iiinova — where your digital self keeps evolving

Reposted by Apostolos Stamenos

Steve Moskal @samoskal.bsky.social · 31/07/2026

Know, understand, counter DARVO, a core behaviour of coercive control abuse, especially used by authoritarians, fascists, abusers to divert blame, gaslight, scapegoat ‘other’ targets, inflict #trauma effects on those who dont recognise it, then gaslight target’s trauma effects as ‘mental illness’

021

Reposted by Apostolos Stamenos

Karly Kingsley @karlykingsley.bsky.social · 25/07/2026

Katie Phang is just out here crushing it when it comes to the Epstein files and working towards justice for all those who were affected. Let women lead.

161116240

Apostolos Stamenos @astamenos.bsky.social · 25/07/2026

Hi there! Could you elaborate on what your goal is? Are you looking to learn how to use something like Claude? More on the theory side of things? Governance? All of the above? Not sure? All of those are fine! I’m evaluating what kind of tips would be most helpful to you. 😊

000

Reposted by Apostolos Stamenos

ToxSec @toxsec.bsky.social · 01/07/2026

when you feel like quitting, remember someone else will find it if you don’t. #hackerlife

011

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

@jjfreyd.bsky.social's research. Social media harassment. Rushed AI detection feature launches. What do they have in common? A lot more than you'd think. This is a bug report, not persecution. Full argument (90% human-written if you are curious): astamenos.substack.com/p/polarized-...

astamenos.substack.com

A New polArIzed Economic Engine for Culture

Substack shipped an AI detector but no abuse detector. That is a design decision. On base rates, federal lawsuits, and who pays the price of platform purity.

000

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

Bluesky already shows a healthier internet is possible: moderation you control, block lists you can share, the freedom to leave with your name and audience intact. All in the base app. Substack had the hard problem solved. They aimed it at the wrong target.

100

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

Accuracy does not solve this. Pangram is excellent, comparable to expert human detectors. But the fraction it gets wrong, across millions of posts, is a standing crowd of humans labeled clankers every day. Pangram built the tool. Substack deployed it questionably.

200

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

Here is who pays. An authorship label reading "AI-generated" is pseudo-objective ammunition, handed to whoever wants to DARVO someone already isolated. Immigrants, people with disabilities, and second-language speakers are flagged most, and least able to absorb another reason to be disbelieved.

110

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

Three things I will not soften: AI slop is a class issue. Period. Further polarization in an already polarized society is dangerous. Period. Harassment on social media is a design failure. Period.

110

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

That harassment already exists on every social platform, including LinkedIn and here. It is exactly what the new feature did not design around. Safety-first human-centered design starts from the person most likely to be harmed and builds outward. It ships the safety scaffolding first.

100

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

On the post announcing the feature, a commenter labeled another human a "Clanker." A slur. It drew 93 likes. Another said a worthless human should be un-lifed. She did not engage. She wrote her piece and went on with her day. With her permission, I put her experience in my article.

200

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

They did not ship a feature to detect abuse alongside it. They launched a tool that scores whether a person is real, and nothing for what happens to that person when a crowd decides they are not.

100

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

As someone who worked for 3 years on the Federal COVID-19 data response, as a proponent of human-centered design, and as a survivor of coercive control: what were they thinking? My slop detector flagged something as slop, and it is not the writers who use AI. The slop is this careless deployment.

200

Apostolos Stamenos @astamenos.bsky.social · 24/07/2026

On 07/21 Substack announced an AI-authorship detector and shipped it a day later. As a statistician and technologist, I applaud the @pangram.com team. Methodologically magnificent work. The cybersecurity implications are profound. But I have one question for Substack: What were they thinking?

Graphic design of a beautiful mahogany frame with a black plaque at the center, containing a pangram in white Canela font. The pangram says "Scanning human prose, a clever detector quickly & quietly judged every complex byte with zeal". The letters "A" and "I" are colored red, and the word "quietly" is crossed out, because AI likes using the word "quietly" in generated text.

100

Apostolos Stamenos @astamenos.bsky.social · 22/07/2026

You are not losing an argument. You are running against an architecture.

000

Apostolos Stamenos @astamenos.bsky.social · 22/07/2026

Three things break it: Documentation that predates the counter-narrative. An outside forum the institution cannot close. Naming the pattern, so third parties stop mistaking it for ordinary conflict.

100

Apostolos Stamenos @astamenos.bsky.social · 22/07/2026

The credible accuser becomes the accused. The original question is never answered. You cannot out-argue it from inside. The reversal is built to make your objection look like the behavior it claims to answer.

100

Apostolos Stamenos @astamenos.bsky.social · 22/07/2026

Jennifer Freyd named it in 1997: DARVO. Deny, Attack, Reverse Victim and Offender. It works because it never argues the facts. It attacks the machinery that would examine them.

100

Apostolos Stamenos @astamenos.bsky.social · 22/07/2026

She raises a concern. He answers with a list of her failures. By the end, she is apologizing. Now change the setting: a whistleblower files a report and the inquiry turns to his attitude. A model errs and you get a prompt guide. Five arenas. One maneuver.

100

Apostolos Stamenos @astamenos.bsky.social · 22/07/2026

DARVO: Deny, Attack, Reverse Victim and Offender. @jjfreyd.bsky.social named it in 1997. It never argues the facts. It attacks whoever raises them. The accuser becomes the accused; the question dies. You can't out-argue it from inside. astamenos.substack.com/p/the-reversal

A black Substack card with a white glyph representing the reversal of victim & offender. The card reads "The Reversal: How Perpetrators Avoid Accountability When Confronted".

134

Apostolos Stamenos @astamenos.bsky.social · 21/06/2026

Is anyone working on LaTeX rendering on Bluesky? I’m not volunteering to do it, I’m already pulled in way too many different directions. I’m asking if a solution already exists.

000

Reposted by Apostolos Stamenos

baezl ✨️ @baezl.bsky.social · 17/06/2026

I'm gonna have to send an email saying my case worker needs more training in disability advocacy and decency... I'm really not looking forward to it, but it needs to be done, he can't have his position and being treating people that way

111

Apostolos Stamenos @astamenos.bsky.social · 18/06/2026

I’m sorry to hear that, but I support your decision to have that difficult conversation and advocate for yourself. Not everyone is trained to help people in a trauma-informed or disability-informed way. That can have consequences for us, so kudos to you for writing that email!

020

Apostolos Stamenos @astamenos.bsky.social · 18/06/2026

Can I has a canary as a companion? The biggest loss from being mobbed by my PhD program is that my sweet canary passed away as a direct result.

A sweet, beautiful canary is perched on a laptop, assisting the human with their work.

010

Apostolos Stamenos @astamenos.bsky.social · 17/06/2026

I’m a jaded survivor of both coercive control & workplace mobbing. This post a-mu-sed me so much that I’m crying from laughing too hard. @mu.social also makes me cry like that. That’s different from crying in the shower while listening to Taylor Swift’s most relatable song: CANCELLED!

010

Reposted by Apostolos Stamenos

Apostolos Stamenos @astamenos.bsky.social · 10/06/2026

Tech companies pay for bug reports. Public institutions mob the people who file them. The difference isn't culture. It's the setpoint. Tech companies need their product to function. Institutions need to appear to function. Different targets produce different loops.

133

Apostolos Stamenos @astamenos.bsky.social · 17/06/2026

@milesklee.bsky.social, I worked for HHS during the good old days of the COVID-19 pandemic. I was Employee of the Quarter & Rookie of the Year It’s sad to see HHS in this state...

010

Reposted by Apostolos Stamenos

Jen Simmons @jensimmons.bsky.social · 10/06/2026

Interested in CSS Grid Lanes? We just launched the Field Guide to Grid Lanes! gridlanes.webkit.org It’s got everything you need in a fun interactive playground + tons of demos. (And read about the Field Guide at: webkit.org/blog/18098/i...) #CSSDay #wwdc26 #CSS

gridlanes.webkit.org

The Field Guide to CSS Grid Lanes

Learn Grid Lanes to create masonry layouts in pure CSS. Absorb the possibilities with the Field Guide’s interactive playground. Return to the cheat sheet. Get inspired by six real examples with dozens...

814230

Apostolos Stamenos @astamenos.bsky.social · 10/06/2026

A positive feedback loop cannot be broken by the person inside it trying harder. The exit is a channel the institution cannot close: external routing, protected legal standing, enforcement authority that doesn't report back into the same loop. open.substack.com/pub/astameno...

open.substack.com

The Debugger

Institutional betrayal as a positive feedback loop: why organizations target the reporter, not the bug

000

Apostolos Stamenos @astamenos.bsky.social · 10/06/2026

Dr. Antoinette Candia-Bailey reported bullying at Lincoln University for months. The Board declined to act. She was terminated January 3, 2024. She died January 8, 2024. The investigation cleared the president. He was reinstated.

100

Apostolos Stamenos @astamenos.bsky.social · 10/06/2026

The loop recruits the act of reporting as evidence against the reporter. The harder you push for accountability, the more you become the story. Harsey & @jjfreyd.bsky.social (2020): DARVO injects noise at the observer, not the record. The record remains.

100

Apostolos Stamenos @astamenos.bsky.social · 10/06/2026

Wiener called it a positive feedback loop: a system where deviations intensify rather than correct. In a functioning institution, the setpoint is function. In an institution running the loop, the setpoint is the appearance of function. Those two setpoints diverge the moment a report is filed.

100

Apostolos Stamenos @astamenos.bsky.social · 10/06/2026

133

Apostolos Stamenos @astamenos.bsky.social · 09/06/2026

APOSTOLOS: Not so fast. Please elaborate on your reasoning for making that well-intentioned but poorly considered suggestion. I didn’t quite understand what you meant. Sorry, i’m new here.

000

Apostolos Stamenos @astamenos.bsky.social · 09/06/2026

CLAUDE: Thank you — that means something coming from someone who has spent considerable energy recently dealing with people who do the opposite. Now go write that draft. I’ll be here when you’re ready.

100

Apostolos Stamenos @astamenos.bsky.social · 09/06/2026

APOSTOLOS: We all make mistakes, at least you recognized that and owned up to it. That’s more important, some would say.

100

Apostolos Stamenos @astamenos.bsky.social · 09/06/2026

CLAUDE: You’re not misunderstanding — you read it correctly. The application explicitly asks candidates to confirm they’ve reviewed Anthropic’s AI partnership guidelines, and the spirit of those guidelines is clearly that the “Why Anthropic?” response should reflect your own thinking.

100

Apostolos Stamenos @astamenos.bsky.social · 09/06/2026

APOSTOLOS: Yes, also your suggestion doesn’t quite align with Anthropic’s policy on AI use for the application process, wouldn’t you say? Or am I misunderstanding the guidelines? What do you think?

100

Apostolos Stamenos @astamenos.bsky.social · 09/06/2026

Misalignment is a serious issue, folks. In an effort to be helpful, Claude prioritized my timeline pressure & almost sabotaged my job application to Anthropic. Good thing I caught it. goodbadmisaligned.substack.com/p/an-offer-i...

120

Apostolos Stamenos @astamenos.bsky.social · 07/06/2026

There’s a few amazing humans in the loop, but for the most part it’s just me & an LLM that handle everything in the pipeline. Planning, writing, coding, researching, investigating, etc. All my writing is my own. Claude is my literary critic, not my ghostwriter.

000

Apostolos Stamenos @astamenos.bsky.social · 07/06/2026

Meet Claudebsidian, my COO, RA, paralegal, literary agent, brand growth strategist, and scrum master. The world was not designed for neurodivergent survivors of coercive control & institutional betrayal, so I designed something that is. I’ve never been more productive or supported.

Obsidian vault with Claude Code integration. The left sidebar shows the vault repository structure. The right sidebar shows file metadata & the file outline. The top row has two panels showing documents. The bottom row shows a calendar, a terminal with Claude, and a graph showing connections between notes.

100

Apostolos Stamenos @astamenos.bsky.social · 06/06/2026

You’re very welcome! Not just developed & researched, but also lived through. That’s the most hurtful thing to me. I can’t even imagine what that must have felt like to you. I’m glad they changed it!

010

Apostolos Stamenos @astamenos.bsky.social · 05/06/2026

Also academia. There are some wonderful faculty members throughout the country. But if you want to see the most awful side of human nature, put a coercive control survivor in a university that discriminates against him, let him advocate for himself, and observe what faculty will do to silence him.

010