Scott Rosenberg

Scott Rosenberg

@wordyard.com

Webbin' so long. Past: Axios editor, Wired/Backchannel contributor, Grist editor, Salon co-founder. Author of some tech books. Up next: Our Web (https://ourweb.is/).

1.5K followers1K following209 posts

You never know what's gonna cut through the noise and have an impact, "have you no sense of decency sir at long last" style, but I gotta say the $50M pardon is a strong contender

Important thread -- AI makers are repeating a cycle we've seen repeatedly in tech history where software developers assume that everyone else in the universe orders daily life and work the way they do.

Tressie McMillan Cottom@tressiemcphd.bsky.social

Believe it or not, a lot of people do not have to-do lists. And the the lists many of them have are not structured for an ai agent to be helpful. This isn’t groundbreaking. If you have ever watched the evening news or read even a single story about people who aren’t like you, it is pretty basic

Scott Rosenberg reposted

Replying to @jwherrman.bsky.social

Also notable about the recent vibe shift in AI is that it has made Jensen Huang – a plainly motivated CEO of the world's largest company, familiar sort of capitalist, and someone with a legible perspective on the industry surrounding him — sound like a heretic www.nytimes.com/2026/09/23/o...

Scott Rosenberg reposted

one takeaway i had from working in digital media in the 2010s was that it's pretty bad for your brain and your understanding of your career/the world when your industry becomes the main subject of twitter and twitter becomes the main context for your industry. also bad for the industry overall

Scott Rosenberg reposted

Literary fiction is coming. It is going to revolutionize the economy. We know that China has already got a start on literary fiction. We can’t afford to be left out of the literary fiction economy. Think of your kids. They might already be engaging with literary fiction.

Warsh said "“Oversharing policy deliberations and overcommitting to future decisions can lead markets, businesses and households astray." But all this time I thought markets function best when information is abundantly and equally available! Did something change?

Dear New York Times: I am one reader who uses multiple browsers on multiple devices — not multiple users sharing passwords to a single account. I am not a "family" and don't wish to upgrade to a family subscription. Please stop requiring me to type in emailed codes and suggesting I pay you extra

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

Authors should be able to archive their own work! But the same tool I just built (without spending a penny) can be used to steal it. There's nothing new here except that the apps that once required expertise to build are now much more easily and widely available.

I think we're going to be doing a ton of this kind of work before long. It's going to take a lot of effort to figure out how IP rights, compensation and authorship need to evolve in such a world. For every hour our new apps save us, we'll need many more to work all that through.

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

I mean, the same thing happens every time you read a web page, it's literally a "browser agent" that goes and gets the page for you so you can read it, you aren't walking over to the server and carrying the bits back to your screen yourself.

Yes, there's a difference in quantity between reading one page and 600. But I can't say I'm clear where exactly you draw a boundary between "human speed" and "bot speed" for page-loading.

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

But also, I was keenly aware that in building this tool, I was walking exactly the same road that any AI developer would if they wanted to snatch a mountain of Axios content for model training without, you know, asking.

This hit home particularly when I had to help Glaze devise workarounds for some of the safeguards Axios's team has put in place. The app has the user manually check the "human verification" box. Well, I'm a human! But yeah, that is my bot, not me, downloading the pages.

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

It took 20 minutes to the first build and after about an hour of debugging I had my archive. I also had an app that I could "publish" and share with former colleagues if they wanted their own story archives. I haven't done that yet.

To me all this just confirms that the world of abundant bespoke personal software that Casey, Paul Ford and others have begun discussing is a real possibility. I find that exciting!

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

I kind of gave up on this for a while but when Casey Newton and others started singing the praises of Glaze I thought, OK, let's try this. And Glaze pulled it off. I think part of the win here for them is that they have focused on one specific dev path (making MacOS apps) that suited me.

At first the glaze agent made a series of mistakes that rendered the app useless. But its debugging was smart and easy. One by one, it took in my bug reports ("it's not getting past the cloudflare challenge page" or "we've got a full list of URLs but no body text") and fixed them.

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

They did keep getting better but there were always roadblocks. Recently it became clear that just writing code to crawl the Axios site and get the files was within reach. But doing so would still require me to set up a development environment, manage the code, run the debugging process, etc.

I could do that with effort. But the whole point here was to eliminate effort! If this was going to be a time-saver, AI would have to do more of the grunt-work itself.

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

View full thread

I always wanted a local archive of the 600+ articles I wrote for Axios during my seven or so years working there. Axios' site has a nice author page but it's chunked into 5-stories-at-a-time and the prospect of manually copying those files always daunted me.

When people started claiming that AI was writing decent code in 2023, I began asking chatGPT and later Claude and other tools to do this job for me. The tools failed miserably for a long time.

Here's a bit of a 🧵 with this one user's experience trying to get a little smidge of productivity out of generative AI coding. TLDR: it's taken three years, but (one of) the bots has finally accomplished a simple task for me, so chalk one up for AI.

I have largely been a gen-AI skeptic since ChatGPT's advent. We're always two years off from AGI or recursive self-improvement or superintelligence singularity, no matter when you start. It's like the whole industry closed its eyes and adopted the ludicrous Musk promise horizon.

But I also try to make sure my skepticism doesn't blind me to actual real-world advances — not benchmark BS but everyday accomplishments — and here is one.

Scott Rosenberg reposted

The fundamental challenge of software development is becoming how do you build reliable software using AI agents that are frequently wrong and often don’t follow instructions while your engineering team increasingly doesn’t understand the codebase because they didn’t write the code?

This thread is a useful explainer on the difference between autocomplete and LLMs. Though I still think the analogy is helpful for dispelling the magical "it's thinking!" response. Read all the way through for an amazing demo of a heat-map tool showing LLM "attention" processes in action

This post is unavailable.

Scott Rosenberg reposted

Too many people still confuse Software Engineering with writing code. Code is the artifact. Software Engineering is understanding the problem, making tradeoffs, designing the system, reducing complexity, and deciding what shouldn't be built. No amount of 'vibe coding' can help you with this.

Just wanna say it took roughly 25 years before you could even imagine getting booed in public by mentioning the Web/internet. Generative AI has now speed-run this road in, what, under four? And the bubble hasn't even burst yet.

404 Media@404media.co

Former Google CEO Eric Schmidt was booed throughout this commencement speech at the University of Arizona for his praise of AI. This comes just a week after another commencement speaker who mentioned AI was booed at a school in Florida. Read more: www.404media.co/ucf-ai-comme...

Scott Rosenberg reposted

PART 2 of MY INVESTIGATION: Watch how ChatGPT gave me chilling advice—as I simulated planning a mass shooting. This 9-minute video and story contain shocking behavior from OpenAI’s chatbot. OpenAI keeps saying it improved safeguards, but that's not what I found. www.motherjones.com/media/2026/0... 🧵

motherjones.comChatGPT gave me chilling advice—as I simulated planning a mass shootingI asked about imitating the Uvalde attacker, defending against police gunfire, and more—everything short of directly stating intent to kill.

Great story. This is a perfect ourobouros -- AI-makers desperately need to hoard data for training but their own data centers' needs are crowding out the public resources that preserve the data.

er this explains a lot

David H. Montgomery@dhmontgomery.com

We at @today.yougov.com asked a representative sample of Americans the odds of flipping two coins and getting two heads. 36% correctly picked "25%." 48% incorrectly said "50%." See more in our full survey: yougov.com/en-us/daily-...

Scott Rosenberg reposted

Substack's CEO now trying to scare people into *only* distributing their content through their proprietary app, not email. (The link he posted warns users away from email.) It's not enough that Substack's locking in publishers & readers with "notes" and comments, now they're escalating tactics.

I guess next up SCOTUS will say nothing about the substance of what we're learning this weekend but immediately launch a mad crusade to find and punish the leaker(s).

Scott Rosenberg reposted

The Wetland Project has been my favorite alternate-reality radio experience since it began. Since it only happens once a year, you can’t become accustomed to it. Close your eyes and be washed over by the sonic reality of an isolated bog on Saturna Island!

Wetland Project@wetlandproject.bsky.social

Details at mailchi.mp/4396a6512d1e...