Recently, AI researcher Simon Willison wanted to add up his charges from using a cloud service, but the payment values and dates he needed were scattered among a dozen separate emails. Inputting them manually would have been tedious, so he turned to a technique he calls "video scraping," which involves feeding a screen recording video into an AI model, similar to ChatGPT, for data extraction purposes.

What he discovered seems simple on its surface, but the quality of the result has deeper implications for the future of AI assistants, which may soon be able to see and interact with what we're doing on our computer screens.

"The other day I found myself needing to add up some numeric values that were scattered across twelve different emails," Willison wrote in a detailed post on his blog. He recorded a 35-second video scrolling through the relevant emails, then fed that video into Google's AI Studio tool, which allows people to experiment with several versions of Google's Gemini 1.5 Pro and Gemini 1.5 Flash AI models.

Willison then asked Gemini to pull the price data from the video and arrange it into a special data format called JSON (JavaScript Object Notation) that included dates and dollar amounts. The AI model successfully extracted the data, which Willison then formatted as CSV (comma-separated values) table for spreadsheet use. After double-checking for errors as part of his experiment, the accuracy of the results—and what the video analysis cost to run—surprised him.

A screenshot of Simon Willison using Google Gemini to extract data from a screen capture video.

Credit: Simon Willison

"The cost [of running the video model] is so low that I had to re-run my calculations three times to make sure I hadn’t made a mistake," he wrote. Willison says the entire video analysis process ostensibly cost less than one-tenth of a cent, using just 11,018 tokens on the Gemini 1.5 Flash 002 model. In the end, he actually paid nothing because Google AI Studio is currently free for some types of use.

Video scraping is just one of many new tricks possible when the latest large language models (LLMs), such as Google's Gemini and GPT-4o, are actually "multimodal" models, allowing audio, video, image, and text input. These models translate any multimedia input into tokens (chunks of data), which they use to make predictions about which tokens should come next in a sequence.

A term like "token prediction model" (TPM) might be more accurate than "LLM" these days for AI models with multimodal inputs and outputs, but a generalized alternative term hasn't really taken off yet. But no matter what you call it, having an AI model that can take video inputs has interesting implications, both good and potentially bad.

Breaking down input barriers

Willison is far from the first person to feed video into AI models to achieve interesting results (more on that below, and here's a 2015 paper that uses the "video scraping" term), but as soon as Gemini launched its video input capability, he began to experiment with it in earnest.

In February, Willison demonstrated another early application of AI video scraping on his blog, where he took a seven-second video of the books on his bookshelves, then got Gemini 1.5 Pro to extract all of the book titles it saw in the video and put them in a structured, or organized, list.

Converting unstructured data into structured data is important to Willison, because he's also a data journalist. Willison has created tools for data journalists in the past, such as the Datasette project, which lets anyone publish data as an interactive website.

To every data journalist's frustration, some sources of data prove resistant to scraping (capturing data for analysis) due to how the data is formatted, stored, or presented. In these cases, Willison delights in the potential for AI video scraping because it bypasses these traditional barriers to data extraction.

"There's no level of website authentication or anti-scraping technology that can stop me from recording a video of my screen while I manually click around inside a web application," Willison noted on his blog. His method works for any visible on-screen content.

Video is the new text

An illustration of a cybernetic eyeball.

Credit: Getty Images

The ease and effectiveness of Willison's technique reflect a noteworthy shift now underway in how some users will interact with token prediction models. Rather than requiring a user to manually paste or type in data in a chat dialog—or detail every scenario to a chatbot as text—some AI applications increasingly work with visual data captured directly on the screen. For example, if you're having trouble navigating a pizza website's terrible interface, an AI model could step in and perform the necessary mouse clicks to order the pizza for you.

In fact, video scraping is already on the radar of every major AI lab, although they are not likely to call it that at the moment. Instead, tech companies typically refer to these techniques as "video understanding" or simply "vision."

In May, OpenAI demonstrated a prototype version of its ChatGPT Mac App with an option that allowed ChatGPT to see and interact with what is on your screen, but that feature has not yet shipped. Microsoft demonstrated a similar "Copilot Vision" prototype concept earlier this month (based on OpenAI's technology) that will be able to "watch" your screen and help you extract data and interact with applications you're running.

Despite these research previews, OpenAI's ChatGPT and Anthropic's Claude have not yet implemented a public video input feature for their models, possibly because it is relatively computationally expensive for them to process the extra tokens from a "tokenized" video stream.

For the moment, Google is heavily subsidizing user AI costs with its war chest from Search revenue and a massive fleet of data centers (to be fair, OpenAI is subsidizing, too, but with investor dollars and help from Microsoft). But costs of AI compute in general are dropping by the day, which will open up new capabilities of the technology to a broader user base over time.

Countering privacy issues

As you might imagine, having an AI model see what you do on your computer screen can have downsides. For now, video scraping is great for Willison, who will undoubtedly use the captured data in positive and helpful ways. But it's also a preview of a capability that could later be used to invade privacy or autonomously spy on computer users on a scale that was once impossible.

A different form of video scraping caused a massive wave of controversy recently for that exact reason. Apps such as the third-party Rewind AI on the Mac and Microsoft's Recall, which is being built into Windows 11, operate by feeding on-screen video into an AI model that stores extracted data into a database for later AI recall. Unfortunately, that approach also introduces potential privacy issues because it records everything you do on your machine and puts it in a single place that could later be hacked.

Credit: Getty Images | Benj Edwards

To that point, although Willison's technique currently involves uploading a video of his data to Google for processing, he is pleased that he can still decide what the AI model sees and when.

"The great thing about this video scraping technique is that it works with anything that you can see on your screen... and it puts you in total control of what you end up exposing to the AI model," Willison explained in his blog post.

It's also possible in the future that a locally run open-weights AI model could pull off the same video analysis method without the need for a cloud connection at all. Microsoft Recall runs locally on supported devices, but it still demands a great deal of unearned trust. For now, Willison is perfectly content to selectively feed video data to AI models when the need arises.

"I expect I’ll be using this technique a whole lot more in the future," he wrote, and perhaps many others will, too, in different forms. If the past is any indication, Willison—who coined the term "prompt injection" in 2022—seems to always be a few steps ahead in exploring novel applications of AI tools. Right now, his attention is on the new implications of AI and video, and yours probably should be, too.

Read full article

Comments

Elise and I recently went on a hike with another couple, and on our way up the mountain I learned from the guy all about how his car has lost $50 grand in value over just a few years. I didn’t think much of it; lots of cars have seen heavy depreciation, especially lately, so I figured he was just exaggerating. But then, when I got home, I recalled the conversation and decided to look up his car: A 2019 Audi E-Tron. And my god was I shocked with what I saw.

Depreciation is a part of life if you buy a new car, with a few exceptions like the Jeep Wrangler, Toyota Tacoma, and pretty much anything bought just before COVID and sold during that car-market nightmare. Among the worst cars when it comes to depreciation are expensive European cars. Buy a new Mercedes S-Class, for example, and you can expect to lose many tens of thousands of dollars in a really short span.

Another segment of the car market that has seen humongous depreciation is electric cars. So what if you blend 1. Expensive German Car with 2. Electric car? Well, you get a depreciation To The Max.

The headline of this article isn’t meant to be a joke, because people losing tens of thousands of dollars is no laughing matter. It can have a huge effect on someone’s livelihood if they end up buying a car whose value tanks just before they have to sell. On one hand, the cars that are depreciating worst are the ones purchased by folks who, at least in theory, can most afford it. On the other hand, I can see how this could blindside someone.

I mean, just look at the reviews of the Audi E-Tron when it came out in 2019; anybody would have thought they were buying a state-of-the-art machine. Motor Trend‘s headline was: “2019 Audi E-Tron Review: What a Way to Glide” and its subheading was “The EV wars are starting in earnest, and Audi has itself a real weapon.” Here’s how Motor Trend described how the E-Tron compared to others in its class, and even described it as “affordable”:

In the showroom wars, the e-tron’s primary enemies are the aforementioned I-Pace and Tesla’s Model X, as well as Mercedes’ upcoming EQC. A little smaller and pricier but quicker and more responsive, the Jag boasts an EPA range of 234 miles. The Model X is by far the costliest of the bunch—when you add desirable options it can soar well past $100K—but it’s also by far the quickest, as it can sprint from zero to 60 mph in 2.8 seconds with the extra-cost Ludicrous Mode. The Tesla also leads with a maximum claimed range of 325 miles. The e-tron, in contrast, is the most “normal” of the trio. Excepting the Mercedes, which starts at $68,895, it’s the most affordable with a base sticker of $75,795, offers a generous 57 cubic feet of cargo space with the rear seats folded down, and while it may not deliver the sizzling straight-line acceleration of the Model X or the halfback-like chassis moves of the I-Pace, it’s designed to charge quickly, glides over the road with unfailing refinement, and is built with battery longevity and unflagging performance as priorities.

CNET’s review was similarly glowing:

In that spirit, today I’d like to celebrate the $74,800 Audi E-Tron, a car I’ve been appreciating for nearly two months now. Audi’s first production electric car takes a subtly different but distinctive path to all-electric glory. There’s nothing ludicrous about this EV SUV and, frankly, I couldn’t care less what its Nurburgring lap time is. What I do know is that this is among the most comfortable, most soothing cars I’ve ever had the privilege of driving, and that makes it something special.

[…]

Configured this way, at $77,290 including destination, this is not a cheap car. But it offers luxury appointments on par with similarly priced premium machines, plus the added benefits of that smooth, quiet, maintenance-free EV lifestyle.

It’s hard to argue with this practicality or with Audi’s wholly rational approach to building an EV. The e-tron is a competent, well-engineered piece that makes few compromises compared to Audi’s gasoline-powered SUVs. But at this point, buying an electric car—especially one that starts at $75,795—is still a bold, somewhat irrational choice, a decision to go against the grain…. But we’re not far enough into the EV era to know what’s right and wrong….

The Tesla Model Y launched just a few months after the Audi E-Tron, and then as other competitors like the Kia EV6, Hyundai Ioniq 5, and a boatload more joined in on the fun, prices tanked. Some of this is a result of EVs being seen as appliances whose value is determined predominantly by a single attribute (range), some of it is a result of early adopters having already bought their EVs and skeptics hesitating to make the plunge given infrastructure issues, part of it is a result of political uncertainty/rebates, and part of it is a result of the crazy price-cuts from Tesla.

In any case, look at what a 2019 Audi E-Tron — whose MSRP was $75,795 for the Premium Plus model and $82,795 for the Prestige model — costs nowadays:

Those cars only have about 50,000 miles on them, meaning they’ve lost over a dollar a mile! My god, a dollar a mile. If I knew my car would lose a dollar of value every mile I drove it, I’d sell it immediately. Check out the value trend over time:

Especially in the last couple of years, after the post-pandemic price-jump, the market has not been friendly to the E-Tron:

It’s worth pointing out that E-Trons and other EVs were eligible for the $7,500 EV rebate. Not to mention, I bet plenty of these were leased, and others were sold with money on the hood. But for those who bought them outright in 2019, even with the EV rebate: Yikes!

Here’s the gasoline-powered Audi Q7, for reference:

Now, you might be thinking that lots of expensive German cars depreciate a lot, and that’s definitely true. A 2022 Mercedes S-Class started between $110,000 and $120,000, and look at how cheap they are now:

That’s about 50 grand in just two years and 35,000 miles! Yikes! But even the mighty S-Class has nothing on the electric version of the S-Class, the EQS. That car started at $102,310 for the EQS 450+ and $119,110 for the EQS 580 4MATIC. (Less the EV rebate).

Now let’s have a look at what these machines are trading for:

That’s among the cheapest ones I’ve seen. Only $39,000 for a car that started at over $100 grand just two years prior! Surely this thing is flying off the shelf, right?

Apparently not! Here’s another EQS that lost 60 grand in 40,000 miles ($1.50 a mile):

And here’s an EQS 580 for good measure (this one actually being sold by a Mercedes dealership). These started at over $119,000, so this is also a car that lost over 60 G’s:

Just look at the pricing trends of the EQS:

Here’s

Here’s the S-Class, in case you’re curious (sorry about the scale from Cargurus):

As you can see, those expensive German cars are depreciating quickly, but the electric expensive German cars are losing value violently.

Yikes!

Thank goodness so many of these were leased or purchased with lots of money on the hood, and hopefully all of them got the $7,500 federal rebate, along with a potential $4000 rebate for the buyer of the used car (though that only applies to sub-$25,000 vehicles).

All Images: Cargurus (Unless otherwise specified)

The post If You Bought A New Audi E-Tron Or Mercedes EQS: I’m So Sorry appeared first on The Autopian.

Measles, whooping cough, polio, tetanus—devastating and sometimes deadly diseases await comebacks in the US as more and more parents are declining routine childhood vaccines that have proved safe and effective.

The vaccination rates among kindergartners have fallen once again, dipping into the range of 92 percent in the 2023–2024 school year, down from about 93 percent the previous school year and 95 percent in 2019–2020. That's according to an analysis of the latest vaccination data published today by the Centers for Disease Control and Prevention.

The analysis also found that vaccination exemptions rose to an all-time high of 3.3 percent, up from 3 percent in the previous school year. The rise in exemptions is nearly entirely driven by non-medical exemptions—in other words, religious or philosophical exemptions. Only 0.2 percent of all vaccination exemptions are medically justified.

The new stats mean that more parents are choosing to decline lifesaving vaccines and, for the fourth consecutive year, the US has remained below the 95 percent vaccination target that would keep vaccine-preventable diseases from spreading within communities. In fact, the country continues to slip further away from that target.

Based on data from 49 states plus the District of Columbia (Montana did not report data), 80 percent of jurisdictions saw declines in vaccinations of all four key vaccines assessed: MMR, against measles, mumps, and rubella; DTaP, against diphtheria, tetanus, and pertussis (whooping cough); VAR, against chickenpox; and polio.

Vulnerable kids

Coverage for MMR fell to 92.7 percent in 2023–2024, down from 93.1 percent in the previous school year. That means that about 280,000 (7.3 percent) kindergartners in the US are at risk of measles, mumps, and rubella infections. Likewise, DTaP coverage fell to 92.3 percent, down from 92.7 percent. Polio vaccination fell to 92.6 percent from 93.1 percent, and VAR was down to 92.4 percent from 92.9 percent.

The declines can partly be explained by the 3.3 percent exemption rate, which is up 0.3 percent from 2022–2023. Since then, 41 states and DC saw increases in the number of kindergartners with one or more vaccine exemptions. Now, 14 states have exemption rates above 5 percent, meaning that even if they vaccinated all non-exempt children, the state would still not be able to reach the 95 percent target. By contrast, during the 2020–2021 school year, there were only two states that couldn't make the 95 percent target due to exemptions.

Of course, these are just big-picture numbers for the country; vaccination rates vary by state, town, zip code, and community. Any areas with pockets of low vaccination rates are susceptible to outbreaks of vaccine-preventable infections. The more susceptible people cluster together, the easier it is for a contagious disease to gain a foothold and spread widely. In recent years, the country has seen upticks in measles outbreaks and whooping cough cases, for instance.

But, even on larger scales, some of the current US numbers are worrying. In Idaho, for example, vaccination rates for the four key vaccines hover around just 80 percent, and the exemption rate in the state is a staggering 14.3 percent.

As the CDC researchers concluded in their analysis: "Among kindergarten students, vaccination coverage continues to decline as exemptions increase, setting the stage for accumulation of clusters of undervaccinated children, which can lead to outbreaks."

The analysis didn't explore the reasons for the declining vaccination rates, but the researchers briefly speculated that the slide toward preventable outbreaks could be due to COVID-19 vaccine hesitancy bleeding over to routine vaccinations, or simply backlash to vaccine mandates during the pandemic. Regardless, the researchers pointed to evidence-based strategies to boost vaccination, including school-based vaccination clinics, enforcement of school vaccination requirements, and strong recommendations from health care providers for routine vaccinations.

Read full article

Comments

I think it’s safe to say that one of the worst driving experiences one can have today is behind the wheel of a large motorhome, like a Class A. I’m not saying this to just be a jerk, that’s just how it is: you can’t have something that is essentially a wheeled mansion and expect it to handle like a wheeled cottage. Motorhomes are also some of the least efficient vehicles you can buy, thanks to the fact that you’re driving around a house on wheels, complete with beds, a kitchen, and, yes, a toilet. They guzzle gas like with all the grace and restraint that a puppy would show if dropped it into a bowl of taco meat. There’s definitely room for improvement, and I’m happy to say that I think I actually experienced some of that improvement a few weeks back when I got a chance to drive the imaginatively named THOR Test Vehicle.

Okay, yes, it was almost a month ago that I went out to the Home of Studebaker, South Bend, Indiana, on the invitation of THOR and EV delivery truck maker Harbinger to try out a new idea that, in hindsight, seems kind of obvious: a Motorhome with a hybrid drivetrain. Yes, yes, it’s been weeks and I haven’t written it up yet, but I’m doing it now! I’ve been busy!

Anyway, let’s talk about this concept in broad strokes just for a moment. Your conventional modern gas-powered Class A (or even a Class C, really) motorhome gets about 5 to 10 mpg, and more than 10 is only possible in pretty much ideal circumstances that you’ll likely only see when you notice the unicorns flying all about you in the sapphire sky.

The Idea

The idea of adapting an electric drivetrain to such a platform has considerable appeal; the inherent torque of an electric motor would help a motorhome’s normally sub-glacial acceleration substantially, and the greater efficiency afforded by an electric drivetrain can’t be ignored.

However, there is an Achilles’ heel here: batteries. Modern battery tech, as advanced as it is, just doesn’t have the energy density to allow for a battery that can give an RV appreciable range. There’s just so much mass involved, and tests of Winnebago’s prototype electric RV have found that a real-world range is about 80 to 90 miles, and for every hour and a half driven, you’ll need to spend at least 45 minutes charging. It’s not great.

The whole point of a motorhome is to go places, and being limited to driving only an hour and a half or so just before having to stop and charge for an extended time doesn’t meet that requirement. An all-electric motorhome is just not a viable solution, at least not yet.

But! There’s a compromise, and that’s a hybrid. A hybrid – in this case a series hybrid, where only the electric drivetrain moves the vehicle and the combustion engine is just used to generate electricity – incorporates the efficiency and high torque of an EV drivetrain with the flexibility and energy density of gasoline.

If you want a walk-through of the THOR Test Vehicle’s chassis and drivetrain, adapted from Harbinger’s electric delivery vehicle chassis, I’d like to invite you to listen to Harbinger co-founder and CEO John Harris explain it all to me:

 

View this post on Instagram

 

A post shared by The Autopian (@theautopian)

Driving A Gas-Powered Motorhome

It’s fascinating stuff, but at the moment we’re here to talk about what the THOR Test Vehicle was like to drive, which I promise, I’m getting to. To get a real sense of the context, arrangements were made for me to drive a new, conventional gas-powered Class A motorhome, an Entrgra Coach Vision:

This elegant beast was, I believe, somewhere around 30 feet, built on a Ford F53 chassis, driven by a 7.3L V8 335 hp gasoline engine making 468 foot-pounds of torque. This is a big, heavy machine, weighing probably somewhere around 20,000 pounds. I mean, you can’t be shocked by that when you realize this thing has a kitchen with marble countertops and tile:

It’s a house on the back of that thing, which is the primary reason why driving one of these things is so, well, terrible. This one I drove is one of the best ones out there, but driving it still feels like you’re trying to guide a hippopotamus on rollerblades by yanking on its ears.

Keeping it in its lane is tricky, because the thing just about fills up an entire lane of traffic completely. If you’re on a road with a prominent crown, keeping the thing tracking straight takes real work. It’s slow to accelerate, slow to stop, and is incredibly top-heavy. Driving a motorhome is something you just have to white-knuckle and endure until you get to your destination, where you can finally relax and enjoy your motorhome.

Incredibly, the driving experience isn’t appreciably better than what driving my old 1977 Class C Dodge Tioga motorhome was like. All of the shittiness felt very familiar – the slowness, the top-heaviness, the nervous lane-keeping, the ponderousness, the constant sub-surface fear that you’re one wrong arm motion away from being upside-down or creating massive rooster tails of sparks as you get too close to a Jersey barrier or something, it’s all still there.

Let’s just say that driving a big RV is no picnic.

Driving The World’s First Hybrid RV

Okay, now the whole point of why I’m here: to drive the first hybrid motorhome. At first glance, it’s not all that different than the gasoline one; the scale and shape are about the same, but there are some big clues that we’re looking at a prototype machine: the interior is only partially built-out, but there is furniture and stuff inside there – this isn’t some empty cargo van, after all.

Also, if you look carefully, you can see the grille is 3D printed, which is a clever solution:

So what does it feel like to drive? Well, I should qualify first that I didn’t get to drive it on real streets, just a little track delineated by traffic cones in a parking lot. I was a little disappointed by this, but I did find that the key elements of what this hybrid RV is like to drive still came out even on this tiny track.

The first thing you notice is the acceleration, which is shockingly decent considering the massive bulk of everything. It’s not whip-your-head-back sports car fast, but it’s still surprisingly quick, given the context. You’re just not used to a whole furnished apartment moving with such urgency.

The bigger difference, though, has to do with weight, or, really, where the weight is. Electrified vehicles are always heavy, because batteries are heavy as hell, and while RVs are already absurdly heavy, in the case of the THOR Test Vehicle, the bulk of that weight is in dense packs of batteries way down low, in the chassis.

This changes the dynamic of how these big beasts drive, because the unsettling top heaviness, that always makes driving a motorhome feel like running with a stack of books or a large vase full of chili on your head, is pretty much eliminated. The hybrid motorhome felt planted, I could take sharp turns without encustardizing my pants, and overall the driving experience was significantly and viscerally better.

Here, watch some video of me driving it, which should give a good idea of what I mean, especially since they let me do some donuts, which I’d never attempt in a conventional motorhome:

 

View this post on Instagram

 

A post shared by The Autopian (@theautopian)

I don’t yet know what sort of fuel economy advantages the hybrid will offer, but I suspect it should be significant. If this can return numbers in, say, the mid-teens, that would be stellar, for a motorhome.

Then, there’s also the considerable advantage of a hybrid in that all that electricity stored in those big, heavy batteries should eliminate the need for a noisy, smelly generator. That’s a huge deal! When you’re out in the middle of beautiful nowhere you want to be able to have electrical power for making margaritas and your various massagers, but who wants to break the numinous silence with the rattle of a diesel generator? Not you! You don’t want to scare away all the sasquatches!

There’s plenty of power in those batteries for all the household needs, and with integrated rooftop solar panels, there’s the possibility of getting some free energy back into the system. Solar is generally useless for actually adding range, but for topping up power from household demands, it can be helpful.

The idea of a hybrid motorhome just makes sense. The advantages to fuel economy and eliminating the need for a generator are incredibly compelling alone, and the driving dynamics improvements are a massive added bonus. I think this needs to happen, and should be one of the most significant and meaningful changes to the otherwise (let’s be honest) pretty stagnant motorhome industry.

I’m impressed. I’ll be curious to see these come to market, and get a better idea of the advantages they bring.

The First Hybrid Motorhome Ever Offers 500 Miles Of Range And I’m Going To Drive It

Here’s How Harbinger And Thor Industries Plan To Change Camping With Motorhomes That’ll Go Around 250 Miles On A Charge

What The Hell Is Harbinger And Why Are They Showing A Boring Delivery Van At The Detroit Auto Show?

The post What It’s Like To Drive The World’s First Hybrid RV appeared first on The Autopian.

Google has signed a deal with California startup Kairos Power for six or seven small modular reactors. The first is due in 2030 and the rest by 2035, for a total of 500 megawatts. [Google blog; press release]

Amazon has also done three deals to fund SMR development. [Press release]

These are deals for vaporware.

Traditional nuclear power plants are massive undertakings. The idea of SMRs is that they’ll be simpler and cheaper to build. Google could put an SMR right next to the data center. 

Everything about commercial SMRs sounds wonderful! Except they don’t really exist yet. “SMR” is press release jargon, not something you can buy. [IAI, 2023]

Only three experimental SMRs exist in the entire world — in Russia, China, and Japan. The Russian and Chinese reactors claim to be in “commercial operation” — though with their intermittent and occasional hours and disconcertingly low load factors, they certainly look experimental. [OECD, PDF; Power]

Like general AI, SMRs are a technology that exists in the fabulous future. SMR advocates will talk all day about the potential of SMRs and gloss over the issues — particularly that SMRs are not yet economically viable.

Kairos doesn’t have an SMR. They have permission to start a non-powered tech demo site in 2027. Will they have an approved and economically viable design by 2030? [Power]

SMRs would be very nice to have — a mass-producible power plant that isn’t dependent on custom engineering for each site, doesn’t pump out CO2, and runs 24/7. The technology is within the range of known science and engineering too, unlike general AI. The US Department of Energy subsidizes SMR development.

The hard part is the economics. You can’t charge 12c–16c/kWh wholesale when wind and solar cost 4c/kWh down to zero. [CleanTechnica, 2023]

SMRs were inspired by nuclear submarine reactors — which are very small and very modular. But the US Navy has the advantage of not having to worry so much about cost.

SMR for commercial power absolutely has to worry about cost. Nuclear reactors are cheaper per kWh when they’re bigger — that’s just the physics. The cost is why actual power companies don’t invest in SMR development. [Economist, archive]

Then there’s the regulatory hurdles. The public expects stringent regulation of nuclear power. An SMR must be licensed by the US Nuclear Regulatory Commission. SMR approvals will take just as long as large reactors do. It takes years from a new working prototype to production — and meanwhile, wind, solar, and batteries get even cheaper. [NRC; NRC]

Nuclear fuel with more than 5% uranium-235 is very restricted in the US. SMR designs tend to use medium-enriched uranium (10–20% U-235) and a few use highly enriched uranium (over 20% U-235). The only country producing highly enriched uranium at present is Russia. Highly enriched uranium is on the path to atom bombs, so permission is unlikely. [NRC; NRC]

Small reactors produce more high-level waste (spent fuel) per kWh than larger reactors.

Will mass production of reactor parts significantly reduce costs? This would rely on a single design being a worldwide success.

There are 98 SMR designs in development around the world — 18 of those in the US — but these are all just on paper as yet.

Not existing yet is an important feature of SMRs. You’ll often see conservative politicians floating SMRs as a fabulous future solution that will fix everything, so we can let fossil fuel power keep running a few years longer. [CleanTechnica, 2023]

Australian opposition leader Peter Dutton is pushing SMRs hard of late — in a way that would just happen to prop up his coal mining donors for another decade — and is ignoring everyone asking him for numbers on costs. He’d likely get half the Greens onside if his numbers worked! [Conversation]

Google and Amazon are doing deals for SMRs that might never be delivered because the big techs are looking into every possible power source for their data centers — especially in the AI bubble. And greenwashing is cheaper than action. [Google blog; Power]

Following an investigation, Elon Musk's X has won its fight to avoid gatekeeper status under the European Union's strict competition law, the Digital Markets Act (DMA).

On Wednesday, the European Commission (EC) announced that "X does indeed not qualify as a gatekeeper in relation to its online social networking service, given that the investigation revealed that X is not an important gateway for business users to reach end users."

Since March, X had strongly opposed the gatekeeper designation by arguing that although X connects advertisers to more than 45 million monthly users, it does not have a "significant impact" on the EU's internal market, a case filing showed.

A gatekeeper "is presumed to have a significant impact on the internal market where it achieves an annual Union turnover equal to or above EUR 7.5 billion in each of the last three financial years," the case filing said. But X submitted evidence showing that its Union turnover was less than that in 2022, the same year that Musk took over Twitter and began alienating advertisers by posting their ads next to extremists' tweets.

Throughout Musk's reign at Twitter/X, the social networking company told the EC, both advertising revenue and users have steadily declined in the EU. In particular, "X Ads has a too small and decreasing scale in terms of share of advertising spend in the Union to constitute an important gateway in the market for online advertising," X argued, further noting that X had a "lack of platform power" to change that anytime soon.

"In the last 15 months, X Ads has faced a decline in number of advertising business users, as well as a decline in pricing," X argued.

In another case filing in the EU, X argued that it's also much smaller than its competitors when it comes to monthly active users in the EU. According to X, it's about 133 percent smaller than Facebook or Instagram, 60 percent smaller than LinkedIn, and 27 percent smaller than TikTok. While the EU noted that X's data "shows considerable discrepancies" based on the argument the app is making, X had submitted enough to show "low and decreasing user engagement."

Additionally, X argued that unlike other DMA gatekeepers like Meta or Google, X primarily deals in brand advertising, "as opposed to direct response advertising which requires superior targeting and measurement of advertising tools." And because X "only offers first-party advertisement on the X online social networking," X can't crunch user data like other gatekeepers tracking users off-platform. This means that X "does not have the same possibility" of "verifying advertising effectiveness across several first and third-party properties" like other companies seemingly posing a greater risk of monopolizing the ad industry, X argued.

"Based on this evidence," as well as "the low and decreasing scale of usage by business users," the Commission concluded that "X Ads is not an important gateway for business users to reach end users."

X may become a gatekeeper if ad boycott ever ends

While this is clearly a win for X—as other big tech companies wish they could avoid DMA compliance too—the continued ad boycott of X over fears of extremist content proliferating on the platform is seemingly keeping X from achieving some of Musk's biggest ambitions since purchasing the platform.

Musk has vowed to turn X into an everything app designed to be the only app that users would need. But progress on that front has seemed to slow after Musk claimed that everyone would be banking on X by the end of 2024.

Instead of rolling out promised new features, X has seemingly been focused this fall on wooing back some major advertisers after Fidelity recently estimated that X is worth about 80 percent less than when Musk initially purchased the app.

Last week, X announced that Unilever would resume advertising, promising that forming that partnership was just "the first part of the ecosystem-wide solution" for advertisers on X.

"We look forward to more resolution across the industry," X said.

For X, wooing back more advertisers might be hard as long as it maintains a lawsuit targeting advertisers for boycotting the platform. In that legal challenge, X accused advertisers of conspiring "to collectively withhold billions of dollars in advertising revenue."

But many advertisers have maintained that Musk's content moderation changes at X have caused more harmful content to spread on the platform, and some even think Musk is part of the problem by seemingly amplifying misleading or objectionable content, including one post condemned by the White House.

Over the weekend, The New York Times published a deep dive showing how once-suspended Twitter conspiracy theorists that Musk reinstated on X (like Alex Jones or Marjorie Taylor Greene) after purchasing Twitter have benefited from Musk engaging with their posts—concluding that "Musk plays a big role in what spreads" on X.

Whether Musk can turn X's revenue around is still unclear as ad revenue hits new lows, and his everything app dream that will supposedly spike revenue seems to be deprioritized. The EU Commission has said that it will continue monitoring changes at X in case ad revenue bounces ever back to a level that would put DMA gatekeeper status back on the table.

Read full article

Comments