how similar the mechanics can be for any matching service (something Facebook must also realise).

It lets an employer assess a potential employee’s compatibility by matching up attributes of the candidate’s personality and values to those of his/her potential manager, as well as to the hiring organization’s culture. It also has algorithms that tell you how well a candidate’s skills and background are suited to a particular job. Given Facebook’s interest and work in recruitment and job finding, it will be interesting to see how and if that product also gets developed.

Along with Steve Carter, we’ve been trying to figure out whether Facebook has been hiring other people to build a relationships team that it could task with building out dating and other products. It turns out that there is definitely some crossover of employees who have worked at various dating sites over the last several years, who are now at Facebook — but most likely, this is mostly a product of the general churn that you get in the tech world, where people are often jumping from one job to another.

But a few hires stand out. Just this month, Facebook happened to hire Badoo’s head of mobile engineering, who appears to be based out of London. This sales director for “e-commerce and dating” started his new role in January 2018, working out of Paris (so I guess there is already some early considering for how to build this as a business too). There are two recent hires from Tinder, one in Los Angeles and one in Menlo Park. And the former data lead from Hinge also joined in the last six months, also based in Los Angeles, and now working in “scaled marketing.”

Italian startup Freeda Media is raising a $10 million Series A round led by Alven Capital. U-Start and business angels are also participating in the round.

Freeda Media runs the popular Freeda Facebook page. With nearly 1.4 million likes, the page has an impressive reach. 24 million unique users see Freeda content every month in Italy, including 100 percent of millennial women who live in Italy. The company is also quite active on Instagram.

In other words, Freeda Media is a new media company that runs mostly on social media platforms. Many of the Facebook posts are short videos, social cuts with subtitles and quick interviews. Some posts link to Freeda’s own website for more traditional articles with text and photos. But there’s no front page per se.

With today’s funding round, the company plans to expand to other markets, starting with Spain. More interestingly, Freeda is still quite young as it took them 15 months to reach this level.

The company compares itself with other media brands for women, such as Elle, Teen Vogue, Vanity Fair, Cosmopolitan or Man Repeller. According to CrowdTangle, Freeda has a much higher engagement rate compared to its direct competitors.

Freeda Media’s business model is branded content and native advertising. But the startup is looking at other potential revenue streams. There’s clearly a risk of branded content overload for social-first media organizations. When you build a Facebook audience based on trust, your reputation can also quickly deteriorate and the algorithms can turn on you. So far, it doesn’t seem to be an issue.

Internationalization is going to be key to foster Freeda’s growth. If Freeda can replicate the same impressive numbers in multiple countries, the startup could end up convincing bigger advertisers and distributing more or less the same content in multiple countries.

NASA’s latest mission to Mars, InSight, is set to launch early Saturday morning in pursuit of a number of historic firsts in space travel and planetology. The lander’s instruments will probe the surface of the planet and monitor its seismic activity with unprecedented precision, while a pair of diminutive cubesats riding shotgun will test the viability of tiny spacecraft for interplanetary travel.

Saturday at 4:05 AM Pacific is the first launch opportunity, but if weather forbids it, they’ll just try again soon after — the chances of clouds sticking around all the way until June 8, when the launch window closes, are slim to none.

InSight isn’t just a pretty name they chose; it stands for Interior Exploration using Seismic Investigations, Geodesy and Heat Transport, at least after massaging the acronym a bit. Its array of instruments will teach us about the Martian interior, granting us insight (see what they did there?) into the past and present of Mars and the other rocky planets in the solar system, including Earth.

Bruce Banerdt, principal investigator for the mission at NASA’s Jet Propulsion Laboratory, has been pushing for this mission for more than two decades, after practically a lifetime working at the place.

“This is the only job I’ve ever had in my life other than working in the tire shop during the summertime,” he said in a recent NASA podcast. He’s worked on plenty of other missions, of course, but his dedication to this one has clearly paid off. It was actually originally scheduled to launch in 2016, but some trouble with an instrument meant they had to wait until the next launch window — now.

InSight is a lander in the style of Phoenix, about the size of a small car, and shot towards Mars faster than a speeding bullet. The launch is a first in itself: NASA has never launched an interplanetary mission from the West coast, but conditions aligned in this case, making California’s Vandenberg air base the best option. It doesn’t even require a gravity assist to get where it’s going.

“Instead of having to go to Florida and using the Earth’s rotation to help slingshot us into orbit… We can blast our way straight out,” Banerdt said in the same podcast. “Plus we get to launch in a way that is gonna be visible to maybe 10 million people in Southern California because this rocket’s gonna go right by LA, right by San Diego. And if people are willing to get up at four o’clock in the morning, they should see a pretty cool light show that day.”

The Atlas V will take it up to orbit and the Centaur will give it its push towards Mars, after which it will cruise for six months or so, arriving late in the Martian afternoon on November 26 (Earth calendar).

Its landing will be as exciting (and terrifying) as Phoenix’s and many others. When it hits the Martian atmosphere, InSight will be going more than 13,000 MPH. It’ll slow down first using the atmosphere itself, losing 90 percent of its velocity as friction against a new, reinforced heat shield. A parachute takes off another 90 percent, but it’ll still be going more than 100 MPH, which would make for an uncomfortable landing. So a couple thousand feet up it will transition to landing jets that will let it touch down at a stately 5.4 MPH at the desired location and orientation.

After the dust has settled (literally) and the lander has confirmed everything is in working order, it will deploy its circular, fanlike solar arrays and get to work.

Robot arms and self-hammering robomoles

InSight’s mission is to get into the geology of Mars with more detail and depth than ever before. To that end it is packing gear for three major experiments.

SEIS is a collection of six seismic sensors (making the name a tidy bilingual, bidirectional pun) that will sit on the ground under what looks like a tiny Kingdome and monitor the slightest movement of the ground underneath. Tiny high-frequency vibrations or longer-period oscillations, they should all be detected.

“Seismology is the method that we’ve used to gain almost everything we know, all the basic information about the interior of the Earth, and we also used it back during the Apollo era to understand and to measure sort of the properties of the inside of the moon,” Banerdt said. “And so, we want to apply the same techniques but use the waves that are generated by Mars quakes, by meteorite impacts to probe deep into the interior of Mars all the way down to its core.”

The heat flow and physical properties probe is an interesting one. It will monitor the temperature of the planet below the surface continually for the duration of the mission — but in order to do so, of course, it has to dig its way down. For that purpose it’s installed with what the team calls a “self-hammering mechanical mole.” Pretty self-explanatory, right?

The “mole” is sort of like a hollow, inch-thick, 16-inch-long nail that will use a spring-loaded tungsten block inside itself to drive itself into the rock. It’s estimated that it will take somewhere between 5,000 and 20,000 strikes to get deep enough to escape the daily and seasonal temperature changes at the surface.

Lastly there’s the Rotation and Interior Structure Experiment, which actually doesn’t need a giant nail, a tiny Kingdome or anything like that. The experiment involves tracking the position of InSight with extreme precision as Mars rotates, using its radio connection with Earth. It can be located to within about four inches, which when you think about it is pretty unbelievable to begin with. The way that position varies may indicate a wobble in the planet’s rotation and consequently shed light on its internal composition. Combined with data from similar experiments in the ’70s and ’90s, it should let planetologists determine how molten the core is.

“In some ways, InSight is like a scientific time machine that will bring back information about the earliest stages of Mars’ formation 4.5 billion years ago,” said Banerdt in an earlier news release. “It will help us learn how rocky bodies form, including Earth, its moon, and even planets in other solar systems.”

In another space first, Insight has a robotic arm that will not just do things like grab rocks to look at, but will grab items from its own inventory and deploy them into its workspace. Its little fingers will grab handles on top of each deployable instrument and grab it just like a human might. Well, maybe a little differently, but the principle is the same. At nearly 8 feet long, it has a bit more reach than the average astronaut.

Cubes riding shotgun

Insight is definitely the main payload, but it’s not the only one. Launching on the same rocket are two CubeSats, known collectively as Mars Cube One, or MarCO. These “briefcase-size” guys will separate from the rocket around the same time as InSight, but take slightly different trajectories. They don’t have the control to adjust their motion and enter an orbit, so they’ll just zoom by Mars right as Insight is landing.

CubeSats launch all the time, though, right? Sure — into Earth orbit. This will be the first attempt to send CubeSats to another planet. If successful there’s no limit to what could be accomplished — assuming you don’t need to pack anything bigger than a breadbox.

The spacecraft aren’t carrying any super-important experiments; there are two in case one fails, and both are only equipped with UHF antennas to send and receive data, and a couple of low-resolution visible-light cameras. The experiment here is really the CubeSats themselves and this launch technique. If they make it to Mars, they might be able to help send InSight’s signal home, and if they keep operating beyond that, it’s just icing on the cake.

You can follow along with InSight’s launch here; there’s also the traditional anthropomorphized Twitter account. We’ll post a link to the live stream as soon as it goes up.

There’s little that’s more annoying on the web than videos that start playing automatically and with their volume up. Over the course of the last few years, Chrome and other browser vendors have started to combat this, but for the most part, those solutions relied on the user explicitly taking action. Now, following the launch of a similar feature on mobile, Chrome on desktop is also getting much smarter about which sites it’ll allow to autoplay and which it’ll block — and it’ll learn from your behavior to personalize this feature.

Google says a “significant number of autoplays” are paused and muted, or have their tabs closed, within six seconds. I’m actually guessing most people close those tabs faster, but six seconds seems like a good enough measure to know whether a user wanted to hear the sound from a video or not.

Going forward, Google will learn from your own browsing behavior and learn which sites you’ll want to mute. For users who aren’t logged in — or who are new to Chrome — the browser will automatically mute about 1,000 sites by default based on that six-second measure it’s using to detect annoying sites.

The company promises that this new system, once you’ve trained it, will block about half of unwanted autoplays. But since none of these systems are perfect, Google admits that it’ll occasionally get things wrong and that you’ll have to manually unmute some sites.

On mobile, Google uses a somewhat different system for allowing autoplays. Here, it’ll allow them when a site has been added to the home screen by a user. I doubt a lot of people do this, even with sites they regularly visit, so on mobile this pretty much amounts to a complete ban on autoplays.

It’s worth noting that this is all about audio. Videos that are muted when they autoplay will still be allowed to autoplay. Google also allows autoplays when a user has tapped or clicked somewhere on the site during a browsing session, which seems like a loophole that the company will hopefully close soon.