A survey estimates that Apple Pay users reached 507M worldwide in September 2020, about 50% of the total iPhone userbase, adding ~66M users from September 2019 (Horace Dediu/Asymco)

IN a recent release, Apple reported that “more than 90% of stores in the US, 85% of stores in the UK, and 99% of stores in Australia accept Apple Pay.” This is encouraging but a very small view for the global Apple Pay picture. How can we assess where Apple Pay is and how do we even measure success? My expectation six years ago was that Apple Pay would be a “$1 billion business” by 2020. Now that 2020 has ended, how was my six year prediction?
I always recommend counting customers rather than (or before) dollars so let’s begin with that. According to one source Apple pay users worldwide reached 507 million by September 2020. This is about 50% of iPhone users and therefore a decent adoption rate (starting as it did six years ago.) The data shows a growth rate of about 66 million in 2020 but a deceleration from 150 million the previous year. This data is obtained through surveys as there is no reporting from any participants.
There is a smattering of other samples about transaction volumes, mostly behind pay walls, but the overall story of adoption is typical: an s-curve where we are either at the point of inflection or slightly after it. If we are half-way then perhaps saturation will occur in another six years.
More concerning is the distribution of these users. A lot of the data that is collected on mobile and online payments (and mobile wallets) is US only. Historically the US was a good place to measure technology adoption as the US was usually the leader. The automobile, PC and many other consumer technologies saw their market defined by the US.
But not so much anymore. The digital mobile phone broke the rule. Today mobile payments is breaking it again. Note that although the iPhone is over-represented in the US (as share of all phones) Apple Pay point-of-sale support is under-represented in the US.
According to one survey, in mid 2018, there were an estimated 38 million Apple Pay users within the United States, and 215 million outside (5.5 times more!) Outside the US mobile payments are nearly 100% accepted by default but in the US the support is patchy and characterized by active intervention to deny.
I’ve used Apple Pay everywhere in Europe without friction. From vending machines to highway tolls to drive-throughs, public transit and parking meters. I can recall one failure to do so in the last six months (a state-owned gas station).
The same cannot be said of the US. Where at WalMart and Home Depot there is the capability of support but an incomprehensible refusal to do so. One rumor about why Home Depot denies it is because they were hacked once and are paranoid about any technology. Walmart may be spiteful due to margin/cost, valuing a few basis points over customer satisfaction, convenience, hygiene and employee productivity. These holdouts are the laggards. Perhaps there is some pride in denying their customers a basic convenience but there certainly isn’t profit in it.
It gets even weirder when it comes to banks. In the US almost all banks immediately supported Apple Pay as an extension of their bank cards. In Europe the banks became the holdouts, rolling out support slowly and perversely.
So when you consider the dependencies of Apple Pay:
Sufficient number of users with iPhones/Watches (check)
Support from Banks (great in US, spotty outside)
Willingness of retailers to accept (sluggish in US, perfect outside)
Availability of points-of-sale equipment (universal by now
It’s an impressive achievement.
My expectation six years ago was that Apple Pay would be a “$1 billion business” by 2020. This was based on a take rate of 15 basis points ($15/$10,000 in transactional value). Juniper Research, which regularly examines payment transaction markets, now expects that Apple will see global Apple Pay transactions of $686 billion by 2024.
At that 15 basis point rate it amounts to $1.03 billion. Thus this particular research suggests that I was off by 4 years, making the $1 billion pay day a 10 year target rather than six.
But maybe that is no fault of Apple’s. The transaction volume is also equivalent to 52% of the proximity mobile payment market. Half the addressable payment market is a pretty good market share for a company holding 25% of the smartphone user base.

Study: smartwatches like Apple Watch and Fitbit, which can measure heart rate variability, could help detect COVID-19 at least a week before symptoms appear (Megan Cerullo/CBS News)

Smartwatches and other wearable devices that continuously measure users’ heart rates, skin temperature and other physiological markers can help spot coronavirus infections days before an individual is diagnosed.Devices like the Apple Watch, Garmin and Fitbit watches can predict whether an individual is positive for COVID-19 even before they are symptomatic or the virus is detectable by tests, according to studies from leading medical and academic institutions, including Mount Sinai Health System in New York and Stanford University in California. Experts say wearable technology could play a vital role in stemming the pandemic and other communicable diseases.
Subtle heartbeat changes
Researchers at Mount Sinai found that the Apple Watch can detect subtle changes in an individual’s heartbeat, which can signal that an individual has the coronavirus, up to seven days before they feel sick or infection is detected through testing. 

“Our goal was to use tools to identify infections at time of infection or before people knew they were sick,” said Rob Hirten, assistant professor of medicine at the Icahn School of Medicine at Mount Sinai in New York City and author of the Warrior Watch study. 
Specifically, the study analyzed a metric called heart rate variability — the variation in time between each heartbeat — which is also a measure of how well a person’s immune system is working. 
“We already knew that heart rate variability markers change as inflammation develops in the body, and Covid is an incredibly inflammatory event,” Hirten told CBS MoneyWatch. “It allows us to predict that people are infected before they know it.”
Individuals with COVID-19 experienced lower heart rate variability, or, in other words, little variation in time between heart beats, in contrast to COVID-negative individuals, the study found. 

High heart rate variability does not reflect an elevated heart rate: It indicates that an individual’s nervous system is active, adaptable and more resilient to stress. 
Investigators followed nearly 300 Mount Sinai health care workers who wore Apple Watches between April 29 and September 29. 

COVID cases surge in California 08:17
Apple did not participate in or fund the study, but is aware of its watches’ capabilities. Tim Cook in September touted the watch’s role in the Mount Sinai study.
The data collected by smartwatches could be instrumental in helping tame the virus, given that more than half of coronavirus cases are spread by asymptomatic carriers, according to a Centers for Disease Control and Prevention model published last week. 
“Right now, we rely on people saying they’re sick and not feeling well, but wearing an Apple Watch doesn’t require any active user input and can identify people who might be asymptomatic. It’s a way to better control infectious diseases,” Hirten said. 
Early warning 
A separate study from Stanford, in which participants wore a variety of different activity trackers from Garmin, Fitbit, Apple and other makers, found that 81% of coronavirus-positive participants experienced changes in their resting heart rates up to nine and a half days prior to the onset of symptoms. An extremely elevated heart rate was indicative of symptom onset, the study reported. 
Researchers used smartwatch data to identify nearly two-thirds of COVID-19 cases four to seven days before people showed symptoms, according to the study, published in Nature Biomedical Engineering in November. The study examined data from 32 people who tested positive for the virus from a pool of more than 5,000 participants. 

The team also has created an alarm system that alerts wearers that their heart rate has been elevated for a sustained period of time. 
“We set the alarm with a certain sensitivity so it will go off every two months or so,” said Stanford University Professor Michael Snyder, who led the study. “Regular fluctuations won’t trigger the alarm — only significant, sustained changes will.”
“It’s a big deal because it’s alerting people not to go out and meet people,” he added. When Snyder’s alarm recently went off, for instance, he cancelled an in-person meeting in case he might be infectious.

Should vaccines be required for employees? 04:43
Such technology could also help compensate for some of the shortcomings with coronavirus testing, Snyder said. “The problem is you can’t do it on people all the time, whereas these devices measure you 24/7. The smartwatch gives you back the data right away, in real time, whereas if you’re lucky you’ll get your test back in a few days.”
Wearable device makers also are looking at how the technology could be used to combat the virus. Oura Health, which makes a smart ring that tracks health data, helped fund a University of California San Diego and University of California San Francisco study that found the device can detect subtle symptoms, like the early onset of fever, that may indicate COVID-19. 
Whoop, which makes a sleep-tracking device, partnered with Australia’s Central Queensland University to author a peer-reviewed paper indicating that its technology can help predict coronavirus infections based on deviations in users’ respiratory rates during nighttime sleep. Healthy individuals experienced little variability in their respiratory rates, while deviations suggested compromised respiratory tract health. 
“All of these studies are jibing in that markers of physiological function collected from devices allow us to identify these conditions and diseases in a non-invasive way,” Hirten said. “They all have limitations but they complement one another.”

How Wikipedia, where neutrality is prized above all, struggled to chronicle the Capitol insurrection in real time (Alex Pasternack/Fast Company)

‘Storming’? ‘Insurrection’? ‘Riot’? ‘Attempted coup’? On Wikipedia, where neutrality is prized above all, volunteers are still searching for the words.

[Source photo: Alex Pasternack; Nohat/Wikimedia Commons]

By Alex Pasternacklong Read

advertisement

advertisement

advertisement

On the afternoon of January 6, as a giant crowd began to swarm the U.S. Capitol, Jason Moore, a 36-year-old digital strategist, was at home in Portland, Oregon, switching between CNN and MSNBC. “I try not to get caught up in the sensationalism of cable news,” he says, but admits he had to watch. Soon, concern became shock. “I could not believe what I was witnessing, and also knew history was being made.”

advertisement

advertisement

So he got to work. Moore is a veteran editor on Wikipedia, spending hours a day creating, shepherding, and policing articles. He started in 2007, ranging across topics of personal interest like music or architecture, but since early last year he’s been focused on the pandemic and political protests. Just after 1:30 p.m. EST, as rioters and police clashed at the bottom of the Capitol steps, he wrote, “On January 6, 2021, thousands of Donald Trump supporters gathered in Washington, D.C., to reject results of the November 2020 presidential election.” He appended links to a couple of sources deemed “reliable” by the community—NPR and The Washington Post—clicked save, and notified some other editors about his article. It was tentatively titled “January 2021 Donald Trump Rally.”
Was this really worthy of its own article, they asked? At that moment, protesters—rioters—were battling with police, both sides spraying chemicals. It was “hard to tell notability in the moment,” Moore wrote under his username, Another Believer. “But what we’re witnessing is unprecedented (like so many things lately).”
While riotous, misinformation-fueled mobs were breaking into the building—forcing lawmakers to evacuate, halting the counting of the Electoral College votes for several hours, and leaving several people dead—another kind of crowd began gathering to build upon Moore’s first sentence. After a brief trickle, Wikipedia veterans and newcomers quickly piled in, scrambling to add details, citations, and photos. On a popular Facebook group for editors, someone posted a warning to Wikipedians in D.C. who had gone to the scene to take photos: “Please please please be safe! Your life is more important than getting the perfect media for Commons.”
One admin soon changed the title from “Rally” to “Protest.” Another placed edit protections on the page to foil vandals. Debates erupted on the article’s Talk page, its public discussion room, as editors wrestled with many of the same hard questions breaking out in newsroom Slack channels across the country. This is no longer just a protest, but what is it?
As facts came in, as editors double-checked and pruned according to Wikipedia standards, the text grew and shrank and grew again, so that only the most relevant verifiable and neutral language remained.  “Once other editors showed up to contribute, I aided, facilitated, and watched eagerly as the article developed,” says Moore.
At the peak of editing, there was a change being saved every 10 seconds, estimates Molly White, a software developer and longtime Wikipedia editor who began working on the article in its earliest minutes. From her desk in Cambridge, Mass., she’s been editing the page for hours every day since. “It was one of those things where I was shocked and horrified at the news as it was unfolding,” she says, “and felt like helping with the article was a more productive way to process everything than just doomscrolling.”

advertisement

About 24 hours after the attack began, she and Moore and 406 other volunteers had crafted a detailed, even-keeled account of an event as it was unfolding—5,000 words long, with 305 references. Those numbers have since mushroomed, along with page views: 1.8 million and counting.
And that was only the English version: By Thursday morning, there were already articles in more than 40 different languages, including Esperanto.

People began looking for images to illustrate the article. The first, a photo of pro-Trump protesters outside of Union Station that was taken just before noon EST, was added to the article at 6:36 EST. https://t.co/CHA6RN0IJL pic.twitter.com/qB1X8CtiAo
— Molly White (@molly0xFFF) January 10, 2021

There’s an old joke about Wikipedia’s crowdsourced competence: Good thing it works in practice, because it sure doesn’t work in theory. “It’s particularly true,” White says, “when it comes to hundreds of people all trying to write about a current event in real time, as sources publish conflicting and sometimes inaccurate information.”
Still, the article—now stretching to more than 15,000 words, or 90 printed pages—is far from perfect. It’s the product of an editing community that tends to skew largely Western, white and male, with all of its biases and blind spots. Reckoning with those issues and testing each sentence for verifiability and neutrality can spark heated, incessant debate. And from the article’s first hours, nothing has been more divisive than the title itself.
The debate over a nameAs police were finally pushing rioters out of the Capitol, a majority of editors agreed that the second title, “2021 Capitol Hill Protests,” had to be changed. But was this a riot, an attack, a siege, a self-coup, an insurrection? “The lack of organization seems to have similarities with the Beer Hall Putsch,” one editor wrote in the hours after the attack. Someone else insisted on “2021 United States coup d’état attempt,” and a few others agreed.

advertisement

A few editors quoted from Wikipedia policy, WP:TITLE, which says articles should be named based on Recognizability, Naturalness, Precision, Conciseness and Consistency. Others pointed to a Wikipedia essay, “WP:COUP,” which explicitly says that the word should be avoided in a title “unless the term is widely used by reliable sources.” That evening, an editor named Spengouli noted, the Associated Press was advising journalists to “not refer to the events as a coup, as they do not see the objectives of the invasion as being overthrowing the government.”
Another editor chimed in with some alternatives: “the New York Times [is] using the words “riot” and “breach” as well as “storm”; CNN is using “riot” and “domestic terror attack”; Fox is calling it “Capitol riots.” (Fox News, Wikipedia’s current policy advises, “is generally reliable for news coverage on topics other than politics and science.”)
In the early hours of Thursday, as Senators reconvened to certify the election, a growing crowd on Wikipedia was pushing for insurrection. Even Senate Majority Leader Mitch McConnell had called it a “failed insurrection” on the floor of the Senate, someone said; soon, others pointed out, NPR and PBS were readily using the term too.
Still, others insisted that per Wikipedia guidance, insurrection is a legal term and should be used only after a ruling by a court or by a successful impeachment vote by the U.S. Senate. As EDG 543, a Chicago-based editor, wrote on Wednesday evening, “Biden, Romney, and a CNN opinion piece calling it an insurrection does not make it factual.” Someone argued the event didn’t meet the definition of insurrection in the Wiktionary, Wikipedia’s sister dictionary: “A violent uprising of part or all of a national population against the government or other authority.”
Except, as more details emerged, others said, it pretty much did meet that definition.

Trying to define exactly what something like this is as it’s happening is probably beyond us.”

Julle
“Trying to define exactly what something like this is as it’s happening is probably beyond us,” Johan Jönsson, who goes by the handle Julle, wrote on Wednesday evening.

advertisement

Frustration stretched the Talk page longer and longer. “Open your eyes!” one anonymous editor said. “This is an armed white supremacist insurrection by a mob intent on overthrowing the incoming democratically elected government and installing God-Emperor Trump as dictator for life, motherfuckers! Why some of you want this to be titled ‘rally,’ ‘protest,’ or ‘peaceful gathering of friends’ is beyond me.”
“Let’s take a deep breath,” wrote DenverCoder9 on Wednesday evening. “The best articles are written with a cool head and we should aspire to that standard.”
History’s crowdsourced front pageWikipedia isn’t supposed to be a source for breaking news—Wikipedians explicitly say that the site is “not a newspaper.” Another oft-cited community guideline, WP:WINARS, insists, “Wikipedia is not a reliable source.”
“Wikipedia is a work in progress,” says Katherine Maher, CEO of the Wikimedia Foundation, the San Francisco-based nonprofit that operates Wikipedia. “And we always say it’s a perfect place to begin learning, but you definitely shouldn’t stop there.”
But many of us do: Wikipedia is now considered reliable enough to serve as something like a central clearinghouse for facts online. Google depends on it to build its knowledge graph, while Facebook and YouTube use it to provide users with contextual information around false content.

Wikipedia is now considered reliable enough to serve as something like a central clearinghouse for facts online.

In fact, Wikipedia began honing its ability to quickly make sense of things during its earliest days, in the aftermath of another shocking event. The website was born 20 years ago this month, a spin-off of a project by two entrepreneurs, Jimmy Wales and Larry Sanger. Nine months later, a group of terrorists crashed passenger jets into the World Trade Center. Someone started a Wikipedia article, and a fledgling, pseudonymous self-built community of editors flooded in. The September 11 attacks were momentous for the site, helping establish and solidify some of its core standards, says Brian Keegan, an assistant professor of information science at the University of Colorado Boulder.

advertisement

Those standards include neutrality and verifiability but also those important rules about “what Wikipedia is not.” (“A Wikipedian’s primary role is as editor, not a compiler or archivist,” Animalparty reminded his colleagues on Monday night.) Twenty years later, says Keegan, coverage of breaking news topics like the coronavirus pandemic are still testing the Wikipedia community, and proving its surprising power.
“It seems even more contradictory when a bunch of volunteers, in the absence of any sort of centralized editing authority or sort of delegation or coordination, is still able to produce these especially high-quality articles,” he says.
When even neutrality can be politicalAs they watched tear gas wafting over the Capitol on TV, White and Moore jumped into ad hoc roles as quasi community organizers, shepherding conversations and handling a growing pile of edit conflicts and requests from users who didn’t have permission to edit the page directly. For sensitive pages like this one, admins can switch on additional safeguards that restrict editing to accounts that are more than 30 days old with more than 500 edits, requiring all other edits to be approved.
That didn’t stop the typical attempts at vandalism, falsehoods, and disinformation. “Mostly there are the anonymous ‘editors’ who vandalize or otherwise troll pages with high traffic,” says Moore, the sorts of bad edits he’d seen around COVID-19 and Black Lives Matter. “But also there are well-meaning people who are genuinely misinformed, and others who introduce bias, purposefully or unknowingly.”
Bad behavior doesn’t go far here. While social platforms like Facebook and Twitter have lately taken a harder approach to policy violations, for instance by banning Trump and others linked with the Capitol attack, Wikipedia has consistently been swift to close the accounts of bad actors. “There’s little appetite for feeding the trolls on the site,” says Moore. “There’s so much more important work to be done.”
On the article’s Talk page, editors shared news articles, aired concerns, and hashed out contentious edits, in theory according to the principles of “assume good faith” and “be polite.” On Wednesday, one visitor wrote a note of thanks. On Friday, someone who had attended the Trump rally beforehand sought to clarify the size of the crowd: “100s to less than 10,000” inside the Capitol, they wrote, and “easily tens to a hundred thousand” outside. By Sunday night, the discussion had flowered to more than 70 topics that ranged from formatting problems to questions about law, semantics, and philosophy. The crowd was processing this unthinkable event in open-source code.

advertisement

The crowd was processing this unthinkable event in open-source code.

With each discussion came more editorial guidance from the sticklers: The names of criminal suspects do not belong in the encyclopedia; only the names of rioters convicted of crimes may be included. George R.R. Martin, a Reddit post, and an on-the-scene Instagram video are not reliable sources; in any case, Wikipedia relies only on secondary sources. Use more neutral, clearer language in general: Words like mob and baseless carry a value judgment; better to stick with rioters and false.
Were the people inside the Capitol best characterized as a “mob” or “rioters”? Were some merely “protesters”? Some editors urged caution with “rioters,” on the grounds that not all participants were violent. “We used the same logic to not call the George Floyd protests the George Floyd riots, because violent rioters do not take away from what peaceful protesters do,” Alfred the Lesser wrote on Thursday morning.
“What a load of horseshit,” wrote SkepticalRaptor, a nine-year Wikipedia veteran, on Sunday. “‘Protestors’ is a weasel word that makes these treasonous insurrectionists appear to be roughly equivalent to BLM protestors (who actually protested). This story is about the attempted coup and the terrorist infiltration of the Capitol. They weren’t protestors, they were terrorists. I even think ‘rioters’ is weasel wording. This seems like whitewashing that we’d find in Conservapedia. Disgusting.”
The battle over what words to use brought into stark relief a central distinction on Wikipedia: between what’s accurate and what fits into an encyclopedia, between what’s “true” and what’s verifiable.
“Wikipedia is about neutrality, so it’s very hard when there’s no neutral word,” DenverCoder9 told me in an email, after they had been furiously editing for spans of hours. “You can see the ungodly amount of edits. I’ve been editing [on Wikipedia] for a while”—at least 20 months— “and I’ve seen nothing like it before.”
But tame neutrality— or the appearance of neutrality— can also be the product of bias or ideology: There may have been a protest, but describing the people raging in and around the Capitol as “protesters” downplays the violence and vileness, their confused and ugly intent. Call a spade a spade, someone said.

advertisement

The problem with ‘storming’At 3 a.m. on Thursday, after more than 200 editors had weighed in, an admin changed the name of the article to “2021 storming of the United States Capitol.” It was a stopgap measure, wrote CaptainEek, not a permanent solution. “We say what sources say, and for the moment they seem to say ‘storming,’” they wrote.
“Whitewashing,” said an editor named Albertaont. “This isn’t some romantic Storming of the Bastille.” Many agreed. On Thursday, Joanne Freeman, a professor of American history at Yale, shared her disapproval on Twitter: “It romanticizes it. There are plenty of other words: Attacked, Mobbed, Vandalized. Use those instead. Words matter.”

So one good idea would be never, ever to call the Sixth of January ‘the Storming of the Capitol.’”

Jill Lepore
By Friday, a few editors pointed out, insurrection was one of the most used terms among reliable sources. Soon, Democrats were distributing articles of impeachment based on a charge of “incitement of insurrection.” A conviction by the Senate could add more credibility to the label.
Anyway, wrote Chronodm, a California-based editor, storming had other problems: “Given Stormfront and The Daily Stormer, not to mention QAnon’s repeated use of ‘storm”,’ I really don’t think it’s a neutral choice.” Someone dropped in a link to a New Yorker essay by Jill Lepore, who was also shaken by the Nazi and QAnon links. “So one good idea,” wrote Lepore, “would be never, ever to call the Sixth of January ‘the Storming of the Capitol.’”
But Lepore doesn’t edit Wikipedia. Other editors insisted that “storming” was an accurate enough description, and that Wikipedia doesn’t bend to Nazis. “We really shouldn’t consider these fringe groups,” DenverCoder9 replied on Friday. “They produce so much nonsense you can find an association for every word, even ‘OK.’ Consider words as meant by the average reader.”
Of course, it’s not always clear how Wikipedia’s average readers interpret words, or even who those readers are. And just as new details emerge, the use and meaning of words change. The point is that words matter, and so the debates and the edits continue.

advertisement

Moore, the article’s first official author, expects the name to change again too, “as media outlets hone in on specific descriptions and words over time,” he says. He doesn’t have a strong opinion about it. “I am confident editors will determine the most appropriate name for the entry based on journalistic secondary coverage, as Wikipedia editors do.”

The Wikipedia article on the tragic 2021 Storming of the United States Capitol is 3 x longer and 3 x more referenced than the Wikipedia article on the triumph of the American Revolution. “Those who tell the stories rule society”-Plato. #Americans #January6th #Politics #BiasMuch pic.twitter.com/x9RR8IRVju
— Wade Burleson (@Wade_Burleson) January 12, 2021

There’s a lot of other work to do, says White: chronicling the injuries and deaths, the litigation, the reactions, the attempts to remove Trump. By Sunday, the article had reached 14,000 words, plus spin-offs, like a timeline of events and a compilation of international reactions. “And as time goes on we will also document if and how the incident has established a lasting place in history,” White says.
Read more: How Wikipedia’s volunteers became the web’s best weapon against misinformation
Like us, future historians will study the article to learn about what happened on January 6. And, as Slate‘s Stephen Harrison and others have previously pointed out, if they look at the behind-the-scenes debates over language, at these first (and second and third) drafts of history, they could also see how we processed the event in real time. The article’s Talk pages and edit histories could reveal things, says Keegan, “that are easily lost in historical accounts that pick up threads with the benefit of hindsight.”
What might those historians find? At a moment of information collapse and violent tribalism, many different people with good intentions could still agree on the tragic reality of what happened—whatever we end up calling it.

On its 20th anniversary, Wikipedians recount how it began and how it grew to become the definitive digital encyclopedia and the default resource for everything (Tom Roston/OneZero )

Tom Roston / OneZero :
On its 20th anniversary, Wikipedians recount how it began and how it grew to become the definitive digital encyclopedia and the default resource for everything  —  hard to imagine the internet without Wikipedia.  Just like the air we breathe, the definitive digital encyclopedia …