SoylentNews

upstart writes:

(A bit on the long side, but well worth the read. --Marty)

The secret sauce for Taiwan's chip superstardom:

When 23-year-old Shih Chin-tay boarded a plane for the United States in the summer of 1969, he was flying to a different world.

He grew up in a fishing village surrounded by sugarcane fields. He had attended university in Taiwan's capital Taipei, then a city of dusty streets and grey apartment buildings where people rarely owned cars.

[...] Today's Taipei is rich and hip. High-speed trains zip passengers along the west coast of the island at 350km/h (218mph). Taipei 101 - briefly the tallest building in the world - towers over the city, an emblem of its prosperity.

Much of that is down to a tiny device no larger than a fingernail. The silicon semiconductor - wafer-thin and best-known now as a chip - sits at the heart of every technology we use, from iPhones to airplanes.

Taiwan now makes more than half the chips that power our lives. Its biggest manufacturer, Taiwan Semiconductor Manufacturing Company (TSMC), is the ninth-most valuable business in the world.

That makes Taiwan nearly irreplaceable - and vulnerable. China, fearing it could be cut off from the most advanced chips, is spending billions to steal Taiwan's crown. Or it could take the island, as it has repeatedly threatened to do.

But Taiwan's path to chip superstardom will not be easy to replicate - the island has a secret sauce, honed through decades of laborious work by its engineers. Plus, the manufacturing relies on a web of economic ties that the escalating US-China rivalry is now trying to undo.

[...] Back then, he quickly realised that taking on US and Japanese giants at their own game was a losing proposition. Instead TSMC would only manufacture chips for others, and not design its own.

This "foundry model", which was unheard of in 1987, changed the landscape of the industry and paved the way for Taiwan to become the pack leader.

[...] "Rule number one at TSMC is don't compete with your customers," Dr Shih says.

Arthur T Knackerbracket has processed the following story:

Although China cannot flood the global market with chips produced with cutting-edge fabrication technologies, strong subsidies for the semiconductor sector in China make it possible for the country to flood the market with chips made on legacy process technologies, thus undercutting much-needed sales that generate revenue that is vital for R&D at Western firms. This tactic could spur the U.S. government to impose tariffs on products using mature processing nodes, reports Bloomberg.

[...] China is known for providing hefty funds to its chipmakers. For example, China-based SMIC invested $24 billion in capital expenditures from 2020 to 2023 with support from banks, local governments, and state-controlled funds, far exceeding its earnings in the period, according to Nikkei. Other semiconductor companies also have generous support from the government, which is how they can quickly expand production capacity using tools that they can procure without any limitations and start producing chips like display driver ICs (DDICs) or power management ICs (PMICs) that are sold in billions of units every year.

[...] The survey's findings are set to guide the U.S. in formulating responses that could include the imposition of tariffs or the use of other trade tools to counteract China's aggressive expansion in the semiconductor industry. Commerce Secretary Gina Raimondo has already indicated that the U.S. is ready to use every tool it has to stop China from flooding the market with low-cost legacy chips. However, she clarified that the most stringent export controls would remain reserved for more advanced process technologies and not for these older generation nodes, so Chinese companies will still be able to procure legacy chipmaking tools.

Original Submission

Motor Trend reports on recent gov't. actions that will lead to drunk & drowsy driving detection built into new cars, https://www.motortrend.com/news/nhtsa-anti-drunk-driving-tech-rules-coming/

The National Highway Traffic Safety Administration (NHTSA) has, after years of voicing its intent, officially submitted an advanced notice of proposed rulemaking surrounding anti-drunk-driving technology. This marks the first formal salvo in its bid to ensure every new vehicle sold in America comes with some form of built-in inebriation detection and possibly an interlock that prevents the car from being driven if the driver is determined to be impaired.

Alcohol-related vehicle crashes are among the leading cause of injury and death on America's roadways, which already have seen a precipitous backsliding in term of safety in recent years. After decades of progress reducing roadway deaths, America has seen car-related fatalities soar, leaving policymakers stumped and scrambling for solutions.

[...] Per NHTSA's notice: "This document initiates rulemaking that would gather the information necessary to develop performance requirements and require that new passenger motor vehicles be equipped with advanced drunk and impaired driving prevention technology through a new Federal Motor Vehicle Safety Standard (FMVSS)." The agency adds that:

"The Infrastructure Investment and Jobs Act (Bipartisan Infrastructure Law or BIL) directs NHTSA to issue a final rule establishing a Federal Motor Vehicle Safety Standard (FMVSS) that requires new passenger vehicles to have 'advanced drunk and impaired driving prevention technology' by 2024. The BIL also provides that an FMVSS should be issued only if it meets the requirements of the National Traffic and Motor Vehicle Safety Act."

Most likely solution are cameras that watch the driver for warning signs, these already exist on some cars. It will be interesting to see if other solutions are also put forward.

MotorTrend then opines,

Of course, this opens many cans of worms. Will NHTSA advocate a warning system onboard that alerts the driver that they're impaired? An ignition interlock that prevents the car from starting and driving at all? What if there are false readings? A sober passenger taking whatever onboard test exists in the driver's place? One also could imagine the sort of arguments advocates of unfettered freedom might come up with: Someone eluding an attacker, but who had previously had a drink, being locked out of their means of escape due to technological misunderstanding. Of course, these concerns must be weighed against the more than 10,000 preventable deaths from drunk drivers annually.

Original Submission

fliptop writes:

Lawmakers want HHS to revise health privacy law to require warrants:

All of the big pharmacy chains in the US hand over sensitive medical records to law enforcement without a warrant—and some will do so without even running the requests by a legal professional, according to a congressional investigation.

The revelation raises grave medical privacy concerns, particularly in a post-Dobbs era in which many states are working to criminalize reproductive health care. Even if people in states with restrictive laws cross state lines for care, pharmacists in massive chains, such as CVS, can access records across borders.

Lawmakers noted the pharmacies' policies for releasing medical records in a letter dated Tuesday to the Department of Health and Human Services (HHS) Secretary Xavier Becerra. The letter—signed by Sen. Ron Wyden (D-Ore.), Rep. Pramila Jayapal (D-Wash.), and Rep. Sara Jacobs (D-Calif.)—said their investigation pulled information from briefings with eight big prescription drug suppliers.

All eight of the pharmacies said they do not require law enforcement to have a warrant prior to sharing private and sensitive medical records, which can include the prescription drugs a person used or uses and their medical conditions. Instead, all the pharmacies hand over such information with nothing more than a subpoena, which can be issued by government agencies and does not require review or approval by a judge.

[...] For now, HIPAA regulations grant patients the right to know who is accessing their health records. But, to do so, patients have to specifically request that information—and almost no one does that. "Last year, CVS Health, the largest pharmacy in the nation by total prescription revenue, only received a single-digit number of such consumer requests," the lawmakers noted.

"The average American is likely unaware that this is even a problem," the lawmakers said.

Originally spotted on Schneier on Security.

Related:

• Websites Selling Abortion Pills are Sharing Sensitive Data With Google
• Meta Faces Lawsuit for Allegedly Collecting Patient Health Data Without Consent
• Google has Access to Detailed Health Records on Tens of Millions of Americans

Original Submission

upstart writes:

Novel Terrapin attack uses prefix truncation to downgrade the security of SSH channels:

Sometime around the start of 1995, an unknown person planted a password sniffer on the network backbone of Finland's Helsinki University of Technology (now known as Aalto University). Once in place, this piece of dedicated hardware surreptitiously inhaled thousands of user names and passwords before it was finally discovered. Some of the credentials belonged to employees of a company run by Tatu Ylönen, who was also a database researcher at the university.

The event proved to be seminal, not just for Ylönen's company but for the entire world. Until that point, people like Ylönen connected to networks using tools which implemented protocols such as Telnet, rlogin, rcp, and rsh. All of these transmitted passwords (and all other data) as plaintext, providing an endless stream of valuable information to sniffers. Ylönen, who at the time knew little about implementing strong cryptography in code, set out to develop the Secure Shell Protocol (SSH) in early 1995, about three months after the discovery of the password sniffer.

[...] Ylönen submitted SSH to the Internet Engineering Taskforce in 1996, and it quickly became an almost ubiquitous tool for remotely connecting computers. Today, it's hard to overstate the importance of the protocol, which underpins the security of apps used inside millions of organizations, including cloud environments crucial to Google, Amazon, Facebook, and other large companies.

[...] Now, nearly 30 years later, researchers have devised an attack with the potential to undermine, if not cripple, cryptographic SSH protections that the networking world takes for granted.

Named Terrapin, the new hack works only when an attacker has an active adversary-in-the middle position on the connection between the admins and the network they remotely connect to. Also known as a man-in-the-middle or MitM attack, this occurs when an attacker secretly positioned between two parties intercepts communications and assumes the identity of both the recipient and the sender. This provides the ability to both intercept and to alter communications. While this position can be difficult for an attacker to achieve, it's one of the scenarios from which SSH was thought to have immunity.

canopic jug writes:

The US Surgeon General has published his 2023 advisory on social control media and youth mental health [warning for PDF]. The report's scope is only on the health and mental health effects, not the weaponized nature of the phenomenon. The body of the report is 17 pages long and includes a call to action.

Extreme, inappropriate, and harmful content continues to be easily and widely accessible by children and adolescents. This can be spread through direct pushes, unwanted content exchanges, and algorithmic designs. In certain tragic cases, childhood deaths have been linked to suicide- and self-harm-related content and risk-taking challenges on social media platforms. This content may be especially risky for children and adolescents who are already experiencing mental health difficulties. Despite social media providing a sense of community for some, a systematic review of more than two dozen studies found that some social media platforms show live depictions of self-harm acts like partial asphyxiation, leading to seizures, and cutting, leading to significant bleeding. Further, these studies found that discussing or showing this content can normalize such behaviors, including through the formation of suicide pacts and posting of self-harm models for others to follow.

Social media may also perpetuate body dissatisfaction, disordered eating behaviors, social comparison, and low self-esteem, especially among adolescent girls. A synthesis of 20 studies demonstrated a significant relationship between social media use and body image concerns and eating disorders, with social comparison as a potential contributing factor. Social comparison driven by social media is associated with body dissatisfaction, disordered eating, and depressive symptoms. When asked about the impact of social media on their body image, nearly half (46%) of adolescents aged 13–17 said social media makes them feel worse, 40% said it makes them feel neither better nor worse, and only 14% said it makes them feel better.

Previously:
(2023) Seattle's Schools are Suing Tech Giants for Harming Young People's Mental Health
(2022) Leaked Documents Reveal Instagram Was Pushing Girls Towards Content That Harmed Mental Health
(2022) Social Media Break Improves Mental Health
(2021) Facebook Documents Show How Toxic Instagram is for Teens, Wall Street Journal Reports

Original Submission

Arthur T Knackerbracket has processed the following story:

When it is very windy, the grid cannot handle the extra power generated. Wind farms are paid to switch off and gas-powered stations are paid to fire up. The cost is passed on to consumers.

The government said major reforms will halve the time it takes to build energy networks to cope with extra wind power.

Energy regulator Ofgem announced new rules in November, which it said would speed up grid connections.

Most of the UK's offshore wind farms are in England - Dogger Bank off the coast of Yorkshire is the largest in the world. Meanwhile, around half of onshore wind farms are in Scotland but most electricity is used in south-east England.

Carbon Tracker said the main problem in getting electricity to where it is needed is a bottleneck in transmission between Scotland and England.

The practice of switching off wind farms and ramping up power stations is known as "wind curtailment" and the costs are passed on to consumers, it said.

[...] "The problem is, there are not enough cables. The logical solution would be to build more grid infrastructure," said Lorenzo Sani, analyst at Carbon Tracker.

"It's not even that expensive," he added, compared with mounting wind curtailment costs.

[...] However, historically it has taken between 10 and 15 years for new transmission cables to be approved.

[...] In November, the government set out a plan to reduce the time it takes to build new infrastructure from 14 to seven years, "speeding up grid connections, supporting thousands of jobs and reducing electricity bills for households across Great Britain", a spokesperson said.

Energy regulator Ofgem said there was a "long queue" of energy projects "which could generate almost 400GW of electricity - well in excess of what is needed to power the entire British energy system".

The watchdog said new rules "will allow stalled or speculative 'zombie' projects to be forced out of the queue, meaning viable projects can be connected quicker".

Mr Sani from Carbon Tracker said it was unclear how much difference these projects would make before 2030.

Original Submission

Arthur T Knackerbracket has processed the following story:

AMD has finally launched its Instinct MI300X accelerators, a new generation of server GPUs designed to provide compelling performance levels for generative AI workloads and other high-performance computing (HPC) applications. MI300X is faster than H100, AMD said earlier this month, but Nvidia tried to refute the competitor's statements with new benchmarks released a couple of days ago.

Nvidia tested its H100 accelerators with TensorRT-LLM, an open-source library and SDK designed to efficiently accelerate generative AI algorithms. According to the GPU company, TensorRT-LLM was able to run 2x faster on H100 than on AMD's MI300X with proper optimizations.

AMD is now providing its own version of the story, refuting Nvidia's statements about H100 superiority. Nvidia used TensorRT-LLM on H100, instead of vLLM used in AMD benchmarks, while comparing performance of FP16 datatype on AMD Instinct MI300X to FP8 datatype on H100. Furthermore, Team Green inverted AMD's published performance data from relative latency numbers to absolute throughput.

AMD suggests that Nvidia tried to rig the game, while it is still busy identifying new paths to unlock performance and raw power on Instinct MI300 accelerators. The company provided the latest performance levels achieved by the Llama 70B chatbot model on MI300X, showing an even higher edge over Nvidia's H100.

By using the vLLM language model for both accelerators, MI300X was able to achieve 2.1x the performance of H100 thanks to the latest optimizations in AMD's software stack (ROCm). The company highlighted a 1.4x performance advantage over H100 (with equivalent datatype and library setup) earlier in December. vLLM was chosen because of its broad adoption within the community and the ability to run on both GPU architectures.

Original Submission

upstart writes:

The technologies, policies, and commitments providing a glimmer of hope in an otherwise gloomy year:

It's been the hottest year on record, with January through November clocking in at 1.46 °C (2.62 °F) warmer on average than preindustrial temperatures. Meanwhile, emissions from fossil fuels hit a new high—36.8 billion metric tons of carbon dioxide, 1.1% more than in 2022.

Scientists are loudly warning that the world is running out of time to avoid dangerous warming levels. The picture is grim. But if you know where to look, there are a few bright spots shining through the darkness.

[...] EVs are on track to make up 15.5% of automotive sales this year, according to BNEF. Between battery electric vehicles and plug-in hybrids, this new growth means there are almost 41 million passenger EVs on the road. China has the largest share of EVs in the world, making up nearly a quarter of the global fleet.

[...] While many signs are positive, it's not all rosy for electric vehicles. Growth in sales slowed between 2022 and 2023, and changing demand has some automakers slowing production for models like the Ford F-150 Lightning. Charging infrastructure isn't available or reliable enough in most markets, a problem that has become one of the biggest barriers to EV adoption.

Cars are being sold at a record pace and road emissions are still going up, so EV sales need to accelerate to make a dent in transportation's climate impact. But EVs' progress so far seems to be an encouraging story of a new climate-friendly technology becoming a mainstream option. Let's hope it keeps going in 2024—all gas, no brakes.

janrinok writes:

Now Photoshop giant needs to cough up that $1B break-up fee:

Adobe has decided to abandon its $20 billion acquisition of Figma in a concession to regulatory pressure in Europe, the United Kingdom, and the United States.

The two software makers said on Monday they had reached a joint decision to cancel the pending deal, which was announced in September, 2022.

"It’s not the outcome we had hoped for, but despite thousands of hours spent with regulators around the world detailing differences between our businesses, our products, and the markets we serve, we no longer see a path toward regulatory approval of the deal," said Dylan Field, co-founder and CEO of Figma, in a statement.

Shantanu Narayen, chair and CEO of Adobe, offered a similar concession speech: "Adobe and Figma strongly disagree with the recent regulatory findings, but we believe it is in our respective best interests to move forward independently."

Doing so will cost Adobe $1 billion – that's the amount of the breakup fee due to Figma for Adobe's failure to complete the deal. Everyone knows Adobe is the home of Photoshop, Illustrator, and other tools; Figma, meanwhile, makes software primarily for planning and sketching out graphic designs in a collaborative way with others.

[...] The cancellation of the acquisition is one of the higher profile antitrust blows struck by regulators against Big Tech in several years of political jawboning on both sides of the Atlantic about the need to tame technology firms.

vux984 writes:

Never submitted a story before, but thought this was interesting, and it was the first I'd heard of it. I thought it was actually a pretty good idea.

https://arstechnica.com/cars/2023/12/turquoise-taillights-tell-you-this-mercedes-is-driving-autonomously/

As some cars gain more autonomy, it's probably helpful if they have a way to signal their intentions to other road users. Concept cars have explored this idea for some time—we've seen demos of cars using their headlights to project crosswalks in front of pedestrians or using exterior panels to communicate to a cyclist that the car has seen them, or just emote to everyone nearby.

As is often the case when an idea goes from concept to reality, the result in practice is a little less futuristic. Mercedes says it picked turquoise for the marker lights in the headlight and taillight clusters as well as the side mirror because the color is differentiated enough from other colored lights that road users might encounter while driving; and human factor research points to turquoise as the optimal color for this application.

https://www.cnn.com/2023/12/19/business/mercedes-adds-new-light-color-blue-for-self-driving/index.html

All other driving assistance systems, including Tesla's so-called Full Self Driving and General Motor's Super Cruise, require the driver to pay attention to what's happening around the vehicle. Mercedes' Drive Pilot allows the driver to take their eyes off the road continuously until the system alerts the driver of a need to take over such as when traffic speeds up. In the meantime, drivers can surf the Internet or play games on the car's big center screen. (Sensors in the car ensure the driver does not fall asleep, though.)

The light blue color was selected because it's eye-catching and it won't be confused with anything else. It's not similar to the color of any other lights on a passenger car but it's also clearly different from the darker blue used by police and other emergency vehicles.

Original Submission

upstart writes:

Law enforcement have long tapped users' location data hoarded by tech giants:

Google will soon allow users to store their location data on their devices rather than on Google's servers, effectively ending a long-running surveillance practice that allowed police and law enforcement to tap Google's vast banks of location data to identify potential criminals.

The use of so-called "geofence warrants" have exploded in recent years, in large part thanks to the ubiquity of smartphones coupled with hungry data companies like Google vacuuming up and storing huge amounts of its users' location data, which becomes obtainable by law enforcement requests.

Police can use geofence warrants (also known as reverse-location warrants) to demand that Google turn over information on which users' devices were in a particular geographic area at a certain point in time.

But critics say geofence warrants are unconstitutional and inherently overly broad, since these demands often also include the information of entirely innocent people who were nearby at a time when a crime was committed. Even the courts cannot agree on whether geofence warrants are legal, likely setting up an eventual challenge at the U.S. Supreme Court.

Google's announcement this week did not mention geofence warrants specifically, saying only that the move to store location data on their devices would give users' "more control" over their data. In reality, the move forces police to seek a search warrant to access that specific device instead, rather than asking Google for the data.

janrinok writes:

Most see these two as a winner-take-all, zero-sum contest:

In almost every discussion about RISC-V's position in the ecosystem, the instruction set architecture (ISA) is often seen as a direct competitor to Arm. Most people view the two ISAs as being engaged in a winner-take-all, zero-sum contest, where only one can prevail. However, we believe that, at least for the foreseeable future, the industry will witness a greater degree of mixing and matching. Both ISAs are likely to coexist side by side within increasingly heterogeneous chips.

Admittedly, the entire history of technology is against us on this one. Over the past 40 years, we have seen a recurring pattern: software ecosystems tend to consolidate around a single platform. No one wants to write the same software twice, and not just write but design, test, and debug. So, developers have generally followed their own paths to maximize profits, which, over time, leads to the dominance of a single platform.

However, this old truism is not entirely accurate... ISAs are not exactly software.

[...] Arm cores are largely fixed in their capabilities, whereas RISC-V cores are marketed as being very 'flexible' (though not entirely, but close enough for our discussion). By mixing and matching the two systems, chip designers can find more optimal paths for their needs.

This is not just theory; we have seen it in practice. Apple's A- and M-Series processors, as well as Google's TPUs, appear to incorporate both Arm and RISC-V cores. This trend is evident in many other chips as well.

[...] Will this remain the case forever? Ask anyone on either side, and they are quick to say no. Highlander rules apply, there can be only one. However, in practice, we are not so certain. Chips are changing, becoming more diverse and heterogeneous, as designers search for ways to cope with the slowing of Moore's Law. This has opened up the door to a rethinking of past rules. Therefore, we expect that for a long time to come, we will see both RISC-V and Arm sitting next to each other in many chips.

Original Submission

upstart writes:

North America's first people may have arrived by sea ice highway as early as 24,000 years ago:

One of the hottest debates in archaeology is how and when humans first arrived in North America. Archaeologists have traditionally argued that people walked through an ice-free corridor that briefly opened between ice sheets an estimated 13,000 years ago.

But a growing number of archaeological and genetic finds—including human footprints in New Mexico dated to around 23,000 years old—suggests that people made their way onto the continent much earlier. These early Americans likely traveled along the Pacific coastline from Beringia, the land bridge between Asia and North America that emerged during the last glacial maximum when ice sheets bound up large amounts of water causing sea levels to fall.

Now, in research to be presented Friday, 15 December at the American Geophysical Union Annual Meeting (AGU23) in San Franciso, paleoclimate reconstructions of the Pacific Northwest hint that sea ice may have been one way for people to move farther south.

[...] To get a fuller picture of ocean conditions during these crucial windows of human migration, Summer Praetorius of the US Geological Survey and her colleagues looked at climate proxies in ocean sediment from the coast. Most of the data came from tiny, fossilized plankton. The abundance and chemistry of these organisms help reconstruct ocean temperatures, salinity, and sea ice cover.

[...] Praetorious' team used climate models and found that ocean currents were more than twice the strength they are today during the height of the last glacial maximum around 20,000 years ago due to glacial winds and lower sea levels. While not impossible to paddle against, these conditions would have made traveling by boat very difficult, Praetorius said.

However, the records also showed that much of the area was home to winter sea ice until around 15,000 years ago. As a cold-adapted people, "rather than having to paddle against this horrible glacial current, maybe they were using the sea ice as a platform," Praetorius said.

https://decrypt.co/208614/chinese-tech-giant-alibaba-shows-off-ai-that-can-animate-anyone

If you can't dance, Animate Anyone can make it happen with just one still photo.

https://techcrunch.com/2023/12/04/animate-anyone-heralds-the-approach-of-full-motion-deepfakes/

As if still-image deepfakes aren't bad enough, we may soon have to contend with generated videos of anyone who dares to put a photo of themselves online: with Animate Anyone, bad actors can puppeteer people better than ever.

[...] The new generative video technique was developed by researchers at Alibaba Group's Institute for Intelligent Computing. It's a big step forward from previous image-to-video systems like DisCo and DreamPose, which were impressive all the way back in summer but are now ancient history.

A video showing many different examples is provided.

Original Submission