Future versions of Apple’s AirPods might be able to measure your body temperature via your ear canal. CC-licensed photo by Ivan Radic 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. Use them wisely. I’m @charlesarthur on Twitter. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
Musk says Twitter is limiting number of posts users can read • The New York Times
Eduardo Medina and Ryan Mac:
»
Elon Musk said on Saturday that Twitter will temporarily limit the number of posts users can read per day to address concerns over data scraping, just hours after thousands of users reported widespread problems using the site.
Many of those users reported that they were getting an error message that they had “exceeded” their “rate limit,” suggesting that they had violated Twitter’s rules and downloaded and viewed too many tweets.
Mr. Musk, who said on Friday that “several hundred organizations” were taking Twitter’s data in a process called scraping and that “it was affecting the real user experience,” did not say how long the limits would last or what could prompt him to lift the restriction.
He originally said that verified accounts would be limited to reading 6,000 posts per day, unverified accounts to 600 posts and new unverified accounts to 300 posts. About two hours later, he bumped those limits to 8,000 for verified, 800 for unverified and 400 for new unverified — before raising them again early Saturday evening to 10,000, 1,000 and 500. “Rate limited due to reading all the posts about rate limits,” Mr. Musk said on Twitter.
…On Saturday, engineers in the company raced to diagnose the problem in private Slack channels, according to two employees. Those people said that Twitter salespeople asked what they should tell their advertising clients as they realized that some ads were not being displayed on the social network.
Twitter’s US advertising revenue for the five weeks from April 1 to the first week of May was $88m, down 59% from a year earlier, according to an internal presentation obtained by The New York Times. The company has regularly fallen short of its US weekly sales projections, sometimes by as much as 30%, the document said.
«
It’s become chaotic. The suggestion is that Twitter switched between AWS and Google Cloud, and also that by blocking unlogged-in tweets he created a DDOS as the hundreds of thousands of embedded tweets on sites all over the web tried to access the site and were rebuked, and tried again, and were rebuked, and tried again… Where Musk and Twitter are concerned, bad ideas never come singly.
unique link to this extract
TikTok-owner ByteDance debuts Ripple music creation app • Engadget
Mariella Moon:
»
ByteDance, TikTok’s Chinese parent company, has debuted a new app designed to make it easier for creators to compose and edit music they could use for their content. The app called Ripple is only available in the US for now, and the company is testing it in a closed beta environment. ByteDance says it can assist creators in the way portable smart digital audio workstations (DAWs) can and is perhaps most useful for beginners and anybody who’d rather not deal with more complex systems. It was also designed to make it easier for creators to add custom soundtracks to their short-form videos for TikTok and other platforms.
Ripple can create songs in various genres based on a melody the user hums. The app prompts them into humming into the phone mic and then generates instrumentals they can use, such as drums, bass and piano. The length of the song output will match the length of the input, though — the app can’t generate a full soundtrack from just a few seconds of humming. Also, Ripple can only generate instrumental music, leaving the vocal work to creators.
ByteDance told us that Ripple’s model was trained on music it owns and music that was licensed to the company. The company also said that it’s committed to respecting the rights of its artists and its rightsholder partners. To note, there have been concerns about the source of data used to train artificial intelligence systems and algorithms.
«
Neat. Music-writing apps are getting closer and closer to the everyday.
unique link to this extract
Climate change making wildfires and smoke worse, scientists call it the ‘new abnormal’ • AP News
Seth Borenstein and Melina Walling:
»
As Earth’s climate continues to change from heat-trapping gases spewed into the air, ever fewer people are out of reach from the billowing and deadly fingers of wildfire smoke, scientists say. Already wildfires are consuming three times more of the United States and Canada each year than in the 1980s and studies predict fire and smoke to worsen.
While many people exposed to bad air may be asking themselves if this is a “new normal,” several scientists told The Associated Press they specifically reject any such idea because the phrase makes it sound like the world has changed to a new and steady pattern of extreme events.
“Is this a new normal? No, it’s a new abnormal,” University of Pennsylvania climate scientist Michael Mann said. “It continues to get worse. If we continue to warm the planet, we don’t settle into some new state. It’s an ever-moving baseline of worse and worse.”
It’s so bad that perhaps the term “wildfire” also needs to be rethought, suggested Woodwell Climate Research Center senior scientist Jennifer Francis.
“We can’t really call them wildfires anymore,” Francis said. “To some extent they’re just not, they’re not wild. They’re not natural anymore. We are just making them more likely. We’re making them more intense.”
Several scientists told the AP that the problem of smoke and wildfires will progressively worsen until the world significantly reduces greenhouse gas emissions, which has not happened despite years of international negotiations and lofty goals.
Fires in North America are generally getting worse, burning more land. Even before July, traditionally the busiest fire month for the country, Canada has set a record for most area burned with 31,432 square miles (81,409 square kilometers), which is nearly 15% higher than the old record.
«
Apple’s next AirPods Pro may check your hearing health and take your temperature • The Verge
Wes Davis:
»
Some AirPods will be gaining a new hearing health feature, supported by iOS 17, that can check yourself for potential hearing issues and may be able to determine your body temperature via your ear canal, according to Bloomberg’s Mark Gurman in today’s Power On newsletter. He also says all of Apple’s new headphones will include USB-C, and that the company is planning new AirPods Pro and AirPods Max models — but he doesn’t think new hardware is coming soon.
AirPods already support audiograms — audio profiles that tell the AirPods where your hearing may be weakest so that they can tune themselves to your hearing abilities. Right now, you can generate an audiogram using the app Mimi, which Apple would be Sherlocking — an infamous Apple tendency to fold third-party features and apps into its operating system — with a built-in hearing test feature.
Other wireless earbuds have similar capability built in as well. The Jabra Elite 75t added a feature in 2020 called MySound that creates custom sound profiles after playing a series of tones in each ear and prompting wearers to tap their screen when they hear a noise, and the Nothing Ear 2 launched with its own hearing test and audio profile feature this year.
…Gurman added that future AirPods could take your temperature with your ear canal. Rumors that Apple would do this go as far back as late 2021, and the Apple Watch Series 8 introduced something similar last year, albeit as a relative temperature that’s only really used for menstrual cycle tracking.
Expanding temperature tracking to the AirPods could be more accurate, he says, and would potentially let Apple more accurately tell a wearer if they’re starting to get sick. Gurman didn’t say whether this would be exclusive to the AirPods Pro, but it seems likely.
«
Gurman is always incredibly vague about when Apple is going to do all these things, except in very rare cases (eg headsets). The vagueness seems to reduce as the launch approaches, so take the imprecision here as indicating it’s many, many months off.
unique link to this extract
Crypto miners seek a new life in AI boom after an implosion in [crypto] mining • WSJ
Tom Dotan and Berber Jin:
»
The boom in demand for high-end chips powering the rise of artificial intelligence has given new life to some of the survivors of the last tech-hype cycle: cryptocurrency miners.
During the crypto upsurge, Satoshi Spain sold and leased out hundreds of souped-up computers—known as mining rigs—equipped with powerful graphics chips. Then, as many of them sat idle starting last year because of the decline of mining currencies, the Spanish company began helping its customers retool the machines for AI.
Today, Satoshi Spain’s machines are handling AI workloads for startups, universities and individual developers in Europe. “You can still make money from your mining rig,” Satoshi Spain founder Alejandro Ibáñez de Pedro said. “It’s mining 2.0.”
Satoshi Spain is one of many reformed—and opportunistic—companies connected to the cryptocurrency business that have turned their attention to the AI boom, for which one of the most crucial assets is access to the graphics chips, known as graphics processing units, or GPUs.
Originally popularized by gamers who needed powerful graphics-processing on computers, the chips are well-suited for the intensive calculations needed to create new units of cryptocurrency. They are also good for the computational workloads needed to train and run AI systems that generate sophisticated text and images.
…The mechanics of mining cryptocurrencies are different from training or running AI models. Mining uses GPUs to solve an increasingly complicated arithmetic problem. Training AI involves easier calculations—but the high volume of data required to generate language or images requires many identical chips working in concert and a lot of memory.
While it isn’t always easy or cheap to convert them, the refurbished mining rigs can be more affordable and accessible than the AI infrastructure offered by the top cloud companies. They are often used by startups and universities that are having trouble getting AI computing power elsewhere. AI companies typically rely on cloud giants such as Microsoft and Amazon to provide computing infrastructure, but the cloud titans are sometimes near full capacity or less interested in smaller orders.
«
Why we shouldn’t hold referendums • Tim Harford
»
Citizens of democracies can be ill-informed and inconsistent, and this often feels like a tragedy or even a crisis. Occasionally, however, one reads something so absurd that it would take a heart of stone not to laugh. Consider a recent survey conducted by the Associated Press-NORC Center for Public Affairs Research (AP-NORC), which finds that 60% of Americans think the government is spending too much.
But, the survey also asks, what exactly is the government spending too much on? Not social security: 62% think the government spends too little on that, versus 7% who think it spends too much. Not Medicare (58% want more spending, 10% want less). Not healthcare (63% want more spending). Not education (65% want more). Not assistance to the poor (59% want more). Military spending is more controversial, but more Americans favour an expansion than a contraction. Add in debt interest, and these areas together cover 91% of US federal government spending last year.
In short, a solid majority of Americans wish their government would spend less money overall, while also spending more on almost everything in its budget.
“That survey is a real classic of the genre,” says Ben Zaranko, an economist at the UK’s Institute for Fiscal Studies. Then he adds, “but it is how governments in the UK behave at spending reviews”.
Spending reviews in the UK usually happen every three years, although we had them in 2019, 2020 and 2021. At these reviews, the government first decides how much it wants to spend overall, then allocates that sum between competing public services, before realising that the overall spending cap implies unpalatable cuts to specific areas. Eventually, the government backtracks and finds extra cash. This has happened in each of the past four spending reviews — most recently, to the tune of £30bn of extra funding, or nearly £500 per person.
«
How much of American democracy did the Supreme Court just destroy? • Eudaimonia and Co
umair haque:
»
The first thought that crossed my mind was this. So de facto segregation is now OK in America again. The Supreme Court had just decided, after all, that on the basis of religious beliefs, it’s OK to discriminate against people. This is the kind of thought that gets me in trouble with Americans. They accuse messengers like me of being alarmists. And yet their democracy is in profound, off the charts trouble now.
These decision aren’t just bad. They are smoking, colossal wrecking balls to centuries of progress. Do Americans understand that? De facto segregation. The first thought that crossed my mind. Alarmism? In fact, it was the central conclusion of Justice Sotomayor’s dissent. Listen to what she has to say, carefully, and really take it in.
»
Although the consequences of today’s decision might be most pressing for the LGBT community, the decision’s logic cannot be limited to discrimination on the basis of sexual orientation or gender identity. The decision threatens to balkanize the market and to allow the exclusion of other groups from many services. A website designer could equally refuse to create a wedding website for an interracial couple, for example.
How quickly we forget that opposition to interracial marriage was often because “‘Almighty God . . . did not intend for the races to mix.’ ” Loving v. Virginia, 388 U. S. 1, 3 (1967). Yet the reason for discrimination need not even be religious, as this case arises under the Free Speech Clause. A stationer could refuse to sell a birth announcement for a disabled couple because she opposes their having a child. A large retail store could reserve its family portrait services for “traditional” families. And so on.
«
Losing LBGTQ rights is bad. But that is not even only what has been lost here. Civil rights have. Period. Full stop. All of them, more or less, as applied to the private sphere. In one fell swoop.
«
Hard not to see how Sotomayor’s argument is correct. Haque isn’t the only one worried in this way. Gorsuch, in particular, with his self-satisfied smug grin, is taking the US back to the 1920s.
unique link to this extract
AI and the automation of work • Benedict Evans
He’s, well, bemused at how overexcited people are – but also holds out the possibility of a lot of change (while not forgetting that we’ve gone through a lot of change regularly):
»
Whatever you think will happen, it will take years, not weeks.
First, the tools that people use for work, and the tasks that might now get a new layer of automation, are complicated and very specialised, and embody a lot of work and institutional knowledge. A lot of people are experimenting with ChatGPT, and seeing what it will do. If you’re reading this, you probably have too. That doesn’t mean that ChatGPT has replaced their existing workflows yet, and replacing or automating any of those tools and tasks is not trivial.
There’s a huge difference between an amazing demo of a transformative technology and something that a big complicated company holding other people’s business can use. You can rarely go to a law firm and sell them an API key to GCP’s translation or sentiment analysis: you need to wrap it in control, security, versioning, management, client privilege and a whole bunch of other things that only legal software companies know about (there’s a graveyard of machine learning companies that learnt this in the last decade). Companies generally can’t buy ‘technology’. Everlaw doesn’t sell translation and People.ai doesn’t sell sentiment analysis – they sell tools and products, and often the AI is only one part of that. I don’t think a text prompt, a ‘go’ button and a black-box, general purpose text generation engine make up a product, and product takes time.
Second, buying tools that manage big complicated things takes time even once the tool is built and has product-market fit. One of the most basic challenges in building an enterprise software startup is that startups run on an 18 month funding cycle and a lot of enterprises run on an 18 month decision cycle.
«
‘It was an accident’: the scientists who have turned humid air into renewable power • The Guardian
Ned Carter Miles:
»
…trying to prove the worth of an early proof-of-concept at conferences had them literally red in the face. He says: “The signal was not stable and it was low. We were able to generate 300 milliwatts, but you had to put all your effort into your lungs in order to breathe enough humidity into the samples.”
They’ve come a long way since then, with Catcher and related projects receiving nearly €5.5m (£4.7m) in funding from the European Innovation Council. The result is a thin grey disc measuring 4cm (1.5in) across. According to the Lyubchyks, one of these devices can generate a relatively modest 1.5 volts and 10 milliamps. However, 20,000 of them stacked into a washing machine-sized cube, they say, could generate 10 kilowatt hours of power a day – roughly the consumption of an average UK household. Even more impressive: they plan to have a prototype ready for demonstration in 2024.
A device that can generate usable electricity from thin (or somewhat muggy) air may sound too good to be true, but Peter Dobson, emeritus professor of engineering science at Oxford University, has been following both the UMass Amherst and Catcher teams’ research, and he’s optimistic.
“When I first heard about it, I thought: ‘Oh yes, another one of those.’ But no, it’s got legs, this one has,” says Dobson. “If you can engineer and scale it, and avoid the thing getting contaminated by atmospheric microbes, it should work.”
He goes on to suggest that preventing microbial contamination is more an “exciting engineering challenge” than a terminal flaw, but there are far greater problems to overcome before this technology is powering our homes.
«
Such as: who builds them and who needs them. But: fascinating proof of concept.
unique link to this extract
How Google Reader died — and why the web misses it more than ever • The Verge
David Pierce:
»
Google’s bad reputation for killing and abandoning products started with Reader [killed 10 years ago] and has only gotten worse over time. But the real tragedy of Reader was that it had all the signs of being something big, and Google just couldn’t see it. Desperate to play catch-up to Facebook and Twitter, the company shut down one of its most prescient projects; you can see in Reader shades of everything from Twitter to the newsletter boom to the rising social web. To executives, Google Reader may have seemed like a humble feed aggregator built on boring technology. But for users, it was a way of organizing the internet, for making sense of the web, for collecting all the things you care about no matter its location or type, and helping you make the most of it.
A decade later, the people who worked on Reader still look back fondly on the project. It was a small group that built the app not because it was a flashy product or a savvy career move — it was decidedly neither — but because they loved trying to find better ways to curate and share the web. They fought through corporate politics and endless red tape just to make the thing they wanted to use. They found a way to make the web better, and all they wanted to do was keep it alive.
“This is going to be the driest story ever,” says Chris Wetherell, when I ask him to describe the beginning of Google Reader. Wetherell wasn’t the first person at Google to ever dream of a better way to read the internet, but he’s the one everyone credits with starting what became Reader. “Okay, here goes: a raging battle between feed formats,” he says when I push. “Does that sound interesting?”
Here’s the short version: one of the most important ways that information moves around the internet is via feeds, which automatically grab a webpage’s most important content and make it available. Feeds are what make podcasts work across apps, and how content shows up in everything from Flipboard to Facebook.
«
This is indeed a fascinating story of what might have been – though how well would we have all worn Google controlling all the feeds pinging around the web, being Twitter and Substack?
And if Wetherell’s name seems familiar: he then went to Twitter where he led the team that built the Retweet feature. Arguably one of the most influential people in the modern social web.
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? Read Social Warming, my latest book, and find answers – and more. |
Errata, corrigenda and ai no corrida: none notified
The Overspill: when there's more that I want to say 