Eric Anderson retweeted
How MAGA hallucinations became real
I pay for the data behind these myself and it take a lot of time and effort to analyze and create these graphics. If you want more of them, SUPPORT LINKS ARE IN BIO
At 1:05 p.m. on Oct 11, Nick Sortor posted that an Ohio mental hospital “CONFIRMS” James Talarico was discharged. That evening the hospital said Talarico never received treatment there. Talarico says he's never even been to Ohio. Rhein Amacher, who made the recorded call, posted that he made up a verification code to get past the front desk.
This story occurred over several days. Ken Paxton called Talarico “missing” on Oct 7. Sara Gonzales teased “unverified information” that would “END his campaign” on Oct 8. On Oct 9, Spencer Pratt said his sources put Talarico in a mental health facility, and a handful of accounts pasted near-identical “credible sources are leaking” posts. At 7:50 a.m. on the 11th, Sortor gave Talarico a deadline, “let us know by 1pm ET today.”
The earlier pushes never fully took off, and acted more as a booster for the posts that came later. Pratt's post got 6.4 million views, but only about 60 accounts posted about the story in the hour after it. Caroline Wren's private-plane theory the next day drew about 40. Before the 11th, the story never topped 80 posts an hour. In the first hour after Sortor's post, 267 accounts posted about it, and by 3 p.m. it was running at almost 500 posts an hour.
The first chart counts each account only once, at its first post about the story. Nothing they posted after that is counted, so it shows who started this and how far those first posts went. The first wave carried the whole thing. Sortor's post and the posts that followed in the next 12 minutes have 9 million views between them. The 55 accounts that joined in the first 20 minutes got 80% of the views in that lane. The other lines never came close. “He lied and covered it up,” the line Elon Musk and Richard Grenell pushed, got about 2 million views across 100 accounts. “He was in an Ohio mental hospital” got about 2 million across 132.
The pushback grew the opposite way, the way a real response does. It started small and kept building all afternoon and evening, with 12 to 46 new accounts joining every hour from 1 p.m. to 10 p.m. Leave out the three biggest posts, which were the hospital's statement, Talarico's reply and Ed Krassenstein's own call to the hospital, and regular accounts pushing back still reached 6 million views. The claim had nearly 90% of its views by 2 p.m. and barely grew after that. By the end of the night, the pushback had more views than the claim, 13 million to 11.4 million.
The second chart is the inorganic behavior, and it covers Oct 8 to 11, not just the day of the "exclusive". Sortor's headline was reposted word for word, or nearly, by 18 other accounts, half of them with under 1,000 followers. Caroline Wren's private-plane post was copied by 14. Four accounts posted the same Eric Daugherty post within 74 seconds of each other. Four more posted the same MAGAVoice line within 54 seconds, four hours after it first went up. Some accounts barely post anything of their own. One copied 21 of its 23 posts about Talarico from other accounts, another 11 of 11. And the spammers flooded the replies. One pasted the same reply 85 times in about two hours, another 15 times in 10 minutes.
Views are each post's total as of early Oct 12. A copy means word for word or at least 90% the same as an earlier post.
I pulled and analyzed about 14,000 posts for this. Sharing it helps more than anything, and if you want to chip in so I can keep doing this, the links are in my bio. It goes a long way.
Eric Anderson retweeted
We just launched Dstl8, the runtime feedback loop for AI-speed engineering - powered by Telemetry Distillation - remember, observability without feedback is just watching controltheory.com/blog/dstl8…
Eric Anderson retweeted
For context, we delivered enough water for about 150 American households annual consumption
In 3 hours
With 2 drones
We will save the Colorado River
This is insanely good - seriously full stack deployed, minutes, stupid simple and easy to approach UI.
Absolutely better than everything else out there right now.
Well done!!
Introducing Nativeline Cloud.
Build native Swift apps with AI. Then build your database with a prompt.
Auth. Storage. Analytics. Everything your app needs — one platform.
No Supabase. No Firebase. No Xcode. No setup.
iPhone. iPad. Mac. From idea to App Store.
nativeline.ai
Over the weekend, I built a handy MCP Client/Inspector that acts as a transparent proxy for debugging, as a local macOS app - did it all in @Nativelineai.
Faster than building it in Claude Code, Nativeline does a bunch of UI stuff for you!
Using it to test @Ctrl_Theory 's AI!
Eric Anderson retweeted
So, here is the comparison with @Nativelineai, really impressed by its design and thinking abilities.
Design: 9.5/10
Functionality: 9/10
- Only one recipe from the link didn't work (should have worked if I provided enough documentation and ways for it, but I didn't provide it to others, so I didn't provide it to this as well to do a fair comparison).
Overall Rating: 9.5/10
I’ll be comparing four apps: @rork, @Replit, @v_computer, and @fastshot_ai, which are included in @RevenueCat’s #shipyard2026. I’ll give each of them a single prompt and evaluate which one performs best based on design, functionality, and completion.
Eric Anderson retweeted
We got tired of AI tools giving us websites when we wanted an iPhone app.
So we built NativeLine.
Native Swift. One platform. Idea to App Store.
It's live today. Watch this ↓
Sometimes, sentiment is all you need.. to observe a system based on logs.
I wrote about it and what we are doing @Ctrl_Theory - read it, and then let's talk about!
controltheory.com/blog/log-s…
#logs #observability #o11y #ai
Eric Anderson retweeted
The easiest way to build a real, native iOS app.
Not cross-platform. Not exported. Not "almost" native.
Built in Swift. For Swift.
If you've ever wanted to ship an iPhone app without touching code, this is it.
Sign up for waitlist: nativeline.ai
Eric Anderson retweeted
Today I built a truly production grade iOS app in less than 3 minutes where I can track habits & see progress with @Nativelineai
Eric Anderson retweeted
Goin’ Gonzo at @Ctrl_Theory booth 1570 #Kubecon - open source log analysis TUI github.com/control-theory/go…
Eric Anderson retweeted
1️⃣ Swift comes to FreeBSD
Swift is now available in preview for FreeBSD 14.3 and later. We want your feedback, bug reports, and contributions to make Swift great on FreeBSD! 🧑💻Read the announcement here: forums.swift.org/t/swift-on-… #FreeBSD
Eric Anderson retweeted
Combine @vercel CLI with a TUI - instantly stream, filter, and analyze logs in real time with an open source terminal UI Gonzo controltheory.com/blog/verce…
Eric Anderson retweeted
VictoriaLogs is a great database for SIEM events and logs:
- It is optimized for storing, querying and analyzing billions of SIEM events and structured logs with big number of fields (hundreds per log entry).
- It automatically parses CEF messages received over Syslog protocol starting from v1.35.0 . docs.victoriametrics.com/vic…
Eric Anderson retweeted
Pointing Gonzo #TUI at AWS CloudWatch logs via live tail - plus AI analysis with your favorite LLM - practical AI for SREs @destari @j_m_reeve controltheory.com/blog/gonzo…
Eric Anderson retweeted
If there's anything I've learned being a maintainer of popular open source tools it's that even if there are rules for contributions, there are no rules for contributions
Eric Anderson retweeted
An interesting tool for logs' exploration from terminal - github.com/control-theory/go…
Eric Anderson retweeted
Gonzo - a real-time log analysis TUI inspired by k9s from @Ctrl_Theory
It automatically detects formats and beautifully prints logs for humans to read
It's exactly what I've been dreaming about as the ideal UX for logs analysis in Uncloud CLI/TUI
Taking heavy inspiration from this project if not straight-up reusing it
Check out Gonzo, the open-source #golang based TUI that analyzes your logs in real time. Supports #opentelemetry and other log formats, and has some AI juice built-in (because why not).
github.com/control-theory/go…
#otel #go #oss



















