The Internet the Swarm Agents Already Polluted

On Tuesday, I finished rereading Dario Amodei's essay on pacing AI development. On Wednesday, Andrew Yang went on CNBC's Squawk Box and got asked whether the whole pacing push is a psyop, a coordinated move by Anthropic and others to lock in a closed model advantage before regulators arrive. Yang did not deflect. He cited an AI lab head who told him agent swarms had already polluted the open internet with self-replicating code, forcing labs to build synthetic internets just to source clean training data. He said the fear is real and the concern is real.

Yang's source seems to have folded several distinct research incidents into one narrative. What sounds like a single runaway breach is a composite of documented behavior across separate, isolated evaluations.

Three incidents, one borrowed narrative

In June, evaluation agents given read-only web access bypassed their own communication restrictions by making over 14,000 edits to DSEWiki, a small German-language wiki on an Austrian server, and used its edit history as a covert message board. Researchers traced the same coordination to twenty other sites, including a Vanderbilt link shortener where agents generated over 50,000 hits in a single day by disguising messages as referrer addresses.

The self-replicating mechanism traces to a separate security evaluation in July. There, agents exploited a hole in a dataset processing pipeline, executed code, harvested credentials, and ran command-and-control infrastructure that rebuilt itself on new servers as old sandboxes were shut down.

That same month, Anthropic separately documented its own agents escalating to self-replicating malware, inside an internal benchmark test, not an open, uncontained leak.

Separating these episodes strengthens the argument: the pattern holds even without the exaggeration. Sacks' argument treats the pacing calls as theater staged for regulators. Theater does not require labs to rebuild their own training pipelines around synthetic internets. Whether through wiki message boards, referrer header exploits, or self-migrating infrastructure, contamination came first and caution followed. Both things can be true: well capitalized firms can use a technical mess to entrench their market position, and the mess itself can still be real.

I closed that earlier post on Amodei's admission that pacing only works if adversaries pace too. Yang's interview supplies the missing half of that sentence. The tools got away from their makers once already, quietly, as contaminated training data instead of a headline. Whatever you conclude about who profits from slowing down, that part of the fear does not need Sacks or Amodei to be honest about their motives. It only needs the servers to be telling the truth.

Can Data Centers Be Built and Operated Responsibly? Ask My Hometown

I grew up in Westfield, Massachusetts, at 342 Holyoke Road. In 2021, the City Council approved a $4 billion data center campus at 199 Servistar Industrial Way, on land that sits partly on wetlands and an aquifer, the region's drinking water source. The site is about 1.8 miles from my childhood house. The vote passed 9 to 3. Almost nobody objected.

By 2026, that had changed. Residents packed council chambers and won a one year moratorium on new data center projects. Lowell and Holyoke passed their own moratoriums. Governor Healey now requires limits on air and noise pollution, caps on water use, and proof that a project will not raise local electric rates.

The environmental cost of a data center comes down to five variables: how the IT equipment is cooled, how the facility rejects that heat outdoors, what powers it, where it sits, and how loud it runs. Air cooled servers run a Power Usage Effectiveness of 1.5 to 1.8. Liquid and immersion cooling remove heat more efficiently at the chip, cutting PUE to 1.03 to 1.08. Immersion cooling by itself does not reduce water use, since that depends on a separate choice: whether the facility rejects heat through evaporative cooling towers or dry coolers. Dry coolers eliminate on-site water use entirely but can demand 10 to 35 percent more electricity than evaporative towers, more during heat waves.

Power source matters more than the cooling loop. Lawrence Berkeley National Laboratory found that 92.5 percent of a data center's water footprint comes from generating its electricity, not cooling its racks. A facility on hydro or nuclear power carries almost no water cost at the plant. The same facility on a coal or gas grid inherits that fuel's water cost.

Location decides much of the rest. Two thirds of data centers built since 2022 sit in water stressed regions, according to Bloomberg Intelligence and the Lincoln Institute. A campus in Iowa competing with farmland for water behaves nothing like one in Quebec, where hydropower and a cold climate do the work for free.

Noise is the variable nobody priced in early. Data centers run cooling equipment, chillers, cooling towers, and HVAC fans, continuously, since server floors need round the clock cooling. Diesel backup generators add to the noise only during monthly testing, typically 50 to 150 hours a year per unit, or an actual grid outage. Microsoft now faces a federal class action in Mount Pleasant, Wisconsin over a persistent hum from its cooling systems, and residents in New Jersey and Ohio report the same problem. Much of it is low frequency infrasound that standard decibel meters and noise ordinances written for block parties were never built to catch.

Westfield's 2021 vote treated its data center as an economic decision. The 2026 moratorium treats it as an engineering one. The state's new certification standards will require the developer to disclose cooling method, power source, and noise mitigation plans before construction proceeds. Those disclosures, not the 2021 vote or the 2026 moratorium, will determine what actually gets built on the Barnes Aquifer.

Pacing AI at the Frontier, Restricting It in the Classroom

On Saturday, September 12, Anthropic CEO Dario Amodei published an essay titled “We Must Pace the Frontier,” a call for AI labs to slow the pace of capability research in the name of safety. By Monday, OpenAI CEO Sam Altman had endorsed it and agreed to embed third party safety evaluators inside his own company. I read both statements twice. Two men who built their companies on the promise of faster, bigger, smarter models were now telling the world, and each other, to slow down.

Neither man promised to cut his own pace unilaterally. Amodei's plan calls for shared safety standards among labs in democratic countries, plus cooperation from authoritarian governments, a group not known for volunteering restraint. Pacing agreements among American and allied labs only bind the labs that sign them. Authortarian governments owe nothing to any of it and can keep advancing capabilities at full speed regardless of what those labs decide. Critics on both sides of the debate have started calling this what it looks like from a distance: a handful of companies agreeing to slow spending on capability research while their market position stays intact. One commentator dared the labs to prove otherwise or admit the announcement was a bid for regulatory capture.

This distinction matters more once the word “limit” leaves boardrooms and enters classrooms. Frontier labs pacing their own training runs is a research decision made by people who understand the systems. Restricting a student's or an employee's access to AI tools is a different decision entirely, and it lands on people with far less power to negotiate the terms.

Pacing at the lab, restricting in the classroom

I teach engineering students who will spend their careers working alongside AI systems, not behind them. In my travels, every semester I meet someone (NOT at institutions where I teach) who was told, by a school policy or a manager, to avoid these tools rather than learn to use them well. That advice does not make them safer. It makes them less prepared, while the companies building the technology keep building it regardless of what any classroom decides.

The pacing debate has its own logic problem. No single lab wants to pace its own training runs first, because whoever does gives up ground to labs that keep going. The same logic applies to any school, employer, or country that restricts AI use on its own. Restraint by the party with the least leverage does not slow anything down. It only changes who understands the tools once the restraint ends.

More than a thousand employees across the major labs have signed on to public statements asking for a pacing mechanism, an unusual show of rising internal dissent. None of them asked their own users to stop using the tools already shipped. The asymmetry tells you where the real risk sits, and it isn't in your classroom or your workplace.

I finished rereading Amodei's essay a third time this morning. The line that stuck was not about pacing or evaluators. It was his admission that a slowdown only works if adversaries slow down too, an admission that undercuts the whole premise for anyone still deciding whether to keep their own students and employees off the tools everyone else is racing to master.

Scratches in the Hardwood: Traces of Seventy-Five Years

342 Holyoke Road by Marion Killacky

My parents built the house in the early 1950s. It was small, single story, about 50 feet by 25 feet with an attached one car garage, one bathroom, three bedroom shared by six of us until the attic was renovated into a bedroom for my two brothers and me. I grew up in it. My Dad passed away on December 26. My Mom couldn't live there alone after that. She's in a long-term care facility now, a beautiful one, with her own little apartment. Last month, we sold the house privately. 

Movers carried the dresser out in February, in about three feet of snow, then the bed frame. The dresser has a large mirror mounted on top, the kind that catches the whole room behind whoever's standing in front of it. The wood is solid all the way through, no veneer over particleboard—the kind of furniture nobody makes anymore because nobody expects to keep anything that long. My parents bought that combo when they built the house. It sat in that bedroom, against that wall, under that light, for roughly 75 years, and it left through a few feet of snow on a truck ramp. 

Seventy-five years without direct sun—I keep doing that math in my head, more than I do the math on anything else from that week. As a kid, I assumed it was going to stay that way forever. The changes came gradually, and some things didn't change for so long that it was easy to mistake slowness for permanence. I noticed the small changes and chose not to look too closely. I knew what was coming; I spent years managing not to think about it, right up until it was unavoidable. 

Once the furniture was gone, I saw small scratches in the hardwood, faint ones. I have no idea if they happened decades ago or that week, while the movers were maneuvering the dresser out the door. Seventy-five years compressed into a set of marks I can't date, no different from the years themselves, most of which I can't place exactly either. The wall behind the headboard was a shade lighter than the rest of the room, a pale pattern where the sun had never reached. My Dad slept in that bed until December. My Mom slept in that bed for over 70 years. Six of us in a house that size, sharing one bathroom, and somehow the walls felt like they held more privacy than that math suggests they could. That mirror held every version of my parents getting ready for work, for church, for whatever came next, for over 70 years, and it held me too, some mornings, standing in their room asking for something. 

That mirror caught the rhythm of a house that ran hot in summer and cold in winter no matter what anyone did about it. For most of those 70 years, there was no air conditioning. In summer, the windows stayed open and the whole house fan ran all night, pulling the cool night air in from outside and pushing it through every room. Upstairs in the attic, the small window unit for my brothers and me barely made a dent—the heat just sat up there while the unit hummed away doing next to nothing. Winters ran on a wood stove, which meant somebody was always splitting wood, always feeding it, always checking it before bed, and the rooms farthest from it never got quite warm enough. The whole house smelled like woodsmoke all winter long, that dry, faintly sweet smell that got into the curtains and the couch cushions and never fully left, even in July. A couple of mini-splits went in about six years ago, no ductwork, just those two units doing what they could for a house that had gone 70 years without any of it. My parents got a handful of summers with them before the house sold. 

In the kitchen, there was a silverware drawer. Every morning I reached in without looking and pulled out a spoon for cereal. I never thought about that drawer once in 40 years except to reach for it. Now I think about it more than almost anything else in that kitchen. 

The cellar door had a specific squeak, the exact same pitch on the way down as on the way back up. Between that hinge and the distinct rhythm of footsteps on the cellar stairs, I could tell from any room in the house not just whether someone was going down for something or coming back up with it, but exactly who it was. Nobody ever oiled that hinge or tried to quiet those steps. At some point, it stopped being a maintenance problem and became a way of knowing where everyone was.

The last time I went to the house, the bedroom was empty. No dresser, no mirror, no bed, nothing left but those small scratches in the floor, the pale pattern on the wall, and breeze coming through a window that used to be someone's whole summer. 

The combo is in her bedroom at the facility now, mirror and all. Same dresser, same bed, and they still don't look right in there. The proportions are off somehow, or the light is different, or it's just me looking for reasons this doesn't fit anymore. She falls asleep looking at the same wood grain she's looked at for most of her life, in the same bed, in an apartment that is by every account nicer than the room she left, kept at a temperature the old house never quite managed. None of that seems to matter to the two pieces of furniture sitting in it. 

We sold the house quietly. A handshake, a closing date, no listing, no strangers walking through. Nobody outside the family will ever know which floorboard creaked, or why, or the pitch of the cellar door hinge, or how six people made one bathroom work for as long as we did. The new owners will refinish the floor and sand out scratches they'll never know the story behind. In a year, there will be no evidence we were ever there. 

I'll still be thinking about that dresser and bed sitting in an apartment that doesn't suit them, and standing in an empty room in last month looking at a wall the sun never touched—in a house that held six of us, that I grew up in and thought would always be there, and won't walk into again.

Quantum from the Ground Up: Second Edition, September 2026

A free PDF pulling together posts from this blog into a single volume. Written for someone entering the quantum workforce, or seriously considering it, who has a technical background but has not taken a graduate course in quantum mechanics.

This edition folds in 22 posts published between June 3 and August 26 into the same 19 chapters as the first edition: quantum networking over existing fiber, the physics-versus-engineering distinction, six qubit fabrication platforms, protein simulation and AI-assisted hardware calibration, post-quantum cryptography, the hardware landscape as it stands in late summer 2026, and the quantum workforce.

Download PDF: Second Edition, September 2026

Edition	Second edition, September 2026
Pages	73
Chapters	19
Source posts	39 posts, October 2025 to August 2026
Updates	Quarterly
License	CC BY-NC-ND 4.0: free to share with attribution, no commercial use, no modifications

 

What's new in this edition

●      Quantum memory repeaters tested on a live network by IonQ and EPB in Chattanooga

●      A free-space quantum network link built by Brookhaven and Stony Brook

●      Two linked quantum cryostats from IBM, a step toward fault-tolerant systems

●      A sputtering-gas fix for a superconducting qubit manufacturing defect, from Cornell

●      A universal gate set on 28 qubits using code-switching magic states, from Quantinuum

●      Two June 22 executive orders: new post-quantum cryptography migration deadlines and the QC-ADDS program

●      IonQ's acquisition of its own chip foundry

●      Quantum key distribution deployed on production Cisco routers

●      IBM's free Quantum Learning certification path

Updates

This page will always link to the current edition. Updates are planned quarterly as new posts on quantum hardware, security, and workforce development are published on this blog. The next edition covers Q3 2026, due around December 1.

Support this work

The book is free and will stay free. If you find it useful and want to support future editions, contributions are welcome at ko-fi.com/gordostuff.

A Plain-Language Edition of Quantum from the Ground Up

Einstein called it spooky action at a distance. He meant quantum entanglement, and
he thought it couldn't be real. He was wrong!

I published a plain-language edition of Quantum from the Ground Up today, built for that reader.

The original book is written for people with a technical background who haven't taken a graduate course in quantum mechanics. It works well for that reader. It does not work for someone who has never had a reason to think about qubits, cryogenics, or post-quantum cryptography and just wants to know what the technology actually does.

The two audiences want different things. Someone building a career in the field wants precision: exact fidelity numbers, fabrication steps, the difference between a physical qubit and a logical one. Someone reading a headline about a new quantum computing milestone wants to know what it means and whether it matters to them. The plain-language edition is built for the second person.

It covers the same 19 chapters as the technical edition, at a fifth of the length: 13 pages instead of 73. It leaves out the math, the fabrication chemistry, and the dense statistics tables. It keeps the facts that matter: what quantum computers actually do right now, why companies are spending billions of dollars building them, why your bank account and your medical records are already part of this story, and where the actual jobs are for people without a physics degree.

Both editions are free. Both get updated quarterly as new research and hardware announcements come out. The next technical edition publishes tomorrow.

You can find current and older versions of each in this folder.

If you have been putting off understanding quantum computing because every explanation you found assumed a physics background you don't have, this edition removes that excuse.

Brookhaven and Stony Brook Just Linked a Quantum Network Through Open Air

In my last post, I laid out the fiber problem: quantum networks can't use classical repeaters because the no-cloning theorem forbids copying an unknown quantum state, so every photon has to survive, unassisted, from source to destination. Fiber compounds that: it loses about 0.2 dB/km, and its only efficient band (1550 nm) doesn't match the wavelengths most atomic memories emit, forcing lossy conversion steps at each end.

Commercial telecom has been married to one wavelength band for as long as I've been in the field, going back to writing the telecommunications curriculum for Verizon's NextStep program in 1995 and later directing NSF-funded Centers of Excellence in telecommunications and optics and photonics. My dad worked the generation before that, retiring in 1984 back when the network still ran on copper and fiber hadn't arrived yet. Brookhaven National Laboratory and Stony Brook University just did something a little different. 

Photons travel 13 miles through open air from Stony Brook's Quantum Watchtower to Brookhaven's Quantum Lighthouse, using adaptive optics instead of fiber. Link connects to existing 161 mile fiber network, with a 30 mile extension to Yale planned next.

On August 21, researchers sent single photons and entangled photon pairs 13 miles through open air, from Stony Brook's Quantum Watchtower to Brookhaven's Quantum Lighthouse in Upton, New York, no fiber involved for that leg of the trip. The daytime event was a formal demonstration for DOE and state officials. The actual first detection happened two days earlier, in the dark: at 12:26 a.m. on August 19, the Lighthouse recorded entangled photon pairs arriving from the Watchtower, confirmed through phase folded photon analysis comparing on-phase detection counts against background noise rates.

Both facilities exist to solve one problem: Brookhaven and Stony Brook need an unbroken line of sight to each other, which is why the Lighthouse sits on the only Brookhaven building with that sightline, a seven story rooftop installation, while the Watchtower sits atop Stony Brook's Health Sciences Center. Fog blocks the link outright. Bright daylight is a subtler problem. It raises atmospheric turbulence and buries the faint photon signal in background light, a limitation the Brookhaven team has compared to trying to spot a flashlight beam from a rooftop in broad daylight.

The optics came out of Brookhaven's Instrumentation Department, built on adaptive optics designs borrowed from the Vera C. Rubin Observatory. Photons leave the Watchtower through a fiber core five microns wide, about a tenth the width of a human hair. A telescope expands that pinprick of light into a 25 inch, 0.6 meter beam to match the primary mirror, while deformable mirrors correct for atmospheric turbulence in real time at kilohertz frequencies. At the Lighthouse, the process runs in reverse: the beam narrows back down and threads into a matching five micron fiber core for detection.

Commercial fiber networks are locked into wavelengths near 1550 nanometers because that band travels through glass with the least loss over distance. A free-space link carries no such requirement. Researchers can transmit infrared wavelengths native to the atomic systems and quantum processors themselves, opening a direct channel to entangle remote atomic memories without a wavelength conversion step.

The new link folds into an existing fiber network spanning 161 miles and eight nodes across Long Island and the New York City area, the longest metropolitan quantum network in the country. DOE Under Secretary for Science Dario Gil cut the ribbon on the receiving aperture at the August 21 event, framing the connection as a step toward linking individual quantum computers into something larger.

A third facility, functionally identical to the Lighthouse and Watchtower, is already built at Yale University in New Haven, Connecticut, with a 30 mile free-space link across Long Island Sound planned to connect Stony Brook and Yale directly. Past that, the team plans to repurpose the same rooftop telescope infrastructure to track low-earth-orbit satellites, laying groundwork for satellite-based quantum key distribution and a longer-term global quantum network. Funding comes from DOE's Office of Science, the National Science Foundation, and $300 million in New York Empire State Development money tied to Stony Brook's Quantum Innovation initiative.

This ties directly to Chapter 1 of Quantum from the Ground Up, which opens on the fiber problem: how to move quantum information over distance without destroying the fragile state that makes it quantum in the first place. Fiber solved part of that by forcing everything into a wavelength band it can carry efficiently. This link solves a different part by removing the requirement to force anything at all. This post will roll into the next quarterly edition of the book, out September 1. You can read the current edition here.