404 Media

2025-09-24 13:56:05

‘Find My Parking Cops’ Tracks Officers Handing Out Tickets All Around San Francisco

With “Find My Parking Cops,” technologist Riley Walz reverse engineered San Francisco’s parking ticket system to place cops on a map seconds after they issue parking citations—in theory, helping people avoid spots where officers are handing out tickets.

“I can see every ticket seconds after it's written,” Walz wrote on X. So I made a website. Find My Friends? AVOID THE PARKING COPS.”

On the Find My Parking Cops site, visitors can see a map of San Francisco with pins locating parking cops, as well as lines tracking their shift routes, how much money they’ve fined people during their shifts, and for what violation the tickets are cited. Most are “street cleaning” violations, where someone didn’t move their car on designated days. Some are for “hill parking,” a citation I, a lifelong East Coast driver, had never heard of until today, that fines parkers for failing to curb the wheels of a vehicle on a hill in a way that would prevent it from rolling. A bunch of expired meter tickets and parking in tow-away areas also pop up on Wednesday morning.

I reverse engineered the San Francisco parking ticket system. I can see every ticket seconds after it's written

So I made a website. Find My Friends? AVOID THE PARKING COPS. pic.twitter.com/67MOWVMleF
— Riley Walz (@rtwlz) September 23, 2025

The site also has a leaderboard showing the officers who fines the most each week: As of this morning, the top cop is Officer 0435, who has given out $17,877 in tickets since Monday.

Walz's site pulls public data from the San Francisco Municipal Transportation Agency (SFMTA) to make the site work. “I discovered that the city website people use to pay their tickets also includes a full copy of the citation,” he wrote on his website. “But you need to know the citation ID number, which presumably you only know if you have the ticket in your hand. I don't have a car, but my roommate does and he got a ticket recently.” He figured out that ticket numbers follow a predictable pattern. “So if an officer just wrote a ticket, I know with certainty the ID after that one will be the next ticket the same officer writes. AND, immediately after a ticket is written, it becomes available to be viewed on the city's website.” He also learned that the city has about 300 parking cops, and that they are issuing tickets with numbers in batches of 100.

“All I have to do is check the first ticket in each batch I don't have in my database,” he wrote on his site. “This approach means I only have to make a request to the city's website every few seconds. Great!”

A few hours after Walz posted the project on X, he said the city changed the way it allows people to pull that public data, and the site went down. But he found a workaround, he wrote, and it went back up. As of Wednesday morning, it's spotty, with the service showing that it was up, then back down.

SFMTA told 404 Media in a statement Tuesday evening: “Citations are a tool to ensure compliance with parking laws, which help keep our streets safe and use our limited curb space efficiently and fairly. We welcome creative uses of technology to encourage legal parking, but we also want to make sure that our employees are able to do their jobs safely, and without disruption.”

As the SF Chronicle noted, the city has recently ramped up parking citations in an effort to make up for a $322 million deficit.

Walz previously made Bop Spotter, a project where he hid a solar-powered Android phone in a box on a pole in San Francisco, set it to record audio, and had it send songs to Shazam’s API to identify what people are listening to in public. He called it “Shot Spotter for music,” referring to the gunshot detection technology.

Hidden ‘BopSpotter’ Microphone Is Constantly Surveilling San Francisco for Good Music

“This is culture surveillance. No one notices, no one consents. But it's not about catching criminals. It's about catching vibes."

^{Jason Koebler (404 Media)}

404 Media

2025-09-24 13:18:26

A Vast ‘Cosmic Web’ Connects the Universe—Really. Now, We Can Emulate It.

🌘
Subscribe to 404 Media to get The Abstract, our newsletter about the most exciting and mind-boggling science news and studies of the week.

You may have noticed that the universe is pretty big—in fact, possibly infinite. These immense scales offer a challenge for scientists who seek to model the “cosmic web,” a network of large-scale structures that link the universe and intersect at nodes where galaxies accumulate into clusters.

That’s why researchers led by Marco Bonici, a cosmologist at the University of Waterloo, have developed an emulator called Effort.jl that can parse cosmic data much faster than traditional models. This approach will accelerate the pace of discoveries about the mysterious cosmic web and help test fundamental theories about the nature of spacetime.

The tool achieves “exceptional computational performance without sacrificing accuracy,” according to the team’s study, which was published last week in the Journal of Cosmology and Astroparticle Physics.

“An emulator is a mimic, in a sense,” Bonici said in a call with 404 Media. “It basically lets you do the same kind of analysis that you would do on a supercomputer in a few days of time, in a bunch of hours on your laptop.”

Emulators can imitate the predictions of more complex models by training on their outputs, he added, without getting bogged down in the underlying physics and repetitive calculations.

“Of course, there is a trade-off in precision,” Bonici said. “This is an approximate method. The goal of my study was to show that for some scenarios, we are able to recover the final result of an analysis, even with the emulator. In this way, we can show that the error that we are introducing is negligible.”

The new emulator is designed to probe the Effective Field Theory of Large-Scale Structure (EFTofLSS), a well-corroborated theoretical model of the universe. Bonici first started developing Effort.jl—an abbreviation of “effective field theory surrogate”—during his work on the European Space Agency’s Euclid space telescope, which launched in 2023.

Euclid, along with the Dark Energy Spectroscopic Instrument (DESI) in Arizona, are producing a flood of exciting new observations about the cosmic web, dark matter, and dark energy that may overturn longstanding assumptions about the universe. Ejjort.jl aims to help interpret these findings more quickly.

Dark matter is an unidentified substance that accounts for most matter in the universe and forms the basis of the cosmic web, while dark energy is the term for whatever the heck is causing the universe to expand at an accelerating rate.

The concept of dark energy dates back to Albert Einstein, who proposed that a “cosmological constant” acts as an anti-gravity force to keep the universe in a static state, preventing it from collapsing under its own gravity. Einstein famously called the constant his “greatest blunder” after Edwin Hubble discovered that the universe was not static, but was rather expanding.

However, the cosmological constant was later repurposed into the standard model of cosmology, also called the Lambda-cold dark matter model (ΛCDM), where Lambda stands for the constant. This constant describes the accelerating expansion of the universe and is now closely associated with dark energy.

This complicated tale now has another exciting twist: DESI’s observations strongly suggest that this constant is not all that constant across the universe’s history. In other words, the effect of dark energy on the universe may change over time, a finding that could upend our basic assumptions about this strange cosmos we inhabit (this seems to happen a lot).

“With DESI’s first data release, we had the first hint that was pointing towards this deviation from the cosmological constant,” Bonici said. “With the second release, that hint was still there.”

“If we have an independent confirmation that this is actually correct, then this will be the first evidence for a behavior beyond the standard model of cosmology,” he continued. “It would be a huge revolution.”

The standard model is still highly accurate in describing cosmic phenomena, but scientists have long wondered about weak spots in its armor, including the unexplained nature of dark matter and dark energy. As Euclid and DESI gather real observations of the cosmic web, these underlying mysteries of our universe will come into sharper focus.

Bonici and his colleagues hope Effort.jl can speed up the process of evaluating the enormous reams of incoming data, acting as a complement to more resource-intensive models that require supercomputers and other expensive and time-consuming approaches.

Though Bonici’s work on the emulator requires a painstaking focus on lines of code, he does occasionally get an opportunity to step back and consider the broad implications of his field. To that end, he recalled that one of his colleagues opened his PhD thesis with the wry observation: “Cosmology is that humble branch of physics which wants to explain the universe as a whole.”

“We are describing literally everything in the universe,” Bonici said. “Most of the time, I just focus on the small details, but when I look at the big picture, to me, it is always exciting to think about, and mind blowing.”

🌘
Subscribe to 404 Media to get The Abstract, our newsletter about the most exciting and mind-boggling science news and studies of the week.

404 Media

^{404 Media}

404 Media

2025-09-24 13:00:44

Podcast: We're Suing ICE. Here's Why

We start this week with some news: we are suing ICE for access to its $2 million contract with a company that sells powerful spyware. Paragon sells tech for remotely breaking into phones and reading messages from encrypted chat apps without a target even clicking a link. After the break, we talk about a couple of stories about AI ‘workslop’ and the engineers who fix peoples’ vibe coding. In the subscribers-only section, we start with a malicious game on Steam stealing cryptocurrency from a cancer patient, then we talk about Silk Song.
playlist.megaphone.fm?e=TBIEA2…
Listen to the weekly podcast on Apple Podcasts,Spotify, or YouTube. Become a paid subscriber for access to this episode's bonus content and to power our journalism. If you become a paid subscriber, check your inbox for an email from our podcast host Transistor for a link to the subscribers-only version! You can also add that subscribers feed to your podcast app of choice and never miss an episode that way. The email should also contain the subscribers-only unlisted YouTube link for the extended video version too. It will also be in the show notes in your podcast player.
youtube.com/embed/NWuDzmKE8kg?…

• We’re Suing ICE for Its $2 Million Spyware Contract
• AI ‘Workslop’ Is Killing Productivity and Making Workers Miserable
• The Software Engineers Paid to Fix Vibe Coded Messes
• Steam Hosted Malware Game that Stole $32,000 from a Cancer Patient Live on Stream
• Does Silksong Seem Unreasonably Hard? You Probably Took a Wrong Turn

The 404 Media Podcast

Tech News Podcast · Updated Weekly · Welcome to the podcast from 404 Media where Joseph, Sam, Emanuel, and Jason catch you up on the stories we published this week. 404 Media is a journalist-owned digital media company exploring the way …

^{Apple Podcasts}

404 Media

2025-09-23 15:41:11

AI ‘Workslop’ Is Killing Productivity and Making Workers Miserable

A joint study by Stanford University researchers and a workplace performance consulting firm published in the Harvard Business Review details the plight of workers who have to fix their colleagues’ AI-generated “workslop,” which they describe as work content that “masquerades as good work, but lacks the substance to meaningfully advance a given task.” The research, based on a survey of 1,150 workers, is the latest analysis to suggest that the injection of AI tools into the workplace has not resulted in some magic productivity boom and instead has just increased the amount of time that workers say they spend fixing low-quality AI-generated “work.”

The Harvard Business Review study came out the day after a Financial Times analysis of hundreds of earnings reports and shareholder meeting transcripts filed by S&P 500 companies that found huge firms are having trouble articulating the specific benefits of widespread AI adoption but have had no trouble explaining the risks and downsides the technology has posed to their businesses: “The biggest US-listed companies keep talking about artificial intelligence. But other than the ‘fear of missing out,’ few appear to be able to describe how the technology is changing their businesses for the better,” the Financial Times found. “Most of the anticipated benefits, such as increased productivity, were vaguely stated and harder to categorize than the risks.”

Other recent surveys and studies also paint a grim picture of AI in the workplace. The main story seems to be that there is widespread adoption of AI, but that it’s not proving to be that useful, has not resulted in widespread productivity gains, and often ends up creating messes that human beings have to clean up. Human workers see their colleagues who use AI as less competent, according to another study published in Harvard Business Review last month. A July MIT report found that “Despite $30–40 billion in enterprise investment into GenAI, this report uncovers a surprising result in that 95% of organizations are getting zero return … Despite high-profile investment, industry-level transformation remains limited.” A June Gallup poll found that AI use among workers doubled over the last two years, and that 40 percent of those polled have used AI at work in some capacity. But the poll found that “many employees are using AI at work without guardrails or guidance,” and that “The benefits of using AI in the workplace are not always obvious. According to employees, the most common AI adoption challenge is ‘unclear use case or value proposition.’”

These studies, anecdotes we have heard from workers, and the rise of industries like “vibe coding cleanup specialists” all suggest that workers are using AI, but that they may not be leading to actual productivity gains for companies. The Harvard Business Review study proposes a possible reason for this phenomenon: Workslop.

The authors of that study, who come from Stanford University and the workplace productivity consulting firm BetterUp, suggest that a growing number of workers are using AI tools to make presentations, reports, write emails, and do other work tasks that they then file to their colleagues or bosses; this work often appears useful but is not: “Workslop uniquely uses machines to offload cognitive work to another human being. When coworkers receive workslop, they are often required to take on the burden of decoding the content, inferring missed or false context. A cascade of effortful and complex decision-making processes may follow, including rework and uncomfortable exchanges with colleagues,” they write.

The researchers say that surveyed workers told them that they are now spending their time trying to figure out if any specific piece of work was created using AI tools, to identify possible hallucinations in the work, and then to manage the employee who turned in workslop. Surveyed workers reported spending time actually fixing the work, but the researchers found that “the most alarming cost may have been interpersonal.”

“Low effort, unhelpful AI generated work is having a significant impact on collaboration at work,” they wrote. “Approximately half of the people we surveyed viewed colleagues who sent workslop as less creative, capable, and reliable than they did before receiving the output. Forty-two percent saw them as less trustworthy, and 37% saw that colleague as less intelligent.”

No single study on AI in the workplace is going to be definitive, but evidence is mounting that AI is affecting people’s work in the same way it’s affecting everything else: It is making it easier to output low-quality slop that other people then have to wade through. Meanwhile, Microsoft researchers who spoke to nurses, financial advisers, and teachers who use AI found that the technology makes people “atrophied and unprepared” cognitively.

Each study I referenced above has several anecdotes about individual workers who have found specific uses of AI that improve their own productivity and several companies have found uses of AI that have helped automate specific tasks, but most of the studies find that the industry- and economy-wide productivity gains that have been promised by AI companies are not happening. The MIT report calls this the “GenAI Divide,” where many companies are pushing expensive AI tools on their workers (and even more workers are using AI without explicit permission), but that few are seeing any actual return from it.

Microsoft Study Finds AI Makes Human Cognition “Atrophied and Unprepared”

Researchers find that the more people use AI at their job, the less critical thinking they use.

^{Emanuel Maiberg (404 Media)}