F'ing drive by trolls. We’ve got a peer-reviewed NEJM study showing meaningful vision restoration for people blinded by macular degeneration — the first of its kind — and this troll wants to derail the conversation by griping about billing codes. The “mandatory payment” this troll is invoking isn’t some blank check to industry — it’s a temporary (at least they got the six months part right) Medicare coverage rule (Transitional Coverage for Emerging Technologies) designed to make sure new devices aren’t withheld from patients while CMS figures out how to price them. It’s about access, not profiteering.

Because physics still matters. I think you missed a chapter in your RF engineering text. RTL-SDR gear is a fun toy, but it’s built around an 8-bit tuner with a 2.4 MHz bandwidth ceiling. The GEO Ku-band transponders in this study are running symbol rates up to 70 MS/s — two orders of magnitude higher. To capture that cleanly, you need a tuner and demodulator that can lock onto and maintain DVB-S2X carriers, apply proper forward-error correction, and handle multi-megabit baseband I/Q streams without dropping bits. That’s what the TBS-5927 and similar professional cards do, and that’s where most of that $750 went. The rest covers a 110 cm dish, a motorized mount for precise orbital alignment, and a low-noise block converter with the gain and linearity to hold lock across dozens of satellites.

Sure, you can “do it in software” — after the hardware front-end delivers a clean, stable stream. But a $30 RTL-SDR dongle isn’t going to demodulate a 70 megabaud DVB-S2X signal any more than a crystal radio can pick up Starlink. I love a good FFT as much as the next RF engineer, but you can’t do math on noise, which is exactly what your toy SDR would be doing.

That’s security Darwinism, and it’s nonsense. This is modern civilization, not Hobbes’ state of nature -- red in tooth and claw, where life is nasty, brutish, and short. Civilization exists precisely so people don’t have to live like prey animals, constantly scanning for predators before they drink at the watering hole. We built laws, standards, and infrastructure so that ordinary citizens can draw a bath without worrying about contamination, flip a light switch without checking the voltage, and withdraw cash from an ATM without wondering if the signal is being broadcast into space because their bank decided that purchasing the crypto module for their satcom link was too expensive.

When those systems fail because the professionals responsible for them treat encryption as optional, or their industry customers see it as too expensive, that isn’t survival of the fittest -- it’s dereliction of duty at the societal level.

A geostationary satellite is a backbone of modern commerce and governance, carrying financial, medical, and logistical lifeblood. Leaving it unencrypted is the moral equivalent of building a city and deciding locks are too expensive.

Civilization runs on invisible trust: clean water, stable voltage, and secure communication. Strip that away, and we’re not clever apes with Wi-Fi — we’re just frightened hominids on the veldt, huddling around the fire, waiting for the next predator to find us.

This is just the latest in a long line of network-security papers I’ve read through the years that are more than a little disturbing. Using less than $750 in consumer hardware — a motorized Ku-band dish, a tuner card, and open-source software — they intercepted real, unencrypted IP traffic from 39 geosynchronous satellites and 411 transponders. What they found was pretty damning: plaintext phone calls, SMS messages, ATM authentication traffic, power-grid control data, and corporate inventory systems, all broadcasting across North America like it was still 1995.

Half the links they scanned had no encryption at all. Even where link-layer “encryption” was enabled, many systems used null ciphers — meaning encryption was technically on but functionally absent. Only about six percent of the networks used IPSec.

The reasons are depressingly familiar: vendor license fees for crypto modules, performance fears (20–30 % throughput loss), ancient export-control defaults, and an industry that still treats satellites as “internal links” rather than global broadcasts. In short, we’re back to security through obscurity — the same mindset Microsoft tried and abandoned a quarter century ago when they discovered that popularity without protection is just an invitation to be pwnd. Satcom companies are going to face the same music Redmond did, but they don't have the we-are-new-at-this excuse Redmond had.

If a grad-student budget and a clear view of the sky are enough to eavesdrop on military, telecom, and financial networks, that’s not a hacker exploit — that’s a structural failure of modern infrastructure. We spent decades encrypting the Web to stop fiber-tap surveillance; meanwhile, a significant fraction of global backhaul traffic is still flying overhead in the clear.

Encryption isn’t expensive anymore. The real cost is complacency.

They not factored in because they are not PHEVs. You are conflating two very different types of hybrids. Toyota’s standard hybrids (HEVs) are efficient because they’re light, purpose-built, and never plug in; the T&E report targets plug-in hybrids (PHEVs)—heavy SUVs that carry both a full engine and a large battery. In Europe those cars look clean on paper because lab tests assume 80 % electric driving, but real-world data from on-board fuel-use monitors shows only about 25–30 %. Their “electric” mode still burns fuel since underpowered motors make the engine kick in. Even the range-extender types (EREVs) average roughly 6.7 L/100 km once the battery’s empty. The problem isn’t with Toyota-style hybrids -- they are awesome and the EU and the rest of the planet could use more of them -- it’s with the regulatory sleight of hand that lets plug-in SUVs pose as zero-emission vehicles while emitting like ordinary petrol ones.

You can holster the flamethrower — it’s not Big Oil behind this one. Transport & Environment is a Brussels-based NGO that’s been pushing for faster EV adoption for decades, funded mostly by climate-philanthropy foundations like Sequoia and Hewlett, not Exxon.

Their motive isn’t to protect fossil fuels; it’s to stop European carmakers from gaming emissions rules by selling plug-in hybrids that look green on paper but run on petrol in practice. Think of it as a pre-emptive strike against a new “SUV loophole” — the same kind of regulatory shell game Detroit played in the 1990s, putting their gas-guzzlers into the EPA's "sport utility vehicle" category, which had far lower emissions and mileage standards.

So if there’s propaganda here, it’s not oil-funded — it’s policy-driven, meant to keep EU regulators from being conned by their own automotive industry, the way Ford and GM gamed the EPA. These clowns can't innovate anything -- even their shell-game tactics are thirty years old.

That’s because you’re talking about a conventional hybrid, not a plug-in hybrid (PHEV). They’re entirely different animals. The study wasn’t about Toyota-style hybrids that self-charge through regenerative braking — those are usually quite efficient.

PHEVs, by contrast, carry both a large battery and a full-size engine. On paper they look brilliant because EU lab tests assume they’re driven mostly on electricity. In real life, most aren’t charged often, especially company cars with free fuel cards. So you get the weight penalty of a battery, the emissions of a combustion engine, and lab results that only a marketing department could love.

The report is calling out that discrepancy, not criticizing efficient non-plug-in hybrids like yours.

Transport & Environment just dropped a report arguing that plug-in hybrids (PHEVs) are a “smoke screen” for real emissions cuts. Their data show that Europe’s latest crop of long-range hybrids emit nearly five times their official CO ratings in real-world use — mostly because the test procedure assumes drivers use electric mode 80% of the time, when the real number is closer to 25–30%.

If that sounds familiar, it should. When the U.S. tightened fuel-economy and emissions rules in the 1990s, Detroit didn’t build cleaner cars — it built them into different categories. SUVs and “light trucks” were regulated under looser standards, so Ford and GM flooded the roads with Explorers and Suburbans that were technically compliant but strategically evasive. The profits rolled in, the CO2 didn’t roll down, and the rest is smog-choked history.

The European version of that dodge is happening right now. Carmakers like BMW, Mercedes, and VW are using flattering lab assumptions to make heavy, powerful PHEVs look virtuous on paper while burning fuel like conventional SUVs on the road. Since EU fleet-average CO2 limits are calculated from these paper numbers, the industry can appear to hit its targets while quietly delaying a full transition to battery EVs.

T&E’s report isn’t fossil-fuel propaganda — it’s the opposite: a warning shot to regulators not to let PHEVs become Europe’s next regulatory fig leaf. The goal is clear: close the loophole before “hybrid compliance cars” wreak the same havoc in the EU that SUVs did in the US.

...so why pretend otherwise? I’ll grant the Drexel Food Lab did their homework — random sampling, coded cups, controlled brewing — but they missed an obvious confounder: coffee, like beer, is an acquired taste.

Nobody actually likes coffee the first time they drink it. It’s bitter, acidic, and smells better than it tastes — until the brain rewires itself to expect the jolt that follows. That means a random sample of “ordinary people” isn’t measuring coffee quality; it’s measuring how well the sample population’s taste buds have been trained to tolerate bitterness.

That would also explain why the world’s largest coffee chain sells more sugar than coffee, and why the coffee aisle in American supermarkets resemble chemistry labs: instant powders beside rows of creamers, sweeteners, and caramel-pumpkin flavor syrups. If you need a caramel macchiato to make your caffiene drinkable, the problem isn’t the roast or the grind on the bean -- it’s evolution telling you to drink water. Instant coffee may have “won,” but it’s a bit like saying lite beer beats stout because more people finished the cup. Taste isn’t democratic; it’s adaptive. Palates evolve for survival, not subtlety.

Easier to clean? I’ve been running the same Bodum cafetière for fifteen years — a veteran of a friend’s coffee shop before it enlisted in my kitchen. I brew a liter a day, rinse the plunger screen, dump the grounds, and call it good. Every few weeks it rides along in the dishwasher if there’s space on the bottom rack. No rituals, no gadgets, no papery puck to admire. One time, I accidentally left the grounds in for a whole week, and used some Cafiza (on the advice and courtesy of that same friend) to rinse the residue out, but that is as elaborate as I ever get with it. I’ve seen the AeroPress brigade in the wild — the plunger, seal, filter cap, paper disc, stir stick, funnel and somehow that’s the easy one to clean? We obviously are working from conflicting definitions of "easy". :) I’m all for clever engineering, but I’ll stick with the two-piece brew system that can survive a decade and a half of my caffeine habit and still look (and taste) good.

NuScale did get further along than anyone else — their VOYGR-6 design is still the only SMR to have full NRC design certification. The problem wasn’t the tech, it was the business model.

Their flagship project, the Carbon Free Power Project (CFPP) in Idaho, was supposed to build a 462 MW plant for the Utah Associated Municipal Power Systems (UAMPS). When the estimate jumped from $58 to $89 per MWh and construction costs ballooned past $9 billion, several municipal utilities dropped out. By late 2023, subscriber commitments fell below the minimum needed for financing, and UAMPS pulled the plug.

NuScale’s design approval is still valid, and they’re shopping it to new partners (think DoD micro-grids and industrial sites), but CFPP was the cautionary tale: first-of-a-kind nuclear still carries first-of-a-kind price tags. Until someone builds one under budget, “being further along” on paper doesn’t guarantee shovels in the ground.

AI slop isn't going to kill the internet. But clickbait headlines surely will. :(

So, according to The Telegraph (amplified by Futurism), Western CEOs are “terrified” after touring Chinese “dark factories” where robots hum along 24/7 assembling EVs in the dark. Cue the usual panic about the West falling behind.

History disagrees.

I remember seeing footage of Datsun’s factory robots -- yes, Datsun -- on a classroom film reel at the tail end of the Nixon administration. The arms were orange, the sparks were blue, and every news anchor swore Japan’s fully automated plants would bury Detroit. Fifty years later, Detroit’s still here, Japan’s still here, and we’re still buying trucks made by humans and robots in a mix that makes sense for economics, not headlines.

Automation is an evolutionary process, not a coup d'etat.

China’s ramp-up is impressive, sure — but it’s also necessary. They’re automating to offset an aging workforce, not to enslave ours. And anyone who’s actually worked in manufacturing knows that “dark factories” are mostly PR stunts. Robots weld, glue, and repeat; humans still debug, calibrate, and fix. Until robots can argue with an OSHA inspector, I am pretty sure the US is safe from Chinese auto-bots. And remember: China was, and still is, overwhelmingly agrarian. After the Communists consolidated power in 1949 (following their civil war with Chiang Kai-shek’s Nationalists), heavy industry came in fits and starts, and the real export-driven manufacturing surge arrived only after the reform era began in the late 1970s. The reality in China is they’re still trying to compress a century and a half's worth of industrial-era catch-up into a few decades. Good luck with that -- it is not going to happen in my lifetime, or my grandkids' either.

Jim Farley’s own comments in The Verge interview (the one Futurism cherry-picked) undermine the whole clickbait line of this story: the problem isn’t Chinese robots — it’s Western complacency, legacy IT systems, and the assumption that cost equals competitiveness. Ford isn’t surrendering to robots; it’s rediscovering the art of making things efficiently -- a curiously pragmatic take, if he's supposed to be "panicking" over chinese industrialization, eh?

So maybe we can skip the “China is terrifying” headlines and focus on something simpler: building things smarter, teaching kids to weld and code, and ignore these BS “end of the West” stories that exist only to draw clicks.

Schneier and Raghavan argue that the same feedback loops that make human agents powerful—observe, orient, decide, act—also make agentic AI vulnerable when fed poisoned or adversarial data. In their words, “The adversary isn’t inside the loop by accident; it’s there by architecture.” Their point is well-taken: modern LLM agents have no privilege separation between data and control. That’s a security nightmare, and their proposed “integrity-first” rethink is overdue. But it just leads to bigger questions: How do you checksum reality? What would a semantic firewall actually look like?

One promising approach is applying Riemannian geometry to the probability distributions that make LLMs work. This approach can’t guarantee truth, but it can move us forward. "How do you checksum a thought?” becomes “check the position of that thought in Hilbert space.” In the vectorized token chains that comprise an LLM's representation space, meaning is derived from the distance and direction to other token chains in the LLM's embedding space. Schneier's semantic firewall could reject instructions that are too far away from known (read: secure) instructions, project out known injection directions, and require multi-encoder agreement before making tool calls, and fall back to a slow verifier tied to provenanced sources. That doesn’t solve everything—geometry can’t certify the world—but it raises the cost of corruption and turns ‘agents on web-scale untrusted data’ into a differential geometry homework exercise.

Agentic AI is a checkpoint that needs to be passed on the road to AGI. But I don’t think these issues are showstoppers for AGI—just growing pains. Humans have been operating on untrusted data since we learned to gossip. We built cognition atop ambiguity, bias, and deception, and somehow we still make progress. Agentic AI will have to do the same. If anything, the essay highlights that the path to AGI isn’t blocked by security—it’s illuminated by it. The challenge isn’t to keep adversaries out of the loop, but to make the loop resilient to untrusted reality.

As a teenager in the 1970s, I had a poster of the black Lamborghini Countach from "Cannonball Run" on my bedroom wall, right next to the Farrah Fawcett poster. I am a big fan of Italian go-fast devices (I am the proud owner of a 2008 Ducati 1098R) so when Ferrari announced their first all-electric sports car, I felt a pang of sympathy for the carburetor crowd. But reading the comment threads is like watching a group therapy session for internal-combustion nostalgics -- denial, anger, bargaining, the whole Kubler-Ross playlist. With Ferrari joining Porsche and Mercedes-Benz, the writing is on the wall for the ICE fan base.

F1 is already tip-toeing toward electrification while pretending not to. The new 2026 power units scrap the MGU-H, boost the MGU-K to 350 kW, and push the electric share of total power to parity with the ICE -- half of every lap powered by electrons. Synthetic fuels are the next fig leaf, but "carbon-neutral combustion" is just the hospice phase of gasoline. The engineering talent is already moving where the current flows. F1's ICE addiction isn't quite at the Cheyne-Stokes breathing phase yet, but it's coming.

The Elettrica doesn't sound like an F355 at redline -- but the clever engineers at Maranello solved that marketing issue by harvesting power train vibrations and amplifying them via resonators so that the Elettrica sounds like it's breathing, not roaring. Let's be honest, most ICE fans think loud = powerful. F1's DNA was never about sound; it was about speed, style, sex -- and extracting impossible performance from finite physics. That's what Ferrari, Mercedes, and Porsche are doing now -- they're just swapping fuel chemistry for battery chemistry.

The day an EV posts a faster lap at Imola than an ICE Ferrari, the old mythology collapses. When an EV finally wins an F1 race -- however they rebrand the series by then -- it won't be a betrayal of the sport's heritage. It will be proof of it. It'll be the moment when the gods of speed and style send an unmistakable message -- evolve or fade into history.

An era comes to a close. I was using a VHS recorder to time- and format-shift television shows I liked (specifically Babylon 5 at the time.) I remember the vice-president of TNT programming specifically calling us pirates for doing that, which we gleefully piled on in rec.arts.sf.tv.b5. :) Then TIVO came along about a year later (late 1999, early 2000, something like that) and the VHS experience went digital. Not only could we fast forward through ads, but TIVO's easily-cracked encryption allowed a whole cottage industry of ad removers to grow and thrive. I had a cron job running on my Slackware-fueled (I know, I'm dating myself) media server that extracted the day's TIVO recordings, decrypted them, fed them through an ad-stripper (that I found on pre-Dice sourceforge) and queued them up for me to watch the next day. With TIVO's fantastic season pass recording, along with mulitple-channel simultaneous recording, it was a good time for consumers of linear TV. Streaming is nice, though I now just pay cash, instead of skull-sweat, for the privilege of going ad-free. I will miss ye, Tivo.