A strike at the New York Times means there won’t be a Wordle today, November 5. Drama! CC-licensed photo by Ken Mayer 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.
There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.
A selection of 10 links for you. Freed! I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
Generative AI has a massive e-waste problem • IEEE Spectrum
Katherine Bourzac:
»
Private investment in generative AI has grown from about US $3bn in 2022 to $25bn in 2023, and about 80% of private companies expect AI to drive their business in the next three years, according to Deloitte. Keeping up with the latest advancements means upgrading GPUs, CPUs, and other electronic equipment in data centers as newer, more advanced chips become available. And that, researchers project, will lead to an explosion in the production of electronic waste.
A study published last week in the journal Nature Computational Science estimates that aggressive adoption of large language models (LLMs) alone will generate 2.5m tonnes of e-waste per year by 2030.
“AI doesn’t exist in a vacuum; it relies on substantial hardware resources that have tangible environmental footprints,” says study coauthor Asaf Tzachor, a sustainability and climate researcher at Reichman University, in Israel. “Awareness of the e-waste issue is crucial for developing strategies that mitigate negative environmental impacts while allowing us to reap the benefits of AI advancements,” he says.
Most research on AI sustainability has focused on these models’ energy and water use and their concomitant carbon emissions. Tzachor worked with Peng Wang and Wei-Qiang Chen, both professors at the Chinese Academy of Sciences, to calculate the potential increase in e-waste associated with generative AI. The study is intended to provide an estimate of the potential scale of the problem, and the researchers hope it will spur companies to adopt more sustainable practices.
«
In its way, classic human behaviour: do it now, worry about the effect later.
unique link to this extract
Scientists may have solved the mystery of sky-high methane emissions • The Washington Post
Shannon Osaka:
»
Almost two decades ago, the atmosphere’s levels of methane — a dangerous greenhouse gas that is over 80 times as potent as carbon dioxide in the short term — started to climb. And climb.
Methane concentrations, which had been stable for years, soared by 5 or 6 parts per billion every year from 2007 onward. Then, in 2020, the growth rate nearly doubled.
Scientists were baffled — and concerned. Methane is the big question mark hanging over the world’s climate estimates; although it breaks down in the atmosphere much faster than carbon dioxide, it is so powerful that higher-than-expected methane levels could shift the world toward much higher temperatures.
But now, a study sheds light on what’s driving record methane emissions. The culprits, scientists believe, are microbes — the tiny organisms that live in cows’ stomachs, agricultural fields and wetlands. And that could mean a dangerous feedback loop — in which these emissions cause warming that releases even more greenhouse gases — is already underway.
“The changes that we saw in the last couple of years — and even since 2007 — are microbial,” said Sylvia Michel, lead author of the paper published last month in the Proceedings of the National Academy of Sciences. “Wetlands, if they are getting warmer and wetter, maybe they’re producing more methane than they used to.”
«
Figured out by isotopic examination of the methane’s carbon atoms. If it’s microbes, this is bad.
unique link to this extract
Why did so many die in Spain? Because Europe still hasn’t accepted the realities of extreme weather • The Guardian
Friederike Otto:
»
I founded World Weather Attribution in 2014 to shift the conversation. Our attribution studies are carried out quickly, over days or weeks, in the immediate aftermath of weather disasters to inform people in real time about the role of the climate crisis.
A quick analysis following the floods in Spain found that the climate emergency made the extreme rainfall about 12% more intense and twice as likely. Despite this, in Paiporta, where at least 62 people have died, the mayor said floods were not common and “people are not afraid”. But the changing climate is making once-rare events more common.
Record-breaking events such as these complicate preparedness – how do you communicate the extreme danger of something someone has never experienced before?
We saw this play out recently after Hurricane Helene made landfall. More than 200 people died in floods in the inland southern Appalachians region of the US. Despite warnings of “catastrophic and life-threatening” flooding ahead of the disaster, people were still caught out when disaster struck, and many could not appreciate how extreme the downpours were going to be.
However, in Spain, people were only warned as it was happening. Warnings were not sent until many people were already trapped in flooded houses or in underground car parks, trying to move their cars to higher ground.
The same happened – or rather didn’t happen – in Germany in 2021. No information was given on how to act and, crucially, no support was given to those who could not help themselves: in the German town of Sinzig, 12 residents of a home for disabled people drowned. Back in Spain, the deaths of the inhabitants of one care home have already been reported and I fear more disturbing stories such as this will emerge in the weeks to come.
World Weather Attribution has studied 30 devastating floods, and in almost all cases, including in developing countries, we’ve found that the rainfall was well forecast. But as we’ve seen in Spain, forecasting is not enough. The warnings, when they finally came, did not include vital information on where to evacuate to and how.
«
404 Media is partnering with Wired • 404 Media
Emanuel Maiberg, Samantha Cole, Jason Koebler and Joseph Coxk:
»
Today we are happy to announce WIRED is going to co-publish two of our articles a month on its website. We are really excited to get our stories in front of a wider audience and grow 404 Media. We also hope to occasionally collaborate with our friends and colleagues at WIRED on stories we can do better together.
To be absolutely clear, 404 Media remains a 100% independent, journalist-owned company managed by the four of us. Our collaboration with WIRED is similar to our collaboration with Court Watch and Capitol Forum, and involves us co-publishing stories on like-minded publications that we think their audience will like and widening our reach beyond what we’re able to do by ourselves. If you’re a paying subscriber reading this, there’s not really anything for you to do or to know—you’ll continue to get ad-free access to all of our articles, including those we co-publish with WIRED. And if you’re reading any of those co-published articles on our site, you will need to be a paying 404 Media subscriber.
Like most people who follow technology, we’ve been fans of WIRED since before we became journalists ourselves, and we’re especially happy to partner with the legendary title now that it is home to many of our talented former co-workers at VICE.
«
Hopeful that this means 404 Media is getting money for this.
unique link to this extract
Gender is a big factor in what a TikTok user sees about politics and the election • The Washington Post
Jeremy B. Merrill, Cristiano Lima-Strong and Caitlin Gilbert:
»
A significant gender gap has emerged in this year’s presidential campaign, with women voters breaking for Vice President Kamala Harris and men for former president Donald Trump. For participants in a unique Washington Post experiment, that gap has also shown up in their TikTok feeds.
This fall, more than 800 American adults shared their TikTok viewing histories with The Post, opening a rare window on how the increasingly popular app presents political news. The Post found that female users received roughly 11% more content about Harris than men did, while men — even liberal ones — were more likely to be shown videos about Trump than women were.
The 800 users who responded to The Post overwhelmingly identified as liberal. But the gender gap also showed up in a Post analysis of 300 TikTok users whose histories were collected in a parallel effort by researchers at Cybersecurity for Democracy, a nonpartisan multi-university project that studies algorithms. In that dataset — where liberals, conservatives and moderates were evenly divided — men were 12.5% more likely than women to see videos about Trump.
The Post found a similar gender imbalance for videos posted directly by the campaigns. In the Cybersecurity for Democracy sample, women of all political persuasions saw Harris campaign videos 40% more often than men did. And men saw Trump campaign videos more than twice as often as women did.
The findings shed light on how TikTok is shaping the way American adults, especially younger ones, get their news — and what they see about politics. The app claims 170 million U.S. users, and 4 in 10 US adults under 30 say they regularly turn to TikTok for news. Both campaigns have said they see the app as a key messaging tool to reach Gen Z and millennial voters.
«
Doesn’t this rather give the lie to the suggestion that TikTok is an Evil Chinese App Trying To Control America’s Politics? Never mind – the forced sale is still scheduled for (Sunday) January 19 next year, the day before the presidential inauguration.
unique link to this extract
Meta’s Threads has 275 million users. It sure doesn’t feel that way • Business Insider
Peter Kafka:
»
Meta says Threads is growing very, very fast and has been adding a million sign-ups a day. Back in April, Meta said Threads had 150 million monthly users. Which means it has grown 83% in half a year.
But all of that is hard to square with my experience on Threads. I’m there quite a bit, posting and reading for fun and for work. And it doesn’t seem like the place has nearly doubled recently.
This isn’t a metrics-based argument — I’m not tracking the engagement for my posts in any serious way. Just a vibes thing. When Threads launched in July 2023, it definitely felt empty, especially compared to what was then still called Twitter. It’s picked up since then.
But it certainly doesn’t seem anything close to what Twitter was like back in 2019, back when lots of people were calling it a hellsite and also checking it every day.
For giggles, I did the classic lazy reporter gambit and asked people on Threads if it felt like the place has gotten a lot more full recently. The results seem split. Special shout out to sami3456567, whose bio says he’s no fan of “incels, present & future cat ladies, anything remotely French.” You’ll always have a place in social media, sami.
One reason I may not be noticing that Threads seems more crowded is that some of that growth is coming from people who are unlikely to be posting in English: Meta says a bunch of its recent users have been coming from Japan and Taiwan.
But the most obvious reason Threads can grow so fast is that Meta surfaces Threads posts in Instagram, which does have gazillions of users. Some of them click on Threads posts they see there, and become Threads users that way.
And you can definitely see, or at least feel, the presence of Instagram users who are logging into Threads. They’re the people who seem to have wandered into the bar not completely knowing why they’re there or what they’re supposed to do, and they are prime targets for my colleague Katie Notopoulos.
«
Notopoulos is notoriously good at the empty ragebait posts which Threads amplifies. But it’s not a news network, and for that reason journalists just don’t like it. (I find it too random.)
unique link to this extract
The end of independent publishing and Giant Freakin Robot • GIANT FREAKIN ROBOT
Joshua Tyler:
»
After relaunching GIANT FREAKIN ROBOT in 2019, the site grew to a readership of more than 20 million a month, through 2021 and 2022. Then Google decided they didn’t want independent publishers around anymore. Most entertainment keywords have now been given to one big company, whose numerous sites own the top slots for nearly every entertainment-related query of any substance.
No one can find our site to read it so that 20 million unique visitors is now a few thousand a month. Nearly every independently owned entertainment news publisher is in the same situation, in one way or another.
Last week I was one of 20 independent publishers invited to an event at Google in Mountain View, California. I didn’t know why I was going, I didn’t know what would happen, and Google didn’t cover all my expenses. But I went, I went because it was the only chance I had to save the jobs of the few employees GIANT FREAKIN ROBOT has left.
Unfortunately, if you’ve read my account of that catastrophe, you know that instead of finding solutions I received a clear signal from Google that they have no intention of allowing anyone to see GIANT FREAKIN ROBOT’s work.
Google’s engineers were adamant that there is no problem with our website or our content. In fact, some of the geeky Googlers there regularly read our articles and seem to genuinely like us. However, my take away was that we don’t fit in with the search monopoly’s new business model, and so we won’t be shown to anyone.
That business model seems to largely revolve around pleasing big brands. During the course of our conversation with Google, an Independent site owner asked why a “Best Shoes To Buy” (not the actual keyword analyzed, just an example) list from a big retailer like Nike always seems to outrank a “Best Shoes To Buy” list from an actual, unbiased reviewer. Google’s response was to ask, “Well if we don’t rank those big brands first, won’t they be mad?”
Unable to help myself I blurted out, “Who cares? Reviewing things isn’t their business. Selling shoes is.” The Googler in charge ignored my answer.
«
His solution: move to YouTube. Somehow I feel this is not going to solve the problem.
unique link to this extract
Vote-counting machines are too dumb to pose a hacking risk • Bloomberg
Austin Carr:
»
As the US presidential election fast approaches, Elon Musk keeps claiming, with scant evidence [correction: ZERO evidence, AS YOUR STORY GOES ON TO SHOW – exasperated Overspill Ed.], that internet-connected voting machines are vulnerable to hacking and that the country must switch back to paper ballots for better security.
Having spent much of my time lately learning what it’d actually take to hack these devices — speaking with election officials and machine makers and visiting a vendor’s factory — I can assure you Musk’s claims are quite unfounded.
It’s not that election computers are advanced to a level of impenetrability; rather, what makes them so ironically safe at scale is that their functions are intentionally limited and dependent on an absurd degree of bureaucratic logistics and analog-world redundancies.
Keep in mind that most voters (98%!) tomorrow will cast their votes on paper, the majority using pens and Sharpies to mark their choice on slips of paper. The precinct tabulators scanning those sheets are offline (often lacking even the hardware for Wi-Fi, Ethernet or Bluetooth) and undergo pre-election logic and accuracy testing. And all ballots are preserved for backup so authorities can confirm the digital and physical counts match. These votes are usually separately recounted at central county offices, anyway.
Since throwing his weight behind Donald Trump, Musk has made voter fraud a key point of his stumping in swing states like Pennsylvania. Unfounded ideas about rigged voting systems are not new among Trump supporters, but Musk’s Tesla Inc. and SpaceX tech background makes him a uniquely effective and highly misleading messenger of these conspiracy theories. “I’m normally someone who favors technology — I’m a super 21st-century technology boy right here,” Musk said at an October campaign stop in Pittsburgh. “And I’m saying: no machines for voting.”
Even if you managed magically to slip by workers and surveillance cameras at a poll site to insert some undetectable, malware-laced thumb drive into a machine to manipulate votes, you’d still have to figure out how to fake loads of corresponding physical ballots while destroying the real ones. And then, so what? You’ve only broken into one machine.
You’d next have to replicate that in-person hack on thousands of additional voting machines. “It’s bull!” said Jerry Feaser, the recently retired elections director of Dauphin County, PA, where Musk has also campaigned, when I asked him about the X Corp. owner’s claims. “There would have to be so much collusion among so many different people and levels.”
«
That an editor at Bloomberg felt compelled to put (or leave) “scant evidence” in that first sentence is beyond annoying. “Musk keeps claiming Earth is flat, with scant evidence.” Would you pass that?
unique link to this extract
Thousands of hacked TP-Link routers used in yearslong account takeover attacks • Ars Technica
Dan Goodin:
»
Hackers working on behalf of the Chinese government are using a botnet of thousands of routers, cameras, and other Internet-connected devices to perform highly evasive password spray attacks against users of Microsoft’s Azure cloud service, the company warned last week.
The malicious network, made up almost entirely of TP-Link routers, was first documented in October 2023 by a researcher who named it Botnet-7777. The geographically dispersed collection of more than 16,000 compromised devices at its peak got its name because it exposes its malicious malware on port 7777.
In July and again in August of this year, security researchers from Sekoia.io and Team Cymru reported the botnet was still operational. All three reports said that Botnet-7777 was being used to skillfully perform password spraying, a form of attack that sends large numbers of login attempts from many different IP addresses. Because each individual device limits the login attempts, the carefully coordinated account-takeover campaign is hard to detect by the targeted service.
Last Thursday, Microsoft reported that CovertNetwork-1658—the name Microsoft uses to track the botnet—is being used by multiple Chinese threat actors in an attempt to compromise targeted Azure accounts. The company said the attacks are “highly evasive” because the botnet—now estimated at about 8,000 strong on average—takes pains to conceal the malicious activity.
«
At first I thought “cameras?” but then realised it meant CCTV. TP-Link and its ilk are a mess that will require cleanup for years and years to come.
unique link to this extract
New York Times’ tech staff threatens strike during Election Day crunch • Semafor
Max Tani:
»
Management says that the Guild has bogged down negotiations with what the paper sees as outlandish, even illegal, proposals. As Semafor previously reported, the Guild proposed a ban on scented products in break rooms, unlimited break time, and accommodations for pet bereavement, as well as mandatory trigger warnings in company meetings discussing events in the news.
Times management has been frustrated by proposals that would provide more money for nonwhite staff and others from underrepresented communities to attend conferences, and language that would prioritize non-citizens in the US on visas in the case of layoffs — both of which the paper pointed out couldn’t be fulfilled because they likely violate employment laws.
While the union has withdrawn or reached agreements with the paper on some of these issues, the paper has been alarmed that the union has continued to push for a provision on journalistic integrity that would allow the non-editorial union to have a say in editorial decisions, including the right to request letters to the editor not be published.
«
Unlimited break time? A say in editorial decisions? Couldn’t they just become journalists if that’s what they want? (Lots of journalists presently find themselves on unlimited breaks, but not through choice.)
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 