Computers really can’t interpret emotions – so why do companies keep trying to insist they can? And should we regulate them? CC-licensed photo by hydra arts on Flickr.
You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.
A selection of 11 links for you. There they are! I’m @charlesarthur on Twitter. Observations and links welcome.
What really caused Facebook’s leak of 500 million users’ data? • WIRED
Lily Hay Newman:
One source of the confusion was that Facebook has had any number of breaches and exposures from which this data could have originated. Was it the 540 million records—including Facebook IDs, comments, likes, and reaction data—exposed by a third party and disclosed by the security firm UpGuard in April 2019? Or was it the 419 million Facebook user records, including hundreds of millions of phone numbers, names, and Facebook IDs, scraped from the social network by bad actors before a 2018 Facebook policy change, that were exposed publicly and reported by TechCrunch in September 2019? Did it have something to do with the Cambridge Analytica third-party data sharing scandal of 2018? Or was this somehow related to the massive 2018 Facebook data breach that compromised access tokens and virtually all personal data from about 30 million users?
In fact, the answer appears to be: none of the above. As Facebook eventually explained in background comments to WIRED and in its Tuesday blogpost, the recently public trove of 533 million records is an entirely different data set that attackers created by abusing a flaw in a Facebook address book contacts import feature. Facebook says it patched the vulnerability in August 2019, but it’s unclear how many times the bug was exploited before then. The information from more than 500 million Facebook users in more than 106 countries contains Facebook IDs, phone numbers, and other information about early Facebook users like Mark Zuckerburg and US secretary of Transportation Pete Buttigieg, as well as the European Union commissioner for data protection, Didier Reynders. Other victims include 61 people who list the “Federal Trade Commission” and 651 people who list “Attorney General” in their details on Facebook.
How surprising that Facebook should misdirect people about whether a data breach was novel.
unique link to this extract
Spotify acquires sports-talk app Locker Room • WSJ
Spotify Technology is making its move into live audio by acquiring the sports-talk app Locker Room and its maker Betty Labs.
The deal values the company, initially backed by Lightspeed Venture Partners, and more recently by Google Ventures and Precursor Ventures, at around $50m, according to a person familiar with the transaction. If certain targets are met the value could climb closer to $80m, this person said.
Locker Room has quickly become the spot for fan chatter around games and sports news, with the likes of Miami Heat forward Andre Iguodala and Philadelphia 76ers guard Seth Curry to podcaster Ant Wright and ESPN’s Jeff Darlington dropping in for conversations as well. It filled a real-time, interactive void for sports fans left by the inability to gather in arenas, stadiums and bars during the Covid-19 lockdowns.
The purchase follows an explosion in demand for live audio apps amid the pandemic. Voice-based social networks, such as Clubhouse, Twitter Spaces, Water Cooler and Locker Room, allow users to converse spontaneously. They are an alternative to podcasts, but they are also a curated amalgamation of podcasts, live streams, conferences and radio. Comedians, artists and business leaders have flocked to these apps’ virtual rooms to perform, chat, debate and network across topics and industries.
For Spotify, which has expanded into podcasting to position itself as the world’s largest audio company—not just a music-streaming giant—the deal is a bet that live audio will last well beyond the pandemic.
This is a smart acquisition: as Ben Thompson and John Gruber have discussed on their Dithering podcast. Unlike Clubhouse, which is “any old audio”, Locker Room knows (or tells you) that you’ve come to listen to/talk about sports, and can funnel towards your interest right when you sign up. Clubhouse, on the other hand, is wildly (over?) valued and trying to cover the entire waterfront. The focus will pay off for Spotify. For contrast: Bloomberg says “Twitter held discussions for $4bn takeover of Clubhouse.” Guess which of these two deals could actually create value.
unique link to this extract
7% of Americans don’t use the internet. Who are they? • Pew Research Center
Andrew Perrin and Sara Atske:
Internet non-adoption is linked to a number of demographic variables, but is strongly connected to age – with older Americans continuing to be one of the least likely groups to use the internet. Today, 25% of adults ages 65 and older report never going online, compared with much smaller shares of adults under the age of 65.
Educational attainment and household income are also indicators of a person’s likelihood to be offline. Some 14% of adults with a high school education or less do not use the internet, but that share falls as the level of educational attainment increases. Adults living in households earning less than $30,000 a year are far more likely than those whose annual household income is $75,000 or more to report not using the internet (14% vs. 1%).
There are no statistically significant differences in non-internet use by gender, race and ethnicity, or community type.
Have to assume that as the cohort ages and dies that this statistic will simply cease to be the case – or will become more and more marginal. For those aged 18-29, 1% aren’t online; aged 30-49 it’s 2%; aged 50-64, it’s 4%. It’s a matter of time – the opposite of people learning to type, which used to be a rare skill that no self-respecting CEO would be seen dead doing.
unique link to this extract
Cycling is ten times more important than electric cars for reaching net-zero cities • The Conversation
Globally, only one in 50 new cars were fully electric in 2020, and one in 14 in the UK. Sounds impressive, but even if all new cars were electric now, it would still take 15-20 years to replace the world’s fossil fuel car fleet.
The emission savings from replacing all those internal combustion engines with zero-carbon alternatives will not feed in fast enough to make the necessary difference in the time we can spare: the next five years. Tackling the climate and air pollution crises requires curbing all motorised transport, particularly private cars, as quickly as possible. Focusing solely on electric vehicles is slowing down the race to zero emissions.
This is partly because electric cars aren’t truly zero-carbon – mining the raw materials for their batteries, manufacturing them and generating the electricity they run on produces emissions.
…In new research, colleagues and I reveal that people who walk or cycle have lower carbon footprints from daily travel, including in cities where lots of people are already doing this. Despite the fact that some walking and cycling happens on top of motorised journeys instead of replacing them, more people switching to active travel [cycling, e-biking and walking] would equate to lower carbon emissions from transport on a daily and trip-by-trip basis.
The handset industry is a flat circle • Digits to Dollars
Today, there are six phone makers left at scale: Apple, Samsung, Huawei, BBK, Xiaomi and Transsion. Apple remains unassailable with the best customers and the majority of industry profits. Samsung survives through scale and integration with other parts of the Samsung chaebol. Xiaomi has built a loyal following through some very solid marketing. Huawei had pulled far ahead, but its future now is not bright. Transsion is mostly a feature phone business, with solid inroads in Africa and now India. And then there’s BBK Group.
As the LG news [that it’s shutting its mobile phone division] broke, we were struck by how many people think the handset market is still fragmented among a dozen vendors. Most people who say that do not realize that a third of the top brands on the market today are owned by a single company, namely BBK. They own Vivo, Oppo, RealMe and OnePlus as well as a few other brands. Depending on who’s counting BBK is now the second or third largest handset vendor on the market. There are a couple of other brands still out there – notably the legacy business of the one-time leaders – HMD (the brand owner of Nokia), Sony (Ericsson) and Lenovo (Motorola), but their collective share is small.
What really strikes us from this list is that we are almost back to the point where we started [over 20 years ago]. Take Huawei off the list because their status is so unclear, and take off Transsion because their smartphone share is tiny – and we are left with four companies.
White House rejects COVID-19 vaccine passports • Poynter
White House press secretary Jen Psaki said Tuesday that the Biden administration does not support asking Americans to carry COVID-19 vaccine passports to prove they have been vaccinated.
“Let me be very clear on this. I know there’s been lots of questions,” Psaki said. “The government is not now, nor will we be, supporting a system that requires Americans to carry a credential.”
Some sort of passport might make it easier to travel internationally or enter sports venues or concert halls. But opponents have raised privacy concerns and questioned whether it would penalize people with underlying health issues who cannot take the vaccines. But, Psaki said, “There will be no federal vaccinations database and no federal mandate requiring everyone to obtain a single vaccination credential.”
That doesn’t, however, mean that individual states can’t do this – and New York looks pretty enthusiastic about the idea. The discussion about “Covid certificates” (passports are for passing through borders, people) is fascinating: there’s almost a horseshoe effect, where the two ends of the political spectrum bend around and meet in agreement against them, for entirely different reasons, while the less extreme (more centrist) ones generally like the idea.
unique link to this extract
Apple opens up its Find My network to third-party devices—no AirTag needed • MacWorld
While we’ve been waiting for the long-rumored AirTags to finally make an appearance, it appears that Apple might have pulled a head fake. Instead of a keychain that you can attach to things, Apple is partnering with third-party device manufacturers to use the Find My app to track down things they have lost.
Much like Apple’s own devices, third-party manufacturers who sign up for the Works with Apple Find My program will be able to tap into the Find My network to see where lost items are on a map, even if they can’t or don’t connect to the internet. The vast Find My network uses end-to-end encryption to crowdsource data from the hundreds of millions of Apple devices around the world to help locate missing items.
Additionally, Works with Apple Find My devices will be able to take advantage of ultra-wideband technology in the iPhone 11, iPhone 12, and Apple Watch Series 6 to track products with greater precision. It’s not clear whether Apple will allow devices to be powered down or remotely wiped as you can with Apple devices.
People have been expecting Apple to release “AirTags” – something you’d stick to an item which would let you track it down – since summer 2019, and in the meantime the antitrust noise, including from Tile (which makes a Bluetooth tag), has ramped up. Letting third-party companies make the hardware is a great way not to have to bear the costs (and potential losses), while reaping the benefits of tying people and third-party companies into your ecosystem to detect the things.
unique link to this extract
New AI technique transforms any image into the style of famous artists • The Next Web
The system morphs an input image towards the suggestion of a text prompt, such as “Salvador Dalí Art.” Over repeated mutations and iterations of each frame, the AI gradually finds features and shapes that match the text description until it produces a final composition.
“The results were like nothing I’ve ever seen as a computer artist for over 30 years,” [computer artist Glenn] Marshall told TNW. “By using any image and any text, the combination of endless possibilities is mind-shattering. And no one else is doing anything like this.”
Each piece was generated with a modified version of the Aleph-Image notebook, which is itself powered by OpenAI’s DALL-E and CLIP models.
Marshall named the technique Chimera, after the mythical beast formed from various animal parts, which has become a byword for something that exists only in the imagination and isn’t possible in reality.
Marshall says the technique is closer to “style distortion” than style transfer. But would the artists he’s distorting appreciate his creations?
AI cannot detect our emotions • OneZero
Evan Selinger talks to Prof Luke Stark, assistant professor in information and media studies at the University of Western Ontario:
Stark: Emotions are simultaneously made up of physiological, mental, psychological, cultural, and individually subjective phenomenological components. No single measurable element of an emotional response is ever going to tell you the whole story. Philosopher Jesse Prinz calls this “the problem of parts.”
To a large degree, then, our emotional responses are inherently interpersonal. By definition, no third party, whether it’s a social media platform or education-technology software, can know for certain how you feel when you’re expressing an emotion. Humans have developed all sorts of culturally specific social conventions to make interpersonal emotional expression more predictable. But several millennia of art and literature make it clear we can’t, as they say, know what’s in someone else’s heart. Some find that fact actively comforting. Others evidently find it frustrating.
Salinger: That’s a fascinating comparison. In everyday life, miscommunication can be vexing, exasperating, and sometimes have deadly consequences. But when given literary expression, the same situations, which we can observe from a somewhat comfortable distance, become dramatically compelling. Misunderstanding is the basis of fascinating plot shifts and nuanced character studies. On the lighter side, it also drives lots of comedy.
Stark: Right, and the resolution of misunderstandings, or reflection on why those resolutions didn’t or couldn’t take place, drives catharsis — releasing and thus getting relief from strong or repressed emotions. I did quite a bit of theater in college. In later chatting about it with a well-known physical computing practitioner, I asked them to observe that theater direction and interaction design are very similar processes. Material media modulate the social expression of emotion, much like dramatic conventions in the theater, which long predate digital technologies.
Time to regulate AI that interprets human emotions • Nature
The polygraph is a useful parallel. This ‘lie detector’ test was invented in the 1920s and used by the FBI and US military for decades, with inconsistent results that harmed thousands of people until its use was largely prohibited by federal law. It wasn’t until 1998 that the US Supreme Court concluded that “there was simply no consensus that polygraph evidence is reliable”.
A formative figure behind the claim that there are universal facial expressions of emotion is the psychologist Paul Ekman. In the 1960s, he travelled the highlands of Papua New Guinea to test his controversial hypothesis that all humans exhibit a small number of ‘universal’ emotions that are innate, cross-cultural and consistent. Early on, anthropologist Margaret Mead disputed this idea, saying that it discounted context, culture and social factors.
But the six emotions Ekman described fit perfectly into the model of the emerging field of computer vision. As I write in my 2021 book Atlas of AI, his theory was adopted because it fit what the tools could do. Six consistent emotions could be standardized and automated at scale — as long as the more complex issues were ignored. Ekman sold his system to the US Transportation Security Administration after the 11 September 2001 terrorist attacks, to assess which airline passengers were showing fear or stress, and so might be terrorists. It was strongly criticized for lacking credibility and for being racially biased. However, many of today’s tools, such as 4 Little Trees, are based on Ekman’s six-emotion categorization. (Ekman maintains that faces do convey universal emotions, but says he’s seen no evidence that automated technologies work.)
Ebang: yet another crypto “China Hustle” absconding with US investor cash • Hindenburg Research
• Ebang is a China-based crypto company that has raised ~$374m from U.S. investors in four offerings since going public in June 2020.
• While the company represented that it would use the majority of its numerous capital proceeds to develop its business operations, our research discovered it instead directed much of the cash out of the company through a series of opaque deals with insiders and questionable counterparties.
• For example, the company directed $103 million, representing ~$11m more than its entire IPO proceeds, into bond purchases linked to its U.S. underwriter, AMTD, which has a track record including (a) fraud and self-dealing allegations levied against it by one of the largest private equity firms in China and (b) listings that have subsequently imploded.
• Ebang claims to be a “leading bitcoin mining machine producer”, yet our research indicates this extraordinary claim is backed by no evidence. Ebang released its final miner in May 2019 and has since seen its sales dwindle to near-zero, delivering only 6,000 total miners in 1H20.
• With its mining machine business failing, Ebang pivoted the story to a cryptocurrency exchange launch called “Ebonex”. Announcements about the exchange added as much as $922 million market capitalization to Ebang.
• We found that Ebang’s exchange appears to be purchased from a white-label crypto exchange provider called Blue Helix that offers out-of-the-box exchanges for as little as no money up-front.
Totally normal, nothing to see here.
unique link to this extract
Errata, corrigenda and ai no corrida: none notified
I’m starting to wonder if the Facebook hacks are desperate Facebook users trying to find a human being to talk to over something Facebook screwed up (it’s a pet peeve as both Physics Today and Physics World woke up a few weeks ago and found <> next to their name in their facebook pages and absolutely no way to reach anyone at Facebook to ask “what the hell did you do?” I mean, we only have 3.7 million followers between us…
That’s “not applicable” which even WordPress knew to remove when I posted my comment 🤦♂️