If you’re wondering why Neil Young is threatening to remove his music from Spotify, a Radiohead song title offers a clue. CC-licensed photo by Otis Historical Archives National Museum of Health and Medicine 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 10 links for you. Listen closely. I’m @charlesarthur on Twitter. Observations and links welcome.
Neil Young posted a since-deleted letter to his management team and record label demanding that they remove his music from Spotify. “I am doing this because Spotify is spreading fake information about vaccines – potentially causing death to those who believe the disinformation being spread by them,” he wrote. “Please act on this immediately today and keep me informed of the time schedule.”
“I want you to let Spotify know immediately TODAY that I want all my music off their platform,” he continued. “They can have [Joe] Rogan or Young. Not both.” Young is referencing the steady stream of misinformation about vaccines that Joe Rogan has peddled on The Joe Rogan Experience. Last month, 270 doctors, physicians, and science educators signed an open letter asking Spotify to stop spreading Rogan’s baseless claims.
“With an estimated 11 million listeners per episode, JRE, which is hosted exclusively on Spotify, is the world’s largest podcast and has tremendous influence,” the letter reads. “Spotify has a responsibility to mitigate the spread of misinformation on its platform, though the company presently has no misinformation policy.”
Young removed most of his music from Spotify several years ago because he felt the sound quality on the service was too low, but he ultimately relented.
Young’s animus here is easily explained: he survived catching polio in 1952, aged seven, in the last big outbreak in Ontario. (Though his animus about sound quality is harder to explain, unless he was chained to a gramophone for a year afterwards.)
Shows though that Spotify can’t escape politics if it chooses to host podcasts. Rogan was always going to bring trouble in his wake.
unique link to this extract
Artificial intelligence is largely a numbers game. When deep neural networks, a form of AI that learns to discern patterns in data, began surpassing traditional algorithms 10 years ago, it was because we finally had enough data and processing power to make full use of them.
Today’s neural networks are even hungrier for data and power. Training them requires carefully tuning the values of millions or even billions of parameters that characterize these networks, representing the strengths of the connections between artificial neurons. The goal is to find nearly ideal values for them, a process known as optimization, but training the networks to reach this point isn’t easy. “Training could take days, weeks or even months,” said Petar Veličković, a staff research scientist at DeepMind in London.
That may soon change. Boris Knyazev of the University of Guelph in Ontario and his colleagues have designed and trained a “hypernetwork” — a kind of overlord of other neural networks — that could speed up the training process. Given a new, untrained deep neural network designed for some task, the hypernetwork predicts the parameters for the new network in fractions of a second, and in theory could make training unnecessary. Because the hypernetwork learns the extremely complex patterns in the designs of deep neural networks, the work may also have deeper theoretical implications.
For now, the hypernetwork performs surprisingly well in certain settings, but there’s still room for it to grow — which is only natural given the magnitude of the problem. If they can solve it, “this will be pretty impactful across the board for machine learning,” said Veličković.
Just putting this on our collective radars: AI building AI has that slightly concerning feel to it.
unique link to this extract
Fashion Nova will pay $4.2m as part of settlement of FTC allegations it blocked negative reviews of products • US Federal Trade Commission
Online fashion retailer Fashion Nova, LLC will be prohibited from suppressing customer reviews of its products and required to pay $4.2m to settle Federal Trade Commission allegations that the company blocked negative reviews of its products from being posted to its website.
The FTC alleged in a complaint that the California-based retailer, which primarily sells its “fast fashion” products online, misrepresented that the product reviews on its website reflected the views of all purchasers who submitted reviews, when in fact it suppressed reviews with ratings lower than four stars out of five. The case is the FTC’s first involving a company’s efforts to conceal negative customer reviews.
“Deceptive review practices cheat consumers, undercut honest businesses, and pollute online commerce,” said Samuel Levine, Director of the FTC’s Bureau of Consumer Protection. “Fashion Nova is being held accountable for these practices, and other firms should take note.”
According to the FTC’s complaint, Fashion Nova used a third-party online product review management interface to automatically post four- and five-star reviews to its website and hold lower-starred reviews for the company’s approval. But from late-2015 until November 2019, Fashion Nova never approved or posted the hundreds of thousands of lower-starred, more negative reviews. Suppressing a product’s negative reviews deprives consumers of potentially useful information and artificially inflates the product’s average star rating.
The FTC also announced that it is sending letters to 10 companies offering review management services, placing them on notice that avoiding the collection or publication of negative reviews violates the FTC Act. In addition, the FTC has released new guidance for online retailers and review platforms to educate them on the agency’s key principles for collecting and publishing customer reviews in ways that do not mislead consumers.
This is the second case the FTC has brought against Fashion Nova in recent years. In April 2020, the FTC announced that Fashion Nova agreed to pay $9.3m to settle allegations that the company failed to properly notify consumers and give them the chance to cancel their orders when it failed to ship merchandise in a timely manner, and that it illegally used gift cards to compensate consumers for unshipped merchandise instead of providing refunds.
This has stirred up certain sections of American Twitter, because it seems to amount to the government telling companies what they can and can’t publish, which moves into First Amendment territory. Maybe Fashion Nova should redefine itself as a social network, and claim Section 230 lets it pick and choose reviews?
unique link to this extract
In terms of legal demands from governments, in the six month period covered in this report, Twitter received 43,387 legal demands to remove content, specifying 196,878 accounts. This is the largest number of accounts ever subject to removal requests in a reporting period since releasing our first transparency report in 2012.
Of the total global volume of legal demands, 95% originated from only five countries (in decreasing order): Japan, Russia, Turkey, India, and South Korea. We withheld or required account holders to remove some or all of the reported content in response to 54% of these global legal demands.
…Twitter required account holders to remove 4.7m Tweets that violated the Twitter Rules. Of the Tweets removed, 68% received fewer than 100 impressions prior to removal, with an additional 24% receiving between 100 and 1,000 impressions. [So 10% got more than 1,000 impressions.] In total, impressions on these violative Tweets accounted for less than 0.1% of all impressions for all Tweets during that time period.
In these six months, Twitter permanently suspended 453,754 unique accounts for violations of our child sexual exploitation (CSE) policy — 89% of those accounts were proactively identified and removed by deploying a range of internal tools and/or by utilizing the industry hash sharing (e.g., PhotoDNA) prior to any reports filed via the designated CSE reporting channel.
In the first half of 2021, Twitter suspended 44,974 unique accounts for promotion of terrorism and violent organizations — 93% of those accounts were proactively identified and removed.
Doctors have discovered an “antibody signature” that can help identify patients most at risk of developing long Covid, a condition where debilitating symptoms of the disease can persist for many months.
Researchers at University hospital Zurich analysed blood from Covid patients and found that low levels of certain antibodies were more common in those who developed long Covid than in patients who swiftly recovered.
When combined with the patient’s age, details of their Covid symptoms and whether or not they had asthma, the antibody signature allowed doctors to predict whether people had a moderate, high or very high risk of developing long-term illness.
“Overall, we think that our findings and identification of an immunoglobulin signature will help early identification of patients that are at increased risk of developing long Covid, which in turn will facilitate research, understanding and ultimately targeted treatments for long Covid,” said Onur Boyman, a professor of immunology who led the research.
The team studied 175 people who tested positive for Covid and 40 healthy volunteers who acted as a control group. To see how their symptoms changed over time, doctors followed 134 of the Covid patients for up to a year after their initial infection.
Blood tests on the participants showed that those who developed long Covid – also known as post-acute Covid-19 syndrome (Pacs) – tended to have low levels of the antibodies IgM and IgG3. When Covid strikes, IgM ramps up rapidly, while IgG antibodies rise later and provide longer-term protection.
The breadth and depth of research that has been done into this disease, in just three years, is amazing really.
unique link to this extract
Kristian Kielhofner, pointing out that bitcoin is turning 13 this month, and comparing it to the launch of the World Wide Web in 1993:
Kids these days may not understand what the early years of the web/internet were like. You needed at least several thousand dollars (in 2021 money) of computer hardware that you had to order from a catalog and wait weeks for and/or be close enough to a store that sold computers. You needed a phone line (back when you paid for even “local” calls) and not to mention an ISP (my first ISP in 1995 was a long distance phone call)! If you don’t know what long distance is back in the day you had to pay per minute when you had to call outside of your local area. When you were on the internet your whole house couldn’t get phone calls unless you added another phone line at great expense.
Then once you got this far you had the fun of Hayes modem commands (potentially IRQ conflicts and jumpers), CHAT scripts, getting a TCP/IP stack to work, etc, etc. Yet, somehow, people saw the value and potential of the web and jumped through all of these ridiculous hoops. My first internet connection was basically 1.5KB/sec — all for very ROUGH web pages that in many cases took literally minutes to load (assuming your modem didn’t randomly drop the connection).
It didn’t take long for things to get better. Internet speeds got faster. More applications were developed. More users got online. All because the internet was delivering so much value and growing so rapidly people (companies) were digging ditches around the world for fiber and dispatching ships across the world’s oceans to lay submarine fiber — all to support the rapidly increasing functionality and use of the internet (web) and laying the groundwork (literally) for all things to come.
Just how much use? Let’s look at some web statistics from 2006, 13 years after the release of the WWW:
• Over 1 billion worldwide users
• 73% of the US Population.
By contrast, best estimates say that Coinbase has about 9 million daily active users, with 11% maret share, hence ~100 million daily active users of crypto exchanges. From a multi-billion addressable market.
unique link to this extract
Josh Zumbrun and Alex Leary:
U.S. manufacturers and other companies that use semiconductors are down to less than five days of inventory for key chips, the Commerce Department said Tuesday, citing the results of a new survey.
In 2019, companies typically maintained 40 days of inventory for key chips, according to the Commerce Department report. Now for the same chips—defined as 160 products that companies identified as being the most challenging to acquire—companies are operating with fewer than five days of inventory, the report said.
Commerce Secretary Gina Raimondo said the survey results show the urgency for Congress to approve the U.S. Innovation and Competition Act, which includes $52bn to boost domestic chip production.
“We aren’t even close to being out of the woods as it relates to the supply problems with semiconductors,” Ms. Raimondo told reporters Tuesday. “The semiconductor supply chain is very fragile, and it is going to remain that way until we can increase chip production.”
…The thin inventories are a source of particular concern because of how a single shutdown can then ripple through the supply chain. With these wafer-thin inventories, a closure of an overseas factory earlier in a company’s supply chain, for more than a few days, can cause it to exhaust its inventories.
After widespread opposition from the rest of the Internet, Google is killing its “FLoC” plans.
The company wants to get rid of the third-party web cookies used for advertising tracking, so it proposed FLoC (“Federated Learning of Cohorts”), which would have let its browser track you for the benefit of advertising companies. With FLoC dead, Google is floating another proposal to track users for advertisers. This time, the system is called the “Topics API.” There are currently no implementation details, but Google has posted info about the Topics API in a blog post, in developer docs, on a GitHub page, and on a “Privacy Sandbox” site.
Google’s Topic API plans are just now being shared with the world, and the company says the next step is to build a trial implementation and gather feedback from the Internet. Hopefully, the EFF, Mozilla, the EU, and other privacy advocates that spoke out about FLoC will chime in on Google’s new plan. The Topics API gives users more control over the tracking process, but if your core complaint was that browser makers should not build user tracking technology directly into the browser for the benefit of advertising companies, you’ll still find fault with Google’s plan. Google is the world’s biggest advertising company, and it’s using its ownership of the world’s biggest browser to insert its business model into Chrome.
…FLoC worked by grouping people with similar browsing histories together into a “cohort” and would make assumptions about that group for advertising purposes. One of the concerns was that these groups could be small enough to individually track users, which is what third-party cookies do today. Google says that Topics should be broad enough to ensure that users are not individually tracked and to further reduce fingerprinting. Google says that “5 [percent] of the time, a random topic (chosen from the full set of topics) is provided.”
Aaron Gordon is – he says – abandoning Android:
for the past six years, Google has made the Pixel line of phones. They are Google-made phones, meaning Google can’t blame discontinuing security updates on other manufacturers, and yet, it announced that’s exactly what it would do.
The planned obsolescence is frustrating enough, and I’m certainly annoyed that I have to spend hundreds of dollars on a new phone when I really shouldn’t have to. It still boggles my mind that we can fashion a bunch of precious metals together to send invisible messages to people anywhere in the world. For millions of years, these metals formed underground, and then, with great ingenuity and exploitation, those metals were mined, transported, and sold as amazing and necessary technology, making Google incredibly rich. And Google has decided it will only put those rocks to use for three measly years before turning those rocks into something even less valuable than rocks. It’s now garbage.
I will recycle the phone when I’m done with it, but I’m also under no illusions about the likelihood that process will yield anything useful. Maybe one day we’ll get better at recycling phones, but because companies like Google want the phones to be as compact, energy efficient, and alluring as possible, they are put together in a way that makes them difficult to take apart whenever Google decides they will no longer function reliably as phones. Which, again, it has done after just three years.
His rationale for switching to an iPhone is longer security update support (up to seven years).
The number of people switching between Android and iOS is tiny compared to the total. The number switching because they’re annoyed at the lack of security updates is even smaller (though probably a good chunk of the former). But it was Gordon’s point about the waste of rare metals that chimed for me. People hang on to their phones for longer and longer now: three years is by no means unusual. Security updates don’t matter for most people; for journalists, it’s different.
unique link to this extract
The ballooning popularity of Bitcoin and other digital currencies has given rise to a strange new political breed: the crypto mayor. Eric Adams, New York’s new mayor, accepted his first paycheck in Bitcoin and another cryptocurrency, Ether. Francis Suarez, Miami’s mayor, headlines crypto conferences. Now even mayors of smaller towns are trying to incorporate crypto into municipal government, courting start-ups and experimenting with buzzy new technologies like nonfungible tokens, or NFTs, to raise money for public projects.
Their growing ranks reflect the increasing mainstream acceptance of digital currencies, which are highly volatile and have fallen in value in recent days. The mayors’ embrace of crypto is also a recognition that its underlying blockchain technology — essentially a distributed ledger system — may create new revenue streams for cities and reshape some basic functions of local government.
“Mayors rationally want to attract high-income citizens who pay their taxes and impose few costs on the municipality,” said Joseph Grundfest, a business professor at Stanford. “Crypto geeks fit this bill perfectly.”
But as with many ambitious crypto projects, it’s unclear whether these local initiatives will ultimately amount to much. So far, most are either largely symbolic or largely theoretical. And the mayors’ aims are partly political: crypto boosterism has a useful bipartisan appeal, garnering popularity among both antigovernment conservatives and socially liberal tech moguls.
“You can do these things because you want to be associated with dudes with AR-15s, or you want to be associated with Meta,” said Finn Brunton, a technology studies professor at the University of California, Davis, who wrote a 2019 book about the history of crypto. “A lot of it is hype and hot air.”
The “rise” seems to consist of a total of three mayors (though New York and Miami aren’t tiddlers). For the opposing view, which isn’t in here, read Mark Headd’s quick thoughts about this. He’s not a fan.
Also, “high-income citizens who pay their taxes”? I though the crypto-libertarian impulse was to avoid taxes at all costs.
unique link to this extract
|• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?
Social Warming, my latest book, and find answers – and more.
Errata, corrigenda and ai no corrida: none notified